From brem.belguebli at gmail.com  Sat Aug  1 00:25:19 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Sat, 1 Aug 2009 02:25:19 +0200
Subject: [Linux-cluster] CLVMD without GFS
In-Reply-To: <1249079337.6494.10.camel@mecatol>
References: <MJ2mY0lYjZGc@NBjvyXzB> <1249079337.6494.10.camel@mecatol>
Message-ID: <29ae894c0907311725k4759e631m332ffd38857a6f99@mail.gmail.com>

Hi Rafael,
Of course, that is what I was telling you the other day, your script works
fine, but I just wanted to twist a little clvm.
The prerequisites are already setup (locking type, dm-mp, lvm, etc...).
The only thing that disturbes me is that you can bypass under certain
conditions the locking.
Thanks for the help
Brem

2009/8/1 Rafael Mic? Miranda <rmicmirregs at gmail.com>

> Hi Brem,
>
> El vie, 31-07-2009 a las 06:09 -0400, crosa at redhat.com escribi?:
> >
> > --- mensagem original ---
> > De: brem belguebli <brem.belguebli at gmail.com>
> > Assunto: Re: [Linux-cluster] CLVMD without GFS
> > Data: 29 de Julho de 2009
> > Hora: 10:1:29
> >
> > Hi Rafael,
> >
> > Just posted the basic tests I'm doing on both linux-cluster and
> linux-lvm.
> >
> > I can't get exclusive activation to work properly, I may be missing some
> > step in my process.
> >
> > Brem
> >
> >
>
> >
> > --
> > Linux-cluster mailing list
> > Linux-cluster at redhat.com
> > https://www.redhat.com/mailman/listinfo/linux-cluster
>
> Sorry, i'm not sure if I missed this mail.
>
> Maybe i need to explain a couple of things first, just to make them
> clear.
>
> As part of the use of the lvm-cluster.sh resource script, you need to:
>
> 1.- Configure properly the lock type into lvm.conf to type 3 (example
> into this link:
>
> https://www.redhat.com/archives/linux-cluster/2009-July/msg00253.html
>
> )
>
> 2.- Configure your multipathing software
>
> 3.- Create your Physical Volumes
>
> 4.- Configure your Volume Groups as clustered Volume Groups
>
> 5.- Create your Logical Volumens into the clustered Volume Groups
>
> 6.- And, and the not so obvious, de-activate all the Logical Volumes you
> plan to use as exclusive Logical Volumes (using exclusive flag).
>
> De-activation can be done with "vgchange -an volgrp01/logvol01" or
> similar command.
>
> If you don't do step 6, you will receive an error message when executing
> "lvchange -aey volgrp01/logvol01". This command is executed (with the
> proper volume group and logical volume names) internally in the resource
> script.
>
> I designed the lvm-cluster.sh resource script to be verbose, maybe you
> can copy here your logs (default to /var/log/messages on RHEL systems).
>
> Cheers and thanks,
>
> Rafael
>
> --
> Rafael Mic? Miranda
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090801/6141400e/attachment.htm>

From alietsantiesteban at gmail.com  Sat Aug  1 18:07:20 2009
From: alietsantiesteban at gmail.com (Aliet Santiesteban Sifontes)
Date: Sat, 1 Aug 2009 14:07:20 -0400
Subject: [Linux-cluster] Lvm2-cluster git repo??
Message-ID: <365467590908011107x195ccaabiedd217907183faab@mail.gmail.com>

Hi, list, just wondering if anybody can point me to the right place to
find the lvm2-cluster-2.02.42-5.el4.src.rpm related to RHBA-2009-1047,
I have looked in rh ftp site but the file there is outdated, I would
like to see the patches included in this srpm release wich affects the
original lvm2 2.02.42 release code.
Thank's in advance.., Aliet



From sergio_gonra at yahoo.es  Mon Aug  3 07:59:43 2009
From: sergio_gonra at yahoo.es (Sergio Gonzalez Ramos)
Date: Mon, 3 Aug 2009 07:59:43 +0000 (GMT)
Subject: [Linux-cluster] System load at 1.00 +
Message-ID: <103431.77947.qm@web28609.mail.ukl.yahoo.com>

Hi There:

Seems that both nodes of the cluster , have  System load at 1.00 + when the cluster resources are running ( mounted Filesystems in a EVA 4400 and ip )

I saw that when there are no cluster resources working, the system load is normal and get reduces near to 0

Is there any issue with cluster / filesystems mounted that get's the load averge increased one point??. Seams that the box is working properly , it's just that confuses me with one more point of load average

i paste here my Red hat version and the cluster.conf

[root at OCEANO1CLUN ~]# cat /etc/cluster/cluster.conf
<?xml version="1.0"?>
<cluster alias="GATTON" config_version="131" name="GATTON">
        <fence_daemon post_fail_delay="0" post_join_delay="3"/>
        <clusternodes>
                <clusternode name="OCEANO1CLUN" nodeid="1" votes="1">
                        <fence/>
                </clusternode>
                <clusternode name="OCEANO2CLUN" nodeid="2" votes="1">
                        <fence/>
                </clusternode>
        </clusternodes>
        <cman expected_votes="1" two_node="1"/>
        <fencedevices/>
        <rm>
                <failoverdomains>
                        <failoverdomain name="BDSIS01_FAIL" ordered="0" restricted="0">
                                <failoverdomainnode name="OCEANO1CLUN" priority="1"/>
                                <failoverdomainnode name="OCEANO2CLUN" priority="1"/>
                        </failoverdomain>
                        <failoverdomain name="GATTO_FAIL" ordered="1">
                                <failoverdomainnode name="OCEANO1CLUN" priority="1"/>
                                <failoverdomainnode name="OCEANO2CLUN" priority="2"/>
                        </failoverdomain>
                </failoverdomains>
                <resources>
                        <ip address="192.168.12.163" monitor_link="1"/>
                        <fs device="/dev/mapper/gatto-gatto" force_fsck="0" force_unmount="1" fsid="8279" fstype="ext3" mountpoint="/orabd/gatto" name="gattobd" options="" self_fence="1"/>
                        <fs device="/dev/mapper/gatto-gatto_h_indices" force_fsck="0" force_unmount="1" fsid="52025" fstype="ext3" mountpoint="/orabd/gatto/h_indices" name="gattoindices" options="" self_fence="1"/>
                        <fs device="/dev/mapper/gatto-gatto_h_tablas" force_fsck="0" force_unmount="1" fsid="2592" fstype="ext3" mountpoint="/orabd/gatto/h_tablas" name="gattotablas" options="" self_fence="1"/>
                        <script file="/etc/init.d/oracle_bd_gatto" name="oracle_bd_gatto"/>
                        <ip address="192.168.12.164" monitor_link="1"/>
                        <fs device="/dev/mapper/bdsis01-bdsis01" force_fsck="0" force_unmount="1" fsid="3496" fstype="ext3" mountpoint="/orabd/bdsis01" name="bdsis01" options="" self_fence="1"/>
                        <script file="/etc/init.d/oracle_bd_bdsis01" name="oracle_bd_bdsis01"/>
                        <fs device="/dev/mapper/gatto_h-gatto_h_tablas_sl" force_fsck="0" force_unmount="1" fsid="1191" fstype="ext3" mountpoint="/orabd/gatto/h_tablas_sl" name="gattotablas_sl" options="" self_fence="1"/>
                        <fs device="/dev/mapper/gatto_h-gatto_h_indices_sl" force_fsck="0" force_unmount="1" fsid="57261" fstype="ext3" mountpoint="/orabd/gatto/h_indices_sl" name="gattoindices_sl" options="" self_fence="1"/>
                        <fs device="/dev/mapper/bdsis01-bdsis01_mediacion" force_fsck="0" force_unmount="1" fsid="27523" fstype="ext2" mountpoint="/orabd/bdsis01/mediacion" name="bdsis01_mediacion" options="" self_fence="1"/>
                </resources>
                <service autostart="1" domain="GATTO_FAIL" name="gatto">
                        <fs ref="gattobd">
                                <fs ref="gattotablas">
                                        <fs ref="gattoindices">
                                                <fs ref="gattotablas_sl">
                                                        <fs ref="gattoindices_sl"/>
                                                        <ip ref="192.168.12.163">
                                                                <script ref="oracle_bd_gatto"/>
                                                        </ip>
                                                </fs>
                                        </fs>
                                </fs>
                        </fs>
                </service>
                <service autostart="1" name="bdsis01">
                        <fs ref="bdsis01">
                                <fs ref="bdsis01_mediacion">
                                        <ip ref="192.168.12.164">
                                                <script ref="oracle_bd_bdsis01"/>
                                        </ip>
                                </fs>
                        </fs>
                </service>
        </rm>
</cluster>


[root at OCEANO1CLUN ~]# cat /etc/redhat-release
Red Hat Enterprise Linux Server release 5.3 (Tikanga)


Regards.
Sergio


      
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090803/3511211b/attachment.htm>

From jds at techma.com  Mon Aug  3 12:04:46 2009
From: jds at techma.com (Simmons, Dan A)
Date: Mon, 3 Aug 2009 08:04:46 -0400
Subject: [Linux-cluster] how to sync rmtab on old cluster
In-Reply-To: <20090801160009.1D79461B1CF@hormel.redhat.com>
References: <20090801160009.1D79461B1CF@hormel.redhat.com>
Message-ID: <35440D75A88EF04C81FDEC03F4F739F7028A4E@tmaemail.techma.com>

Hi All,
I have a very old RHEL3u3 two node NFS cluster and I have recovered one of
the nodes with a backup from a year ago.  I now have entries in the rmtab on
the recovered system that conflict with recent client mounts on the newer
node.  As a result, when I move the nfs service to the recovered node, many
of my nfs client workstations cannot mount the exports.  

What is the correct way to sync the clustered rmtab in this situation?

Is there some way to reset clurmtab to clear out all the old entries?  

Any advice would be appreciated.

J.Dan 



From rpeterso at redhat.com  Tue Aug  4 12:58:01 2009
From: rpeterso at redhat.com (Bob Peterson)
Date: Tue, 4 Aug 2009 08:58:01 -0400 (EDT)
Subject: [Linux-cluster] Re: GFS2 corruption/withdrawal/crash (Bob
	Peterson)
In-Reply-To: <2040479433.1003341248797259502.JavaMail.root@zmail06.collab.prod.int.phx2.redhat.com>
Message-ID: <2064630003.133081249390681257.JavaMail.root@zmail06.collab.prod.int.phx2.redhat.com>

----- "Bob Peterson" <rpeterso at redhat.com> wrote:
| ----- "Wendell Dingus" <wendell at bisonline.com> wrote:
| | Absolutely.. We're doing a gfs2_edit savemeta right now and at 22%
| | it's already over 4.5GB in size. So how can I get that to you when
| | it completes? 
| | 
| | Thanks!
| 
| Hi Wendell,
| 
| What would probably work best is if you use bzip2 to compress the
| metadata then upload it to some web or ftp server for me to download.
| There are a couple free ones out there.
| 
| Regards,
| 
| Bob Peterson
| Red Hat File Systems

Hi Wendell,

What's the status on this?  I haven't heard back from you.
Did the metadata save finish?  If so, where can I download it?

Regards,

Bob Peterson
Red Hat File Systems



From wendell at bisonline.com  Tue Aug  4 13:54:18 2009
From: wendell at bisonline.com (Wendell Dingus)
Date: Tue, 4 Aug 2009 09:54:18 -0400 (EDT)
Subject: [Linux-cluster] Re: GFS2 corruption/withdrawal/crash (Bob
	Peterson)
In-Reply-To: <8911529.6621249394015230.JavaMail.wendell@philco>
Message-ID: <23884039.6641249394055066.JavaMail.wendell@philco>

I had emailed this to you directly Bob but perhaps you didn't get it. The URL is an IP address and some folks SPAM/malware filters might balk at that. So I'll be a bit more vague in how I pass it on this time in hopes of that not happening... 

Here is the GFS2 metadata file. Be forewarned though, it's 2.8GB compressed. 

Server, accessible via HTTP, but indexes not enabled so put filename on URL. Server IP is 204.87.213.57 The file on there is raid1_gfs2_metadata.txt.bz2 

Thanks. 

PS. You have impeccable timing :) This equipment had been up and running fine for quite a few days. I had begun to think the gfs2_fsck had resolved the issue or at least mostly resolved it. Until just _now_ that is... The first sign we have that something is wrong is kind of odd, it's lack of all IP connectivity on one or more of our virtual machines. They have a private network between them and the physical server they're hosted on. The physical server mounts a GFS2 filesystem and NFS shares it and it's mounted by whichever VM(s) are running on it. If the GFS filesystem has issues the passed-through NFS share of course has issues as well. It's odd how it manifests itself though, *all* IP connectivity on multiple VMs is what we see, including external-facing public IPs not associated with the NFS share. Strange... 



----- "Bob Peterson" <rpeterso at redhat.com> wrote: 
> ----- "Bob Peterson" <rpeterso at redhat.com> wrote: 
> | ----- "Wendell Dingus" <wendell at bisonline.com> wrote: 
> | | Absolutely.. We're doing a gfs2_edit savemeta right now and at 22% 
> | | it's already over 4.5GB in size. So how can I get that to you when 
> | | it completes? 
> | | 
> | | Thanks! 
> | 
> | Hi Wendell, 
> | 
> | What would probably work best is if you use bzip2 to compress the 
> | metadata then upload it to some web or ftp server for me to download. 
> | There are a couple free ones out there. 
> | 
> | Regards, 
> | 
> | Bob Peterson 
> | Red Hat File Systems 
> 
> Hi Wendell, 
> 
> What's the status on this? I haven't heard back from you. 
> Did the metadata save finish? If so, where can I download it? 
> 
> Regards, 
> 
> Bob Peterson 
> Red Hat File Systems 
> 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090804/254f3079/attachment.htm>

From marko.jung at oucs.ox.ac.uk  Tue Aug  4 15:43:23 2009
From: marko.jung at oucs.ox.ac.uk (Marko Jung)
Date: Tue, 04 Aug 2009 16:43:23 +0100
Subject: [Linux-cluster] MySQL with datadir on GFS volume
Message-ID: <4A78571B.2020506@oucs.ox.ac.uk>

Hi,

has someone set-up an MySQL server (failover, 1 active node) using GFSv2 as 
filesystem for the datadir? I would be interested in the required changes to 
the configuration and startup scripts to get this running.

Thank you for your comments.
Marko



From corey.kovacs at gmail.com  Tue Aug  4 15:49:50 2009
From: corey.kovacs at gmail.com (Corey Kovacs)
Date: Tue, 4 Aug 2009 16:49:50 +0100
Subject: [Linux-cluster] how to sync rmtab on old cluster
In-Reply-To: <35440D75A88EF04C81FDEC03F4F739F7028A4E@tmaemail.techma.com>
References: <20090801160009.1D79461B1CF@hormel.redhat.com>
	<35440D75A88EF04C81FDEC03F4F739F7028A4E@tmaemail.techma.com>
Message-ID: <8EE13E7D-C696-4131-B739-925FE1663508@gmail.com>

Hey, it's Tuesday, wan't to go to hooters? Oh wait, mine is in tenerife!

Kiss my arse!

Yet bud

Corey

Sent from my iPod

On Aug 3, 2009, at 13:04, "Simmons, Dan A" <jds at techma.com> wrote:

> Hi All,
> I have a very old RHEL3u3 two node NFS cluster and I have recovered  
> one of
> the nodes with a backup from a year ago.  I now have entries in the  
> rmtab on
> the recovered system that conflict with recent client mounts on the  
> newer
> node.  As a result, when I move the nfs service to the recovered  
> node, many
> of my nfs client workstations cannot mount the exports.
>
> What is the correct way to sync the clustered rmtab in this situation?
>
> Is there some way to reset clurmtab to clear out all the old entries?
>
> Any advice would be appreciated.
>
> J.Dan
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster



From gordan at bobich.net  Tue Aug  4 15:59:02 2009
From: gordan at bobich.net (Gordan Bobic)
Date: Tue, 04 Aug 2009 16:59:02 +0100
Subject: [Linux-cluster] MySQL with datadir on GFS volume
In-Reply-To: <4A78571B.2020506@oucs.ox.ac.uk>
References: <4A78571B.2020506@oucs.ox.ac.uk>
Message-ID: <0f5c995e65a46913cd7cb155e93524f0@localhost>

> has someone set-up an MySQL server (failover, 1 active node) using GFSv2
as
> 
> filesystem for the datadir? I would be interested in the required changes
> to 
> the configuration and startup scripts to get this running.

If you are only using it in fail-over mode (i.e. you define a mysqld script
resource tied to a fail-over IP in cluster.conf), and you will never start
it up simultaneously on more than one machine, then it should just work out
of the box.

If you want to run it on multiple nodes at the same time, I seem to
remember that you have to set the instances to use external locking - but
the performance will suffer quite significantly in that case.

Performance-wise, you are still probably best off with round-robin
replication on a standard file system if you nead near-real-time
redundancy.

Gordan



From marko.jung at oucs.ox.ac.uk  Tue Aug  4 16:07:48 2009
From: marko.jung at oucs.ox.ac.uk (Marko Jung)
Date: Tue, 04 Aug 2009 17:07:48 +0100
Subject: [Linux-cluster] MySQL with datadir on GFS volume
In-Reply-To: <0f5c995e65a46913cd7cb155e93524f0@localhost>
References: <4A78571B.2020506@oucs.ox.ac.uk>
	<0f5c995e65a46913cd7cb155e93524f0@localhost>
Message-ID: <4A785CD4.10501@oucs.ox.ac.uk>

Hi.

Gordan Bobic wrote:
>> has someone set-up an MySQL server (failover, 1 active node) using GFSv2
>> as filesystem for the datadir? I would be interested in the required changes
>> to the configuration and startup scripts to get this running.
> 
> If you are only using it in fail-over mode (i.e. you define a mysqld script
> resource tied to a fail-over IP in cluster.conf), and you will never start
> it up simultaneously on more than one machine, then it should just work out
> of the box.

Unfortunately not, it complains not to be able to open and lock the 
privilege tables:
"[ERROR] Fatal error: Can't open and lock privilege tables: Table 'host' is
  read only"
despite permissions and paths are right. Am I missing something regarding 
locking on GFSv2?

Thanks a lot!
Marko



From gordan at bobich.net  Tue Aug  4 16:29:24 2009
From: gordan at bobich.net (Gordan Bobic)
Date: Tue, 04 Aug 2009 17:29:24 +0100
Subject: [Linux-cluster] MySQL with datadir on GFS volume
In-Reply-To: <4A785CD4.10501@oucs.ox.ac.uk>
References: <4A78571B.2020506@oucs.ox.ac.uk>	<0f5c995e65a46913cd7cb155e93524f0@localhost>
	<4A785CD4.10501@oucs.ox.ac.uk>
Message-ID: <203e89bbd8cce02b50a6abe7aa6fb6f5@localhost>

>>> has someone set-up an MySQL server (failover, 1 active node) using
GFSv2
>>> as filesystem for the datadir? I would be interested in the required
>>> changes
>>> to the configuration and startup scripts to get this running.
>> 
>> If you are only using it in fail-over mode (i.e. you define a mysqld
>> script
>> resource tied to a fail-over IP in cluster.conf), and you will never
>> start
>> it up simultaneously on more than one machine, then it should just work
>> out
>> of the box.
> 
> Unfortunately not, it complains not to be able to open and lock the 
> privilege tables:
> "[ERROR] Fatal error: Can't open and lock privilege tables: Table 'host'
is
>   read only"
> despite permissions and paths are right. Am I missing something regarding

> locking on GFSv2?

The two things that spring to mind are:
1) SELinux - try disabling it.
2) You may have bumped into a GFS2 bug. What distro/cluster version are you
running? GFS2 was "preview only" until very recently, and the pace of
development and fixes in it are still a tad too high for a production
system for my liking. Is there any particular reason why you can't use
GFS1?

Gordan



From marko.jung at oucs.ox.ac.uk  Tue Aug  4 16:35:33 2009
From: marko.jung at oucs.ox.ac.uk (Marko Jung)
Date: Tue, 04 Aug 2009 17:35:33 +0100
Subject: [Linux-cluster] MySQL with datadir on GFS volume
In-Reply-To: <203e89bbd8cce02b50a6abe7aa6fb6f5@localhost>
References: <4A78571B.2020506@oucs.ox.ac.uk>	<0f5c995e65a46913cd7cb155e93524f0@localhost>	<4A785CD4.10501@oucs.ox.ac.uk>
	<203e89bbd8cce02b50a6abe7aa6fb6f5@localhost>
Message-ID: <4A786355.8070701@oucs.ox.ac.uk>

Gordan Bobic wrote:
>>>> has someone set-up an MySQL server (failover, 1 active node) using
>>>> GFSv2as filesystem for the datadir? I would be interested in the required
>>>> changes to the configuration and startup scripts to get this running.
 >>>
>>> If you are only using it in fail-over mode (i.e. you define a mysqld
>>> script resource tied to a fail-over IP in cluster.conf), and you will never
>>> start it up simultaneously on more than one machine, then it should just work
>>> out of the box.
 >>
>> Unfortunately not, it complains not to be able to open and lock the 
>> privilege tables:
>> "[ERROR] Fatal error: Can't open and lock privilege tables: Table 'host'is
>> read only" despite permissions and paths are right. Am I missing something 
 >> regarding locking on GFSv2?
> 
> The two things that spring to mind are:
> 1) SELinux - try disabling it.

It is turned off. Running RHEL cluster suite with SELinux on is quite tough.

> 2) You may have bumped into a GFS2 bug. What distro/cluster version are you
> running? GFS2 was "preview only" until very recently, and the pace of
> development and fixes in it are still a tad too high for a production
> system for my liking. Is there any particular reason why you can't use
> GFS1?

It is CentOS 5.3 final without any fancy external updates or similar stuff. 
We chose GFS2 because it is the default with RHEL 5.3.

Thanks again!
Marko



From marko.jung at oucs.ox.ac.uk  Tue Aug  4 16:41:14 2009
From: marko.jung at oucs.ox.ac.uk (Marko Jung)
Date: Tue, 04 Aug 2009 17:41:14 +0100
Subject: [Linux-cluster] MySQL with datadir on GFS volume
In-Reply-To: <4A786355.8070701@oucs.ox.ac.uk>
References: <4A78571B.2020506@oucs.ox.ac.uk>	<0f5c995e65a46913cd7cb155e93524f0@localhost>	<4A785CD4.10501@oucs.ox.ac.uk>	<203e89bbd8cce02b50a6abe7aa6fb6f5@localhost>
	<4A786355.8070701@oucs.ox.ac.uk>
Message-ID: <4A7864AA.1020006@oucs.ox.ac.uk>

Re.

Marko Jung wrote:
> Gordan Bobic wrote:
>> 2) You may have bumped into a GFS2 bug. What distro/cluster version are you
>> running? GFS2 was "preview only" until very recently, and the pace of
>> development and fixes in it are still a tad too high for a production
>> system for my liking. Is there any particular reason why you can't use
>> GFS1?
> 
> It is CentOS 5.3 final without any fancy external updates or similar stuff. 
> We chose GFS2 because it is the default with RHEL 5.3.

I thought I should be more verbose on what I am trying to achieve MySQL 
associated to a floating IP running on cluster as failover. Storage and 
configuration should reside on a huge GFSv2 Filesystem.

I used the distribution provided MySQL packages and used the following my.cnf:

------%<-----------------------------------
[mysqld]
datadir=/san/mysql
socket=/var/lib/mysql/mysql.sock
user=mysql
skip-innodb
enable-locking

[mysqld_safe]
log-error=/var/log/mysqld.log
pid-file=/var/run/mysqld/mysqld.pid
------%<-----------------------------------

In addition to this i modified the /usr/bin/mysqld_safe to use 
--external-locking as option.

mysql.log shows the following error:
------%<-----------------------------------
090804 17:13:25  mysqld started
090804 17:13:25 [ERROR] bdb:  /san/mysql: Permission denied
090804 17:13:25 [ERROR] bdb:  /san/mysql/log.0000000001: Permission denied
090804 17:13:25 [ERROR] bdb:  PANIC: Permission denied
090804 17:13:25 [ERROR] bdb:  PANIC: DB_RUNRECOVERY: Fatal error, run 
database recovery
090804 17:13:25 [ERROR] bdb:  fatal region error detected; run recovery
090804 17:13:25 [ERROR] bdb:  /san/mysql: Permission denied
090804 17:13:25 [ERROR] Fatal error: Can't open and lock privilege tables: 
Table 'host' is read only
------%<-----------------------------------

But permissions are right (everything owned by mysql user, group mysql).
I am a bit confused - the identical configuration on an ext3 filesystem 
works fine.

Marko



From marko.jung at oucs.ox.ac.uk  Tue Aug  4 17:18:02 2009
From: marko.jung at oucs.ox.ac.uk (Marko Jung)
Date: Tue, 04 Aug 2009 18:18:02 +0100
Subject: [Linux-cluster] MySQL with datadir on GFS volume
In-Reply-To: <4A7864AA.1020006@oucs.ox.ac.uk>
References: <4A78571B.2020506@oucs.ox.ac.uk>	<0f5c995e65a46913cd7cb155e93524f0@localhost>	<4A785CD4.10501@oucs.ox.ac.uk>	<203e89bbd8cce02b50a6abe7aa6fb6f5@localhost>	<4A786355.8070701@oucs.ox.ac.uk>
	<4A7864AA.1020006@oucs.ox.ac.uk>
Message-ID: <4A786D4A.3010606@oucs.ox.ac.uk>

Marko Jung wrote:
> Marko Jung wrote:
>> Gordan Bobic wrote:
>>> 2) You may have bumped into a GFS2 bug. What distro/cluster version are you
>>> running? GFS2 was "preview only" until very recently, and the pace of
>>> development and fixes in it are still a tad too high for a production
>>> system for my liking. Is there any particular reason why you can't use
>>> GFS1?
>> It is CentOS 5.3 final without any fancy external updates or similar stuff. 
>> We chose GFS2 because it is the default with RHEL 5.3.
> 
> I thought I should be more verbose on what I am trying to achieve MySQL 
> associated to a floating IP running on cluster as failover. Storage and 
> configuration should reside on a huge GFSv2 Filesystem.

Gordan, thank you very much for your support, but the problem was quite 
easy, the mysqld_safe script overrides the external_locking configuration 
directive.

Marko



From wendell at bisonline.com  Tue Aug  4 17:50:10 2009
From: wendell at bisonline.com (Wendell Dingus)
Date: Tue, 4 Aug 2009 13:50:10 -0400 (EDT)
Subject: [Linux-cluster] Re: GFS2 corruption/withdrawal/crash
In-Reply-To: <30764099.7001249408032363.JavaMail.wendell@philco>
Message-ID: <21123838.7021249408207452.JavaMail.wendell@philco>

> Hi Wendell, 
> 
> What's the status on this? I haven't heard back from you. 
> Did the metadata save finish? If so, where can I download it? 
> 
> Regards, 
> 
> Bob Peterson 
> Red Hat File Systems 

As Bob mentioned in a separate email, perhaps someone else can suggest what might be causing the complete loss of IP connectivity which *appears* to be related to the GFS2 issues. Even though that *should* not happen...

Three physical servers with clvmd, Xen kernels, GFS, shared FC storage, multiple virtual machine instances. Each physical server mounts 3 fairly large GFS2 filesystems. Each one has 2 physical NICs and a virtual NIC. One physical NIC gets a 192.168.x.y private IP and connects to a good gigE switch for server-to-server clustering/gfs communications. The virtual NIC is only visible between a VM running on a physical host and that physical host it runs on. Each physical host server NFS shares those GFS filesystems and a number of the VMs mount those, via the virtual NIC that allows it to see the physical node it's running on only.

Our first indication of a problem is generally loss of IP connectivity to one or more of the VMs. This morning the one that we first noticed couldn't be accessed was running on physical server3. We immediately started checking to see if we could see what was going on and noticed that doing an "ls" on server1 of one of the GFS filesystems gave an I/O error. Doing so against the 2nd GFS2 fielsystem hung. Both server2 and server3 appeared able to fully access those filesystems. The VM that lost IP connectivity was running on server3.

So a strange incongruity there. Server3 appears healthy, nothing odd in logs, can access GFS filesystems just fine, but at about the time server1 starts failing on a GFS filesystem a VM running on server3 loses all IP connectivity. Has to be related but I'm just not sure how/why.

Our last crash was around 9:30 this morning. Just now I noticed server2 has this in /var/log/messages (see below). But, as of right now I *can* access files on all 3 GFS2 mounted filesystems on server2 without issue it would appear. That's probably a sign that it's going to belly up here in a bit, maybe(?)

Oh, one other thing, you mentioning needing to run gfs2_fsck *twice* to fix some problems Bob. I did not know that and did not run it twice on the first attempt to repair the filesystem I think is causing the grief there. If you don't complete changes to gfs2_fsck so as to not require two passes like that by this weekend would you recommend we umount and run it twice? 

Thanks.

Aug 4 09:40:39 server2 ntpd[5125]: synchronized to 192.168.77.1, stratum 3 
Aug 4 13:01:01 server2 kernel: kswapd0: page allocation failure. order:0, mode:0xd0 
Aug 4 13:01:01 server2 kernel: 
Aug 4 13:01:01 server2 kernel: Call Trace: 
Aug 4 13:01:01 server2 kernel: [<ffffffff8020f6b0>] __alloc_pages+0x2b5/0x2ce 
Aug 4 13:01:01 server2 kernel: [<ffffffff8020c6e3>] do_generic_mapping_read+0x389/0x3de 
Aug 4 13:01:01 server2 kernel: [<ffffffff88554a7a>] :gfs2:gfs2_read_actor+0x0/0x7b 
Aug 4 13:01:01 server2 kernel: [<ffffffff8025dfb7>] cache_alloc_refill+0x267/0x4a9 
Aug 4 13:01:01 server2 kernel: [<ffffffff8020afb1>] kmem_cache_alloc+0x50/0x6d 
Aug 4 13:01:01 server2 kernel: [<ffffffff8854a910>] :gfs2:gfs2_glock_get+0x9f/0x29d 
Aug 4 13:01:01 server2 kernel: [<ffffffff8855c5f5>] :gfs2:read_rindex_entry+0x317/0x32a 
Aug 4 13:01:01 server2 kernel: [<ffffffff8855c920>] :gfs2:gfs2_rindex_hold+0x101/0x153 
Aug 4 13:01:01 server2 kernel: [<ffffffff80225cd7>] find_or_create_page+0x3c/0xab 
Aug 4 13:01:01 server2 kernel: [<ffffffff88543714>] :gfs2:do_strip+0xba/0x358 
Aug 4 13:01:01 server2 kernel: [<ffffffff88551d20>] :gfs2:gfs2_meta_read+0x17/0x65 
Aug 4 13:01:01 server2 kernel: [<ffffffff885427e2>] :gfs2:recursive_scan+0xf2/0x175 
Aug 4 13:01:01 server2 kernel: [<ffffffff885428fe>] :gfs2:trunc_dealloc+0x99/0xe7 
Aug 4 13:01:01 server2 kernel: [<ffffffff8854365a>] :gfs2:do_strip+0x0/0x358 
Aug 4 13:01:01 server2 kernel: [<ffffffff80290000>] do_sysctl+0x18f/0x26e 
Aug 4 13:01:01 server2 kernel: [<ffffffff885588b9>] :gfs2:gfs2_delete_inode+0xe3/0x18d 
Aug 4 13:01:01 server2 kernel: [<ffffffff8855881c>] :gfs2:gfs2_delete_inode+0x46/0x18d 
Aug 4 13:01:01 server2 kernel: [<ffffffff885587d6>] :gfs2:gfs2_delete_inode+0x0/0x18d 
Aug 4 13:01:01 server2 kernel: [<ffffffff802303a5>] generic_delete_inode+0xc6/0x143 
Aug 4 13:01:01 server2 kernel: [<ffffffff802d9566>] prune_one_dentry+0x4d/0x76 
Aug 4 13:01:01 server2 kernel: [<ffffffff8022f865>] prune_dcache+0x10f/0x149 
Aug 4 13:01:01 server2 kernel: [<ffffffff802d95a6>] shrink_dcache_memory+0x17/0x30 
Aug 4 13:01:01 server2 kernel: [<ffffffff80240951>] shrink_slab+0xdc/0x154 
Aug 4 13:01:01 server2 kernel: [<ffffffff802599c6>] kswapd+0x347/0x447 
Aug 4 13:01:01 server2 kernel: [<ffffffff8026defe>] monotonic_clock+0x35/0x7b 
Aug 4 13:01:01 server2 kernel: [<ffffffff80299fe8>] autoremove_wake_function+0x0/0x2e 
Aug 4 13:01:01 server2 kernel: [<ffffffff80299dd0>] keventd_create_kthread+0x0/0xc4 
Aug 4 13:01:01 server2 kernel: [<ffffffff8025967f>] kswapd+0x0/0x447 
Aug 4 13:01:01 server2 kernel: [<ffffffff80299dd0>] keventd_create_kthread+0x0/0xc4 
Aug 4 13:01:01 server2 kernel: [<ffffffff802334b4>] kthread+0xfe/0x132 
Aug 4 13:01:01 server2 kernel: [<ffffffff8025fb2c>] child_rip+0xa/0x12 
Aug 4 13:01:01 server2 kernel: [<ffffffff80299dd0>] keventd_create_kthread+0x0/0xc4 
Aug 4 13:01:01 server2 kernel: [<ffffffff802333b6>] kthread+0x0/0x132 
Aug 4 13:01:01 server2 kernel: [<ffffffff8025fb22>] child_rip+0x0/0x12 
Aug 4 13:01:01 server2 kernel: 
Aug 4 13:01:01 server2 kernel: Mem-info: 
Aug 4 13:01:01 server2 kernel: DMA per-cpu: 
Aug 4 13:01:01 server2 kernel: cpu 0 hot: high 186, batch 31 used:0 
Aug 4 13:01:01 server2 kernel: cpu 0 cold: high 62, batch 15 used:48 
Aug 4 13:01:01 server2 kernel: cpu 1 hot: high 186, batch 31 used:21 
Aug 4 13:01:01 server2 kernel: cpu 1 cold: high 62, batch 15 used:8 
Aug 4 13:01:01 server2 kernel: cpu 2 hot: high 186, batch 31 used:176 
Aug 4 13:01:01 server2 kernel: cpu 2 cold: high 62, batch 15 used:9 
Aug 4 13:01:01 server2 kernel: cpu 3 hot: high 186, batch 31 used:128 
Aug 4 13:01:01 server2 kernel: cpu 3 cold: high 62, batch 15 used:3 
Aug 4 13:01:01 server2 kernel: cpu 4 hot: high 186, batch 31 used:156 
Aug 4 13:01:01 server2 kernel: cpu 4 cold: high 62, batch 15 used:10 
Aug 4 13:01:01 server2 kernel: cpu 5 hot: high 186, batch 31 used:28 
Aug 4 13:01:01 server2 kernel: cpu 5 cold: high 62, batch 15 used:50 
Aug 4 13:01:01 server2 kernel: cpu 6 hot: high 186, batch 31 used:157 
Aug 4 13:01:01 server2 kernel: cpu 6 cold: high 62, batch 15 used:9 
Aug 4 13:01:01 server2 kernel: cpu 7 hot: high 186, batch 31 used:13 
Aug 4 13:01:01 server2 kernel: cpu 7 cold: high 62, batch 15 used:2 
Aug 4 13:01:01 server2 kernel: DMA32 per-cpu: empty 
Aug 4 13:01:01 server2 kernel: Normal per-cpu: empty 
Aug 4 13:01:01 server2 kernel: HighMem per-cpu: empty 
Aug 4 13:01:01 server2 kernel: Free pages: 0kB (0kB HighMem) 
Aug 4 13:01:01 server2 kernel: Active:131632 inactive:475042 dirty:89 writeback:0 unstable:0 free:0 slab:129221 mapped-file:31572 mapped-anon:75077 pagetables:4941 
Aug 4 13:01:01 server2 kernel: DMA free:0kB min:7100kB low:8872kB high:10648kB active:526528kB inactive:1900168kB present:3153920kB pages_scanned:0 all_unreclaimable? no 
Aug 4 13:01:01 server2 kernel: lowmem_reserve[]: 0 0 0 0 
Aug 4 13:01:01 server2 kernel: DMA32 free:0kB min:0kB low:0kB high:0kB active:0kB inactive:0kB present:0kB pages_scanned:0 all_unreclaimable? no 
Aug 4 13:01:01 server2 kernel: lowmem_reserve[]: 0 0 0 0 
Aug 4 13:01:01 server2 kernel: Normal free:0kB min:0kB low:0kB high:0kB active:0kB inactive:0kB present:0kB pages_scanned:0 all_unreclaimable? no 
Aug 4 13:01:01 server2 kernel: lowmem_reserve[]: 0 0 0 0 
Aug 4 13:01:01 server2 kernel: HighMem free:0kB min:128kB low:128kB high:128kB active:0kB inactive:0kB present:0kB pages_scanned:0 all_unreclaimable? no 
Aug 4 13:01:01 server2 kernel: lowmem_reserve[]: 0 0 0 0 
Aug 4 13:01:01 server2 kernel: DMA: 0*4kB 0*8kB 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 0kB 
Aug 4 13:01:01 server2 kernel: DMA32: empty 
Aug 4 13:01:01 server2 kernel: Normal: empty 
Aug 4 13:01:01 server2 kernel: HighMem: empty 
Aug 4 13:01:01 server2 kernel: 531563 pagecache pages 
Aug 4 13:01:01 server2 kernel: Swap cache: add 2, delete 0, find 0/0, race 0+0 
Aug 4 13:01:01 server2 kernel: Free swap = 16777200kB 
Aug 4 13:01:01 server2 kernel: Total swap = 16777208kB 
Aug 4 13:01:01 server2 kernel: Free swap: 16777200kB 
Aug 4 13:01:01 server2 kernel: 788480 pages of RAM 
Aug 4 13:01:01 server2 kernel: 35847 reserved pages 
Aug 4 13:01:01 server2 kernel: 123018 pages shared 
Aug 4 13:01:01 server2 kernel: 2 pages swap cached 
Aug 4 13:01:01 server2 kernel: GFS2: fsid=Peace:raid2.0: gfs2_delete_inode: -12 

From curtis at athabascau.ca  Tue Aug  4 20:48:05 2009
From: curtis at athabascau.ca (Curtis Collicutt)
Date: Tue, 04 Aug 2009 14:48:05 -0600
Subject: [Linux-cluster] Redhat 5.3 clvmd hang
Message-ID: <1249418823-sup-3686@beaker.cs.athabascau.ca>

Hi,

I've been experiencing hangs with clvmd (and all associated commands: lvs, vgdsplay, ect) on a 3 node cluster.

I am using cman and clvmd to create clustered lvs from a san backend.

I recently upgraded from 5.2 to 5.3 because occasionally commands began to hang, and would not work until I rebooted either 1 or 2 of the nodes, and there seemed to be several fixes in 5.3.

These are the versions of cman and lvm2-cluster I am using. I'm using the lvm2-cluster from centos, rather than redhat.

[root at cluster6 ~]# cat /etc/redhat-release 
Red Hat Enterprise Linux Server release 5.3 (Tikanga)
[root at cluster6 ~]# rpm -qa cman
cman-2.0.98-1.el5_3.4
[root at cluster6 ~]# rpm -qa lvm2-cluster
lvm2-cluster-2.02.40-7.el5 <- from centos packages

and this is what nodes there are:

[root at cluster6 ~]# cman_tool nodes
Node  Sts   Inc   Joined               Name
   1   M   2336   2009-08-04 14:39:42  cluster7
   2   M   2336   2009-08-04 14:39:42  cluster5
   3   M   2332   2009-08-04 14:39:19  cluster6
[root at cluster6 ~]# cman_tool services
type             level name     id       state       
fence            0     default  00010003 none        
[1 2 3]
dlm              1     clvmd    00020003 none        
[1 2 3]

and this is what's hanging after "service clvmd start"

 7056 pts/0    S+     0:00 /usr/sbin/vgscan

it just sits there forever, and I can't run any "lvm" type commands. If I kill it the clvmd init script moves onto vgchange -ayl which also hangs.

Any thoughts? Is there some kind of human error misconfiguration that could be going on here?

Thanks,
Curtis.

__ 
    This communication is intended for the use of the recipient to whom it
    is addressed, and may contain confidential, personal, and or privileged
    information. Please contact us immediately if you are not the intended
    recipient of this communication, and do not copy, distribute, or take
    action relying on it. Any communications received in error, or
    subsequent reply, should be deleted or destroyed.
---



From ccaulfie at redhat.com  Wed Aug  5 07:01:42 2009
From: ccaulfie at redhat.com (Christine Caulfield)
Date: Wed, 05 Aug 2009 08:01:42 +0100
Subject: [Linux-cluster] Redhat 5.3 clvmd hang
In-Reply-To: <1249418823-sup-3686@beaker.cs.athabascau.ca>
References: <1249418823-sup-3686@beaker.cs.athabascau.ca>
Message-ID: <4A792E56.80006@redhat.com>

On 04/08/09 21:48, Curtis Collicutt wrote:
> Hi,
>
> I've been experiencing hangs with clvmd (and all associated commands: lvs, vgdsplay, ect) on a 3 node cluster.
>
> I am using cman and clvmd to create clustered lvs from a san backend.
>
> I recently upgraded from 5.2 to 5.3 because occasionally commands began to hang, and would not work until I rebooted either 1 or 2 of the nodes, and there seemed to be several fixes in 5.3.
>
> These are the versions of cman and lvm2-cluster I am using. I'm using the lvm2-cluster from centos, rather than redhat.
>
> [root at cluster6 ~]# cat /etc/redhat-release
> Red Hat Enterprise Linux Server release 5.3 (Tikanga)
> [root at cluster6 ~]# rpm -qa cman
> cman-2.0.98-1.el5_3.4
> [root at cluster6 ~]# rpm -qa lvm2-cluster
> lvm2-cluster-2.02.40-7.el5<- from centos packages
>
> and this is what nodes there are:
>
> [root at cluster6 ~]# cman_tool nodes
> Node  Sts   Inc   Joined               Name
>     1   M   2336   2009-08-04 14:39:42  cluster7
>     2   M   2336   2009-08-04 14:39:42  cluster5
>     3   M   2332   2009-08-04 14:39:19  cluster6
> [root at cluster6 ~]# cman_tool services
> type             level name     id       state
> fence            0     default  00010003 none
> [1 2 3]
> dlm              1     clvmd    00020003 none
> [1 2 3]
>
> and this is what's hanging after "service clvmd start"
>
>   7056 pts/0    S+     0:00 /usr/sbin/vgscan
>
> it just sits there forever, and I can't run any "lvm" type commands. If I kill it the clvmd init script moves onto vgchange -ayl which also hangs.
>
> Any thoughts? Is there some kind of human error misconfiguration that could be going on here?
>

If you have RHEL then I suggest you reinstall the RHEL packages and 
contact support. There is an outstanding BZ for some similar symptoms 
and I'm sure they'd be glad of some extra evidence.

Chrissie



From gianluca.cecchi at gmail.com  Wed Aug  5 09:20:06 2009
From: gianluca.cecchi at gmail.com (Gianluca Cecchi)
Date: Wed, 5 Aug 2009 11:20:06 +0200
Subject: [Linux-cluster] info on modclusterd in cluster 3
Message-ID: <561c252c0908050220s7342b6f3je1b59dacd1446a3f@mail.gmail.com>

Hello,
I have a cluster 3 on 2 x F11 x86_64 nodes.
main components are:
fence-agents-3.0.0-14.fc11.x86_64
clusterlib-3.0.0-20.fc11.x86_64
resource-agents-3.0.0-12.fc11.noarch
lvm2-cluster-2.02.48-2.fc11.x86_64
modcluster-0.16.1-1.fc11.x86_64
corosynclib-1.0.0-1.fc11.x86_64
cman-3.0.0-20.fc11.x86_64
corosync-1.0.0-1.fc11.x86_64
rgmanager-3.0.0-20.fc11.x86_64

It seems that on both nodes modclusterd is not able to start
with this log in /var/log/clumond.log
Mon Aug  3 17:04:31 2009 clumond: started
Mon Aug  3 17:04:31 2009 clumond: unhandled exception in main():
ServerSocket(port=16851): socket() failed: Address family not
supported by protocol
Mon Aug  3 17:04:31 2009 clumond: died

If I try to manual restart it (with cman, rgmanager, ricci aleady
started and up), I get:
[root ~]# service modclusterd start
Starting Cluster Module - cluster monitor: Setting verbosity level to LogBasic
                                                           [FAILED]

Is this service supposed to start automatically (aka chkconfiged on)
in cluster 3?

In rh 5.3 my chkconfig situation is this:

cman           	0:off	1:off	2:on	3:on	4:on	5:on	6:off
modclusterd    	0:off	1:off	2:on	3:on	4:on	5:on	6:off
openais        	0:off	1:off	2:off	3:off	4:off	5:off	6:off
qdiskd         	0:off	1:off	2:on	3:on	4:on	5:on	6:off
rgmanager      	0:off	1:off	2:on	3:on	4:on	5:on	6:off
ricci          	0:off	1:off	2:on	3:on	4:on	5:on	6:off

and when I have it:
clvmd          	0:off	1:off	2:on	3:on	4:on	5:on	6:off

So that when I have to do maintenance on a node, leaving it open but
without cluster infrastructure active I do :
1) eventual relocation of active services
2) stop of related services:
service modclusterd stop
service rgmanager stop
service ricci stop
service clvmd stop (if present)
service qdiskd stop
service cman stop

At this moment in my F11 nodes I have the same chkconfig setup for
these services (apart from qdisk that I have not configured at the
moment).

Has been changed anything in startup scripts config setup?
Also,
Description for modclusterd init script:
# description:  Starts and stops Ricci Cluster Module - cluster monitor

Description for ricci init script:
# description:  Starts and stops Red Hat Cluster and Storage Remote \
#               Configuration Interface (ricci)

Not very clear (at least to me) what are the relative functions of
these two services.

Thanks,
Gianluca



From gianluca.cecchi at gmail.com  Wed Aug  5 10:52:20 2009
From: gianluca.cecchi at gmail.com (Gianluca Cecchi)
Date: Wed, 5 Aug 2009 12:52:20 +0200
Subject: [Linux-cluster] Re: info on modclusterd in cluster 3
In-Reply-To: <561c252c0908050220s7342b6f3je1b59dacd1446a3f@mail.gmail.com>
References: <561c252c0908050220s7342b6f3je1b59dacd1446a3f@mail.gmail.com>
Message-ID: <561c252c0908050352v7d264b1ci1ba08133706453a5@mail.gmail.com>

On Wed, Aug 5, 2009 at 11:20 AM, Gianluca
Cecchi<gianluca.cecchi at gmail.com> wrote:
> Hello,
> I have a cluster 3 on 2 x F11 x86_64 nodes.

I forgot to mention that also ricci doesn't start:
[root ~]# service ricci start
Starting ricci: main.cpp:228: exception: ServerSocket(port=11111):
socket() failed: Address family not supported by protocol
                                                           [FAILED]

Gianluca



From Alain.Moulle at bull.net  Wed Aug  5 13:07:43 2009
From: Alain.Moulle at bull.net (Alain.Moulle)
Date: Wed, 05 Aug 2009 15:07:43 +0200
Subject: [Linux-cluster] Best quick way to identify CS4 / CS5
Message-ID: <4A79841F.30301@bull.net>

Hi

What is the best way in a script to quickly identify if the CS is CS4 or 
CS5 ?

(I got  "ccs_test -V" but perhaps is there a better solution ?)

Thanks a lot
Alain
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090805/312d899f/attachment.htm>

From ccaulfie at redhat.com  Wed Aug  5 13:19:44 2009
From: ccaulfie at redhat.com (Christine Caulfield)
Date: Wed, 05 Aug 2009 14:19:44 +0100
Subject: [Linux-cluster] Best quick way to identify CS4 / CS5
In-Reply-To: <4A79841F.30301@bull.net>
References: <4A79841F.30301@bull.net>
Message-ID: <4A7986F0.90700@redhat.com>

On 05/08/09 14:07, Alain.Moulle wrote:
> Hi
>
> What is the best way in a script to quickly identify if the CS is CS4 or
> CS5 ?
>

Probably the best way is to use 'cman_tool status' and look for the 
'Version:' string. For RHEL4 it's 5.x.x, for RHEL5 it's 6.[01].x and 
the STABLE3 code returns 6.2.x

chrissie



From rpeterso at redhat.com  Wed Aug  5 14:11:59 2009
From: rpeterso at redhat.com (Bob Peterson)
Date: Wed, 5 Aug 2009 10:11:59 -0400 (EDT)
Subject: [Linux-cluster] Re: GFS2 corruption/withdrawal/crash (Bob
	Peterson)
In-Reply-To: <23884039.6641249394055066.JavaMail.wendell@philco>
Message-ID: <886718136.203981249481519966.JavaMail.root@zmail06.collab.prod.int.phx2.redhat.com>

----- "Wendell Dingus" <wendell at bisonline.com> wrote:
| I had emailed this to you directly Bob but perhaps you didn't get it.
| The URL is an IP address and some folks SPAM/malware filters might
| balk at that. So I'll be a bit more vague in how I pass it on this
| time in hopes of that not happening...
| 
| Here is the GFS2 metadata file. Be forewarned though, it's 2.8GB
| compressed.

Hi Wendell,

I just wanted to send you a progress report: After running all last
night, my fsck.gfs2 on your metadata is 39% into pass1.  So far it
has not found anything wrong, but it has a long way to go.

Fwiw, I recently wrote some performance improvements for gfs1's fsck,
(they're not available yet, except as a patch in the bugzilla)
and I hope to port them to gfs2's fsck when I can.  Unfortunately,
I need to get through this set of patches first before I can start
with that.

Regards,

Bob Peterson
Red Hat File Systems



From matt at bravenet.com  Wed Aug  5 16:39:05 2009
From: matt at bravenet.com (Matthew Kent)
Date: Wed, 05 Aug 2009 09:39:05 -0700
Subject: [Linux-cluster] Fwd: CLVM exclusive mode
In-Reply-To: <29ae894c0907311229w7921e7e5o8c37e058a1e01d8e@mail.gmail.com>
References: <29ae894c0907290553r5c51b938v4da9a0f944d75af2@mail.gmail.com>
	<29ae894c0907290655l17053b05j22a5e942985d714a@mail.gmail.com>
	<4A705979.1040605@redhat.com>
	<1cafab770907292242m1c4253b1o60be8a58856365db@mail.gmail.com>
	<4A713CF5.707@redhat.com>
	<29ae894c0907300015r7ee0d114q8dca432e9491496b@mail.gmail.com>
	<1248963304.6459.3.camel@mecatol>
	<29ae894c0907301702m452bbfb5u2178565cc127b4da@mail.gmail.com>
	<29ae894c0907311229w7921e7e5o8c37e058a1e01d8e@mail.gmail.com>
Message-ID: <1249490345.4103.380.camel@fuego>

On Fri, 2009-07-31 at 21:29 +0200, brem belguebli wrote:
> Hi,
>  
> Same behaviour as the one from Rafael.
>  
> Everything is coherent as long as you use the exclusive flag from the
> rogue node, the locking does the job. Deactivating an already opened
> VG (mounted lvol) is not possible either. How could this behave in
> case one used raw devices instead of FS ? 
>  
> But when you come to ignore the exclusive flag on the rogue node
> (vgchange -a y vgXX) the locking is completely bypassed. It's
> definitely here that the watchdog has to be (within the tools
> lvchange, vgchange, or at dlm level).

Is there an open bugzilla # for this? Would like to follow this issue.

-- 
Matthew Kent \ SA \ bravenet.com



From jjzhang at novell.com  Thu Aug  6 01:52:30 2009
From: jjzhang at novell.com (Jia Ju Zhang)
Date: Wed, 05 Aug 2009 19:52:30 -0600
Subject: [Linux-cluster] Fwd: CLVM exclusive mode
Message-ID: <4A7AA7DE02000027000536E3@lucius.provo.novell.com>

Just RFC:
I noticed that 'vgchange -ay' can convert the lock which locked by 'vgchange -aey' 
from EX to CR. Is that acceptable to change the logic into always allocating a new lock
rather than converting an existing lock?
In that case, 'vgchange -ay' won't change the result of 'vgchange -aey'. But if we really
want to convert the lock, we can firstly invoke 'vgchange -aen' to release the EX lock,
then invoke the 'vgchange -ay'.

Does this make sense? Or what side effect it may introduce?

Thanks,
Jiaju

>>> On 8/6/2009 at  9:39 AM, in message <4A7A346B.A94 : 39 : 18251>, Jia Ju Zhang
wrote: 
> On Fri, 2009-07-31 at 21:29 +0200, brem belguebli wrote:
>> Hi,
>>  
>> Same behaviour as the one from Rafael.
>>  
>> Everything is coherent as long as you use the exclusive flag from the
>> rogue node, the locking does the job. Deactivating an already opened
>> VG (mounted lvol) is not possible either. How could this behave in
>> case one used raw devices instead of FS ? 
>>  
>> But when you come to ignore the exclusive flag on the rogue node
>> (vgchange -a y vgXX) the locking is completely bypassed. It's
>> definitely here that the watchdog has to be (within the tools
>> lvchange, vgchange, or at dlm level).
> 
> Is there an open bugzilla # for this? Would like to follow this issue.
> 





From vikigoyal at gmail.com  Thu Aug  6 07:04:31 2009
From: vikigoyal at gmail.com (Vikram Goyal)
Date: Thu, 6 Aug 2009 12:34:31 +0530
Subject: [Linux-cluster] Requesting help in regard to clustering as drbd
	unavailable in FC11
Message-ID: <ee9497fc0908060004i33d53609mbcb22b3313358587@mail.gmail.com>

Hello All,

I need to implement a two node cluster through heartbeat and drbd as a
redundant solution to one of our db application.

Since I am running fc11, I did not find drbd rpm in the repos. Drbd
module can be compiled and installed but I want to remain with the rpms
available through repos. Also since the module is not available in repos
then something else must have replaced it with better functionality.

Could you please suggest what has replaced drbd as I could not find
any such info on the net or better if any pointers are given for the
latest info on the net as to how to go about it in FC11, maybe with
nbd etc.

Thanks for your help,

Regards!
Vikram Goyal
--
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090806/a8a9dfd7/attachment.htm>

From ccaulfie at redhat.com  Thu Aug  6 07:12:00 2009
From: ccaulfie at redhat.com (Christine Caulfield)
Date: Thu, 06 Aug 2009 08:12:00 +0100
Subject: [Linux-cluster] Fwd: CLVM exclusive mode
In-Reply-To: <4A7AA7DE02000027000536E3@lucius.provo.novell.com>
References: <4A7AA7DE02000027000536E3@lucius.provo.novell.com>
Message-ID: <4A7A8240.7000206@redhat.com>

On 06/08/09 02:52, Jia Ju Zhang wrote:
> Just RFC:
> I noticed that 'vgchange -ay' can convert the lock which locked by 'vgchange -aey'
> from EX to CR. Is that acceptable to change the logic into always allocating a new lock
> rather than converting an existing lock?
> In that case, 'vgchange -ay' won't change the result of 'vgchange -aey'. But if we really
> want to convert the lock, we can firstly invoke 'vgchange -aen' to release the EX lock,
> then invoke the 'vgchange -ay'.
>
> Does this make sense? Or what side effect it may introduce?


I think it makes no sense at all, and have already said so on this list. 
As far as I know there is no bugzilla for this problem and therefore it 
isn't being worked on.

So ... if you care about this ... you know what to do ;-)

Chrissie

>>>> On 8/6/2009 at  9:39 AM, in message<4A7A346B.A94 : 39 : 18251>, Jia Ju Zhang
> wrote:
>> On Fri, 2009-07-31 at 21:29 +0200, brem belguebli wrote:
>>> Hi,
>>>
>>> Same behaviour as the one from Rafael.
>>>
>>> Everything is coherent as long as you use the exclusive flag from the
>>> rogue node, the locking does the job. Deactivating an already opened
>>> VG (mounted lvol) is not possible either. How could this behave in
>>> case one used raw devices instead of FS ?
>>>
>>> But when you come to ignore the exclusive flag on the rogue node
>>> (vgchange -a y vgXX) the locking is completely bypassed. It's
>>> definitely here that the watchdog has to be (within the tools
>>> lvchange, vgchange, or at dlm level).
>> Is there an open bugzilla # for this? Would like to follow this issue.
>>
>
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster



From wendell at bisonline.com  Thu Aug  6 19:08:02 2009
From: wendell at bisonline.com (Wendell Dingus)
Date: Thu, 6 Aug 2009 15:08:02 -0400 (EDT)
Subject: [Linux-cluster] Re: GFS2 corruption/withdrawal/crash
In-Reply-To: <886718136.203981249481519966.JavaMail.root@zmail06.collab.prod.int.phx2.redhat.com>
Message-ID: <6386746.9121249585677585.JavaMail.wendell@philco>

Thanks very much for the update. Hopefully by now it might have completed. 

Since it takes so long to run (here) I'm hoping to be able to
correct it's ills over the weekend. If you are not complete or your 
results are inconclusive do you think running the already existing
fsck.gfs2 against the filesystem *twice* would be worthwhile?

Thanks...

> Hi Wendell, 
> 
> I just wanted to send you a progress report: After running all last 
> night, my fsck.gfs2 on your metadata is 39% into pass1. So far it 
> has not found anything wrong, but it has a long way to go. 
> 
> Fwiw, I recently wrote some performance improvements for gfs1's fsck, 
> (they're not available yet, except as a patch in the bugzilla) 
> and I hope to port them to gfs2's fsck when I can. Unfortunately, 
> I need to get through this set of patches first before I can start 
> with that. 

From jamesc at exa.com  Thu Aug  6 21:02:46 2009
From: jamesc at exa.com (James Chamberlain)
Date: Thu, 6 Aug 2009 17:02:46 -0400
Subject: [Linux-cluster] GFS1 filesystem consistency error
Message-ID: <9B003828-ACE3-4E2D-BBCF-D36D6FE3B764@exa.com>

Hi all,

While I'm waiting for a gfs_fsck to complete, I thought I'd send this  
in the list's direction and ask if anyone else had any thoughts about  
it.

Aug  6 04:51:13 s12n01 kernel: 493 [RAIDarray.mpp]FastT1:1:0:3 Cmnd  
failed-retry the same path. vcmnd SN 18724261484 pdev H3:C0:T1:L3  
0x00/0x00/0x00 0x00020000 mpp_status:2
Aug  6 04:51:13 s12n01 kernel: 493 [RAIDarray.mpp]FastT1:1:0:3 Cmnd  
failed-retry the same path. vcmnd SN 18724261488 pdev H3:C0:T1:L3  
0x00/0x00/0x00 0x00020000 mpp_status:2
Aug  6 04:51:13 s12n01 kernel: 493 [RAIDarray.mpp]FastT1:1:0:3 Cmnd  
failed-retry the same path. vcmnd SN 18724261490 pdev H3:C0:T1:L3  
0x00/0x00/0x00 0x00020000 mpp_status:2
[...]
Aug  6 05:01:14 s12n01 kernel: GFS: fsid=s12:scratch13.2: fatal:  
filesystem consistency error
Aug  6 05:01:14 s12n01 kernel: GFS: fsid=s12:scratch13.2:   inode =  
4918461516/4918461516
Aug  6 05:01:14 s12n01 kernel: GFS: fsid=s12:scratch13.2:   function =  
dinode_dealloc
Aug  6 05:01:14 s12n01 kernel: GFS: fsid=s12:scratch13.2:   file = / 
builddir/build/BUILD/gfs-kernel-2.6.9-75/smp/src/gfs/inode.c, line = 529
Aug  6 05:01:14 s12n01 kernel: GFS: fsid=s12:scratch13.2:   time =  
1249549274
Aug  6 05:01:14 s12n01 kernel: GFS: fsid=s12:scratch13.2: about to  
withdraw from the cluster
Aug  6 05:01:14 s12n01 kernel: GFS: fsid=s12:scratch13.2: waiting for  
outstanding I/O
Aug  6 05:01:14 s12n01 kernel: GFS: fsid=s12:scratch13.2: telling LM  
to withdraw
Aug  6 05:01:15 s12n03 kernel: GFS: fsid=s12:scratch13.0: jid=2:  
Trying to acquire journal lock...
Aug  6 05:01:15 s12n03 kernel: GFS: fsid=s12:scratch13.0: jid=2:  
Looking at journal...
Aug  6 05:01:15 s12n02 kernel: GFS: fsid=s12:scratch13.1: jid=2:  
Trying to acquire journal lock...
Aug  6 05:01:15 s12n02 kernel: GFS: fsid=s12:scratch13.1: jid=2: Busy
Aug  6 05:01:15 s12n03 kernel: GFS: fsid=s12:scratch13.0: jid=2:  
Acquiring the transaction lock...
Aug  6 05:01:15 s12n03 kernel: GFS: fsid=s12:scratch13.0: jid=2:  
Replaying journal...
Aug  6 05:01:21 s12n03 kernel: GFS: fsid=s12:scratch13.0: jid=2:  
Replayed 10050 of 11671 blocks
Aug  6 05:01:21 s12n03 kernel: GFS: fsid=s12:scratch13.0: jid=2:  
replays = 10050, skips = 472, sames = 1149
Aug  6 05:01:21 s12n03 kernel: GFS: fsid=s12:scratch13.0: jid=2:  
Journal replayed in 7s
Aug  6 05:01:21 s12n03 kernel: GFS: fsid=s12:scratch13.0: jid=2: Done
Aug  6 05:01:21 s12n01 kernel: lock_dlm: withdraw abandoned memory
Aug  6 05:01:21 s12n01 kernel: GFS: fsid=s12:scratch13.2: withdrawn
Aug  6 05:01:21 s12n01 kernel:   mh_magic = 0x01161970
Aug  6 05:01:22 s12n01 kernel:   mh_type = 4
Aug  6 05:01:22 s12n01 kernel:   mh_generation = 133
Aug  6 05:01:22 s12n01 kernel:   mh_format = 400
Aug  6 05:01:22 s12n01 kernel:   mh_incarn = 0
Aug  6 05:01:22 s12n01 kernel:   no_formal_ino = 4918461516
Aug  6 05:01:22 s12n01 kernel:   no_addr = 4918461516
Aug  6 05:01:22 s12n01 kernel:   di_mode = 0664
Aug  6 05:01:22 s12n01 kernel:   di_uid = 690
Aug  6 05:01:22 s12n01 kernel:   di_gid = 2017
Aug  6 05:01:22 s12n01 kernel:   di_nlink = 0
Aug  6 05:01:22 s12n01 kernel:   di_size = 0
Aug  6 05:01:22 s12n01 kernel:   di_blocks = 119
Aug  6 05:01:22 s12n01 kernel:   di_atime = 1248334920
Aug  6 05:01:22 s12n01 kernel:   di_mtime = 1249549274
Aug  6 05:01:22 s12n01 kernel:   di_ctime = 1249549274
Aug  6 05:01:22 s12n01 kernel:   di_major = 0
Aug  6 05:01:22 s12n01 kernel:   di_minor = 0
Aug  6 05:01:22 s12n01 kernel:   di_rgrp = 4918433973
Aug  6 05:01:22 s12n01 kernel:   di_goal_rgrp = 4918433973
Aug  6 05:01:22 s12n01 kernel:   di_goal_dblk = 27528
Aug  6 05:01:22 s12n01 kernel:   di_goal_mblk = 27528
Aug  6 05:01:22 s12n01 kernel:   di_flags = 0x00000000
Aug  6 05:01:22 s12n01 kernel:   di_payload_format = 0
Aug  6 05:01:22 s12n01 kernel:   di_type = 1
Aug  6 05:01:22 s12n01 kernel:   di_height = 0
Aug  6 05:01:22 s12n01 kernel:   di_incarn = 0
Aug  6 05:01:22 s12n01 kernel:   di_pad = 0
Aug  6 05:01:22 s12n01 kernel:   di_depth = 0
Aug  6 05:01:22 s12n01 kernel:   di_entries = 0
Aug  6 05:01:22 s12n01 kernel:   no_formal_ino = 0
Aug  6 05:01:22 s12n01 kernel:   no_addr = 0
Aug  6 05:01:22 s12n01 kernel:   di_eattr = 0
Aug  6 05:01:22 s12n01 kernel:   di_reserved =
Aug  6 05:01:22 s12n01 kernel: 00 00 00 00 00 00 00 00 00 00 00 00 00  
00 00 00
Aug  6 05:01:22 s12n01 last message repeated 2 times
Aug  6 05:01:22 s12n01 kernel: 00 00 00 00 00 00 00 00
Aug  6 05:01:35 s12n01 clurgmgrd: [6943]: <err> clusterfs:gfs- 
scratch13: Mount point is not accessible!
Aug  6 05:01:35 s12n01 clurgmgrd[6943]: <notice> status on  
clusterfs:gfs-scratch13 returned 1 (generic error)
Aug  6 05:01:35 s12n01 clurgmgrd[6943]: <notice> Stopping service  
scratch13
Aug  6 05:01:35 s12n01 clurgmgrd: [6943]: <info> Removing IPv4 address  
10.14.12.5 from bond0
Aug  6 05:01:45 s12n01 clurgmgrd: [6943]: <err> /scratch13 is not a  
directory
Aug  6 05:01:45 s12n01 clurgmgrd[6943]: <notice> stop on nfsclient:nfs- 
scratch13 returned 2 (invalid argument(s))
Aug  6 05:01:45 s12n01 clurgmgrd[6943]: <crit> #12: RG scratch13  
failed to stop; intervention required
Aug  6 05:01:45 s12n01 clurgmgrd[6943]: <notice> Service scratch13 is  
failed

I don't think that the FastT messages I've included caused the  
problem, since I've seen them at times without the file system  
crashing.  It's not great to be getting those sorts of messages, but  
the file system didn't crash for another 10 minutes after that.  I  
can't rule it out, though.

A bit later, I tried to disable the scratch13 service so I could work  
on its associated file system.  The "clusvcadm -d" failed, as below,  
but the service was still disabled.  Any thoughts?

Aug  6 05:50:10 s12n01 clurgmgrd[6943]: <notice> Stopping service  
scratch13
Aug  6 05:50:10 s12n01 clurgmgrd: [6943]: <err> /scratch13 is not a  
directory
Aug  6 05:50:10 s12n01 clurgmgrd[6943]: <notice> stop on nfsclient:nfs- 
scratch13 returned 2 (invalid argument(s))
Aug  6 05:50:10 s12n01 clurgmgrd[6943]: <alert> Marking scratch13 as  
'disabled', but some resources may still be allocated!
Aug  6 05:50:10 s12n01 clurgmgrd[6943]: <notice> Service scratch13 is  
disabled

I've made it through pass1, 1b, 1c, and 2 in gfs_fsck, and I believe  
I'm in pass3 right now.  I'm getting pages and pages of the following  
messages for the last few hours... but then, I always have, whenever  
I've needed to run gfs_fsck.  I'm not too worried since I've seen this  
before, but I'd still like to understand it better.  Could someone  
enlighten me as to their meaning, and if I should be more concerned?

Converting 366 unused metadata blocks to free data blocks...
Converting 192 unused metadata blocks to free data blocks...
Converting 88 unused metadata blocks to free data blocks...
Converting 87 unused metadata blocks to free data blocks...
Converting 681 unused metadata blocks to free data blocks...
Converting 339 unused metadata blocks to free data blocks...
Converting 256 unused metadata blocks to free data blocks...
Converting 441 unused metadata blocks to free data blocks...
Converting 375 unused metadata blocks to free data blocks...
Converting 315 unused metadata blocks to free data blocks...
Converting 173 unused metadata blocks to free data blocks...
Converting 118 unused metadata blocks to free data blocks...
Converting 69 unused metadata blocks to free data blocks...
Converting 396 unused metadata blocks to free data blocks...
Converting 331 unused metadata blocks to free data blocks...
Converting 397 unused metadata blocks to free data blocks...
Converting 275 unused metadata blocks to free data blocks...
Converting 439 unused metadata blocks to free data blocks...

Anyone have any insight they could share with me?  It's been since  
December that I last had major problems with a GFS file system, but  
it's already happened three times this week.  This storage cluster is  
running CentOS 4.6 and GFS1.

Thanks,

James



From binder.christian at gmx.de  Fri Aug  7 16:02:42 2009
From: binder.christian at gmx.de (Christian Binder)
Date: Fri, 07 Aug 2009 18:02:42 +0200
Subject: [Linux-cluster] RHEL 5.2 - ccsd. Invalid descriptor specified (-52)
Message-ID: <20090807160242.165080@gmx.net>

hello,

we are running a two node cluster with RHEL5.2 without problems for a long time. Yesterday we noticed some messages in /var/log/messages, which we cannot fully interpret.

---
Aug  5 14:54:33 dbcn01 auditd[10729]: Audit daemon rotating log files
Aug  6 02:45:16 dbcn01 ccsd[11379]: Invalid descriptor specified (-52).
Aug  6 02:45:16 dbcn01 ccsd[11379]: Someone may be attempting something evil.
Aug  6 02:45:16 dbcn01 ccsd[11379]: Error while processing get: Invalid request descriptor
Aug  6 02:45:16 dbcn01 ccsd[11379]: Invalid descriptor specified (-52).
Aug  6 02:45:16 dbcn01 ccsd[11379]: Someone may be attempting something evil.
Aug  6 02:45:16 dbcn01 ccsd[11379]: Error while processing get: Invalid request descriptor
Aug  6 02:45:16 dbcn01 ccsd[11379]: Invalid descriptor specified (-22).
Aug  6 02:45:16 dbcn01 ccsd[11379]: Someone may be attempting something evil.
Aug  6 02:45:16 dbcn01 ccsd[11379]: Error while processing disconnect: Invalid request descriptor
Aug  6 04:14:26 dbcn01 auditd[10729]: Audit daemon rotating log files
--

Is this a indication of a network-error ? (we have two Fibre Switche and use Ethernetbonding on the hosts). 

There is no error logged on the bond-device:

[root at dbcn01 ~]# cat /proc/net/bonding/bond0
Ethernet Channel Bonding Driver: v3.2.4 (January 28, 2008)

Bonding Mode: load balancing (round-robin)
MII Status: up
MII Polling Interval (ms): 100
Up Delay (ms): 0
Down Delay (ms): 0

Slave Interface: eth4
MII Status: up
Link Failure Count: 0
Permanent HW addr: 00:15:17:77:bf:82

Slave Interface: eth6
MII Status: up
Link Failure Count: 0
Permanent HW addr: 00:15:17:77:c1:90

Because I go on holidays next week I would like to ask, if I (or my colleague) have to by worried about this. 
There is no error since this event and unfortunaly we have no access for check the Fibre-Ethernetswitches.

Thank you for your help

Christian
-- 
Jetzt kostenlos herunterladen: Internet Explorer 8 und Mozilla Firefox 3 -
sicherer, schneller und einfacher! http://portal.gmx.net/de/go/atbrowser



From wendell at bisonline.com  Sat Aug  8 23:19:25 2009
From: wendell at bisonline.com (Wendell Dingus)
Date: Sat, 8 Aug 2009 19:19:25 -0400 (EDT)
Subject: [Linux-cluster] Re: GFS2 corruption/withdrawal/crash
In-Reply-To: <838974665.138471249773461957.JavaMail.root@mail.bisonline.com>
Message-ID: <247362321.138491249773565280.JavaMail.root@mail.bisonline.com>

Well, I just ran fsck.gfs2 against this filesystem twice with a 10-minute 
pause between them. As such:
# fsck -C -t gfs2 -y /dev/mapper/VGIMG0-LVIMG0

Output of second run:
fsck 1.39 (29-May-2006)
Initializing fsck
Recovering journals (this may take a while)...
Journal recovery complete.
Validating Resource Group index.
Level 1 RG check.
(level 1 passed)
Starting pass1
Pass1 complete
Starting pass1b
Pass1b complete
Starting pass1c
Pass1c complete
Starting pass2
Pass2 complete
Starting pass3
Pass3 complete
Starting pass4
Pass4 complete
Starting pass5
Unlinked block found at block 37974707 (0x24372b3), left unchanged.
..snip about 30 total of these..
Unlinked block found at block 96603710 (0x5c20e3e), left unchanged.
Pass5 complete
Writing changes to disk
gfs2_fsck complete

When it was done I remounted the filesystem and tried to "rm -rf /raid1/bad"
which is a subdir in the root of this filesystem that contains the zero-byte
file that was the focal point of this grief to start with. 

Results:

Aug  8 19:09:43 server1 kernel: GFS2: fsid=clustname:raid1.0: fatal: invalid metad
ata block
Aug  8 19:09:43 server1 kernel: GFS2: fsid=clustname:raid1.0:   bh = 1633350398 (m
agic number)
Aug  8 19:09:43 server1 kernel: GFS2: fsid=clustname:raid1.0:   function = gfs2_me
ta_indirect_buffer, file = /builddir/build/BUILD/gfs2-kmod-1.92/_kmod_build_xen/
meta_io.c, line = 334
Aug  8 19:09:43 server1 kernel: GFS2: fsid=clustname:raid1.0: about to withdraw th
is file system
Aug  8 19:09:43 server1 kernel: GFS2: fsid=clustname:raid1.0: telling LM to withdr
aw
Aug  8 19:09:44 server1 kernel: GFS2: fsid=clustname:raid1.0: withdrawn
Aug  8 19:09:44 server1 kernel:
Aug  8 19:09:44 server1 kernel: Call Trace:
Aug  8 19:09:44 server1 kernel:  [<ffffffff8854b91a>] :gfs2:gfs2_lm_withdraw+0xc1/0xd0
Aug  8 19:09:44 server1 kernel:  [<ffffffff80262907>] __wait_on_bit+0x60/0x6e
Aug  8 19:09:44 server1 kernel:  [<ffffffff80215780>] sync_buffer+0x0/0x3f
Aug  8 19:09:44 server1 kernel:  [<ffffffff80262981>] out_of_line_wait_on_bit+0x6c/0x78
Aug  8 19:09:44 server1 kernel:  [<ffffffff8029a016>] wake_bit_function+0x0/0x23
Aug  8 19:09:44 server1 kernel:  [<ffffffff8021a7f6>] submit_bh+0x10a/0x111
Aug  8 19:09:44 server1 kernel:  [<ffffffff8855d627>] :gfs2:gfs2_meta_check_ii
+0x2c/0x38
Aug  8 19:09:44 server1 kernel:  [<ffffffff8854f168>] :gfs2:gfs2_meta_indirect_buffer+0x104/0x160
Aug  8 19:09:44 server1 kernel:  [<ffffffff8853f786>] :gfs2:recursive_scan+0x96/0x175
Aug  8 19:09:44 server1 kernel:  [<ffffffff8853f82c>] :gfs2:recursive_scan+0x13c/0x175
Aug  8 19:09:44 server1 kernel:  [<ffffffff8854065a>] :gfs2:do_strip+0x0/0x358
Aug  8 19:09:44 server1 kernel:  [<ffffffff88548d21>] :gfs2:glock_work_func+0x0/0xa8
Aug  8 19:09:44 server1 kernel:  [<ffffffff8853f8fe>] :gfs2:trunc_dealloc+0x99/0xe7
Aug  8 19:09:44 server1 kernel:  [<ffffffff8854065a>] :gfs2:do_strip+0x0/0x358
Aug  8 19:09:44 server1 kernel:  [<ffffffff80286595>] deactivate_task+0x28/0x5f
Aug  8 19:09:44 server1 kernel:  [<ffffffff8853fa99>] :gfs2:gfs2_truncatei_resume+0x10/0x1f
Aug  8 19:09:44 server1 kernel:  [<ffffffff8854734d>] :gfs2:do_promote+0x9a/0x117
Aug  8 19:09:44 server1 kernel:  [<ffffffff885485a1>] :gfs2:finish_xmote+0x28c/0x2b2
Aug  8 19:09:44 server1 kernel:  [<ffffffff88548d3e>] :gfs2:glock_work_func+0x1d/0xa8
Aug  8 19:09:44 server1 kernel:  [<ffffffff8024ee76>] run_workqueue+0x94/0xe4
Aug  8 19:09:44 server1 kernel:  [<ffffffff8024b781>] worker_thread+0x0/0x122
Aug  8 19:09:44 server1 kernel:  [<ffffffff80299dd0>] keventd_create_kthread+0x0/0xc4
Aug  8 19:09:44 server1 kernel:  [<ffffffff8024b871>] worker_thread+0xf0/0x122
Aug  8 19:09:44 server1 kernel:  [<ffffffff80286d8b>] default_wake_function+0x0/0xe
Aug  8 19:09:44 server1 kernel:  [<ffffffff80299dd0>] keventd_create_kthread+0x0/0xc4
Aug  8 19:09:44 server1 kernel:  [<ffffffff80299dd0>] keventd_create_kthread+0x0/0xc4
Aug  8 19:09:44 server1 kernel:  [<ffffffff802334b4>] kthread+0xfe/0x132
Aug  8 19:09:44 server1 kernel:  [<ffffffff8025fb2c>] child_rip+0xa/0x12
Aug  8 19:09:44 server1 kernel:  [<ffffffff80299dd0>] keventd_create_kthread+0x0/0xc4
Aug  8 19:09:44 server1 kernel:  [<ffffffff802333b6>] kthread+0x0/0x132
Aug  8 19:09:44 server1 kernel:  [<ffffffff8025fb22>] child_rip+0x0/0x12

<frown>

Any suggestions?

Thanks...


> Hi Wendell, 
> 
> Actually, the first run is still at 59% into pass1. My hardware 
> might be a touch slower than yours though. So far everything has 
> checked out and it hasn't found and/or fixed any problems. 
> I'm hoping it's done by Monday. 
> 
> Regards, 
> 
> Bob Peterson 
> Red Hat File Systems 



From wen.gang.wang at oracle.com  Mon Aug 10 04:57:27 2009
From: wen.gang.wang at oracle.com (Wengang Wang)
Date: Mon, 10 Aug 2009 12:57:27 +0800
Subject: [Linux-cluster] which kernel fits gfs2 STABLE branch
Message-ID: <4A7FA8B7.5080708@oracle.com>

Hi experts,

I downloaded the gfs2 STABLE branch from
http://git.fedorahosted.org/git/cluster.git?p=cluster.git;a=shortlog;h=refs/heads/STABLE2

I compile with 2.6.18 kernel, but fails since it support >=2.6.29
however when I compile with 2.6.29.6(fedora kernel), mainline 2.6.29 
tags, or the 2.6.30.4 kernel, I got following error like:

----------------
/home/wwg/dir/wwg/work/gfs2.git/gnbd-kernel/src/gnbd.c:1054: error: 
?struct gendisk? has no member named ?dev?
make[3]: *** [/home/wwg/dir/wwg/work/gfs2.git/gnbd-kernel/src/gnbd.o] 
Error 1
make[2]: *** [_module_/home/wwg/dir/wwg/work/gfs2.git/gnbd-kernel/src] 
Error 2
----------------

so which kernel I should compile with?

regards,
wengang.

-- 
--just begin to learn, you are never too late...



From mad at wol.de  Mon Aug 10 09:29:13 2009
From: mad at wol.de (Marc - A. Dahlhaus [ Administration | Westermann GmbH ])
Date: Mon, 10 Aug 2009 11:29:13 +0200
Subject: [Linux-cluster] which kernel fits gfs2 STABLE branch
In-Reply-To: <4A7FA8B7.5080708@oracle.com>
References: <4A7FA8B7.5080708@oracle.com>
Message-ID: <1249896553.19917.2.camel@marc>

http://sourceware.org/cluster

you can easily spot the dependencies there...

Am Montag, den 10.08.2009, 12:57 +0800 schrieb Wengang Wang:
> Hi experts,
> 
> I downloaded the gfs2 STABLE branch from
> http://git.fedorahosted.org/git/cluster.git?p=cluster.git;a=shortlog;h=refs/heads/STABLE2
> 
> I compile with 2.6.18 kernel, but fails since it support >=2.6.29
> however when I compile with 2.6.29.6(fedora kernel), mainline 2.6.29 
> tags, or the 2.6.30.4 kernel, I got following error like:
> 
> ----------------
> /home/wwg/dir/wwg/work/gfs2.git/gnbd-kernel/src/gnbd.c:1054: error: 
> ?struct gendisk? has no member named ?dev?
> make[3]: *** [/home/wwg/dir/wwg/work/gfs2.git/gnbd-kernel/src/gnbd.o] 
> Error 1
> make[2]: *** [_module_/home/wwg/dir/wwg/work/gfs2.git/gnbd-kernel/src] 
> Error 2
> ----------------
> 
> so which kernel I should compile with?
> 
> regards,
> wengang.




From esggrupos at gmail.com  Mon Aug 10 10:24:25 2009
From: esggrupos at gmail.com (ESGLinux)
Date: Mon, 10 Aug 2009 12:24:25 +0200
Subject: [Linux-cluster] do I have a fence DRAC device?
Message-ID: <3128ba140908100324l6cdb4c34ra5f5edb39c6903e9@mail.gmail.com>

Hi all,
I was designing a 2 node cluster and I was going to use 2 servers DELL
PowerEdge 1950. I was going to buy a DRAC card to use for fencing but
running several commands in the servers I have noticed that when I run this
command:
#ipmitool lan print
Set in Progress : Set Complete
Auth Type Support : NONE MD2 MD5 PASSWORD
Auth Type Enable : Callback : MD2 MD5
                        : User : MD2 MD5
                        : Operator : MD2 MD5
                        : Admin : MD2 MD5
                        : OEM : MD2 MD5
IP Address Source : Static Address
IP Address : 0.0.0.0
Subnet Mask : 0.0.0.0
MAC Address : 00:1e:c9:ae:6f:7e
SNMP Community String : public
IP Header : TTL=0x40 Flags=0x40 Precedence=0x00 TOS=0x10
Default Gateway IP : 0.0.0.0
Default Gateway MAC : 00:00:00:00:00:00
Backup Gateway IP : 0.0.0.0
Backup Gateway MAC : 00:00:00:00:00:00
802.1q VLAN ID : Disabled
802.1q VLAN Priority : 0
RMCP+ Cipher Suites : 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14
Cipher Suite Priv Max : aaaaaaaaaaaaaaa
                        : X=Cipher Suite Unused
                        : c=CALLBACK
                        : u=USER
                        : o=OPERATOR
                        : a=ADMIN
                        : O=OEM

does this mean that I already have an ipmi card (not configured) that I can
use for fencing? if the anwser is yes, where hell must I configure it? I
don?t see wher can I do it.

If I haven?t a fencing device which one do you recommed to use?

Thanks in advance

ESG
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090810/91d5eb5c/attachment.htm>

From swhiteho at redhat.com  Mon Aug 10 10:46:44 2009
From: swhiteho at redhat.com (Steven Whitehouse)
Date: Mon, 10 Aug 2009 11:46:44 +0100
Subject: [Linux-cluster] Re: GFS2 corruption/withdrawal/crash
In-Reply-To: <247362321.138491249773565280.JavaMail.root@mail.bisonline.com>
References: <247362321.138491249773565280.JavaMail.root@mail.bisonline.com>
Message-ID: <1249901204.3405.3.camel@localhost.localdomain>

Hi,

On Sat, 2009-08-08 at 19:19 -0400, Wendell Dingus wrote:
> Well, I just ran fsck.gfs2 against this filesystem twice with a 10-minute 
> pause between them. As such:
> # fsck -C -t gfs2 -y /dev/mapper/VGIMG0-LVIMG0
> 
> Output of second run:
> fsck 1.39 (29-May-2006)
> Initializing fsck
> Recovering journals (this may take a while)...
> Journal recovery complete.
> Validating Resource Group index.
> Level 1 RG check.
> (level 1 passed)
> Starting pass1
> Pass1 complete
> Starting pass1b
> Pass1b complete
> Starting pass1c
> Pass1c complete
> Starting pass2
> Pass2 complete
> Starting pass3
> Pass3 complete
> Starting pass4
> Pass4 complete
> Starting pass5
> Unlinked block found at block 37974707 (0x24372b3), left unchanged.
> ..snip about 30 total of these..
> Unlinked block found at block 96603710 (0x5c20e3e), left unchanged.
> Pass5 complete
> Writing changes to disk
> gfs2_fsck complete
> 
> When it was done I remounted the filesystem and tried to "rm -rf /raid1/bad"
> which is a subdir in the root of this filesystem that contains the zero-byte
> file that was the focal point of this grief to start with. 
> 
> Results:
> 
That looks like a bug in fsck at least as it should be dealing with the
unlinked blocks that it finds, not ignoring them. Chances are that the
block which is causing the issues belongs to one of the unlinked blocks
(inodes I think it should say)

Steve.




From mad at wol.de  Mon Aug 10 11:08:20 2009
From: mad at wol.de (Marc - A. Dahlhaus [ Administration | Westermann GmbH ])
Date: Mon, 10 Aug 2009 13:08:20 +0200
Subject: [Linux-cluster] which kernel fits gfs2 STABLE branch
In-Reply-To: <4A7FEE8D.8070400@oracle.com>
References: <4A7FA8B7.5080708@oracle.com> <1249896553.19917.2.camel@marc>
	<4A7FEC6F.2050405@oracle.com>  <4A7FEE8D.8070400@oracle.com>
Message-ID: <1249902500.20848.11.camel@marc>

Am Montag, den 10.08.2009, 17:55 +0800 schrieb Wengang Wang:
> Wengang Wang wrote:
> > Hi,
> > 
> > the brach:
> > cluster.git STABLE2 branch: 
> > http://git.fedorahosted.org/git/cluster.git?p=cluster.git;a=shortlog;h=refs/heads/STABLE2 
> > 
> > 
> >     * use with current kernel.org stable release
> > 
> > I think it works with kernel.org, which I say it mainline, but I got error:
> > 
> >>> ----------------
> >>> /home/wwg/dir/wwg/work/gfs2.git/gnbd-kernel/src/gnbd.c:1054: error: 
> >>> ?struct gendisk? has no member named ?dev?
> >>> make[3]: *** [/home/wwg/dir/wwg/work/gfs2.git/gnbd-kernel/src/gnbd.o] 
> >>> Error 1
> >>> make[2]: *** 
> >>> [_module_/home/wwg/dir/wwg/work/gfs2.git/gnbd-kernel/src] Error 2
> 
> what I did to get that is:
> 
> ./configure
> make
> 
> regards,
> wengang
> 
> > do you have any idea?
> > 
> > regards,
> > wengang.

Please don't reply offlist...

stable2 got a little bit behind with respect to mainline kernel support
lately as the developers spend there time on finalizing stable3
series...

Stable2 should work with 2.6.29 if you use a git checkout...

Do you really need the deprecated gnbd services?

You can simply disable gnbd on configure...


Marc



From rpeterso at redhat.com  Mon Aug 10 14:05:29 2009
From: rpeterso at redhat.com (Bob Peterson)
Date: Mon, 10 Aug 2009 10:05:29 -0400 (EDT)
Subject: [Linux-cluster] Re: GFS2 corruption/withdrawal/crash
In-Reply-To: <1249901204.3405.3.camel@localhost.localdomain>
Message-ID: <151206104.389651249913129665.JavaMail.root@zmail06.collab.prod.int.phx2.redhat.com>

----- "Steven Whitehouse" <swhiteho at redhat.com> wrote:
| Hi,
| 
| On Sat, 2009-08-08 at 19:19 -0400, Wendell Dingus wrote:
| > Well, I just ran fsck.gfs2 against this filesystem twice with a
| 10-minute 
| > pause between them. As such:
| > # fsck -C -t gfs2 -y /dev/mapper/VGIMG0-LVIMG0
| > 
| > Output of second run:
| > fsck 1.39 (29-May-2006)
| > Initializing fsck
| > Recovering journals (this may take a while)...
| > Journal recovery complete.
| > Validating Resource Group index.
| > Level 1 RG check.
| > (level 1 passed)
| > Starting pass1
| > Pass1 complete
| > Starting pass1b
| > Pass1b complete
| > Starting pass1c
| > Pass1c complete
| > Starting pass2
| > Pass2 complete
| > Starting pass3
| > Pass3 complete
| > Starting pass4
| > Pass4 complete
| > Starting pass5
| > Unlinked block found at block 37974707 (0x24372b3), left unchanged.
| > ..snip about 30 total of these..
| > Unlinked block found at block 96603710 (0x5c20e3e), left unchanged.
| > Pass5 complete
| > Writing changes to disk
| > gfs2_fsck complete
| > 
| > When it was done I remounted the filesystem and tried to "rm -rf
| /raid1/bad"
| > which is a subdir in the root of this filesystem that contains the
| zero-byte
| > file that was the focal point of this grief to start with. 
| > 
| > Results:
| > 
| That looks like a bug in fsck at least as it should be dealing with
| the
| unlinked blocks that it finds, not ignoring them. Chances are that
| the
| block which is causing the issues belongs to one of the unlinked
| blocks
| (inodes I think it should say)
| 
| Steve.

Hi,

The "Unlinked block found...left unchanged." messages are harmless.
This merely means that fsck.gfs2 found some blocks that were
marked as "unlinked metadata" that should be automatically
reassigned by gfs2's kernel code when needed.  At some point, we 
made the decision not to fix the bitmaps for various reasons.  I don't
remember the details, but I remember discussing it anyway.  Lately
I've been thinking that we made the wrong decision and I should make
fsck.gfs2 fix them rather than ignore them.

In theory, those blocks should not have caused the kernel withdraw
problem you saw.  Do any of the blocks in the fsck output correspond
to the block complained about in the kernel withdraw output?  That
might be an important clue what happened.

Regards,

Bob Peterson
Red Hat File Systems



From swhiteho at redhat.com  Mon Aug 10 14:09:42 2009
From: swhiteho at redhat.com (Steven Whitehouse)
Date: Mon, 10 Aug 2009 15:09:42 +0100
Subject: [Linux-cluster] Re: GFS2 corruption/withdrawal/crash
In-Reply-To: <151206104.389651249913129665.JavaMail.root@zmail06.collab.prod.int.phx2.redhat.com>
References: <151206104.389651249913129665.JavaMail.root@zmail06.collab.prod.int.phx2.redhat.com>
Message-ID: <1249913382.3298.157.camel@localhost.localdomain>

Hi,

On Mon, 2009-08-10 at 10:05 -0400, Bob Peterson wrote:
> ----- "Steven Whitehouse" <swhiteho at redhat.com> wrote:
> | Hi,
> | 
> | On Sat, 2009-08-08 at 19:19 -0400, Wendell Dingus wrote:
> | > Well, I just ran fsck.gfs2 against this filesystem twice with a
> | 10-minute 
> | > pause between them. As such:
> | > # fsck -C -t gfs2 -y /dev/mapper/VGIMG0-LVIMG0
> | > 
> | > Output of second run:
> | > fsck 1.39 (29-May-2006)
> | > Initializing fsck
> | > Recovering journals (this may take a while)...
> | > Journal recovery complete.
> | > Validating Resource Group index.
> | > Level 1 RG check.
> | > (level 1 passed)
> | > Starting pass1
> | > Pass1 complete
> | > Starting pass1b
> | > Pass1b complete
> | > Starting pass1c
> | > Pass1c complete
> | > Starting pass2
> | > Pass2 complete
> | > Starting pass3
> | > Pass3 complete
> | > Starting pass4
> | > Pass4 complete
> | > Starting pass5
> | > Unlinked block found at block 37974707 (0x24372b3), left unchanged.
> | > ..snip about 30 total of these..
> | > Unlinked block found at block 96603710 (0x5c20e3e), left unchanged.
> | > Pass5 complete
> | > Writing changes to disk
> | > gfs2_fsck complete
> | > 
> | > When it was done I remounted the filesystem and tried to "rm -rf
> | /raid1/bad"
> | > which is a subdir in the root of this filesystem that contains the
> | zero-byte
> | > file that was the focal point of this grief to start with. 
> | > 
> | > Results:
> | > 
> | That looks like a bug in fsck at least as it should be dealing with
> | the
> | unlinked blocks that it finds, not ignoring them. Chances are that
> | the
> | block which is causing the issues belongs to one of the unlinked
> | blocks
> | (inodes I think it should say)
> | 
> | Steve.
> 
> Hi,
> 
> The "Unlinked block found...left unchanged." messages are harmless.
> This merely means that fsck.gfs2 found some blocks that were
> marked as "unlinked metadata" that should be automatically
> reassigned by gfs2's kernel code when needed.  At some point, we 
> made the decision not to fix the bitmaps for various reasons.  I don't
> remember the details, but I remember discussing it anyway.  Lately
> I've been thinking that we made the wrong decision and I should make
> fsck.gfs2 fix them rather than ignore them.
> 
We should not be ignoring them, certainly. Either they should be checked
just like "normal" inodes, or they should be removed by fsck. I don't
think it matters too much which we do, but I rather suspect that they
are not being checked like the other inodes are,

Steve.




From bergman at merctech.com  Mon Aug 10 16:16:13 2009
From: bergman at merctech.com (bergman at merctech.com)
Date: Mon, 10 Aug 2009 12:16:13 -0400
Subject: [Linux-cluster] cluster architecture/filesystem suggestions wanted
Message-ID: <13580.1249920973@mirchi>


Hello,

I've been testing RHCS (CentOS 5.2) cluster with GFS1 for a while, and
I'm about to transition the cluster to production, and I'd appreciate
a quick review of the architecture and filesystem choices. I've got some 
concerns about GFS (1 & 2) stability and performance vrs. ext3fs, but the 
increased flexibility in a clustered filesystem has a lot of advantages.

If there are fundamental stability advantages to a design that does not
cluster the filesystems (ie., that uses GFS in lock_nolock mode or ext3fs), 
that would override any performance consideration.

Assuming that stability is not an issue, my basic question in terms of choosing
an architecture is whether there is better performance through using GFS with
multiple cluster nodes (gaining some CPU and network load balancing at the cost
of the GFS locking performance penalty) accessing the same data, or whether
serving each volume from a single server via NFS (using RCHS solely for
fail-over) is more efficient? Obviously, I don't expect anyone to provide
definitive answers or data that's unique to our environment, but I'd highly
appreciate your view on the architecture choices.


Background:
    Our lab does basic science research on software to process medical
    images. There are about 40 lab members, with about 15~25 logged in
    at any given time. Most people will be logged into multiple servers
    at once, with their home directory and all data directories provided
    via NFS at this time.

    The workload is divided between a software development environment
    (compile/test cycles) and image processing. The software development
    process is interactive, and includes algorithm testing which requires
    reading/writing multi-MB files.  There's a reasonably high performance
    expectation for interactive work, less so for the testing phase.

    Many lab members also SAMBA mount filesystems from the servers
    to their desktop machines, for which there is a high performance
    expectation.

    The image processing is very strongly CPU-bound, but involves reading
    many image files in the 1 to 50MB range, and writing results files
    in the same range, along with smaller index and metadata files. The
    image processing is largely non-interactive, so the I/O performance
    is not critical.

The RHCS cluster will be used for infrastructure services (not as a
compute resource for image processing, not as login servers, not as
compilation servers). The primary services to be run on the clustered
machines are:

	network file sharing (NFS, Samba)
	SVN repository
	backup server (bacula, to fibre-attached tape drive)
	Wiki
	nagios

None of those services require a lot of CPU. The network file sharing
could benefit from load balancing, so that the NFS and SAMBA clients have
multiple network paths to the storage, but the NFS and SAMBA protocols
are not well suited for using RHCS as a load balancer, so this may not
be possible (using LVS or a front-end hardware load balancer is not an
option at this time...HA is much more important than load balancing).

The goals of using RHCS and clustering those functions are (in order of
importance):

	stability of applications
	high availability of applications
	performance
	expandability of filesystems (ie., expand volumes at the SAN, LUN,
		LVM, and filesystem layers)
	expandability of servers (add more servers to the cluster, with
		machines dedicated to functions, as a crude form of load
		balancing)
	
The computing environment consists of:
	2 RHCS servers
		fibre attached to storage and backup tape device

	~15TB EMC fibre-attached storage
	~14TB fibre and iSCSI attached storage in the near future

	4 compute servers
		currently accessing storage via NFS, could be
		fibre-attached and configured as cluster members

	35 compute servers
		NFS-only access to storage, possibly iSCSI in the
		future, no chance of fibre attachment


As I see it, there are 3 possible architecture choices:

	[1] infrastructure only-GFS+NFS
		the 2 cluster nodes share storage via GFS, and
		act as NFS servers to all compute servers

		+ load balancing of some services
		- complexity of GFS
		- performance of shared GFS storage

	[2] shared storage/NFS
		2 cluster nodes and 4 fibre-attached compute servers
		share storage via GFS (all machines are RHCS nodes, but
		the compute nodes do not provide infrastructure services,
		just use cluster membership for GFS file access)

		each GFS node is potentially an NFS server (via a VIP) to
		the 35 compute servers that are not on the fibre SAN

		+ potentially faster access to data for 4 fibre-attached
		  compute servers

		- potentially slow accesss to data for 4 fibre-attached
		  compute servers due to GFS locking

		+ increased stability over 2 node cluster
		- increased complexity
		
	[3] exclusive storage/NFS
		filesystems are formatted as ext3fs, exclusively mounted
		to one of the 2 infrastructure cluster nodes at a time,
		each filesystem mount also includes a child (dependent)
		function for the node to be an NFS server, all compute nodes
		access data via NFS

		+ reliability of filesystem
		+ performance of filesystem
		- potential for corruption in case of non-exclusive access
		- decreased flexibility due to exclusive use
		- no potential for load balancing across cluster nodes

I'm very interested in getting your opinion of the choices, and would like
to learn about other ideas that I may have overlooked.

Thanks,

Mark


----
Mark Bergman                              voice: 215-662-7310
mark.bergman at uphs.upenn.edu                 fax: 215-614-0266
System Administrator     Section of Biomedical Image Analysis
Department of Radiology            University of Pennsylvania
      PGP Key: https://www.rad.upenn.edu/sbia/bergman



From brem.belguebli at gmail.com  Mon Aug 10 17:11:28 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Mon, 10 Aug 2009 19:11:28 +0200
Subject: [Linux-cluster] rgmanager ignoring monitor action for a custom
	resource script
Message-ID: <29ae894c0908101011i33f6bb9bl9dd0427fcb877701@mail.gmail.com>

Hi,

I have written a custom resource script that starts,  stops and "meta-data"
's with the correct return  value (no error).

Everything seems to work fine except the fact that rgmanager never runs the
custom monitor action defined in my resource metadata.

I have lowered the interval value for monitor action to 10 (  are the
interval and timeout units in seconds by default ?) but nothing happens.

I have added a trace to the monitor case of my resource script that ocf_log
debug "monitor started at `date +%H:%M:%S`" but it never gets there:

----------------------------------------------8<--------------------------------------

case $1 in

start)                 ....................

                       ;;

stop)                ..........................

                      ;;

monitor)            ocf_log debug "at `/bin/date +%H:%M:%S`, launching
monitor"
                        check_conf
                        check_status
                        check_md_state
                        if [ $? -eq 33 ]; then
                                repair
                        fi
                        ;;

meta-data)        cat `echo $0 | sed 's/^\(.*\)\.sh$/\1.metadata/'`
                        exit $OCF_SUCCESS
                        ;;

*) echo "usage: $0 {start|stop|monitor|meta-data}"
                ;;
esac
---------------------------------------->8------------------------------------------------

Any idea ? Thanks

attached my meta-data
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090810/f4c6e440/attachment.htm>
-------------- next part --------------
<?xml version="1.0"?>
<resource-agent name="mdmonitor" version="rgmanager 2.0">
    <version>1.0</version>

    <longdesc lang="en">
        Resource for activating, deactivating and monitoring MD devices
    </longdesc>
    <shortdesc lang="en">
        Resource for activating, deactivating and monitoring MD devices
    </shortdesc>

    <parameters>

        <parameter name="raidconf" required="1">
            <longdesc lang="en">
                Name of the file containing the description of the array(s)
                Format is strict, must be "servicename"-raid.conf
                It will be  placed in /etc/cluster/"servicename"
            </longdesc>
            <shortdesc lang="en">
                raid.conf  filename
            </shortdesc>
            <content type="string"/>
        </parameter>

        <parameter name="policy" required="0">
           <longdesc lang="en">
                Policy to adopt at start, strict will prevent
                assembling any array partially thus causing the
                resource to fail.
                Quorum will allow partial start. If not defined,
                defaults to quorum
           </longdesc>
           <shortdesc lang="en">
                Assembly Policy
            </shortdesc>
            <content type="string"/>
        </parameter>

    </parameters>

    <actions>
        <action name="start" timeout="5"/>
        <action name="stop" timeout="5"/>

        <action name="monitor" interval="10" timeout="5" />

        <action name="meta-data" timeout="5"/>
    </actions>

    <special tag="rgmanager">
        <attributes maxinstances="1"/>
    </special>

</resource-agent>

From pradhanparas at gmail.com  Mon Aug 10 18:34:52 2009
From: pradhanparas at gmail.com (Paras pradhan)
Date: Mon, 10 Aug 2009 13:34:52 -0500
Subject: [Linux-cluster] do I have a fence DRAC device?
In-Reply-To: <3128ba140908100324l6cdb4c34ra5f5edb39c6903e9@mail.gmail.com>
References: <3128ba140908100324l6cdb4c34ra5f5edb39c6903e9@mail.gmail.com>
Message-ID: <8b711df40908101134t69b8e12cof6cc551809421e45@mail.gmail.com>

On Mon, Aug 10, 2009 at 5:24 AM, ESGLinux<esggrupos at gmail.com> wrote:
> Hi all,
> I was designing a 2 node cluster and I was going to use 2 servers DELL
> PowerEdge 1950. I was going to buy a DRAC card to use for fencing but
> running several commands in the servers I have noticed that when I run this
> command:
> #ipmitool lan print
> Set in Progress : Set Complete
> Auth Type Support : NONE MD2 MD5 PASSWORD
> Auth Type Enable : Callback : MD2 MD5
> ????????????????????????: User : MD2 MD5
> ????????????????????????: Operator : MD2 MD5
> ????????????????????????: Admin : MD2 MD5
> ????????????????????????: OEM : MD2 MD5
> IP Address Source : Static Address
> IP Address : 0.0.0.0
> Subnet Mask : 0.0.0.0
> MAC Address : 00:1e:c9:ae:6f:7e
> SNMP Community String : public
> IP Header : TTL=0x40 Flags=0x40 Precedence=0x00 TOS=0x10
> Default Gateway IP : 0.0.0.0
> Default Gateway MAC : 00:00:00:00:00:00
> Backup Gateway IP : 0.0.0.0
> Backup Gateway MAC : 00:00:00:00:00:00
> 802.1q VLAN ID : Disabled
> 802.1q VLAN Priority : 0
> RMCP+ Cipher Suites : 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14
> Cipher Suite Priv Max : aaaaaaaaaaaaaaa
> ????????????????????????: X=Cipher Suite Unused
> ????????????????????????: c=CALLBACK
> ????????????????????????: u=USER
> ????????????????????????: o=OPERATOR
> ????????????????????????: a=ADMIN
> ????????????????????????: O=OEM
> does this mean that I already have an ipmi card (not configured) that I can
> use for fencing? if the anwser is yes, where hell must I configure it? I
> don?t see wher can I do it.
> If I haven?t a fencing device which one do you recommed to use?
> Thanks in advance
> ESG
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>

Yes you have IPMI and if you are using 1950 Dell, DRAC should be there
too. You can see if you have DRAC or not when the server starts and
before the loading of the OS.

I have 1850s and I am using DRAC for fencing.


Paras.



From wendell at bisonline.com  Tue Aug 11 02:24:11 2009
From: wendell at bisonline.com (Wendell Dingus)
Date: Mon, 10 Aug 2009 22:24:11 -0400 (EDT)
Subject: [Linux-cluster] Re: Linux-cluster Digest, Vol 64, Issue 10
In-Reply-To: <1823022474.145311249957220903.JavaMail.root@mail.bisonline.com>
Message-ID: <625461635.145331249957451147.JavaMail.root@mail.bisonline.com>

> Hi, 
> 
> The "Unlinked block found...left unchanged." messages are harmless. 
> This merely means that fsck.gfs2 found some blocks that were 
> marked as "unlinked metadata" that should be automatically 
> reassigned by gfs2's kernel code when needed. At some point, we 
> made the decision not to fix the bitmaps for various reasons. I don't 
> remember the details, but I remember discussing it anyway. Lately 
> I've been thinking that we made the wrong decision and I should make 
> fsck.gfs2 fix them rather than ignore them. 
> 
> In theory, those blocks should not have caused the kernel withdraw 
> problem you saw. Do any of the blocks in the fsck output correspond 
> to the block complained about in the kernel withdraw output? That 
> might be an important clue what happened. 
> 
> Regards, 
> 
> Bob Peterson 
> Red Hat File Systems 

Well, here's the entire list of blocks it ignored and the entire message section.
Perhaps I'm just overlooking it but I'm not seeing anything in the messages
that appears to be a block number. Maybe 1633350398 but if so it is not a match.
Anyway, since you didn't specifically say a new/fixed version of fsck was 
imminent and that it would likely fix this we began plan B today. We plugged
in another drive, placed a GFS2 filesystem on it and am actively copying files
off to it now. Fingers crossed that nothing will hit a disk block that causes
this again but I could be so lucky probably...

Thanks.

Unlinked block found at block 37974707 (0x24372b3), left unchanged.
Unlinked block found at block 38041678 (0x244784e), left unchanged.
Unlinked block found at block 38178682 (0x2468f7a), left unchanged.
Unlinked block found at block 90090083 (0x55eaa63), left unchanged.
Unlinked block found at block 90090332 (0x55eab5c), left unchanged.
Unlinked block found at block 90090827 (0x55ead4b), left unchanged.
Unlinked block found at block 90092020 (0x55eb1f4), left unchanged.
Unlinked block found at block 90092617 (0x55eb449), left unchanged.
Unlinked block found at block 90093214 (0x55eb69e), left unchanged.
Unlinked block found at block 90096087 (0x55ec1d7), left unchanged.
Unlinked block found at block 90097524 (0x55ec774), left unchanged.
Unlinked block found at block 90099311 (0x55ece6f), left unchanged.
Unlinked block found at block 90161279 (0x55fc07f), left unchanged.
Unlinked block found at block 90201225 (0x5605c89), left unchanged.
Unlinked block found at block 90233675 (0x560db4b), left unchanged.
Unlinked block found at block 90241237 (0x560f8d5), left unchanged.
Unlinked block found at block 90243594 (0x561020a), left unchanged.
Unlinked block found at block 90247295 (0x561107f), left unchanged.
Unlinked block found at block 90251642 (0x561217a), left unchanged.
Unlinked block found at block 90253816 (0x56129f8), left unchanged.
Unlinked block found at block 90256999 (0x5613667), left unchanged.
Unlinked block found at block 90259800 (0x5614158), left unchanged.
Unlinked block found at block 96600646 (0x5c20246), left unchanged.
Unlinked block found at block 96603710 (0x5c20e3e), left unchanged.
Pass5 complete
Writing changes to disk
gfs2_fsck complete



Aug  8 19:09:43 serenity1 kernel: GFS2: fsid=Peace:raid1.0: fatal: invalid metad
ata block
Aug  8 19:09:43 serenity1 kernel: GFS2: fsid=Peace:raid1.0:   bh = 1633350398 (m
agic number)
Aug  8 19:09:43 serenity1 kernel: GFS2: fsid=Peace:raid1.0:   function = gfs2_me
ta_indirect_buffer, file = /builddir/build/BUILD/gfs2-kmod-1.92/_kmod_build_xen/
meta_io.c, line = 334
Aug  8 19:09:43 serenity1 kernel: GFS2: fsid=Peace:raid1.0: about to withdraw th
is file system
Aug  8 19:09:43 serenity1 kernel: GFS2: fsid=Peace:raid1.0: telling LM to withdr
aw
Aug  8 19:09:44 serenity1 kernel: GFS2: fsid=Peace:raid1.0: withdrawn
Aug  8 19:09:44 serenity1 kernel:
Aug  8 19:09:44 serenity1 kernel: Call Trace:
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff8854b91a>] :gfs2:gfs2_lm_withdraw+0xc1/0xd0
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff80262907>] __wait_on_bit+0x60/0x6e
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff80215780>] sync_buffer+0x0/0x3f
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff80262981>] out_of_line_wait_on_bit+0x6c/0x78
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff8029a016>] wake_bit_function+0x0/0x23
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff8021a7f6>] submit_bh+0x10a/0x111
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff8855d627>] :gfs2:gfs2_meta_check_ii
+0x2c/0x38
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff8854f168>] :gfs2:gfs2_meta_indirect_buffer+0x104/0x160
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff8853f786>] :gfs2:recursive_scan+0x96/0x175
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff8853f82c>] :gfs2:recursive_scan+0x13c/0x175
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff8854065a>] :gfs2:do_strip+0x0/0x358
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff88548d21>] :gfs2:glock_work_func+0x0/0xa8
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff8853f8fe>] :gfs2:trunc_dealloc+0x99/0xe7
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff8854065a>] :gfs2:do_strip+0x0/0x358
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff80286595>] deactivate_task+0x28/0x5f
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff8853fa99>] :gfs2:gfs2_truncatei_resume+0x10/0x1f
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff8854734d>] :gfs2:do_promote+0x9a/0x117
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff885485a1>] :gfs2:finish_xmote+0x28c/0x2b2
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff88548d3e>] :gfs2:glock_work_func+0x1d/0xa8
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff8024ee76>] run_workqueue+0x94/0xe4
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff8024b781>] worker_thread+0x0/0x122
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff80299dd0>] keventd_create_kthread+0x0/0xc4
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff8024b871>] worker_thread+0xf0/0x122
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff80286d8b>] default_wake_function+0x0/0xe
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff80299dd0>] keventd_create_kthread+0x0/0xc4
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff80299dd0>] keventd_create_kthread+0x0/0xc4
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff802334b4>] kthread+0xfe/0x132
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff8025fb2c>] child_rip+0xa/0x12
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff80299dd0>] keventd_create_kthread+0x0/0xc4
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff802333b6>] kthread+0x0/0x132
Aug  8 19:09:44 serenity1 kernel:  [<ffffffff8025fb22>] child_rip+0x0/0x12



From jason at monsterjam.org  Tue Aug 11 03:16:39 2009
From: jason at monsterjam.org (jason at monsterjam.org)
Date: Mon, 10 Aug 2009 23:16:39 -0400
Subject: [Linux-cluster] sanity check
Message-ID: <20090811031639.GA728@monsterjam.org>

can someone please look at the attached config and let me know if anything looks 
blatantly wrong.. This cluster config was working earlier for my httpd and postfix 
services , but now the virtual cluster ip address (.7) wont come up any more and the 
httpd and postfix will therefore not start. Im not sure why.

Jason

-------------- next part --------------
<?xml version="1.0"?>
<cluster config_version="25" name="mycluster">
	<fence_daemon clean_start="0" post_fail_delay="0" post_join_delay="3"/>
	<clusternodes>
		<clusternode name="server1" votes="1">
			<fence>
				<method name="1">
					<device name="apc_power_switch" option="off" port="1" switch="1"/>
					<device name="apc_power_switch" option="off" port="2" switch="1"/>
					<device name="apc_power_switch" option="on" port="1" switch="1"/>
					<device name="apc_power_switch" option="on" port="2" switch="1"/>
				</method>
			</fence>
		</clusternode>
		<clusternode name="server2" votes="1">
			<fence>
				<method name="1">
					<device name="apc_power_switch" option="off" port="3" switch="1"/>
					<device name="apc_power_switch" option="off" port="4" switch="1"/>
					<device name="apc_power_switch" option="on" port="3" switch="1"/>
					<device name="apc_power_switch" option="on" port="4" switch="1"/>
				</method>
			</fence>
		</clusternode>
	</clusternodes>
	<cman expected_votes="1" two_node="1"/>
	<fencedevices>
		<fencedevice agent="fence_apc" ipaddr="192.168.1.8" login="xxxxxxx" name="apc_power_switch" passwd="xxxxxxxxxxxxx"/>
	</fencedevices>
	<rm>
		<failoverdomains>
			<failoverdomain name="httpd" ordered="1" restricted="1">
				<failoverdomainnode name="server1" priority="1"/>
				<failoverdomainnode name="server2" priority="2"/>
			</failoverdomain>
		</failoverdomains>
		<resources>
			<script file="/etc/init.d/httpd" name="cluster_apache"/>
			<ip address="192.168.1.7" monitor_link="1"/>
			<script file="/etc/init.d/postfix" name="cluster_posstfix"/>
		</resources>
		<service autostart="1" domain="httpd" name="Apache Service">
			<ip ref="192.168.1.7"/>
			<script ref="cluster_apache"/>
		</service>
		<service autostart="1" domain="httpd" name="postfix service">
			<ip ref="192.168.1.7"/>
			<script ref="cluster_posstfix"/>
		</service>
	</rm>
</cluster>

From brem.belguebli at gmail.com  Tue Aug 11 10:23:20 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Tue, 11 Aug 2009 12:23:20 +0200
Subject: [Linux-cluster] Re: rgmanager ignoring monitor action for a custom
	resource script
In-Reply-To: <29ae894c0908101011i33f6bb9bl9dd0427fcb877701@mail.gmail.com>
References: <29ae894c0908101011i33f6bb9bl9dd0427fcb877701@mail.gmail.com>
Message-ID: <29ae894c0908110323v6ee8033fh5117559262a34890@mail.gmail.com>

Problem resolved,

rgmanager seems to rely on the 'status' action rather than on the 'monitor'
one.

Which doesn't seem to conform to the opencf API (
http://www.opencf.org/cgi-bin/viewcvs.cgi/*checkout*/specs/ra/resource-agent-api.txt?rev=1.10)
chapter 3.4.3 (is this doc outdated ?) that states that 'start' 'stop'
'monitor' and 'meta-data' are mandatory.


Regards



2009/8/10, brem belguebli <brem.belguebli at gmail.com>:
>
> Hi,
>
> I have written a custom resource script that starts,  stops and "meta-data"
> 's with the correct return  value (no error).
>
> Everything seems to work fine except the fact that rgmanager never runs the
> custom monitor action defined in my resource metadata.
>
> I have lowered the interval value for monitor action to 10 (  are the
> interval and timeout units in seconds by default ?) but nothing happens.
>
> I have added a trace to the monitor case of my resource script that ocf_log
> debug "monitor started at `date +%H:%M:%S`" but it never gets there:
>
>
> ----------------------------------------------8<--------------------------------------
>
> case $1 in
>
> start)                 ....................
>
>                        ;;
>
> stop)                ..........................
>
>                       ;;
>
> monitor)            ocf_log debug "at `/bin/date +%H:%M:%S`, launching
> monitor"
>                         check_conf
>                         check_status
>                         check_md_state
>                         if [ $? -eq 33 ]; then
>                                 repair
>                         fi
>                         ;;
>
> meta-data)        cat `echo $0 | sed 's/^\(.*\)\.sh$/\1.metadata/'`
>                         exit $OCF_SUCCESS
>                         ;;
>
> *) echo "usage: $0 {start|stop|monitor|meta-data}"
>                 ;;
> esac
>
> ---------------------------------------->8------------------------------------------------
>
> Any idea ? Thanks
>
> attached my meta-data
>
>
>
>
>
>
>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090811/5315cb3f/attachment.htm>

From bob.marcan at gmail.com  Tue Aug 11 11:10:30 2009
From: bob.marcan at gmail.com (Bob Marcan)
Date: Tue, 11 Aug 2009 13:10:30 +0200
Subject: [Linux-cluster] VirtualBox - fence agent
In-Reply-To: <20090811031639.GA728@monsterjam.org>
References: <20090811031639.GA728@monsterjam.org>
Message-ID: <20090811131030.0b13dac7@smicro.interstudio.homeunix.net>

Hi.
Does anybody write a fence agent for the VirtualBox?
Can we get this into the mainstream?

BR, Bob



From esggrupos at gmail.com  Tue Aug 11 12:03:18 2009
From: esggrupos at gmail.com (ESGLinux)
Date: Tue, 11 Aug 2009 14:03:18 +0200
Subject: [Linux-cluster] do I have a fence DRAC device?
In-Reply-To: <8b711df40908101134t69b8e12cof6cc551809421e45@mail.gmail.com>
References: <3128ba140908100324l6cdb4c34ra5f5edb39c6903e9@mail.gmail.com>
	<8b711df40908101134t69b8e12cof6cc551809421e45@mail.gmail.com>
Message-ID: <3128ba140908110503m27532c62rce2844564c35da06@mail.gmail.com>

Thanks
I?ll check it when I could reboot the server.

greetings,

ESG

2009/8/10 Paras pradhan <pradhanparas at gmail.com>

> On Mon, Aug 10, 2009 at 5:24 AM, ESGLinux<esggrupos at gmail.com> wrote:
> > Hi all,
> > I was designing a 2 node cluster and I was going to use 2 servers DELL
> > PowerEdge 1950. I was going to buy a DRAC card to use for fencing but
> > running several commands in the servers I have noticed that when I run
> this
> > command:
> > #ipmitool lan print
> > Set in Progress : Set Complete
> > Auth Type Support : NONE MD2 MD5 PASSWORD
> > Auth Type Enable : Callback : MD2 MD5
> >                         : User : MD2 MD5
> >                         : Operator : MD2 MD5
> >                         : Admin : MD2 MD5
> >                         : OEM : MD2 MD5
> > IP Address Source : Static Address
> > IP Address : 0.0.0.0
> > Subnet Mask : 0.0.0.0
> > MAC Address : 00:1e:c9:ae:6f:7e
> > SNMP Community String : public
> > IP Header : TTL=0x40 Flags=0x40 Precedence=0x00 TOS=0x10
> > Default Gateway IP : 0.0.0.0
> > Default Gateway MAC : 00:00:00:00:00:00
> > Backup Gateway IP : 0.0.0.0
> > Backup Gateway MAC : 00:00:00:00:00:00
> > 802.1q VLAN ID : Disabled
> > 802.1q VLAN Priority : 0
> > RMCP+ Cipher Suites : 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14
> > Cipher Suite Priv Max : aaaaaaaaaaaaaaa
> >                         : X=Cipher Suite Unused
> >                         : c=CALLBACK
> >                         : u=USER
> >                         : o=OPERATOR
> >                         : a=ADMIN
> >                         : O=OEM
> > does this mean that I already have an ipmi card (not configured) that I
> can
> > use for fencing? if the anwser is yes, where hell must I configure it? I
> > don?t see wher can I do it.
> > If I haven?t a fencing device which one do you recommed to use?
> > Thanks in advance
> > ESG
> >
> > --
> > Linux-cluster mailing list
> > Linux-cluster at redhat.com
> > https://www.redhat.com/mailman/listinfo/linux-cluster
> >
>
> Yes you have IPMI and if you are using 1950 Dell, DRAC should be there
> too. You can see if you have DRAC or not when the server starts and
> before the loading of the OS.
>
> I have 1850s and I am using DRAC for fencing.
>
>
> Paras.
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090811/8631f08c/attachment.htm>

From robejrm at gmail.com  Tue Aug 11 12:14:03 2009
From: robejrm at gmail.com (Juan Ramon Martin Blanco)
Date: Tue, 11 Aug 2009 14:14:03 +0200
Subject: [Linux-cluster] do I have a fence DRAC device?
In-Reply-To: <3128ba140908110503m27532c62rce2844564c35da06@mail.gmail.com>
References: <3128ba140908100324l6cdb4c34ra5f5edb39c6903e9@mail.gmail.com>
	<8b711df40908101134t69b8e12cof6cc551809421e45@mail.gmail.com>
	<3128ba140908110503m27532c62rce2844564c35da06@mail.gmail.com>
Message-ID: <8a5668960908110514u4703a951qa49938375d0a7e52@mail.gmail.com>

On Tue, Aug 11, 2009 at 2:03 PM, ESGLinux <esggrupos at gmail.com> wrote:

> Thanks
> I?ll check it when I could reboot the server.
>
> greetings,
>
You have a BMC ipmi in the first network interface, it can be configured at
boot time (I don't  remember if inside the BIOS or pressing cntrl+something
during boot)

Greetings,
Juanra

>
> ESG
>
> 2009/8/10 Paras pradhan <pradhanparas at gmail.com>
>
> On Mon, Aug 10, 2009 at 5:24 AM, ESGLinux<esggrupos at gmail.com> wrote:
>> > Hi all,
>> > I was designing a 2 node cluster and I was going to use 2 servers DELL
>> > PowerEdge 1950. I was going to buy a DRAC card to use for fencing but
>> > running several commands in the servers I have noticed that when I run
>> this
>> > command:
>> > #ipmitool lan print
>> > Set in Progress : Set Complete
>> > Auth Type Support : NONE MD2 MD5 PASSWORD
>> > Auth Type Enable : Callback : MD2 MD5
>> >                         : User : MD2 MD5
>> >                         : Operator : MD2 MD5
>> >                         : Admin : MD2 MD5
>> >                         : OEM : MD2 MD5
>> > IP Address Source : Static Address
>> > IP Address : 0.0.0.0
>> > Subnet Mask : 0.0.0.0
>> > MAC Address : 00:1e:c9:ae:6f:7e
>> > SNMP Community String : public
>> > IP Header : TTL=0x40 Flags=0x40 Precedence=0x00 TOS=0x10
>> > Default Gateway IP : 0.0.0.0
>> > Default Gateway MAC : 00:00:00:00:00:00
>> > Backup Gateway IP : 0.0.0.0
>> > Backup Gateway MAC : 00:00:00:00:00:00
>> > 802.1q VLAN ID : Disabled
>> > 802.1q VLAN Priority : 0
>> > RMCP+ Cipher Suites : 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14
>> > Cipher Suite Priv Max : aaaaaaaaaaaaaaa
>> >                         : X=Cipher Suite Unused
>> >                         : c=CALLBACK
>> >                         : u=USER
>> >                         : o=OPERATOR
>> >                         : a=ADMIN
>> >                         : O=OEM
>> > does this mean that I already have an ipmi card (not configured) that I
>> can
>> > use for fencing? if the anwser is yes, where hell must I configure it? I
>> > don?t see wher can I do it.
>> > If I haven?t a fencing device which one do you recommed to use?
>> > Thanks in advance
>> > ESG
>> >
>> > --
>> > Linux-cluster mailing list
>> > Linux-cluster at redhat.com
>> > https://www.redhat.com/mailman/listinfo/linux-cluster
>> >
>>
>> Yes you have IPMI and if you are using 1950 Dell, DRAC should be there
>> too. You can see if you have DRAC or not when the server starts and
>> before the loading of the OS.
>>
>> I have 1850s and I am using DRAC for fencing.
>>
>>
>> Paras.
>>
>> --
>> Linux-cluster mailing list
>> Linux-cluster at redhat.com
>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090811/748afaa4/attachment.htm>

From robert at exa-omicron.nl  Tue Aug 11 12:59:13 2009
From: robert at exa-omicron.nl (Robert Verspuy)
Date: Tue, 11 Aug 2009 14:59:13 +0200
Subject: [Linux-cluster] Succesfull installation on centos 5.3 with live kvm
	migration
Message-ID: <4A816B21.2040907@exa-omicron.nl>

Getting cluster software including kvm virtual machines with live 
migration working,
can be a very difficult task, with many obstacles.

But I would like to mention to the mailing list, that I just booked some 
succes.
And because nobody is around to tell the wonderfull news,
I would like to share my hapiness here ;)

My setup:
2 NAS servers and 1 Supermicro bladeserver with 5 blades.

The NAS servers are running Openfiler 2.3
both NAS servers have:
1 Transcend IDE 4Gbyte flashcard (on the ide port on the mainboard).
3 x Transcend 4Gbyte usb sticks
8 SATA disks.

The IDE flashcard is setup in a raid-1 mirroring (md0) with one USB 
stick providing the root FS voor openfiler
The other 2 USB sticks have 5 partitions: 4 x 500 MB and 1 x 2GB.
those are mirrored with raid-1 together. (md5 until md8 are the 500mb 
partions, and md9 is the 2Gb partition).
Then the 8 harddisks are also tied together per 2 as mirroring raid1 
(md1 until md4).

Then I used DRBD (8.2.7) to mirror the 4 raid-1's of the disks (md1 
until md4) and the 2GB mirror (md9)
over the network to the other NAS server. (drbd1, drbd2, drbd3 and drbd4)
The 500mb raid-1's are used to store metadata of the 4 disk raid-1's.
The 2gb drbd (drbd0) has internal metadata.

The 2gb drbd (drbd0) is mounted as ext3 on one only server and is used 
to store all kinds
of openfiler information that is needed on both nas servers,
like the openfiler config (mostly), dhcp leases database, openldap database.
And heartbeat makes sure that one NAS server is running all the 
software, and with any problems,
it can switch over very easily.

The drbd1 til 4 are setup as a LVM PV, and bound together in one big VG.
 From that VG, I created a 5 x 5GB LV to be used as root device for 
blade1 til blade5
These LV's are stripped accross 2 PV for speed (altough that's still my 
only bottleneck at the moment, but more later about this...).
These LV's are setup as iSCSI

I also created one big LV of around 600GB, which can be mounted through NFS.

Then a few more LV's are created (around 10GB, also iscsi) for every VM 
I want.
For every iSCSI LV I create a separate target.

The Supermicro blades can boot from an iscsi device.
The exact scsi device is given through a DHCP option.
I only setup a initiatior name in the iscsi bios of the blade.

On the blade LV's I installed CentOS 5.3 (latest updates).
But with a few modifications.

I changes a few things in the initrd, to bound eth0 to br0 during the 
linux boot,
and before linux is taking over the iscsi from the bios, because
when you have a linux root through iscsi, and try to attach eth0 to br0,
you loose networkconnectivity for a moment, and could crash the linux,
because everything it uses, comes from the network (iscsi root).
I also added a little script to the initrd to call iscsiadm with a fixed 
iscsi
target, because unfortunately iscsiadm can't read the iscsi settings 
from dhcp
or the supermicro firmware.

When the blades are booted, they all join one redhat cluster with 3 
nodes to be quorum.
Because I have 5 blades, two can fail before everything stops working.

Then I compiled the following software my own, because the ones in the 
centos repo,
and the testing repo didn't function correctly:
libvirt 0.7.0 (./configure --prefix=/usr)
kvm-88 (./configure --prefix=/usr --disable-xen)

The /usr/share/cluster/vm.sh from the default centos repo is still based 
on xen.
I downloaded the latest from 
https://bugzilla.redhat.com/show_bug.cgi?id=412911
but it appears that that one is not working correctly either.
I made some changes myself.

And now it's working all together very nicely

I just ran a VM on blade1, and while this VM was running bonnie++ on a 
NFS mount to the NAS server,
I live-migrated it about 10 times to blade2 and back.

During this bonnie++ run and live migrations, I pinged the device.
And where the normal ping times are around 20-35 ms (I pinged through a 
VPN line from my home to the data center).
I only saw one or 2 pings just around the end of the live migration that 
were around 40-60ms.
but no drops, and no errors in bonnie++.

I will write some more information about the complete setup, and post it 
somewhere on my blog or someting,
But I just wanted to let everybody know, that it can be done ;)

If you have any questions, let me know.

The only 'problem' I still have is the speed to and from the disks.
When I update any settings on the bladeserver. I always do this on blade1.
Then shut it down, On the NAS server I copy the content of the iscsi LV 
to an image file on the ext3 LV.
Then I can power up blade1, wait until it reenters the cluster,
and then one by one shut down the next blade, On the NAS copy the image 
from the ext3 LV to the blade LV.
And start the blade again.

I use the drbd1 til drbd4 as 4 PV's for a VG.
The speed (hdparm -t on the NAS) of all PV's are around 75 MB/sec 
(except for one which is 45MB/sec)

The blade LV (/dev/vg0/blade1 for example) is striped over 2 PV's.
The Speed (hdparm -t) of /dev/vg0/blade1 is 122MB/sec.

The ext3 LV (/dev/vg0/data0) is striped over 4 PV's.
The Speed (hdparm -t) of /dev/vg0/data0 is 227 MB/sec.

But when copying from the blade LV to the ext3 LV:
dd if=/dev/vg0/blade1 of=/mnt/vg0/data0/vm/image/blade_v2.7.img
it takes about 70 seconds, which is about 75MB/sec.

but when copying back:
dd if=/mnt/vg0/data0/vm/image/blade_v2.7.img of=/dev/vg0/blade1
It takes about 390 seconds, which is about 13MB/sec

I think it has something to do with the striped over 4 PV's of the LVM.
So I will try to create a new ext3 LV stiped accross 2 PV's and see if 
this is faster.

Robert Verspuy

-- 
*Exa-Omicron*
Patroonsweg 10
3892 DB Zeewolde
Tel.: 088-OMICRON (66 427 66)
http://www.exa-omicron.nl



From akapp at fnds3000.com  Tue Aug 11 12:58:38 2009
From: akapp at fnds3000.com (akapp)
Date: Tue, 11 Aug 2009 14:58:38 +0200
Subject: [Linux-cluster] Centos 5.3 X64 a& luci
Message-ID: <34CB6BB9-6856-4E52-B946-5465F90B081C@fnds3000.com>

Good day

I have a Sun x4100 server running Centos 5.3 X64 - patch to latest and  
greatest.

When trying to start luci, it simply fails, no error in /var/log  and  
nothing in /var/lib/luci/log

I have re-installed luci and ricci a couple of times now. Cleaned out / 
var/lib/luci & /rici between installations.

I have even tried the complete yum grouremove "Clustering" "Cluster  
Storage" and re-installed the complete package again.

Used ricci/luci combination with great success in 5.2, but both  
servers giving the same problem.

Any pointers will be appreciated.


Here is screen snipped of problem:



Installed: luci.x86_64 0:0.12.1-7.3.el5.centos.1
Complete!
[root at clu1 luci]# luci_admin init
Initializing the luci server


Creating the 'admin' user

Enter password:
Confirm password:

Please wait...
The admin password has been successfully set.
Generating SSL certificates...
The luci server has been successfully initialized


You must restart the luci server for changes to take effect.

Run "service luci restart" to do so

[root at clu1 luci]# service luci restart
Shutting down luci:                                        [  OK  ]
Starting luci: Generating https SSL certificates...  done
                                                            [FAILED]
[root at clu1 luci]#






Tks
Andre



From rpeterso at redhat.com  Tue Aug 11 13:24:14 2009
From: rpeterso at redhat.com (Bob Peterson)
Date: Tue, 11 Aug 2009 09:24:14 -0400 (EDT)
Subject: [Linux-cluster] Re: Linux-cluster Digest, Vol 64, Issue 10
In-Reply-To: <625461635.145331249957451147.JavaMail.root@mail.bisonline.com>
Message-ID: <1891903567.443711249997054068.JavaMail.root@zmail06.collab.prod.int.phx2.redhat.com>

----- "Wendell Dingus" <wendell at bisonline.com> wrote:
| Well, here's the entire list of blocks it ignored and the entire
| message section.
| Perhaps I'm just overlooking it but I'm not seeing anything in the
| messages
| that appears to be a block number. Maybe 1633350398 but if so it is
| not a match.

Your assumption is correct.  The block number was 1633350398, which
is labeled "bh = " for some reason.

| Anyway, since you didn't specifically say a new/fixed version of fsck
| was 
| imminent and that it would likely fix this we began plan B today. We

Yesterday I pushed a newer gfs_fsck and fsck.gfs2 to their appropriate
git source repositories.  So you can build that version from source if
you need it right away.  But it sounds like it wouldn't have helped
your problem anyway.  What would really be nice is if there is a way
to recreate the problem in our lab.  In theory, this error could be
caused by a hardware problem too.

| plugged
| in another drive, placed a GFS2 filesystem on it and am actively
| copying files
| off to it now. Fingers crossed that nothing will hit a disk block that
| causes
| this again but I could be so lucky probably...

It's hard to say whether you'll hit it again.

Regards,

Bob Peterson
Red Hat File Systems



From rhurst at bidmc.harvard.edu  Tue Aug 11 14:55:48 2009
From: rhurst at bidmc.harvard.edu (Robert Hurst)
Date: Tue, 11 Aug 2009 10:55:48 -0400
Subject: [Linux-cluster] RHEL 4.7 fenced fails -- stuck join state: S-2,2,1
Message-ID: <1250002549.2782.36.camel@WSBID06223.bidmc.harvard.edu>

Simple 4-node cluster, 2-nodes have a GFS shared home directory mounted
for over a month.  Today, I wanted to mount /home on a 3rd node, so:

# service fenced start                [failed]

Weird.  Checking /var/log/messages show:

Aug 11 10:19:06 cerberus kernel: Lock_Harness 2.6.9-80.9.el4_7.10 (built
Jan 22 2009 18:39:16) installed
Aug 11 10:19:06 cerberus kernel: GFS 2.6.9-80.9.el4_7.10 (built Jan 22
2009 18:39:32) installed
Aug 11 10:19:06 cerberus kernel: GFS: Trying to join cluster "lock_dlm",
"ccc_cluster47:home"
Aug 11 10:19:06 cerberus kernel: Lock_DLM (built Jan 22 2009 18:39:18)
installed
Aug 11 10:19:06 cerberus kernel: lock_dlm: fence domain not found; check
fenced
Aug 11 10:19:06 cerberus kernel: GFS: can't mount proto = lock_dlm,
table = ccc_cluster47:home, hostdata = 

# cman_tool services
Service          Name                              GID LID State
Code
Fence Domain:    "default"                           0   2 join
S-2,2,1
[]

So, a fenced process is now hung:

root     28302  0.0  0.0  3668  192 ?        Ss   10:19   0:00 fenced -t
120 -w

Q: Any idea how to "recover" from this state, without rebooting?

The other two servers are unaffected by this (thankfully) and show
normal operations:

$ cman_tool services

Service          Name                              GID LID State
Code
Fence Domain:    "default"                           2   2 run       -
[1 12]

DLM Lock Space:  "home"                              5   5 run       -
[1 12]

GFS Mount Group: "home"                              6   6 run       -
[1 12]

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090811/7b46b120/attachment.htm>

From bergman at merctech.com  Tue Aug 11 15:39:39 2009
From: bergman at merctech.com (bergman at merctech.com)
Date: Tue, 11 Aug 2009 11:39:39 -0400
Subject: [Linux-cluster] do I have a fence DRAC device?
In-Reply-To: Your message of "Tue, 11 Aug 2009 14:14:03 +0200."
	<8a5668960908110514u4703a951qa49938375d0a7e52@mail.gmail.com>
References: <8a5668960908110514u4703a951qa49938375d0a7e52@mail.gmail.com>
	<3128ba140908100324l6cdb4c34ra5f5edb39c6903e9@mail.gmail.com>
	<8b711df40908101134t69b8e12cof6cc551809421e45@mail.gmail.com>
	<3128ba140908110503m27532c62rce2844564c35da06@mail.gmail.com>
Message-ID: <27649.1250005179@mirchi>



In the message dated: Tue, 11 Aug 2009 14:14:03 +0200,
The pithy ruminations from Juan Ramon Martin Blanco on 
<Re: [Linux-cluster] do I have a fence DRAC device?> were:
=> --===============1917368601==
=> Content-Type: multipart/alternative; boundary=0016364c7c07663f600470dca3b8
=> 
=> --0016364c7c07663f600470dca3b8
=> Content-Type: text/plain; charset=ISO-8859-1
=> Content-Transfer-Encoding: quoted-printable
=> 
=> On Tue, Aug 11, 2009 at 2:03 PM, ESGLinux <esggrupos at gmail.com> wrote:
=> 
=> > Thanks
=> > I=B4ll check it when I could reboot the server.
=> >
=> > greetings,
=> >
=> You have a BMC ipmi in the first network interface, it can be configured at
=> boot time (I don't  remember if inside the BIOS or pressing cntrl+something
=> during boot)
=> 

Based on my notes, here's how I configured the DRAC interface on a Dell 1950
for use as a fence device:

	Configuring the card from Linux depending on the installation of Dell's
	OMSA package. Once that's installed, use the following 
commands:

		racadm config -g cfgSerial -o cfgSerialTelnetEnable 1
		racadm config -g cfgLanNetworking -o cfgDNSRacName HOSTNAME_FOR_INTERFACE
		racadm config -g cfgDNSDomainName DOMAINNAME_FOR_INTERFACE
		racadm config -g cfgUserAdmin -o cfgUserAdminPassword -i 2 PASSWORD
		racadm config -g cfgNicEnable 1
		racadm config -g cfgNicIpAddress WWW.XXX.YYY.ZZZ
		racadm config -g cfgNicNetmask WWW.XXX.YYY.ZZZ
		racadm config -g cfgNicGateway WWW.XXX.YYY.ZZZ
		racadm config -g cfgNicUseDhcp 0


	I also save a backup of the configuration with:

		racadm getconfig -f ~/drac_config


Hope this helps,

Mark

----
Mark Bergman                              voice: 215-662-7310
mark.bergman at uphs.upenn.edu                 fax: 215-614-0266
System Administrator     Section of Biomedical Image Analysis
Department of Radiology            University of Pennsylvania
      PGP Key: https://www.rad.upenn.edu/sbia/bergman 


=> Greetings,
=> Juanra
=> 
=> >
=> > ESG
=> >
=> > 2009/8/10 Paras pradhan <pradhanparas at gmail.com>
=> >
=> > On Mon, Aug 10, 2009 at 5:24 AM, ESGLinux<esggrupos at gmail.com> wrote:
=> >> > Hi all,
=> >> > I was designing a 2 node cluster and I was going to use 2 servers DELL
=> >> > PowerEdge 1950. I was going to buy a DRAC card to use for fencing but
=> >> > running several commands in the servers I have noticed that when I run
=> >> this
=> >> > command:
=> >> > #ipmitool lan print
=> >> > Set in Progress : Set Complete
=> >> > Auth Type Support : NONE MD2 MD5 PASSWORD
=> >> > Auth Type Enable : Callback : MD2 MD5
=> >> >                         : User : MD2 MD5
=> >> >                         : Operator : MD2 MD5
=> >> >                         : Admin : MD2 MD5
=> >> >                         : OEM : MD2 MD5
=> >> > IP Address Source : Static Address
=> >> > IP Address : 0.0.0.0
=> >> > Subnet Mask : 0.0.0.0
=> >> > MAC Address : 00:1e:c9:ae:6f:7e
=> >> > SNMP Community String : public
=> >> > IP Header : TTL=0x40 Flags=0x40 Precedence=0x00 TOS=0x10
=> >> > Default Gateway IP : 0.0.0.0
=> >> > Default Gateway MAC : 00:00:00:00:00:00
=> >> > Backup Gateway IP : 0.0.0.0
=> >> > Backup Gateway MAC : 00:00:00:00:00:00
=> >> > 802.1q VLAN ID : Disabled
=> >> > 802.1q VLAN Priority : 0
=> >> > RMCP+ Cipher Suites : 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14
=> >> > Cipher Suite Priv Max : aaaaaaaaaaaaaaa
=> >> >                         : X=Cipher Suite Unused
=> >> >                         : c=CALLBACK
=> >> >                         : u=USER
=> >> >                         : o=OPERATOR
=> >> >                         : a=ADMIN
=> >> >                         : O=OEM
=> >> > does this mean that I already have an ipmi card (not configured) that 
=> I
=> >> can
=> >> > use for fencing? if the anwser is yes, where hell must I configure it?
=>  I
=> >> > don=B4t see wher can I do it.
=> >> > If I haven=B4t a fencing device which one do you recommed to use?
=> >> > Thanks in advance
=> >> > ESG
=> >> >
=> >> > --
=> >> > Linux-cluster mailing list
=> >> > Linux-cluster at redhat.com
=> >> > https://www.redhat.com/mailman/listinfo/linux-cluster
=> >> >
=> >>
=> >> Yes you have IPMI and if you are using 1950 Dell, DRAC should be there
=> >> too. You can see if you have DRAC or not when the server starts and
=> >> before the loading of the OS.
=> >>
=> >> I have 1850s and I am using DRAC for fencing.
=> >>
=> >>
=> >> Paras.
=> >>
=> >> --
=> >> Linux-cluster mailing list
=> >> Linux-cluster at redhat.com
=> >> https://www.redhat.com/mailman/listinfo/linux-cluster
=> >>
=> >
=> >





From pradhanparas at gmail.com  Tue Aug 11 19:30:17 2009
From: pradhanparas at gmail.com (Paras pradhan)
Date: Tue, 11 Aug 2009 14:30:17 -0500
Subject: [Linux-cluster] quorum handling
Message-ID: <8b711df40908111230q45d1faafp839238d25d6c4e26@mail.gmail.com>

I have a 3 nodes xen cluser under centos5.3 that host linux virtual
machines. Node1 and Node2 got virtual machines but not on Node3. Node
3 is basically used when Node1 or Node2 goes down. This is working OK.
Now when I turn off both  Node1 and Node2. Then quorum is dissolved
and my cluser fails. I do not see any virtual machines in Node3 if the
other 2 nodes die.

So, how to handle this situation? I need node 3 to be alive a working
to host virtual machines when Node 1 and Node 2 is shut down.


Thanks !
Paras.



From ironludo at free.fr  Wed Aug 12 07:08:20 2009
From: ironludo at free.fr (LEROUX Ludovic)
Date: Wed, 12 Aug 2009 09:08:20 +0200
Subject: [Linux-cluster] fence device
Message-ID: <EE5A9AA37E4A40D5BD7B22FED6A61FE9@siim94.local>

i have a linux redhat cluster with two redhat 5.3 nodes.
I have a shared iscsi disk with gfs.
And i created an oracle service which mounts the gfs file system and start an oracle instance in fail over mode.
I have two questions:
1?) I don't have a fence device. Is it possible to create a cluster without a fence device? 
2?) is it possible to create a high available cluster with two or more nodes which works together (for example http server).
Thanks.
Ludo
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090812/108f18ac/attachment.htm>

From wen.gang.wang at oracle.com  Wed Aug 12 08:19:47 2009
From: wen.gang.wang at oracle.com (Wengang Wang)
Date: Wed, 12 Aug 2009 16:19:47 +0800
Subject: [Linux-cluster] where to get libtool 2.2.7
In-Reply-To: <1249902500.20848.11.camel@marc>
References: <4A7FA8B7.5080708@oracle.com> <1249896553.19917.2.camel@marc>	
	<4A7FEC6F.2050405@oracle.com> <4A7FEE8D.8070400@oracle.com>
	<1249902500.20848.11.camel@marc>
Message-ID: <4A827B23.7050107@oracle.com>

Hi Marc,

I am using the git master, while running ./autogen.sh I got

autoreconf: Entering directory `.'
autoreconf: configure.ac: not using Gettext
autoreconf: running: aclocal -I m4
configure.ac:7: error: Libtool version 2.2.7 or higher is required
/usr/share/aclocal/libtool.m4:46: LT_PREREQ is expanded from...
configure.ac:7: the top level
autom4te: /usr/bin/m4 failed with exit status: 63
aclocal: autom4te failed with exit status: 63
autoreconf: aclocal failed with exit status: 63

it requires Libtool versin 2.2.7.
I checked http://ftp.gnu.org/gnu/libtool/, the highest version is 
2.2.6a. so could you point me where to get the 2.2.7 libtool?

regards,
wengang.



From veliogluh at itu.edu.tr  Wed Aug 12 09:08:57 2009
From: veliogluh at itu.edu.tr (Hakan VELIOGLU)
Date: Wed, 12 Aug 2009 12:08:57 +0300
Subject: [Linux-cluster] Is there any backup heartbeat channel
In-Reply-To: <4A827B23.7050107@oracle.com>
References: <4A7FA8B7.5080708@oracle.com> <1249896553.19917.2.camel@marc>
	<4A7FEC6F.2050405@oracle.com> <4A7FEE8D.8070400@oracle.com>
	<1249902500.20848.11.camel@marc> <4A827B23.7050107@oracle.com>
Message-ID: <20090812120857.14055bi0l6l01okp@webmail.itu.edu.tr>

Hi list,

I am trying a two node cluster with RH 5.3 on Sun X4150 hardware. I  
use a private network for heartbeat channel and fence_ipmilan for  
fencing the two node. When I unplug the heartbeat channel cable, both  
cluster nodes shut themselves down with fence_ipmilan.

I think the single point of failure for this design is private  
heartbeat channel. I made a search on the web and there seems no  
backup heartbeat channel (like LVS) on cluster suite.

How can I overcome this single point of failure problem?

One solution is I think, bonding heartbeat interfaces, however I use  
all available interfaces and there is no extra interface for private  
heartbeat channel. For this reason this solution doesn't suit me.

Anybody has an idea to overcome heartbeat channel failure problem.


Hakan Velio?lu




From robejrm at gmail.com  Wed Aug 12 09:21:00 2009
From: robejrm at gmail.com (Juan Ramon Martin Blanco)
Date: Wed, 12 Aug 2009 11:21:00 +0200
Subject: [Linux-cluster] Is there any backup heartbeat channel
In-Reply-To: <20090812120857.14055bi0l6l01okp@webmail.itu.edu.tr>
References: <4A7FA8B7.5080708@oracle.com> <1249896553.19917.2.camel@marc>
	<4A7FEC6F.2050405@oracle.com> <4A7FEE8D.8070400@oracle.com>
	<1249902500.20848.11.camel@marc> <4A827B23.7050107@oracle.com>
	<20090812120857.14055bi0l6l01okp@webmail.itu.edu.tr>
Message-ID: <8a5668960908120221n7810e30dhd1b771a7fc4ad84b@mail.gmail.com>

2009/8/12 Hakan VELIOGLU <veliogluh at itu.edu.tr>

> Hi list,
>
> I am trying a two node cluster with RH 5.3 on Sun X4150 hardware. I use a
> private network for heartbeat channel and fence_ipmilan for fencing the two
> node. When I unplug the heartbeat channel cable, both cluster nodes shut
> themselves down with fence_ipmilan.
>
> I think the single point of failure for this design is private heartbeat
> channel. I made a search on the web and there seems no backup heartbeat
> channel (like LVS) on cluster suite.
>
> How can I overcome this single point of failure problem?
>
> One solution is I think, bonding heartbeat interfaces, however I use all
> available interfaces and there is no extra interface for private heartbeat
> channel. For this reason this solution doesn't suit me.
>
Why do you need a private heartbeat channel? Configure two network
interfaces in a bonding one, connect the two of them to two different
switches (with a redundant interconnection), make sure switches let
multicast traffic through. Configure bonding interface IP to fit your needs.

Greetings,
Juanra

>
> Anybody has an idea to overcome heartbeat channel failure problem.
>
>
> Hakan Velio?lu
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090812/a3fb7638/attachment.htm>

From veliogluh at itu.edu.tr  Wed Aug 12 10:01:22 2009
From: veliogluh at itu.edu.tr (Hakan VELIOGLU)
Date: Wed, 12 Aug 2009 13:01:22 +0300
Subject: [Linux-cluster] Is there any backup heartbeat channel
In-Reply-To: <8a5668960908120221n7810e30dhd1b771a7fc4ad84b@mail.gmail.com>
References: <4A7FA8B7.5080708@oracle.com> <1249896553.19917.2.camel@marc>
	<4A7FEC6F.2050405@oracle.com> <4A7FEE8D.8070400@oracle.com>
	<1249902500.20848.11.camel@marc> <4A827B23.7050107@oracle.com>
	<20090812120857.14055bi0l6l01okp@webmail.itu.edu.tr>
	<8a5668960908120221n7810e30dhd1b771a7fc4ad84b@mail.gmail.com>
Message-ID: <20090812130122.802620ghy522dkhe@webmail.itu.edu.tr>


----- robejrm at gmail.com den ileti ---------
   Tarih: Wed, 12 Aug 2009 11:21:00 +0200
  Kimden:  Juan Ramon Martin Blanco <robejrm at gmail.com>
Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
    Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
    Kime: linux clustering <linux-cluster at redhat.com>


> 2009/8/12 Hakan VELIOGLU <veliogluh at itu.edu.tr>
>
>> Hi list,
>>
>> I am trying a two node cluster with RH 5.3 on Sun X4150 hardware. I use a
>> private network for heartbeat channel and fence_ipmilan for fencing the two
>> node. When I unplug the heartbeat channel cable, both cluster nodes shut
>> themselves down with fence_ipmilan.
>>
>> I think the single point of failure for this design is private heartbeat
>> channel. I made a search on the web and there seems no backup heartbeat
>> channel (like LVS) on cluster suite.
>>
>> How can I overcome this single point of failure problem?
>>
>> One solution is I think, bonding heartbeat interfaces, however I use all
>> available interfaces and there is no extra interface for private heartbeat
>> channel. For this reason this solution doesn't suit me.
>>
> Why do you need a private heartbeat channel? Configure two network
> interfaces in a bonding one, connect the two of them to two different
> switches (with a redundant interconnection), make sure switches let
> multicast traffic through. Configure bonding interface IP to fit your needs.

This is another good solution which already mentioned in the list. I  
use this cluster design for two xen hosts. We have VLAN networking and  
I got two interfaces which are trunk. Then I made bridge vlan  
interfaces on this bond trunk network for virtual machines. So there  
left two interfaces. One of them is for management network which is in  
a DMZ zone. And another for private heartbeat. I can use management  
for heartbeat but what changes? When I unplug the management network  
cable this problem will occur again. I cant bond the private and  
management interfaces because of limitations on network hardware?

On the other hand this solution is good for other cluster desings but  
not this I think. I am really confused why there is no backup  
heartbeat channel in cluster suite?

>
> Greetings,
> Juanra
>
>>
>> Anybody has an idea to overcome heartbeat channel failure problem.
>>
>>
>> Hakan Velio?lu
>>
>>
>> --
>> Linux-cluster mailing list
>> Linux-cluster at redhat.com
>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>
>


----- robejrm at gmail.com den iletiyi bitir -----





From kbphillips80 at gmail.com  Wed Aug 12 11:53:08 2009
From: kbphillips80 at gmail.com (Kaerka Phillips)
Date: Wed, 12 Aug 2009 07:53:08 -0400
Subject: [Linux-cluster] do I have a fence DRAC device?
In-Reply-To: <27649.1250005179@mirchi>
References: <3128ba140908100324l6cdb4c34ra5f5edb39c6903e9@mail.gmail.com>
	<8b711df40908101134t69b8e12cof6cc551809421e45@mail.gmail.com>
	<3128ba140908110503m27532c62rce2844564c35da06@mail.gmail.com>
	<8a5668960908110514u4703a951qa49938375d0a7e52@mail.gmail.com>
	<27649.1250005179@mirchi>
Message-ID: <e558921b0908120453x302228bck3b4aa66b797820a5@mail.gmail.com>

If you have a drac, it should either be visible with the OpenManage
interface, commands, or as a 10/100 port on the far left of the back of the
server; and also should give you a setup screen during bootup.  If you don't
have one though, you can still use ipmi fencing to the on-board BMC/ESM
card.  We've found the only real drawback to this is that sometimes the ipmi
fencing may have difficulty communicating to a locked up host or a host
which has kernel panic'd.
The drac is better though, as it allows for ssh and full power-off/power-on
functionality.  Also, if your drac is a drac5 and has firmware greater than
1.20 on it (such as 1.33), you'll need to use the fence_drac5 module which
is included but not fully documented.

On Tue, Aug 11, 2009 at 11:39 AM, <bergman at merctech.com> wrote:

>
>
> In the message dated: Tue, 11 Aug 2009 14:14:03 +0200,
> The pithy ruminations from Juan Ramon Martin Blanco on
> <Re: [Linux-cluster] do I have a fence DRAC device?> were:
> => --===============1917368601==
> => Content-Type: multipart/alternative;
> boundary=0016364c7c07663f600470dca3b8
> =>
> => --0016364c7c07663f600470dca3b8
> => Content-Type: text/plain; charset=ISO-8859-1
> => Content-Transfer-Encoding: quoted-printable
> =>
> => On Tue, Aug 11, 2009 at 2:03 PM, ESGLinux <esggrupos at gmail.com> wrote:
> =>
> => > Thanks
> => > I=B4ll check it when I could reboot the server.
> => >
> => > greetings,
> => >
> => You have a BMC ipmi in the first network interface, it can be configured
> at
> => boot time (I don't  remember if inside the BIOS or pressing
> cntrl+something
> => during boot)
> =>
>
> Based on my notes, here's how I configured the DRAC interface on a Dell
> 1950
> for use as a fence device:
>
>        Configuring the card from Linux depending on the installation of
> Dell's
>        OMSA package. Once that's installed, use the following
> commands:
>
>                racadm config -g cfgSerial -o cfgSerialTelnetEnable 1
>                racadm config -g cfgLanNetworking -o cfgDNSRacName
> HOSTNAME_FOR_INTERFACE
>                racadm config -g cfgDNSDomainName DOMAINNAME_FOR_INTERFACE
>                racadm config -g cfgUserAdmin -o cfgUserAdminPassword -i 2
> PASSWORD
>                racadm config -g cfgNicEnable 1
>                racadm config -g cfgNicIpAddress WWW.XXX.YYY.ZZZ
>                racadm config -g cfgNicNetmask WWW.XXX.YYY.ZZZ
>                racadm config -g cfgNicGateway WWW.XXX.YYY.ZZZ
>                racadm config -g cfgNicUseDhcp 0
>
>
>        I also save a backup of the configuration with:
>
>                racadm getconfig -f ~/drac_config
>
>
> Hope this helps,
>
> Mark
>
> ----
> Mark Bergman                              voice: 215-662-7310
> mark.bergman at uphs.upenn.edu                 fax: 215-614-0266
> System Administrator     Section of Biomedical Image Analysis
> Department of Radiology            University of Pennsylvania
>      PGP Key: https://www.rad.upenn.edu/sbia/bergman
>
>
> => Greetings,
>  => Juanra
> =>
> => >
> => > ESG
> => >
> => > 2009/8/10 Paras pradhan <pradhanparas at gmail.com>
> => >
> => > On Mon, Aug 10, 2009 at 5:24 AM, ESGLinux<esggrupos at gmail.com> wrote:
> => >> > Hi all,
> => >> > I was designing a 2 node cluster and I was going to use 2 servers
> DELL
> => >> > PowerEdge 1950. I was going to buy a DRAC card to use for fencing
> but
> => >> > running several commands in the servers I have noticed that when I
> run
> => >> this
> => >> > command:
> => >> > #ipmitool lan print
> => >> > Set in Progress : Set Complete
> => >> > Auth Type Support : NONE MD2 MD5 PASSWORD
> => >> > Auth Type Enable : Callback : MD2 MD5
> => >> >                         : User : MD2 MD5
> => >> >                         : Operator : MD2 MD5
> => >> >                         : Admin : MD2 MD5
> => >> >                         : OEM : MD2 MD5
> => >> > IP Address Source : Static Address
> => >> > IP Address : 0.0.0.0
> => >> > Subnet Mask : 0.0.0.0
> => >> > MAC Address : 00:1e:c9:ae:6f:7e
> => >> > SNMP Community String : public
> => >> > IP Header : TTL=0x40 Flags=0x40 Precedence=0x00 TOS=0x10
> => >> > Default Gateway IP : 0.0.0.0
> => >> > Default Gateway MAC : 00:00:00:00:00:00
> => >> > Backup Gateway IP : 0.0.0.0
> => >> > Backup Gateway MAC : 00:00:00:00:00:00
> => >> > 802.1q VLAN ID : Disabled
> => >> > 802.1q VLAN Priority : 0
> => >> > RMCP+ Cipher Suites : 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14
> => >> > Cipher Suite Priv Max : aaaaaaaaaaaaaaa
> => >> >                         : X=Cipher Suite Unused
> => >> >                         : c=CALLBACK
> => >> >                         : u=USER
> => >> >                         : o=OPERATOR
> => >> >                         : a=ADMIN
> => >> >                         : O=OEM
> => >> > does this mean that I already have an ipmi card (not configured)
> that
> => I
> => >> can
> => >> > use for fencing? if the anwser is yes, where hell must I configure
> it?
> =>  I
> => >> > don=B4t see wher can I do it.
> => >> > If I haven=B4t a fencing device which one do you recommed to use?
>  => >> > Thanks in advance
> => >> > ESG
> => >> >
> => >> > --
> => >> > Linux-cluster mailing list
> => >> > Linux-cluster at redhat.com
> => >> > https://www.redhat.com/mailman/listinfo/linux-cluster
> => >> >
> => >>
> => >> Yes you have IPMI and if you are using 1950 Dell, DRAC should be
> there
> => >> too. You can see if you have DRAC or not when the server starts and
> => >> before the loading of the OS.
> => >>
> => >> I have 1850s and I am using DRAC for fencing.
> => >>
> => >>
> => >> Paras.
> => >>
> => >> --
> => >> Linux-cluster mailing list
> => >> Linux-cluster at redhat.com
> => >> https://www.redhat.com/mailman/listinfo/linux-cluster
> => >>
> => >
> => >
>
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090812/1426af32/attachment.htm>

From cthulhucalling at gmail.com  Wed Aug 12 14:09:30 2009
From: cthulhucalling at gmail.com (Ian Hayes)
Date: Wed, 12 Aug 2009 07:09:30 -0700
Subject: [Linux-cluster] fence device
In-Reply-To: <EE5A9AA37E4A40D5BD7B22FED6A61FE9@siim94.local>
References: <EE5A9AA37E4A40D5BD7B22FED6A61FE9@siim94.local>
Message-ID: <36df569a0908120709h6c8c2a5dta8be54af806496bf@mail.gmail.com>

1) You can use fence_manual if you don't have a real fence device. Not
recommended, of course.

2) I believe you want to look at Pirhana

On Wed, Aug 12, 2009 at 12:08 AM, LEROUX Ludovic <ironludo at free.fr> wrote:

>  i have a linux redhat cluster with two redhat 5.3 nodes.
> I have a shared iscsi disk with gfs.
> And i created an oracle service which mounts the gfs file system and
> start an oracle instance in fail over mode.
> I have two questions:
> 1?) I don't have a fence device. Is it possible to create a cluster without
> a fence device?
> 2?) is it possible to create a high available cluster with two or more
> nodes which works together (for example http server).
> Thanks.
> Ludo
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090812/dbd756c2/attachment.htm>

From deJongm at TEOCO.com  Wed Aug 12 16:15:53 2009
From: deJongm at TEOCO.com (de Jong, MarkJan)
Date: Wed, 12 Aug 2009 12:15:53 -0400
Subject: [Linux-cluster] RHEL 5.3: Joining fence domain hangs when selinux
	is enabled
Message-ID: <5E3DCAE61C95FA4397679425D7275D26091A547C@HQ-MX03.us.teo.earth>

It seems that with selinux enabled, fencing hangs during 'service cman start'.

When selinux is set to enforcing, the cman startup script hangs at "Starting fencing ...." and never times out.
There are NO logs related to the event in /var/log/audit/audit.log, nor anything telling in /var/log/messages. 'fence_tool dump' also does not provide any further details.

After setting selinux to permissive, fencing starts up without incident.

I'm using the following packages:

kernel-xen-2.6.18-128.4.1.el5
cman-2.0.98-1.el5_3.4

Let me know if I can provide any further info.

thanks,
Mark de Jong





________________________________
PRIVILEGED AND CONFIDENTIAL
PLEASE NOTE: The information contained in this message is privileged and confidential, and is intended only for the use of the individual to whom it is addressed and others who have been specifically authorized to receive it. If you are not the intended recipient, you are hereby notified that any dissemination, distribution or copying of this communication is strictly prohibited. If you have received this communication in error, or if any problems occur with transmission, please contact sender. Thank you.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090812/97e833a2/attachment.htm>

From cthulhucalling at gmail.com  Wed Aug 12 17:26:20 2009
From: cthulhucalling at gmail.com (Ian Hayes)
Date: Wed, 12 Aug 2009 10:26:20 -0700
Subject: [Linux-cluster] RHEL 5.3: Joining fence domain hangs when selinux
	is enabled
In-Reply-To: <5E3DCAE61C95FA4397679425D7275D26091A547C@HQ-MX03.us.teo.earth>
References: <5E3DCAE61C95FA4397679425D7275D26091A547C@HQ-MX03.us.teo.earth>
Message-ID: <36df569a0908121026od26a5e7uc50c1bf2f83bd9f8@mail.gmail.com>

I'm assuming that you're running the Targeted policy and not the strict
policy...

RHEL5 has a module for ccs, but I haven't taken it apart. The files for
fencing may be incorrectly labeled or the policy doesn't allow fenced to run
correctly.

Look at your /var/log/audit/audit.log files and see what's being denied. You
may want to install sealert and setroubleshootd so you can browse the
messages. First, check the file contexts of the files that are appearing in
your audit logs.  Nothing should be default_t. If anything looks out of
whack, try restoring the correct file contexts with restorecon and see if
the file contexts have changed.  If you're feeling brave, you can start
writing a custom policy module to permit fenced to start up.

The audit logs will tell you everything, and where you will need to start. I
managed to knock out a policy for 389Server in about an hour, but I had the
benefit of just coming back from Redhat's SELinux class.


On Wed, Aug 12, 2009 at 9:15 AM, de Jong, MarkJan <deJongm at teoco.com> wrote:

>  It seems that with selinux enabled, fencing hangs during ?service cman
> start?.
>
>
>
> When selinux is set to enforcing, the cman startup script hangs at
> ?Starting fencing ?.? and never times out.
>
> There are NO logs related to the event in /var/log/audit/audit.log, nor
> anything telling in /var/log/messages. ?fence_tool dump? also does not
> provide any further details.
>
>
>
> After setting selinux to permissive, fencing starts up without incident.
>
>
>
> I?m using the following packages:
>
>
>
> kernel-xen-2.6.18-128.4.1.el5
>
> cman-2.0.98-1.el5_3.4
>
>
>
> Let me know if I can provide any further info.
>
>
>
> thanks,
>
> Mark de Jong
>
>
>
>
>
>
>
>
>
> ------------------------------
> PRIVILEGED AND CONFIDENTIAL
> PLEASE NOTE: The information contained in this message is privileged and
> confidential, and is intended only for the use of the individual to whom it
> is addressed and others who have been specifically authorized to receive it.
> If you are not the intended recipient, you are hereby notified that any
> dissemination, distribution or copying of this communication is strictly
> prohibited. If you have received this communication in error, or if any
> problems occur with transmission, please contact sender. Thank you.
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090812/78fba72a/attachment.htm>

From deJongm at TEOCO.com  Wed Aug 12 17:50:03 2009
From: deJongm at TEOCO.com (de Jong, MarkJan)
Date: Wed, 12 Aug 2009 13:50:03 -0400
Subject: [Linux-cluster] RHEL 5.3: Joining fence domain hangs when
	selinux	is enabled
In-Reply-To: <36df569a0908121026od26a5e7uc50c1bf2f83bd9f8@mail.gmail.com>
References: <5E3DCAE61C95FA4397679425D7275D26091A547C@HQ-MX03.us.teo.earth>
	<36df569a0908121026od26a5e7uc50c1bf2f83bd9f8@mail.gmail.com>
Message-ID: <5E3DCAE61C95FA4397679425D7275D26091A5482@HQ-MX03.us.teo.earth>

That's just it. There are no logs being generated in audit.log. I'm pretty well versed in creating custom SELInux policies.

I've reported on issues in the past where SELinux does not generate logs. It was a while ago and have since forgotten what the resolution was but I was fixed by the devs.. I'll be more than happy to file a bug report.

Thx,
M

From: linux-cluster-bounces at redhat.com [mailto:linux-cluster-bounces at redhat.com] On Behalf Of Ian Hayes
Sent: Wednesday, August 12, 2009 1:26 PM
To: linux clustering
Subject: Re: [Linux-cluster] RHEL 5.3: Joining fence domain hangs when selinux is enabled

I'm assuming that you're running the Targeted policy and not the strict policy...

RHEL5 has a module for ccs, but I haven't taken it apart. The files for fencing may be incorrectly labeled or the policy doesn't allow fenced to run correctly.

Look at your /var/log/audit/audit.log files and see what's being denied. You may want to install sealert and setroubleshootd so you can browse the messages. First, check the file contexts of the files that are appearing in your audit logs.  Nothing should be default_t. If anything looks out of whack, try restoring the correct file contexts with restorecon and see if the file contexts have changed.  If you're feeling brave, you can start writing a custom policy module to permit fenced to start up.

The audit logs will tell you everything, and where you will need to start. I managed to knock out a policy for 389Server in about an hour, but I had the benefit of just coming back from Redhat's SELinux class.

On Wed, Aug 12, 2009 at 9:15 AM, de Jong, MarkJan <deJongm at teoco.com<mailto:deJongm at teoco.com>> wrote:

It seems that with selinux enabled, fencing hangs during 'service cman start'.



When selinux is set to enforcing, the cman startup script hangs at "Starting fencing ...." and never times out.

There are NO logs related to the event in /var/log/audit/audit.log, nor anything telling in /var/log/messages. 'fence_tool dump' also does not provide any further details.



After setting selinux to permissive, fencing starts up without incident.



I'm using the following packages:



kernel-xen-2.6.18-128.4.1.el5

cman-2.0.98-1.el5_3.4



Let me know if I can provide any further info.



thanks,

Mark de Jong









________________________________
PRIVILEGED AND CONFIDENTIAL
PLEASE NOTE: The information contained in this message is privileged and confidential, and is intended only for the use of the individual to whom it is addressed and others who have been specifically authorized to receive it. If you are not the intended recipient, you are hereby notified that any dissemination, distribution or copying of this communication is strictly prohibited. If you have received this communication in error, or if any problems occur with transmission, please contact sender. Thank you.

--
Linux-cluster mailing list
Linux-cluster at redhat.com<mailto:Linux-cluster at redhat.com>
https://www.redhat.com/mailman/listinfo/linux-cluster


________________________________
PRIVILEGED AND CONFIDENTIAL
PLEASE NOTE: The information contained in this message is privileged and confidential, and is intended only for the use of the individual to whom it is addressed and others who have been specifically authorized to receive it. If you are not the intended recipient, you are hereby notified that any dissemination, distribution or copying of this communication is strictly prohibited. If you have received this communication in error, or if any problems occur with transmission, please contact sender. Thank you.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090812/aa35e593/attachment.htm>

From cthulhucalling at gmail.com  Wed Aug 12 18:08:35 2009
From: cthulhucalling at gmail.com (Ian Hayes)
Date: Wed, 12 Aug 2009 11:08:35 -0700
Subject: [Linux-cluster] RHEL 5.3: Joining fence domain hangs when selinux
	is enabled
In-Reply-To: <5E3DCAE61C95FA4397679425D7275D26091A5482@HQ-MX03.us.teo.earth>
References: <5E3DCAE61C95FA4397679425D7275D26091A547C@HQ-MX03.us.teo.earth>
	<36df569a0908121026od26a5e7uc50c1bf2f83bd9f8@mail.gmail.com>
	<5E3DCAE61C95FA4397679425D7275D26091A5482@HQ-MX03.us.teo.earth>
Message-ID: <36df569a0908121108s7a5f01fdg5989b3b7f0ab92db@mail.gmail.com>

There may be a dontaudit clause in either the base or the ccs modules. Try
inserting the enableaudit.pp module and see what happens. I'm not sure if
RHEL5.3 supports "semodule -DB" but that would turn off dontaudit also.

On Wed, Aug 12, 2009 at 10:50 AM, de Jong, MarkJan <deJongm at teoco.com>wrote:

>  That?s just it. There are no logs being generated in audit.log. I?m
> pretty well versed in creating custom SELInux policies.
>
>
>
> I?ve reported on issues in the past where SELinux does not generate logs.
> It was a while ago and have since forgotten what the resolution was but I
> was fixed by the devs.. I?ll be more than happy to file a bug report.
>
>
>
> Thx,
>
> M
>
>
>
> *From:* linux-cluster-bounces at redhat.com [mailto:
> linux-cluster-bounces at redhat.com] *On Behalf Of *Ian Hayes
> *Sent:* Wednesday, August 12, 2009 1:26 PM
> *To:* linux clustering
> *Subject:* Re: [Linux-cluster] RHEL 5.3: Joining fence domain hangs when
> selinux is enabled
>
>
>
> I'm assuming that you're running the Targeted policy and not the strict
> policy...
>
> RHEL5 has a module for ccs, but I haven't taken it apart. The files for
> fencing may be incorrectly labeled or the policy doesn't allow fenced to run
> correctly.
>
> Look at your /var/log/audit/audit.log files and see what's being denied.
> You may want to install sealert and setroubleshootd so you can browse the
> messages. First, check the file contexts of the files that are appearing in
> your audit logs.  Nothing should be default_t. If anything looks out of
> whack, try restoring the correct file contexts with restorecon and see if
> the file contexts have changed.  If you're feeling brave, you can start
> writing a custom policy module to permit fenced to start up.
>
> The audit logs will tell you everything, and where you will need to start.
> I managed to knock out a policy for 389Server in about an hour, but I had
> the benefit of just coming back from Redhat's SELinux class.
>
>  On Wed, Aug 12, 2009 at 9:15 AM, de Jong, MarkJan <deJongm at teoco.com>
> wrote:
>
> It seems that with selinux enabled, fencing hangs during ?service cman
> start?.
>
>
>
> When selinux is set to enforcing, the cman startup script hangs at
> ?Starting fencing ?.? and never times out.
>
> There are NO logs related to the event in /var/log/audit/audit.log, nor
> anything telling in /var/log/messages. ?fence_tool dump? also does not
> provide any further details.
>
>
>
> After setting selinux to permissive, fencing starts up without incident.
>
>
>
> I?m using the following packages:
>
>
>
> kernel-xen-2.6.18-128.4.1.el5
>
> cman-2.0.98-1.el5_3.4
>
>
>
> Let me know if I can provide any further info.
>
>
>
> thanks,
>
> Mark de Jong
>
>
>
>
>
>
>
>
>
>
>  ------------------------------
>
> PRIVILEGED AND CONFIDENTIAL
> PLEASE NOTE: The information contained in this message is privileged and
> confidential, and is intended only for the use of the individual to whom it
> is addressed and others who have been specifically authorized to receive it.
> If you are not the intended recipient, you are hereby notified that any
> dissemination, distribution or copying of this communication is strictly
> prohibited. If you have received this communication in error, or if any
> problems occur with transmission, please contact sender. Thank you.
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
>
>
> ------------------------------
> PRIVILEGED AND CONFIDENTIAL
> PLEASE NOTE: The information contained in this message is privileged and
> confidential, and is intended only for the use of the individual to whom it
> is addressed and others who have been specifically authorized to receive it.
> If you are not the intended recipient, you are hereby notified that any
> dissemination, distribution or copying of this communication is strictly
> prohibited. If you have received this communication in error, or if any
> problems occur with transmission, please contact sender. Thank you.
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090812/66c178d7/attachment.htm>

From jumanjiman at gmail.com  Wed Aug 12 18:14:36 2009
From: jumanjiman at gmail.com (Paul Morgan)
Date: Wed, 12 Aug 2009 13:14:36 -0500
Subject: [Linux-cluster] RHEL 5.3: Joining fence domain hangs when selinux
	is enabled
In-Reply-To: <36df569a0908121026od26a5e7uc50c1bf2f83bd9f8@mail.gmail.com>
References: <5E3DCAE61C95FA4397679425D7275D26091A547C@HQ-MX03.us.teo.earth>
	<36df569a0908121026od26a5e7uc50c1bf2f83bd9f8@mail.gmail.com>
Message-ID: <2D75352F-D24D-4F85-BE1B-8D36E43B6F8C@gmail.com>



On Aug 12, 2009, at 12:26, Ian Hayes <cthulhucalling at gmail.com> wrote:

> I'm assuming that you're running the Targeted policy and not the  
> strict policy...
>
> RHEL5 has a module for ccs, but I haven't taken it apart. The files  
> for fencing may be incorrectly labeled or the policy doesn't allow  
> fenced to run correctly.
>
> Look at your /var/log/audit/audit.log files and see what's being  
> denied. You may want to install sealert and setroubleshootd so you  
> can browse the messages. First, check the file contexts of the files  
> that are appearing in your audit logs.  Nothing should be default_t.  
> If anything looks out of whack, try restoring the correct file  
> contexts with restorecon and see if the file contexts have changed.

Definitely run restorecon as suggested, but add a "-vv" option (two  
levels of verbosity) to see whether it's ignoring any customizable  
types. If so, run again and additionally use the -F option to force.

Hth,
-paul



From brem.belguebli at gmail.com  Wed Aug 12 22:45:29 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Thu, 13 Aug 2009 00:45:29 +0200
Subject: [Linux-cluster] Qdisk question
Message-ID: <29ae894c0908121545r3a92cf8fk59d5c10c162cdfad@mail.gmail.com>

Hi,
I have a question about the qdisk concept.

My understanding of qdisk is that it is used as a tie-breaker, but it looks
like it is more a heatbeat vector than a simple tie-breaker.

My setup consists of 4 nodes located on 2 different production sites (2+2)
using SAN shared storage (2 disk frames, 1 per site).

The qdisk is a iscsi shared lun from a third site that I expected to use as
a tie-breaker in case 1 of my 2 prod sites was experiencing network problems
and gets completely isolated.

Until here, no real problem indeed, if the site gets apart from the other
prod site and also from the third site (hosting the iscsi target qdisk) the
2 nodes from the failing site get evicted from the cluster.

But, what if my third site gets isolated while the 2 prod ones are fine ?

The real  question is what happens in case all the nodes loose access to the
qdisk while they're still able to see each others ?

The 4 nodes have each 1 vote and the qdisk 1 vote. The expected quorum is 3.

When the cluster is running with all of its nodes and the qdisk, the number
of votes is 5.

If I loose the qdisk, the number of votes falls to 4, the cluster is quorate
(4>3) but it looks like everything goes bad, each node deactivate itself as
it can't write its alive status (--> heartbeat vector) to the qdisk even if
the network heartbeating is working fine.

I have tried to configure heuristics (ping a node on the third site) without
qdisk device but they seem to be ignored.

Any comments or tips ?

Regards
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090813/2c6ce078/attachment.htm>

From raju.rajsand at gmail.com  Thu Aug 13 03:27:15 2009
From: raju.rajsand at gmail.com (Rajagopal Swaminathan)
Date: Thu, 13 Aug 2009 08:57:15 +0530
Subject: [Linux-cluster] Is there any backup heartbeat channel
In-Reply-To: <20090812120857.14055bi0l6l01okp@webmail.itu.edu.tr>
References: <4A7FA8B7.5080708@oracle.com> <1249896553.19917.2.camel@marc>
	<4A7FEC6F.2050405@oracle.com> <4A7FEE8D.8070400@oracle.com>
	<1249902500.20848.11.camel@marc> <4A827B23.7050107@oracle.com>
	<20090812120857.14055bi0l6l01okp@webmail.itu.edu.tr>
Message-ID: <8786b91c0908122027g79716885j7ed933c3287c889b@mail.gmail.com>

Greetings,

2009/8/12 Hakan VELIOGLU <veliogluh at itu.edu.tr>:
> Hi list,
>
> I am trying a two node cluster with RH 5.3 on Sun X4150 hardware. I use a

IIRC, Sun x4150 has four ethernet ports. Two can be used for outside
networking and two can be bonded and used for heartbeat.

ALOM can be used for fencing and can be on a seperate subnet if required.

Regards

Rajagopal



From akapp at fnds3000.com  Thu Aug 13 03:50:55 2009
From: akapp at fnds3000.com (akapp)
Date: Thu, 13 Aug 2009 05:50:55 +0200
Subject: [Linux-cluster] Re: Linux-cluster Digest, Vol 64, Issue 12
In-Reply-To: <20090811160009.8EFB3619D6E@hormel.redhat.com>
References: <20090811160009.8EFB3619D6E@hormel.redhat.com>
Message-ID: <F08EE546-FF9F-43BD-B748-EFD4D46BDF9C@fnds3000.com>

Hi all

Just an update on the CentOS 5.3 and lucci.

When installing the network and enabling IPV6, the default , and well  
known, 127.0.0.1 localhost.localdomain  localhost is replaced with a new
1:: localhost.localdomain localhost.

Luci lives on 127.0.0.1, so if that is not in /etc/hosts, luci will  
not start!

I think the biggest frustration with this was the problem that no  
proper logfile entry is written.

Tks
Andre


# Do not remove the following line, or various programs
# that require network functionality will fail.
::1	localhost.localdomain	localhost	apollo
127.0.0.1	localhost.localdomain	localhost



On Aug 11, 2009, at 6:00 PM, linux-cluster-request at redhat.com wrote:

> Send Linux-cluster mailing list submissions to
> 	linux-cluster at redhat.com
>
> To subscribe or unsubscribe via the World Wide Web, visit
> 	https://www.redhat.com/mailman/listinfo/linux-cluster
> or, via email, send a message with subject or body 'help' to
> 	linux-cluster-request at redhat.com
>
> You can reach the person managing the list at
> 	linux-cluster-owner at redhat.com
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Linux-cluster digest..."
>
>
> Today's Topics:
>
>   1. Succesfull installation on centos 5.3 with live kvm	migration
>      (Robert Verspuy)
>   2. Centos 5.3 X64 a& luci (akapp)
>   3. Re: Linux-cluster Digest, Vol 64, Issue 10 (Bob Peterson)
>   4. RHEL 4.7 fenced fails -- stuck join state: S-2,2,1 (Robert Hurst)
>   5. Re: do I have a fence DRAC device? (bergman at merctech.com)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Tue, 11 Aug 2009 14:59:13 +0200
> From: Robert Verspuy <robert at exa-omicron.nl>
> Subject: [Linux-cluster] Succesfull installation on centos 5.3 with
> 	live kvm	migration
> To: linux clustering <linux-cluster at redhat.com>
> Message-ID: <4A816B21.2040907 at exa-omicron.nl>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
>
> Getting cluster software including kvm virtual machines with live
> migration working,
> can be a very difficult task, with many obstacles.
>
> But I would like to mention to the mailing list, that I just booked  
> some
> succes.
> And because nobody is around to tell the wonderfull news,
> I would like to share my hapiness here ;)
>
> My setup:
> 2 NAS servers and 1 Supermicro bladeserver with 5 blades.
>
> The NAS servers are running Openfiler 2.3
> both NAS servers have:
> 1 Transcend IDE 4Gbyte flashcard (on the ide port on the mainboard).
> 3 x Transcend 4Gbyte usb sticks
> 8 SATA disks.
>
> The IDE flashcard is setup in a raid-1 mirroring (md0) with one USB
> stick providing the root FS voor openfiler
> The other 2 USB sticks have 5 partitions: 4 x 500 MB and 1 x 2GB.
> those are mirrored with raid-1 together. (md5 until md8 are the 500mb
> partions, and md9 is the 2Gb partition).
> Then the 8 harddisks are also tied together per 2 as mirroring raid1
> (md1 until md4).
>
> Then I used DRBD (8.2.7) to mirror the 4 raid-1's of the disks (md1
> until md4) and the 2GB mirror (md9)
> over the network to the other NAS server. (drbd1, drbd2, drbd3 and  
> drbd4)
> The 500mb raid-1's are used to store metadata of the 4 disk raid-1's.
> The 2gb drbd (drbd0) has internal metadata.
>
> The 2gb drbd (drbd0) is mounted as ext3 on one only server and is used
> to store all kinds
> of openfiler information that is needed on both nas servers,
> like the openfiler config (mostly), dhcp leases database, openldap  
> database.
> And heartbeat makes sure that one NAS server is running all the
> software, and with any problems,
> it can switch over very easily.
>
> The drbd1 til 4 are setup as a LVM PV, and bound together in one big  
> VG.
> From that VG, I created a 5 x 5GB LV to be used as root device for
> blade1 til blade5
> These LV's are stripped accross 2 PV for speed (altough that's still  
> my
> only bottleneck at the moment, but more later about this...).
> These LV's are setup as iSCSI
>
> I also created one big LV of around 600GB, which can be mounted  
> through NFS.
>
> Then a few more LV's are created (around 10GB, also iscsi) for every  
> VM
> I want.
> For every iSCSI LV I create a separate target.
>
> The Supermicro blades can boot from an iscsi device.
> The exact scsi device is given through a DHCP option.
> I only setup a initiatior name in the iscsi bios of the blade.
>
> On the blade LV's I installed CentOS 5.3 (latest updates).
> But with a few modifications.
>
> I changes a few things in the initrd, to bound eth0 to br0 during the
> linux boot,
> and before linux is taking over the iscsi from the bios, because
> when you have a linux root through iscsi, and try to attach eth0 to  
> br0,
> you loose networkconnectivity for a moment, and could crash the linux,
> because everything it uses, comes from the network (iscsi root).
> I also added a little script to the initrd to call iscsiadm with a  
> fixed
> iscsi
> target, because unfortunately iscsiadm can't read the iscsi settings
> from dhcp
> or the supermicro firmware.
>
> When the blades are booted, they all join one redhat cluster with 3
> nodes to be quorum.
> Because I have 5 blades, two can fail before everything stops working.
>
> Then I compiled the following software my own, because the ones in the
> centos repo,
> and the testing repo didn't function correctly:
> libvirt 0.7.0 (./configure --prefix=/usr)
> kvm-88 (./configure --prefix=/usr --disable-xen)
>
> The /usr/share/cluster/vm.sh from the default centos repo is still  
> based
> on xen.
> I downloaded the latest from
> https://bugzilla.redhat.com/show_bug.cgi?id=412911
> but it appears that that one is not working correctly either.
> I made some changes myself.
>
> And now it's working all together very nicely
>
> I just ran a VM on blade1, and while this VM was running bonnie++ on a
> NFS mount to the NAS server,
> I live-migrated it about 10 times to blade2 and back.
>
> During this bonnie++ run and live migrations, I pinged the device.
> And where the normal ping times are around 20-35 ms (I pinged  
> through a
> VPN line from my home to the data center).
> I only saw one or 2 pings just around the end of the live migration  
> that
> were around 40-60ms.
> but no drops, and no errors in bonnie++.
>
> I will write some more information about the complete setup, and  
> post it
> somewhere on my blog or someting,
> But I just wanted to let everybody know, that it can be done ;)
>
> If you have any questions, let me know.
>
> The only 'problem' I still have is the speed to and from the disks.
> When I update any settings on the bladeserver. I always do this on  
> blade1.
> Then shut it down, On the NAS server I copy the content of the iscsi  
> LV
> to an image file on the ext3 LV.
> Then I can power up blade1, wait until it reenters the cluster,
> and then one by one shut down the next blade, On the NAS copy the  
> image
> from the ext3 LV to the blade LV.
> And start the blade again.
>
> I use the drbd1 til drbd4 as 4 PV's for a VG.
> The speed (hdparm -t on the NAS) of all PV's are around 75 MB/sec
> (except for one which is 45MB/sec)
>
> The blade LV (/dev/vg0/blade1 for example) is striped over 2 PV's.
> The Speed (hdparm -t) of /dev/vg0/blade1 is 122MB/sec.
>
> The ext3 LV (/dev/vg0/data0) is striped over 4 PV's.
> The Speed (hdparm -t) of /dev/vg0/data0 is 227 MB/sec.
>
> But when copying from the blade LV to the ext3 LV:
> dd if=/dev/vg0/blade1 of=/mnt/vg0/data0/vm/image/blade_v2.7.img
> it takes about 70 seconds, which is about 75MB/sec.
>
> but when copying back:
> dd if=/mnt/vg0/data0/vm/image/blade_v2.7.img of=/dev/vg0/blade1
> It takes about 390 seconds, which is about 13MB/sec
>
> I think it has something to do with the striped over 4 PV's of the  
> LVM.
> So I will try to create a new ext3 LV stiped accross 2 PV's and see if
> this is faster.
>
> Robert Verspuy
>
> -- 
> *Exa-Omicron*
> Patroonsweg 10
> 3892 DB Zeewolde
> Tel.: 088-OMICRON (66 427 66)
> http://www.exa-omicron.nl
>
>
>
> ------------------------------
>
> Message: 2
> Date: Tue, 11 Aug 2009 14:58:38 +0200
> From: akapp <akapp at fnds3000.com>
> Subject: [Linux-cluster] Centos 5.3 X64 a& luci
> To: linux-cluster at redhat.com
> Message-ID: <34CB6BB9-6856-4E52-B946-5465F90B081C at fnds3000.com>
> Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes
>
> Good day
>
> I have a Sun x4100 server running Centos 5.3 X64 - patch to latest and
> greatest.
>
> When trying to start luci, it simply fails, no error in /var/log  and
> nothing in /var/lib/luci/log
>
> I have re-installed luci and ricci a couple of times now. Cleaned  
> out /
> var/lib/luci & /rici between installations.
>
> I have even tried the complete yum grouremove "Clustering" "Cluster
> Storage" and re-installed the complete package again.
>
> Used ricci/luci combination with great success in 5.2, but both
> servers giving the same problem.
>
> Any pointers will be appreciated.
>
>
> Here is screen snipped of problem:
>
>
>
> Installed: luci.x86_64 0:0.12.1-7.3.el5.centos.1
> Complete!
> [root at clu1 luci]# luci_admin init
> Initializing the luci server
>
>
> Creating the 'admin' user
>
> Enter password:
> Confirm password:
>
> Please wait...
> The admin password has been successfully set.
> Generating SSL certificates...
> The luci server has been successfully initialized
>
>
> You must restart the luci server for changes to take effect.
>
> Run "service luci restart" to do so
>
> [root at clu1 luci]# service luci restart
> Shutting down luci:                                        [  OK  ]
> Starting luci: Generating https SSL certificates...  done
>                                                            [FAILED]
> [root at clu1 luci]#
>
>
>
>
>
>
> Tks
> Andre
>
>
>
> ------------------------------
>
> Message: 3
> Date: Tue, 11 Aug 2009 09:24:14 -0400 (EDT)
> From: Bob Peterson <rpeterso at redhat.com>
> Subject: [Linux-cluster] Re: Linux-cluster Digest, Vol 64, Issue 10
> To: Wendell Dingus <wendell at bisonline.com>
> Cc: linux-cluster at redhat.com
> Message-ID:
> 	<1891903567.443711249997054068.JavaMail.root at zmail06.collab.prod.int.phx2.redhat.com 
> >
> 	
> Content-Type: text/plain; charset=utf-8
>
> ----- "Wendell Dingus" <wendell at bisonline.com> wrote:
> | Well, here's the entire list of blocks it ignored and the entire
> | message section.
> | Perhaps I'm just overlooking it but I'm not seeing anything in the
> | messages
> | that appears to be a block number. Maybe 1633350398 but if so it is
> | not a match.
>
> Your assumption is correct.  The block number was 1633350398, which
> is labeled "bh = " for some reason.
>
> | Anyway, since you didn't specifically say a new/fixed version of  
> fsck
> | was
> | imminent and that it would likely fix this we began plan B today. We
>
> Yesterday I pushed a newer gfs_fsck and fsck.gfs2 to their appropriate
> git source repositories.  So you can build that version from source if
> you need it right away.  But it sounds like it wouldn't have helped
> your problem anyway.  What would really be nice is if there is a way
> to recreate the problem in our lab.  In theory, this error could be
> caused by a hardware problem too.
>
> | plugged
> | in another drive, placed a GFS2 filesystem on it and am actively
> | copying files
> | off to it now. Fingers crossed that nothing will hit a disk block  
> that
> | causes
> | this again but I could be so lucky probably...
>
> It's hard to say whether you'll hit it again.
>
> Regards,
>
> Bob Peterson
> Red Hat File Systems
>
>
>
> ------------------------------
>
> Message: 4
> Date: Tue, 11 Aug 2009 10:55:48 -0400
> From: "Robert Hurst" <rhurst at bidmc.harvard.edu>
> Subject: [Linux-cluster] RHEL 4.7 fenced fails -- stuck join state:
> 	S-2,2,1
> To: "linux clustering" <linux-cluster at redhat.com>
> Message-ID: <1250002549.2782.36.camel at WSBID06223.bidmc.harvard.edu>
> Content-Type: text/plain; charset="us-ascii"
>
> Simple 4-node cluster, 2-nodes have a GFS shared home directory  
> mounted
> for over a month.  Today, I wanted to mount /home on a 3rd node, so:
>
> # service fenced start                [failed]
>
> Weird.  Checking /var/log/messages show:
>
> Aug 11 10:19:06 cerberus kernel: Lock_Harness 2.6.9-80.9.el4_7.10  
> (built
> Jan 22 2009 18:39:16) installed
> Aug 11 10:19:06 cerberus kernel: GFS 2.6.9-80.9.el4_7.10 (built Jan 22
> 2009 18:39:32) installed
> Aug 11 10:19:06 cerberus kernel: GFS: Trying to join cluster  
> "lock_dlm",
> "ccc_cluster47:home"
> Aug 11 10:19:06 cerberus kernel: Lock_DLM (built Jan 22 2009 18:39:18)
> installed
> Aug 11 10:19:06 cerberus kernel: lock_dlm: fence domain not found;  
> check
> fenced
> Aug 11 10:19:06 cerberus kernel: GFS: can't mount proto = lock_dlm,
> table = ccc_cluster47:home, hostdata =
>
> # cman_tool services
> Service          Name                              GID LID State
> Code
> Fence Domain:    "default"                           0   2 join
> S-2,2,1
> []
>
> So, a fenced process is now hung:
>
> root     28302  0.0  0.0  3668  192 ?        Ss   10:19   0:00  
> fenced -t
> 120 -w
>
> Q: Any idea how to "recover" from this state, without rebooting?
>
> The other two servers are unaffected by this (thankfully) and show
> normal operations:
>
> $ cman_tool services
>
> Service          Name                              GID LID State
> Code
> Fence Domain:    "default"                           2   2 run       -
> [1 12]
>
> DLM Lock Space:  "home"                              5   5 run       -
> [1 12]
>
> GFS Mount Group: "home"                              6   6 run       -
> [1 12]
>
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: https://www.redhat.com/archives/linux-cluster/attachments/20090811/7b46b120/attachment.html
>
> ------------------------------
>
> Message: 5
> Date: Tue, 11 Aug 2009 11:39:39 -0400
> From: bergman at merctech.com
> Subject: Re: [Linux-cluster] do I have a fence DRAC device?
> To: linux clustering <linux-cluster at redhat.com>
> Message-ID: <27649.1250005179 at mirchi>
> Content-Type: text/plain; charset=us-ascii
>
>
>
> In the message dated: Tue, 11 Aug 2009 14:14:03 +0200,
> The pithy ruminations from Juan Ramon Martin Blanco on
> <Re: [Linux-cluster] do I have a fence DRAC device?> were:
> => --===============1917368601==
> => Content-Type: multipart/alternative;  
> boundary=0016364c7c07663f600470dca3b8
> =>
> => --0016364c7c07663f600470dca3b8
> => Content-Type: text/plain; charset=ISO-8859-1
> => Content-Transfer-Encoding: quoted-printable
> =>
> => On Tue, Aug 11, 2009 at 2:03 PM, ESGLinux <esggrupos at gmail.com>  
> wrote:
> =>
> => > Thanks
> => > I=B4ll check it when I could reboot the server.
> => >
> => > greetings,
> => >
> => You have a BMC ipmi in the first network interface, it can be  
> configured at
> => boot time (I don't  remember if inside the BIOS or pressing cntrl 
> +something
> => during boot)
> =>
>
> Based on my notes, here's how I configured the DRAC interface on a  
> Dell 1950
> for use as a fence device:
>
> 	Configuring the card from Linux depending on the installation of  
> Dell's
> 	OMSA package. Once that's installed, use the following
> commands:
>
> 		racadm config -g cfgSerial -o cfgSerialTelnetEnable 1
> 		racadm config -g cfgLanNetworking -o cfgDNSRacName  
> HOSTNAME_FOR_INTERFACE
> 		racadm config -g cfgDNSDomainName DOMAINNAME_FOR_INTERFACE
> 		racadm config -g cfgUserAdmin -o cfgUserAdminPassword -i 2 PASSWORD
> 		racadm config -g cfgNicEnable 1
> 		racadm config -g cfgNicIpAddress WWW.XXX.YYY.ZZZ
> 		racadm config -g cfgNicNetmask WWW.XXX.YYY.ZZZ
> 		racadm config -g cfgNicGateway WWW.XXX.YYY.ZZZ
> 		racadm config -g cfgNicUseDhcp 0
>
>
> 	I also save a backup of the configuration with:
>
> 		racadm getconfig -f ~/drac_config
>
>
> Hope this helps,
>
> Mark
>
> ----
> Mark Bergman                              voice: 215-662-7310
> mark.bergman at uphs.upenn.edu                 fax: 215-614-0266
> System Administrator     Section of Biomedical Image Analysis
> Department of Radiology            University of Pennsylvania
>      PGP Key: https://www.rad.upenn.edu/sbia/bergman
>
>
> => Greetings,
> => Juanra
> =>
> => >
> => > ESG
> => >
> => > 2009/8/10 Paras pradhan <pradhanparas at gmail.com>
> => >
> => > On Mon, Aug 10, 2009 at 5:24 AM, ESGLinux<esggrupos at gmail.com>  
> wrote:
> => >> > Hi all,
> => >> > I was designing a 2 node cluster and I was going to use 2  
> servers DELL
> => >> > PowerEdge 1950. I was going to buy a DRAC card to use for  
> fencing but
> => >> > running several commands in the servers I have noticed that  
> when I run
> => >> this
> => >> > command:
> => >> > #ipmitool lan print
> => >> > Set in Progress : Set Complete
> => >> > Auth Type Support : NONE MD2 MD5 PASSWORD
> => >> > Auth Type Enable : Callback : MD2 MD5
> => >> >                         : User : MD2 MD5
> => >> >                         : Operator : MD2 MD5
> => >> >                         : Admin : MD2 MD5
> => >> >                         : OEM : MD2 MD5
> => >> > IP Address Source : Static Address
> => >> > IP Address : 0.0.0.0
> => >> > Subnet Mask : 0.0.0.0
> => >> > MAC Address : 00:1e:c9:ae:6f:7e
> => >> > SNMP Community String : public
> => >> > IP Header : TTL=0x40 Flags=0x40 Precedence=0x00 TOS=0x10
> => >> > Default Gateway IP : 0.0.0.0
> => >> > Default Gateway MAC : 00:00:00:00:00:00
> => >> > Backup Gateway IP : 0.0.0.0
> => >> > Backup Gateway MAC : 00:00:00:00:00:00
> => >> > 802.1q VLAN ID : Disabled
> => >> > 802.1q VLAN Priority : 0
> => >> > RMCP+ Cipher Suites : 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14
> => >> > Cipher Suite Priv Max : aaaaaaaaaaaaaaa
> => >> >                         : X=Cipher Suite Unused
> => >> >                         : c=CALLBACK
> => >> >                         : u=USER
> => >> >                         : o=OPERATOR
> => >> >                         : a=ADMIN
> => >> >                         : O=OEM
> => >> > does this mean that I already have an ipmi card (not  
> configured) that
> => I
> => >> can
> => >> > use for fencing? if the anwser is yes, where hell must I  
> configure it?
> =>  I
> => >> > don=B4t see wher can I do it.
> => >> > If I haven=B4t a fencing device which one do you recommed to  
> use?
> => >> > Thanks in advance
> => >> > ESG
> => >> >
> => >> > --
> => >> > Linux-cluster mailing list
> => >> > Linux-cluster at redhat.com
> => >> > https://www.redhat.com/mailman/listinfo/linux-cluster
> => >> >
> => >>
> => >> Yes you have IPMI and if you are using 1950 Dell, DRAC should  
> be there
> => >> too. You can see if you have DRAC or not when the server  
> starts and
> => >> before the loading of the OS.
> => >>
> => >> I have 1850s and I am using DRAC for fencing.
> => >>
> => >>
> => >> Paras.
> => >>
> => >> --
> => >> Linux-cluster mailing list
> => >> Linux-cluster at redhat.com
> => >> https://www.redhat.com/mailman/listinfo/linux-cluster
> => >>
> => >
> => >
>
>
>
>
>
> ------------------------------
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
> End of Linux-cluster Digest, Vol 64, Issue 12
> *********************************************
>



From michael.osullivan at auckland.ac.nz  Thu Aug 13 04:17:55 2009
From: michael.osullivan at auckland.ac.nz (Michael O'Sullivan)
Date: Thu, 13 Aug 2009 16:17:55 +1200
Subject: [Linux-cluster] GFS2 + multipath iscsi problems
Message-ID: <4A8393F3.6020301@auckland.ac.nz>

Hi everyone,

I hope someone can help me. I am have created a DRBD device with 2 
servers and present this device to 2 other servers in a RedHat cluster 
using 2 iSCSI paths from each DRBD server to each cluster node (i.e., 4 
paths per cluster node, 8 paths in total). I then use multipath so that 
each cluster node identifies the paths as belonging to the same device. 
Finally, I create a GFS2 filesystem on the device. This was all going 
very well and I was experimenting with different settings for the round 
robin behaviour of multipath until I decided to carve the DRBD device 
into smaller chunks. After some playing around I managed this, but now I 
can only get the GFS2 system to mount properly on both cluster nodes if 
the round robin switching parameter (rr_min_io) is set to 1000. I had 
previously been able to use values of 100, 50, 2, 1 and many others, but 
these settings now cause GFS2 to hang or refuse to mount. By looking 
through the various mailing lists I have been able to update to kernel 
2.6.18-162.el5 which has stopped the hanging, but the GFS2 system still 
refuses to mount at times (multiple gfs2_fsck calls seem to help 
sometimes here) and will withdraw after a few IOs (at least thats what 
dmesg tells me). This is pure speculation, but I am wondering if there 
are some timers I need to set to allow GFS2 to coordinate better with 
lower rr_min_io. I'm happy to provide output, error messages, etc but 
I'm not sure at this stage what would be useful.

Thanks in advance for any help. Kind regards, Mike O'S



From wen.gang.wang at oracle.com  Thu Aug 13 06:22:11 2009
From: wen.gang.wang at oracle.com (Wengang Wang)
Date: Thu, 13 Aug 2009 14:22:11 +0800
Subject: [Linux-cluster] gfs2 mount hangs
Message-ID: <4A83B113.60609@oracle.com>

Hi,

I'm hitting a problem that mounting on a gfs2 volume hangs.

ENV:
1) cluster.conf
[wwg at cool gfs2]$ cat /etc/cluster/cluster.conf
<?xml version="1.0"?>
<cluster name="testgfs2" config_version="1">

<clusternodes>
<clusternode name="cool" nodeid="1">
        <fence>
                <method name="1">
                        <device name="manual" nodename="cool"/>
                </method>
        </fence>
</clusternode>

<clusternode name="desk" nodeid="2">
         <fence>
                 <method name="1">
                         <device name="manual" nodename="desk"/>
                 </method>
         </fence>
</clusternode>

</clusternodes>

<fencedevices>
         <fencedevice name="manual" agent="fence_manual"/>
</fencedevices>

<cman two_node="1" expected_votes="2"/>

</cluster>

2) I format the volume /dev/sdb as
mkfs.gfs2 -t testgfs2:1 -j 8 /dev/sdb
#though -j 2 is enough. testing againt 2 journal volumes has the same 
result.

3) kernel version
[wwg at cool gfs2]$ uname -r
2.6.31-rc5
I also tested with the mainline git 2.6.30.4

4) gfs2 module
tested with the one included in 2.6.31-rc5, in 2.6.30.4 or in 
git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-2.6-nmw.git
they have same result.

the hang happens when the other node already mounted the volume. and if 
the other node unmount the volume, the mount on this node successes.
by checking dmesg, the successful mount causes logs like:

GFS2: fsid=testgfs2:1.0: Joined cluster. Now mounting FS...
GFS2: fsid=testgfs2:1.0: jid=0, already locked for use
GFS2: fsid=testgfs2:1.0: jid=0: Looking at journal...
GFS2: fsid=testgfs2:1.0: jid=0: Done
GFS2: fsid=testgfs2:1.0: jid=1: Trying to acquire journal lock...
GFS2: fsid=testgfs2:1.0: jid=1: Looking at journal...
GFS2: fsid=testgfs2:1.0: jid=1: Done
GFS2: fsid=testgfs2:1.0: jid=2: Trying to acquire journal lock...
GFS2: fsid=testgfs2:1.0: jid=2: Looking at journal...
GFS2: fsid=testgfs2:1.0: jid=2: Done
GFS2: fsid=testgfs2:1.0: jid=3: Trying to acquire journal lock...
GFS2: fsid=testgfs2:1.0: jid=3: Looking at journal...
GFS2: fsid=testgfs2:1.0: jid=3: Done
GFS2: fsid=testgfs2:1.0: jid=4: Trying to acquire journal lock...
GFS2: fsid=testgfs2:1.0: jid=4: Looking at journal...
GFS2: fsid=testgfs2:1.0: jid=4: Done
GFS2: fsid=testgfs2:1.0: jid=5: Trying to acquire journal lock...
GFS2: fsid=testgfs2:1.0: jid=5: Looking at journal...
GFS2: fsid=testgfs2:1.0: jid=5: Done
GFS2: fsid=testgfs2:1.0: jid=6: Trying to acquire journal lock...
GFS2: fsid=testgfs2:1.0: jid=6: Looking at journal...
GFS2: fsid=testgfs2:1.0: jid=6: Done
GFS2: fsid=testgfs2:1.0: jid=7: Trying to acquire journal lock...
GFS2: fsid=testgfs2:1.0: jid=7: Looking at journal...
GFS2: fsid=testgfs2:1.0: jid=7: Done

and the hang one stops after:

GFS2: fsid=testgfs2:1.0: Joined cluster. Now mounting FS...

so I guess it's waiting for the other node to release locks on journals...

sysrq-t shows the mount stack:

mount.gfs2    D 000000e2     0  2889   2888 0x00000080
  ed4adc6c 00000082 a24475bb 000000e2 ed4e4f74 c09a5260 c09a9b60 ed4e4f74
  ed4adc3c c09a9b60 c09a9b60 f00019c0 dac8a400 ed4adc58 ed4adc44 00000001
  a2443568 000000e2 ed4e4ce0 ed4adc5c ed4adc5c 00000246 c1f98cd4 c1f98cd8
Call Trace:
  [<f8c5a6e6>] gfs2_glock_holder_wait+0xd/0x11 [gfs2]
  [<c0742897>] __wait_on_bit+0x39/0x60
  [<f8c5a6d9>] ? gfs2_glock_holder_wait+0x0/0x11 [gfs2]
  [<f8c5a6d9>] ? gfs2_glock_holder_wait+0x0/0x11 [gfs2]
  [<c074295e>] out_of_line_wait_on_bit+0xa0/0xa8
  [<c044779d>] ? wake_bit_function+0x0/0x3c
  [<f8c5d487>] wait_on_bit.clone.1+0x1c/0x28 [gfs2]
  [<f8c5d4fe>] gfs2_glock_wait+0x31/0x37 [gfs2]
  [<f8c5d774>] gfs2_glock_nq+0x270/0x278 [gfs2]
  [<f8c5d8a0>] gfs2_glock_nq_num+0x4c/0x6c [gfs2]
  [<f8c66567>] init_journal+0x2b2/0x675 [gfs2]
  [<c04c76bb>] ? wake_up_inode+0x1c/0x1e
  [<c04c76fa>] ? unlock_new_inode+0x3d/0x40
  [<c04c6fe1>] ? d_alloc+0x23/0x15e
  [<c04dd3a0>] ? inotify_d_instantiate+0x17/0x3a
  [<f8c65e5f>] ? gfs2_glock_nq_init+0x13/0x31 [gfs2]
  [<f8c6694f>] init_inodes+0x25/0x152 [gfs2]
  [<f8c6750d>] fill_super+0xa91/0xc10 [gfs2]
  [<f8c5d899>] ? gfs2_glock_nq_num+0x45/0x6c [gfs2]
  [<c04bade0>] get_sb_bdev+0xdc/0x119
  [<c04b4aea>] ? pcpu_alloc+0x352/0x38b
  [<f8c65ad8>] gfs2_get_sb+0x18/0x1a [gfs2]
  [<f8c66a7c>] ? fill_super+0x0/0xc10 [gfs2]
  [<c04baacc>] vfs_kern_mount+0x82/0xf0
  [<c04bab89>] do_kern_mount+0x38/0xc3
  [<c04cc03f>] do_mount+0x68c/0x6e4
  [<c0492dc1>] ? __get_free_pages+0x24/0x26
  [<c04cc0fd>] sys_mount+0x66/0x98
  [<c0402a28>] sysenter_do_call+0x12/0x27

anybody could please help me out?

regards,
wengang.

-- 
--just begin to learn, you are never too late...



From ccaulfie at redhat.com  Thu Aug 13 06:38:32 2009
From: ccaulfie at redhat.com (Christine Caulfield)
Date: Thu, 13 Aug 2009 07:38:32 +0100
Subject: [Linux-cluster] RHEL 5.3: Joining fence domain hangs when selinux
	is enabled
In-Reply-To: <36df569a0908121026od26a5e7uc50c1bf2f83bd9f8@mail.gmail.com>
References: <5E3DCAE61C95FA4397679425D7275D26091A547C@HQ-MX03.us.teo.earth>
	<36df569a0908121026od26a5e7uc50c1bf2f83bd9f8@mail.gmail.com>
Message-ID: <4A83B4E8.9040204@redhat.com>

On 12/08/09 18:26, Ian Hayes wrote:
> I'm assuming that you're running the Targeted policy and not the strict
> policy...
>
> RHEL5 has a module for ccs, but I haven't taken it apart. The files for
> fencing may be incorrectly labeled or the policy doesn't allow fenced to
> run correctly.
>

It's largely an accident that RHCS ran with the SELinux targeted policy, 
and it's never been supported.

For 5.3 some changes were made to the way aisexec communicates with its 
clients that was not covered by the existing ccs policy, so it's now 
broken in that configuration.

Support have some workarounds.

Christine



From veliogluh at itu.edu.tr  Thu Aug 13 08:14:01 2009
From: veliogluh at itu.edu.tr (Hakan VELIOGLU)
Date: Thu, 13 Aug 2009 11:14:01 +0300
Subject: [Linux-cluster] Is there any backup heartbeat channel
In-Reply-To: <8786b91c0908122027g79716885j7ed933c3287c889b@mail.gmail.com>
References: <4A7FA8B7.5080708@oracle.com> <1249896553.19917.2.camel@marc>
	<4A7FEC6F.2050405@oracle.com> <4A7FEE8D.8070400@oracle.com>
	<1249902500.20848.11.camel@marc> <4A827B23.7050107@oracle.com>
	<20090812120857.14055bi0l6l01okp@webmail.itu.edu.tr>
	<8786b91c0908122027g79716885j7ed933c3287c889b@mail.gmail.com>
Message-ID: <20090813111401.15196fqfjh2gts21@webmail.itu.edu.tr>

----- raju.rajsand at gmail.com den ileti ---------
   Tarih: Thu, 13 Aug 2009 08:57:15 +0530
  Kimden:  Rajagopal Swaminathan <raju.rajsand at gmail.com>
Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
    Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
    Kime: linux clustering <linux-cluster at redhat.com>


> Greetings,
>
> 2009/8/12 Hakan VELIOGLU <veliogluh at itu.edu.tr>:
>> Hi list,
>>
>> I am trying a two node cluster with RH 5.3 on Sun X4150 hardware. I use a
>
> IIRC, Sun x4150 has four ethernet ports. Two can be used for outside
> networking and two can be bonded and used for heartbeat.
I think, I couldn't explain my networking. I use two ethernet ports  
for xen vm which are trunk and bonded ports. Then there left two. Our  
network topology (which is out of my control) available for one port  
for server control (SSH). I use the other one with a cross over cable  
for heartbeat. So there is no way for bonding these two interfaces. Of  
course if I buy an extra switch I may do this.
I don't realy understand why there is no backup heartbeat channel. LVS  
and MS cluster has this ability.
>
> ALOM can be used for fencing and can be on a seperate subnet if required.
I used this for fencing_ipmilan.
>
> Regards
>
> Rajagopal
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>


----- raju.rajsand at gmail.com den iletiyi bitir -----





From brem.belguebli at gmail.com  Thu Aug 13 08:20:09 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Thu, 13 Aug 2009 10:20:09 +0200
Subject: [Linux-cluster] Qdisk question
In-Reply-To: <29ae894c0908121545r3a92cf8fk59d5c10c162cdfad@mail.gmail.com>
References: <29ae894c0908121545r3a92cf8fk59d5c10c162cdfad@mail.gmail.com>
Message-ID: <29ae894c0908130120l3840282av205bf54d793b7325@mail.gmail.com>

Hi,

I have a question about the qdisk concept.


My understanding of qdisk is that it is used as a tie-breaker, but it looks
like it is more a heatbeat vector than a simple tie-breaker.


My setup consists of 4 nodes located on 2 different production sites (2+2)
using SAN shared storage (2 disk frames, 1 per site).


The qdisk is a iscsi shared lun from a third site that I expected to use as
a tie-breaker in case 1 of my 2 prod sites was experiencing network problems
and gets completely isolated.


Until here, no real problem indeed, if the site gets apart from the other
prod site and also from the third site (hosting the iscsi target qdisk) the
2 nodes from the failing site get evicted from the cluster.


But, what if my third site gets isolated while the 2 prod ones are fine ?


The real  question is what happens in case all the nodes loose access to the
qdisk while they're still able to see each others ?


The 4 nodes have each 1 vote and the qdisk 1 vote. The expected quorum is 3.


When the cluster is running with all of its nodes and the qdisk, the number
of votes is 5.


If I loose the qdisk, the number of votes falls to 4, the cluster is quorate
(4>3) but it looks like everything goes bad, each node deactivate itself as
it can't write its alive status (--> heartbeat vector) to the qdisk even if
the network heartbeating is working fine.


I have tried to configure heuristics (ping a node on the third site) without
qdisk device but they seem to be ignored.


Any comments or tips ?


Regards
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090813/60119fd0/attachment.htm>

From robejrm at gmail.com  Thu Aug 13 08:42:11 2009
From: robejrm at gmail.com (Juan Ramon Martin Blanco)
Date: Thu, 13 Aug 2009 10:42:11 +0200
Subject: [Linux-cluster] Is there any backup heartbeat channel
In-Reply-To: <20090813111401.15196fqfjh2gts21@webmail.itu.edu.tr>
References: <4A7FA8B7.5080708@oracle.com> <1249896553.19917.2.camel@marc>
	<4A7FEC6F.2050405@oracle.com> <4A7FEE8D.8070400@oracle.com>
	<1249902500.20848.11.camel@marc> <4A827B23.7050107@oracle.com>
	<20090812120857.14055bi0l6l01okp@webmail.itu.edu.tr>
	<8786b91c0908122027g79716885j7ed933c3287c889b@mail.gmail.com>
	<20090813111401.15196fqfjh2gts21@webmail.itu.edu.tr>
Message-ID: <8a5668960908130142x58e5f393y157c46ff89097bbe@mail.gmail.com>

2009/8/13 Hakan VELIOGLU <veliogluh at itu.edu.tr>

> ----- raju.rajsand at gmail.com den ileti ---------
>  Tarih: Thu, 13 Aug 2009 08:57:15 +0530
>  Kimden:  Rajagopal Swaminathan <raju.rajsand at gmail.com>
> Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
>   Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
>   Kime: linux clustering <linux-cluster at redhat.com>
>
>
>  Greetings,
>>
>> 2009/8/12 Hakan VELIOGLU <veliogluh at itu.edu.tr>:
>>
>>> Hi list,
>>>
>>> I am trying a two node cluster with RH 5.3 on Sun X4150 hardware. I use a
>>>
>>
>> IIRC, Sun x4150 has four ethernet ports. Two can be used for outside
>> networking and two can be bonded and used for heartbeat.
>>
> I think, I couldn't explain my networking. I use two ethernet ports for xen
> vm which are trunk and bonded ports. Then there left two. Our network
> topology (which is out of my control) available for one port for server
> control (SSH).

So you can't use a bonded port for both server management  and cluster
communications, can you? You can configure an active-passive bonding and
then you can have many virtual interfaces on top of that, i.e: bond0:0,
bond0:1 and assign them the ip addesses you need.


I use the other one with a cross over cable for heartbeat. So there is no
> way for bonding these two interfaces. Of course if I buy an extra switch I
> may do this.

You can connect them to the same switch (though you lost kind of
redundancy), or you can use two crossover cables and move the management IP
to the same ports you are using for the vm's.

Greetings,
Juanra

>
> I don't realy understand why there is no backup heartbeat channel. LVS and
> MS cluster has this ability.
>
>>
>> ALOM can be used for fencing and can be on a seperate subnet if required.
>>
> I used this for fencing_ipmilan.
>
>>
>> Regards
>>
>> Rajagopal
>>
>> --
>> Linux-cluster mailing list
>> Linux-cluster at redhat.com
>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>
>>
>
> ----- raju.rajsand at gmail.com den iletiyi bitir -----
>
>
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090813/9bd1ca71/attachment.htm>

From brem.belguebli at gmail.com  Thu Aug 13 09:48:01 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Thu, 13 Aug 2009 11:48:01 +0200
Subject: [Linux-cluster] Qdisk question
Message-ID: <29ae894c0908130248k2d139f1cgd762e141dd1ef44e@mail.gmail.com>

Hi,
I have a question about the qdisk concept.

My understanding of qdisk is that it is used as a tie-breaker, but it looks
like it is more a heatbeat vector than a simple tie-breaker.

My setup consists of 4 nodes located on 2 different production sites (2+2)
using SAN shared storage (2 disk frames, 1 per site).

The qdisk is a iscsi shared lun from a third site that I expected to use as
a tie-breaker in case 1 of my 2 prod sites was experiencing network problems
and gets completely isolated.

Until here, no real problem indeed, if the site gets apart from the other
prod site and also from the third site (hosting the iscsi target qdisk) the
2 nodes from the failing site get evicted from the cluster.

But, what if my third site gets isolated while the 2 prod ones are fine ?

The real  question is what happens in case all the nodes loose access to the
qdisk while they're still able to see each others ?

The 4 nodes have each 1 vote and the qdisk 1 vote. The expected quorum is 3.

When the cluster is running with all of its nodes and the qdisk, the number
of votes is 5.

If I loose the qdisk, the number of votes falls to 4, the cluster is quorate
(4>3) but it looks like everything goes bad, each node deactivate itself as
it can't write its alive status (--> heartbeat vector) to the qdisk even if
the network heartbeating is working fine.

I have tried to configure heuristics (ping a node on the third site) without
qdisk device but they seem to be ignored.

Any comments or tips ?

Regards

PS: added the [Linux-cluster] flag :-)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090813/516db0b4/attachment.htm>

From esggrupos at gmail.com  Thu Aug 13 11:27:16 2009
From: esggrupos at gmail.com (ESGLinux)
Date: Thu, 13 Aug 2009 13:27:16 +0200
Subject: [Linux-cluster] do I have a fence DRAC device?
In-Reply-To: <27649.1250005179@mirchi>
References: <3128ba140908100324l6cdb4c34ra5f5edb39c6903e9@mail.gmail.com>
	<8b711df40908101134t69b8e12cof6cc551809421e45@mail.gmail.com>
	<3128ba140908110503m27532c62rce2844564c35da06@mail.gmail.com>
	<8a5668960908110514u4703a951qa49938375d0a7e52@mail.gmail.com>
	<27649.1250005179@mirchi>
Message-ID: <3128ba140908130427i6ab85406ye6da34073e6a6e97@mail.gmail.com>

Hi,
I couldn?t reboot my system yet but I have installed the openmanage
packages:

srvadmin-omacore-5.4.0-260
srvadmin-iws-5.4.0-260
srvadmin-syscheck-5.4.0-260
srvadmin-rac5-components-5.4.0-260
srvadmin-deng-5.4.0-260
srvadmin-ipmi-5.4.0-260.DUP
srvadmin-racadm5-5.4.0-260
srvadmin-omauth-5.4.0-260.rhel5
srvadmin-hapi-5.4.0-260
srvadmin-cm-5.4.0-260
srvadmin-racdrsc5-5.4.0-260
srvadmin-omilcore-5.4.0-260
srvadmin-isvc-5.4.0-260
srvadmin-storage-5.4.0-260
srvadmin-jre-5.4.0-260
srvadmin-omhip-5.4.0-260

Now I have the command racadm but when I try to execut it I get this:

racadm config -g cfgSerial -o cfgSerialTelnetEnable 1
ERROR: RACADM is unable to process the requested subcommand because there is
no
local RAC configuration to communicate with.

Local RACADM subcommand execution requires the following:

 1. A Remote Access Controller (RAC) must be present on the managed server
 2. Appropriate managed node software must be installed and running on the
    server


What do I need to install/start? or until I configure the bios I can?t get
this work?

Greetings

ESG


2009/8/11 <bergman at merctech.com>

>
>
> In the message dated: Tue, 11 Aug 2009 14:14:03 +0200,
> The pithy ruminations from Juan Ramon Martin Blanco on
> <Re: [Linux-cluster] do I have a fence DRAC device?> were:
> => --===============1917368601==
> => Content-Type: multipart/alternative;
> boundary=0016364c7c07663f600470dca3b8
> =>
> => --0016364c7c07663f600470dca3b8
> => Content-Type: text/plain; charset=ISO-8859-1
> => Content-Transfer-Encoding: quoted-printable
> =>
> => On Tue, Aug 11, 2009 at 2:03 PM, ESGLinux <esggrupos at gmail.com> wrote:
> =>
> => > Thanks
> => > I=B4ll check it when I could reboot the server.
> => >
> => > greetings,
> => >
> => You have a BMC ipmi in the first network interface, it can be configured
> at
> => boot time (I don't  remember if inside the BIOS or pressing
> cntrl+something
> => during boot)
> =>
>
> Based on my notes, here's how I configured the DRAC interface on a Dell
> 1950
> for use as a fence device:
>
>        Configuring the card from Linux depending on the installation of
> Dell's
>        OMSA package. Once that's installed, use the following
> commands:
>
>                racadm config -g cfgSerial -o cfgSerialTelnetEnable 1
>                racadm config -g cfgLanNetworking -o cfgDNSRacName
> HOSTNAME_FOR_INTERFACE
>                racadm config -g cfgDNSDomainName DOMAINNAME_FOR_INTERFACE
>                racadm config -g cfgUserAdmin -o cfgUserAdminPassword -i 2
> PASSWORD
>                racadm config -g cfgNicEnable 1
>                racadm config -g cfgNicIpAddress WWW.XXX.YYY.ZZZ
>                racadm config -g cfgNicNetmask WWW.XXX.YYY.ZZZ
>                racadm config -g cfgNicGateway WWW.XXX.YYY.ZZZ
>                racadm config -g cfgNicUseDhcp 0
>
>
>        I also save a backup of the configuration with:
>
>                racadm getconfig -f ~/drac_config
>
>
> Hope this helps,
>
> Mark
>
> ----
> Mark Bergman                              voice: 215-662-7310
> mark.bergman at uphs.upenn.edu                 fax: 215-614-0266
> System Administrator     Section of Biomedical Image Analysis
> Department of Radiology            University of Pennsylvania
>      PGP Key: https://www.rad.upenn.edu/sbia/bergman
>
>
> => Greetings,
> => Juanra
> =>
> => >
> => > ESG
> => >
> => > 2009/8/10 Paras pradhan <pradhanparas at gmail.com>
> => >
> => > On Mon, Aug 10, 2009 at 5:24 AM, ESGLinux<esggrupos at gmail.com> wrote:
> => >> > Hi all,
> => >> > I was designing a 2 node cluster and I was going to use 2 servers
> DELL
> => >> > PowerEdge 1950. I was going to buy a DRAC card to use for fencing
> but
> => >> > running several commands in the servers I have noticed that when I
> run
> => >> this
> => >> > command:
> => >> > #ipmitool lan print
> => >> > Set in Progress : Set Complete
> => >> > Auth Type Support : NONE MD2 MD5 PASSWORD
> => >> > Auth Type Enable : Callback : MD2 MD5
> => >> >                         : User : MD2 MD5
> => >> >                         : Operator : MD2 MD5
> => >> >                         : Admin : MD2 MD5
> => >> >                         : OEM : MD2 MD5
> => >> > IP Address Source : Static Address
> => >> > IP Address : 0.0.0.0
> => >> > Subnet Mask : 0.0.0.0
> => >> > MAC Address : 00:1e:c9:ae:6f:7e
> => >> > SNMP Community String : public
> => >> > IP Header : TTL=0x40 Flags=0x40 Precedence=0x00 TOS=0x10
> => >> > Default Gateway IP : 0.0.0.0
> => >> > Default Gateway MAC : 00:00:00:00:00:00
> => >> > Backup Gateway IP : 0.0.0.0
> => >> > Backup Gateway MAC : 00:00:00:00:00:00
> => >> > 802.1q VLAN ID : Disabled
> => >> > 802.1q VLAN Priority : 0
> => >> > RMCP+ Cipher Suites : 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14
> => >> > Cipher Suite Priv Max : aaaaaaaaaaaaaaa
> => >> >                         : X=Cipher Suite Unused
> => >> >                         : c=CALLBACK
> => >> >                         : u=USER
> => >> >                         : o=OPERATOR
> => >> >                         : a=ADMIN
> => >> >                         : O=OEM
> => >> > does this mean that I already have an ipmi card (not configured)
> that
> => I
> => >> can
> => >> > use for fencing? if the anwser is yes, where hell must I configure
> it?
> =>  I
> => >> > don=B4t see wher can I do it.
> => >> > If I haven=B4t a fencing device which one do you recommed to use?
> => >> > Thanks in advance
> => >> > ESG
> => >> >
> => >> > --
> => >> > Linux-cluster mailing list
> => >> > Linux-cluster at redhat.com
> => >> > https://www.redhat.com/mailman/listinfo/linux-cluster
> => >> >
> => >>
> => >> Yes you have IPMI and if you are using 1950 Dell, DRAC should be
> there
> => >> too. You can see if you have DRAC or not when the server starts and
> => >> before the loading of the OS.
> => >>
> => >> I have 1850s and I am using DRAC for fencing.
> => >>
> => >>
> => >> Paras.
> => >>
> => >> --
> => >> Linux-cluster mailing list
> => >> Linux-cluster at redhat.com
> => >> https://www.redhat.com/mailman/listinfo/linux-cluster
> => >>
> => >
> => >
>
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090813/a4558d27/attachment.htm>

From mark.benschop.lists at gmail.com  Thu Aug 13 12:45:13 2009
From: mark.benschop.lists at gmail.com (mark benschop)
Date: Thu, 13 Aug 2009 14:45:13 +0200
Subject: [Linux-cluster] clusterservice stays in 'recovering' state
Message-ID: <f97c3a70908130545n11ce442ej17d74c9cdc450e45@mail.gmail.com>

Hi All,

I've a problem with a clusterservice. The service was started up while one
of the resources, an NFS, export was not accessible.
Therefore the service never started up right but got into the 'recovering'
state.
In the mean time the NFS exports are setup properly but to no avail.
Stopping the clusterservice, using clusvcadm -d <service>, will result in
the service going down but staying in the 'recovering' state.
Starting it again doesn't work. The service doesn't start and stays in the
recovery status.
I'm suspecting rgmanager lost it somehow.

Anybody had any ideas on what could be the problem and how to resolve it ?

Thanks in advance,
Mark
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090813/31731cd3/attachment.htm>

From veliogluh at itu.edu.tr  Thu Aug 13 13:13:12 2009
From: veliogluh at itu.edu.tr (Hakan VELIOGLU)
Date: Thu, 13 Aug 2009 16:13:12 +0300
Subject: [Linux-cluster] Is there any backup heartbeat channel
In-Reply-To: <8a5668960908130142x58e5f393y157c46ff89097bbe@mail.gmail.com>
References: <4A7FA8B7.5080708@oracle.com> <1249896553.19917.2.camel@marc>
	<4A7FEC6F.2050405@oracle.com> <4A7FEE8D.8070400@oracle.com>
	<1249902500.20848.11.camel@marc> <4A827B23.7050107@oracle.com>
	<20090812120857.14055bi0l6l01okp@webmail.itu.edu.tr>
	<8786b91c0908122027g79716885j7ed933c3287c889b@mail.gmail.com>
	<20090813111401.15196fqfjh2gts21@webmail.itu.edu.tr>
	<8a5668960908130142x58e5f393y157c46ff89097bbe@mail.gmail.com>
Message-ID: <20090813161312.11546h2sp6psr814@webmail.itu.edu.tr>

Thanks for all the answers.

I think there is realy no backup heartbeat channel. Maybe the reason  
is GFS. DLM works on the heartbeat channel. If you lost your heartbeat  
you lose your lock consistency so it is better to fence the other  
node. For this reason I think if you don't have enough network  
interface on server and switch, loosing the heartbeat network may shut  
all the cluster members.

Hakan VEL?O?LU


----- robejrm at gmail.com den ileti ---------
   Tarih: Thu, 13 Aug 2009 10:42:11 +0200
  Kimden:  Juan Ramon Martin Blanco <robejrm at gmail.com>
Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
    Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
    Kime: linux clustering <linux-cluster at redhat.com>


> 2009/8/13 Hakan VELIOGLU <veliogluh at itu.edu.tr>
>
>> ----- raju.rajsand at gmail.com den ileti ---------
>>  Tarih: Thu, 13 Aug 2009 08:57:15 +0530
>>  Kimden:  Rajagopal Swaminathan <raju.rajsand at gmail.com>
>> Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
>>   Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
>>   Kime: linux clustering <linux-cluster at redhat.com>
>>
>>
>>  Greetings,
>>>
>>> 2009/8/12 Hakan VELIOGLU <veliogluh at itu.edu.tr>:
>>>
>>>> Hi list,
>>>>
>>>> I am trying a two node cluster with RH 5.3 on Sun X4150 hardware. I use a
>>>>
>>>
>>> IIRC, Sun x4150 has four ethernet ports. Two can be used for outside
>>> networking and two can be bonded and used for heartbeat.
>>>
>> I think, I couldn't explain my networking. I use two ethernet ports for xen
>> vm which are trunk and bonded ports. Then there left two. Our network
>> topology (which is out of my control) available for one port for server
>> control (SSH).
>
> So you can't use a bonded port for both server management  and cluster
> communications, can you? You can configure an active-passive bonding and
> then you can have many virtual interfaces on top of that, i.e: bond0:0,
> bond0:1 and assign them the ip addesses you need.
>
>
> I use the other one with a cross over cable for heartbeat. So there is no
>> way for bonding these two interfaces. Of course if I buy an extra switch I
>> may do this.
>
> You can connect them to the same switch (though you lost kind of
> redundancy), or you can use two crossover cables and move the management IP
> to the same ports you are using for the vm's.
>
> Greetings,
> Juanra
>
>>
>> I don't realy understand why there is no backup heartbeat channel. LVS and
>> MS cluster has this ability.
>>
>>>
>>> ALOM can be used for fencing and can be on a seperate subnet if required.
>>>
>> I used this for fencing_ipmilan.
>>
>>>
>>> Regards
>>>
>>> Rajagopal
>>>
>>> --
>>> Linux-cluster mailing list
>>> Linux-cluster at redhat.com
>>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>>
>>>
>>
>> ----- raju.rajsand at gmail.com den iletiyi bitir -----
>>
>>
>>
>>
>> --
>> Linux-cluster mailing list
>> Linux-cluster at redhat.com
>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>
>


----- robejrm at gmail.com den iletiyi bitir -----





From robejrm at gmail.com  Thu Aug 13 13:29:43 2009
From: robejrm at gmail.com (Juan Ramon Martin Blanco)
Date: Thu, 13 Aug 2009 15:29:43 +0200
Subject: [Linux-cluster] Is there any backup heartbeat channel
In-Reply-To: <20090813161312.11546h2sp6psr814@webmail.itu.edu.tr>
References: <4A7FA8B7.5080708@oracle.com> <4A7FEC6F.2050405@oracle.com>
	<4A7FEE8D.8070400@oracle.com> <1249902500.20848.11.camel@marc>
	<4A827B23.7050107@oracle.com>
	<20090812120857.14055bi0l6l01okp@webmail.itu.edu.tr>
	<8786b91c0908122027g79716885j7ed933c3287c889b@mail.gmail.com>
	<20090813111401.15196fqfjh2gts21@webmail.itu.edu.tr>
	<8a5668960908130142x58e5f393y157c46ff89097bbe@mail.gmail.com>
	<20090813161312.11546h2sp6psr814@webmail.itu.edu.tr>
Message-ID: <8a5668960908130629n6ec05a88n463a3b03da331dae@mail.gmail.com>

2009/8/13 Hakan VELIOGLU <veliogluh at itu.edu.tr>

> Thanks for all the answers.
>
> I think there is realy no backup heartbeat channel. Maybe the reason is
> GFS. DLM works on the heartbeat channel. If you lost your heartbeat you lose
> your lock consistency so it is better to fence the other node. For this
> reason I think if you don't have enough network interface on server and
> switch, loosing the heartbeat network may shut all the cluster members.
>
There is no backup heartbeat channel because you should do the backup at a
operating system level, i.e: bonding
That's why you should use a bonded interface for the heartbeat channel with
at least 2 ethernet slaves; going further (for better redundancy) each of
the slaves should be on a different network card and you should connect the
each slave to a different switch.
But what I am trying to explain, is that you can use that bonded logical
interface also for things different from hearbeat. ;)

Greetings,
Juanra


> Hakan VEL?O?LU
>
>
> ----- robejrm at gmail.com den ileti ---------
>  Tarih: Thu, 13 Aug 2009 10:42:11 +0200
>  Kimden:  Juan Ramon Martin Blanco <robejrm at gmail.com>
>
> Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
>   Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
>   Kime: linux clustering <linux-cluster at redhat.com>
>
>
>  2009/8/13 Hakan VELIOGLU <veliogluh at itu.edu.tr>
>>
>>  ----- raju.rajsand at gmail.com den ileti ---------
>>>  Tarih: Thu, 13 Aug 2009 08:57:15 +0530
>>>  Kimden:  Rajagopal Swaminathan <raju.rajsand at gmail.com>
>>> Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
>>>  Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
>>>  Kime: linux clustering <linux-cluster at redhat.com>
>>>
>>>
>>>  Greetings,
>>>
>>>>
>>>> 2009/8/12 Hakan VELIOGLU <veliogluh at itu.edu.tr>:
>>>>
>>>>  Hi list,
>>>>>
>>>>> I am trying a two node cluster with RH 5.3 on Sun X4150 hardware. I use
>>>>> a
>>>>>
>>>>>
>>>> IIRC, Sun x4150 has four ethernet ports. Two can be used for outside
>>>> networking and two can be bonded and used for heartbeat.
>>>>
>>>>  I think, I couldn't explain my networking. I use two ethernet ports for
>>> xen
>>> vm which are trunk and bonded ports. Then there left two. Our network
>>> topology (which is out of my control) available for one port for server
>>> control (SSH).
>>>
>>
>> So you can't use a bonded port for both server management  and cluster
>> communications, can you? You can configure an active-passive bonding and
>> then you can have many virtual interfaces on top of that, i.e: bond0:0,
>> bond0:1 and assign them the ip addesses you need.
>>
>>
>> I use the other one with a cross over cable for heartbeat. So there is no
>>
>>> way for bonding these two interfaces. Of course if I buy an extra switch
>>> I
>>> may do this.
>>>
>>
>> You can connect them to the same switch (though you lost kind of
>> redundancy), or you can use two crossover cables and move the management
>> IP
>> to the same ports you are using for the vm's.
>>
>> Greetings,
>> Juanra
>>
>>
>>> I don't realy understand why there is no backup heartbeat channel. LVS
>>> and
>>> MS cluster has this ability.
>>>
>>>
>>>> ALOM can be used for fencing and can be on a seperate subnet if
>>>> required.
>>>>
>>>>  I used this for fencing_ipmilan.
>>>
>>>
>>>> Regards
>>>>
>>>> Rajagopal
>>>>
>>>> --
>>>> Linux-cluster mailing list
>>>> Linux-cluster at redhat.com
>>>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>>>
>>>>
>>>>
>>> ----- raju.rajsand at gmail.com den iletiyi bitir -----
>>>
>>>
>>>
>>>
>>> --
>>> Linux-cluster mailing list
>>> Linux-cluster at redhat.com
>>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>>
>>>
>>
>
> ----- robejrm at gmail.com den iletiyi bitir -----
>
>
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090813/cd4dc079/attachment.htm>

From nehemiasjahcob at gmail.com  Thu Aug 13 14:07:13 2009
From: nehemiasjahcob at gmail.com (Nehemias Jahcob)
Date: Thu, 13 Aug 2009 10:07:13 -0400
Subject: [Linux-cluster] RHCS on KVM
Message-ID: <5f61ab380908130707q5c936504k7351d0d6b3459090@mail.gmail.com>

Hi.

 How to create a cluster of 2 nodes in rhel5.4 (or Fedora 10) with KVM?

With XEN follow this guide:
http://sources.redhat.com/cluster/wiki/VMClusterCookbook?highlight =
(CategoryHowTo).

Do you have a guide to implementation of RHCS in KVM?

Thank you all.
NJ
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090813/f3a69a80/attachment.htm>

From ntoughe at hotmail.com  Thu Aug 13 14:16:47 2009
From: ntoughe at hotmail.com (NTOUGHE GUY-SERGE)
Date: Thu, 13 Aug 2009 14:16:47 +0000
Subject: [Linux-cluster] Cman hang
Message-ID: <BAY119-W410E2F250E8B461752CFC9A5050@phx.gbl>



Hi gurus,

i installed  RHEL 5.3 on 2 servers which participating to a cluster composed of these 2 nodes:
kernel version:
kernel-headers-2.6.18-128.el5
kernel-devel-2.6.18-128.el5
kernel-2.6.18-128.el5
cman-devel-2.0.98-1.el5_3.1
cman-2.0.98-1.el5_3.1
cluster-cim-0.12.1-2.el5
lvm2-cluster-2.02.40-7.el5
cluster-snmp-0.12.1-2.el5
modcluster-0.12.1-2.el5
When i want to start cman the following message is sent:
cman not started: Multicast and node address families differ. /usr/sbin/cman_tool: aisexec daemon didn't start
                                                           [FAILED]

I trier to mount gfs2
and i got theses messages:
# mount -t gfs2 /dev/VolGroup01/LogVol01 /appli/prod --o lockTablename=arvclust:/appli/prod, Lockproto=lock_dlm

/sbin/mount.gfs2: can't connect to gfs_controld: Connection refused

/sbin/mount.gfs2: can't connect to gfs_controld: Connection refused

/sbin/mount.gfs2: can't connect to gfs_controld: Connection refused

/sbin/mount.gfs2: can't connect to gfs_controld: Connection refused

do you have any clues?
Please it's an hurry, i waste long time to lok for solution help
regards








ntoughe at hotmail.com


_________________________________________________________________
With Windows Live, you can organize, edit, and share your photos.
http://www.microsoft.com/middleeast/windows/windowslive/products/photo-gallery-edit.aspx
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090813/0a55101d/attachment.htm>

From teigland at redhat.com  Thu Aug 13 14:14:24 2009
From: teigland at redhat.com (David Teigland)
Date: Thu, 13 Aug 2009 09:14:24 -0500
Subject: [Linux-cluster] gfs2 mount hangs
In-Reply-To: <4A83B113.60609@oracle.com>
References: <4A83B113.60609@oracle.com>
Message-ID: <20090813141424.GA8148@redhat.com>

On Thu, Aug 13, 2009 at 02:22:11PM +0800, Wengang Wang wrote:
> <cman two_node="1" expected_votes="2"/>

That's not a valid combination, two_node="1" requires expected_votes="1".

You didn't mention which userspace cluster version/release you're using, or
include any status about the cluster.  Before trying to mount gfs on either
node, collect from both nodes:

  cman_tool status
  cman_tool nodes
  group_tool

Then mount on the first node and collect the same information, then try
mounting on the second node, collect the same information, and look for any
errors in /var/log/messages.

Since you're using new kernels, you need to be using the cluster 3.0 userspace
code.  You're using the old manual fencing config.  There is no more
fence_manual; the new way to configure manual fencing is to not configure any
fencing at all.  So, your cluster.conf should look like this:

<?xml version="1.0"?>
<cluster name="testgfs2" config_version="1">
<cman two_node="1" expected_votes="1"/>
<clusternodes>
<clusternode name="cool" nodeid="1"/>
<clusternode name="desk" nodeid="2"/>
</clusternodes>
</cluster>

Dave



From lhh at redhat.com  Thu Aug 13 14:39:46 2009
From: lhh at redhat.com (Lon Hohberger)
Date: Thu, 13 Aug 2009 10:39:46 -0400
Subject: [Linux-cluster] Qdisk question
In-Reply-To: <29ae894c0908121545r3a92cf8fk59d5c10c162cdfad@mail.gmail.com>
References: <29ae894c0908121545r3a92cf8fk59d5c10c162cdfad@mail.gmail.com>
Message-ID: <1250174386.23376.1440.camel@localhost.localdomain>

On Thu, 2009-08-13 at 00:45 +0200, brem belguebli wrote:

> My understanding of qdisk is that it is used as a tie-breaker, but it
> looks like it is more a heatbeat vector than a simple tie-breaker.

Right, it's a secondary membership algorithm.


> Until here, no real problem indeed, if the site gets apart from the
> other prod site and also from the third site (hosting the iscsi target
> qdisk) the 2 nodes from the failing site get evicted from the cluster.
> 
> 
> But, what if my third site gets isolated while the 2 prod ones are
> fine ?

Qdisk votes will not be presented to CMAN any more, but the two sites
should remain online if they still have a "majority" of votes.


> The real  question is what happens in case all the nodes loose access
> to the qdisk while they're still able to see each others ?

Qdisk is just a vote like other voting mechanisms.  If all nodes lose
access at the same time, it should behave like a node death.  However,
the default action if _one_ node loses access is to kill that node (even
if CMAN still sees it).


> The 4 nodes have each 1 vote and the qdisk 1 vote. The expected quorum
> is 3.


> If I loose the qdisk, the number of votes falls to 4, the cluster is
> quorate (4>3) but it looks like everything goes bad, each node
> deactivate itself as it can't write its alive status (--> heartbeat
> vector) to the qdisk even if the network heartbeating is working
> fine. 

What happens specifically?  Most of the actions qdiskd performs are
configurable.  For example, if the nodes are rebooting, you can turn
that behavior off.



I wrote a simple 'ping' tiebreaker based the behaviors in RHEL3.  It
functions in many ways in the same manner as qdiskd with respect to vote
advertisement to CMAN, but without needing a disk - maybe you would find
it useful?

http://people.redhat.com/lhh/qnet.tar.gz

-- Lon



From robejrm at gmail.com  Thu Aug 13 15:02:15 2009
From: robejrm at gmail.com (Juan Ramon Martin Blanco)
Date: Thu, 13 Aug 2009 17:02:15 +0200
Subject: [Linux-cluster] Cman hang
In-Reply-To: <BAY119-W410E2F250E8B461752CFC9A5050@phx.gbl>
References: <BAY119-W410E2F250E8B461752CFC9A5050@phx.gbl>
Message-ID: <8a5668960908130802p4f5168cbueda86d1e6f1324bb@mail.gmail.com>

On Thu, Aug 13, 2009 at 4:16 PM, NTOUGHE GUY-SERGE <ntoughe at hotmail.com>wrote:

>
> Hi gurus,
>
> i installed  RHEL 5.3 on 2 servers which participating to a cluster
> composed of these 2 nodes:
> kernel version:
> kernel-headers-2.6.18-128.el5
> kernel-devel-2.6.18-128.el5
> kernel-2.6.18-128.el5
> cman-devel-2.0.98-1.el5_3.1
> cman-2.0.98-1.el5_3.1
> cluster-cim-0.12.1-2.el5
> lvm2-cluster-2.02.40-7.el5
> cluster-snmp-0.12.1-2.el5
> modcluster-0.12.1-2.el5
> When i want to start cman the following message is sent:
> cman not started: Multicast and node address families differ.
> /usr/sbin/cman_tool: aisexec daemon didn't start
>   [FAILED]
>
Please, show us your cluster.conf file so we can help.

Regards,
Juanra

>
> I trier to mount gfs2
> and i got theses messages:
> # mount -t gfs2 /dev/VolGroup01/LogVol01 /appli/prod --o
> lockTablename=arvclust:/appli/prod, Lockproto=lock_dlm
>
> /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
>
> /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
>
> /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
>
> /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
>
> do you have any clues?
> Please it's an hurry, i waste long time to lok for solution help
> regards
>
>
>
>
>
>
>
>
> ntoughe at hotmail.com
>
>
>
> ------------------------------
> With Windows Live, you can organize, edit, and share your photos.<http://www.microsoft.com/middleeast/windows/windowslive/products/photo-gallery-edit.aspx>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090813/9ecbcab1/attachment.htm>

From ntoughe at hotmail.com  Thu Aug 13 15:13:32 2009
From: ntoughe at hotmail.com (NTOUGHE GUY-SERGE)
Date: Thu, 13 Aug 2009 15:13:32 +0000
Subject: [Linux-cluster] Cman hang
In-Reply-To: <20090813150236.B2D7461A5CE@hormel.redhat.com>
References: <20090813150236.B2D7461A5CE@hormel.redhat.com>
Message-ID: <BAY119-W294291EA707ECD00D02139A5050@phx.gbl>


Hi, this is my cluster.conf
<?xml version="1.0"?>
<cluster alias="arevclust" config_version="21" name="arevclust">
        <fence_daemon clean_start="0" post_fail_delay="0" post_join_delay="3"/>
        <clusternodes>
                <clusternode name="host1" nodeid="1" votes="1">
                        <fence>
                                <method name="2">
                                        <device name=""/>
                                </method>
                        </fence>
                        <multicast addr="" interface=""/>
                </clusternode>
                <clusternode name="host2" nodeid="2" votes="1">
                        <fence>
                                <method name="1">
                                        <device name=""/>
                                </method>
                                <method name=""/>
                        </fence>
                        <multicast addr="" interface=""/>
                </clusternode>
        </clusternodes>
        <cman expected_votes="" two_node="">
                <multicast addr=""/>
        </cman>
        <fencedevices>
                <fencedevice agent="fence_brocade" ipaddr="" login="" name="" passwd=""/>
        </fencedevices>
        <rm>
                <failoverdomains>
                </failoverdomains>
                <resources>
                </resources>
        </rm>
</cluster>

Regards




ntoughe at hotmail.com



> From: linux-cluster-request at redhat.com
> Subject: Linux-cluster Digest, Vol 64, Issue 16
> To: linux-cluster at redhat.com
> Date: Thu, 13 Aug 2009 11:02:36 -0400
> 
> Send Linux-cluster mailing list submissions to
> 	linux-cluster at redhat.com
> 
> To subscribe or unsubscribe via the World Wide Web, visit
> 	https://www.redhat.com/mailman/listinfo/linux-cluster
> or, via email, send a message with subject or body 'help' to
> 	linux-cluster-request at redhat.com
> 
> You can reach the person managing the list at
> 	linux-cluster-owner at redhat.com
> 
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Linux-cluster digest..."
> 
> 
> Today's Topics:
> 
>    1. Re: do I have a fence DRAC device? (ESGLinux)
>    2. clusterservice stays in 'recovering' state (mark benschop)
>    3. Re: Is there any backup heartbeat channel (Hakan VELIOGLU)
>    4. Re: Is there any backup heartbeat channel
>       (Juan Ramon Martin Blanco)
>    5. RHCS on KVM (Nehemias Jahcob)
>    6. Cman hang (NTOUGHE GUY-SERGE)
>    7. Re: gfs2 mount hangs (David Teigland)
>    8. Re: Qdisk question (Lon Hohberger)
>    9. Re: Cman hang (Juan Ramon Martin Blanco)
> 
> 
> ----------------------------------------------------------------------
> 
> Message: 1
> Date: Thu, 13 Aug 2009 13:27:16 +0200
> From: ESGLinux <esggrupos at gmail.com>
> Subject: Re: [Linux-cluster] do I have a fence DRAC device?
> To: linux clustering <linux-cluster at redhat.com>
> Message-ID:
> 	<3128ba140908130427i6ab85406ye6da34073e6a6e97 at mail.gmail.com>
> Content-Type: text/plain; charset="iso-8859-1"
> 
> Hi,
> I couldn?t reboot my system yet but I have installed the openmanage
> packages:
> 
> srvadmin-omacore-5.4.0-260
> srvadmin-iws-5.4.0-260
> srvadmin-syscheck-5.4.0-260
> srvadmin-rac5-components-5.4.0-260
> srvadmin-deng-5.4.0-260
> srvadmin-ipmi-5.4.0-260.DUP
> srvadmin-racadm5-5.4.0-260
> srvadmin-omauth-5.4.0-260.rhel5
> srvadmin-hapi-5.4.0-260
> srvadmin-cm-5.4.0-260
> srvadmin-racdrsc5-5.4.0-260
> srvadmin-omilcore-5.4.0-260
> srvadmin-isvc-5.4.0-260
> srvadmin-storage-5.4.0-260
> srvadmin-jre-5.4.0-260
> srvadmin-omhip-5.4.0-260
> 
> Now I have the command racadm but when I try to execut it I get this:
> 
> racadm config -g cfgSerial -o cfgSerialTelnetEnable 1
> ERROR: RACADM is unable to process the requested subcommand because there is
> no
> local RAC configuration to communicate with.
> 
> Local RACADM subcommand execution requires the following:
> 
>  1. A Remote Access Controller (RAC) must be present on the managed server
>  2. Appropriate managed node software must be installed and running on the
>     server
> 
> 
> What do I need to install/start? or until I configure the bios I can?t get
> this work?
> 
> Greetings
> 
> ESG
> 
> 
> 2009/8/11 <bergman at merctech.com>
> 
> >
> >
> > In the message dated: Tue, 11 Aug 2009 14:14:03 +0200,
> > The pithy ruminations from Juan Ramon Martin Blanco on
> > <Re: [Linux-cluster] do I have a fence DRAC device?> were:
> > => --===============1917368601==
> > => Content-Type: multipart/alternative;
> > boundary=0016364c7c07663f600470dca3b8
> > =>
> > => --0016364c7c07663f600470dca3b8
> > => Content-Type: text/plain; charset=ISO-8859-1
> > => Content-Transfer-Encoding: quoted-printable
> > =>
> > => On Tue, Aug 11, 2009 at 2:03 PM, ESGLinux <esggrupos at gmail.com> wrote:
> > =>
> > => > Thanks
> > => > I=B4ll check it when I could reboot the server.
> > => >
> > => > greetings,
> > => >
> > => You have a BMC ipmi in the first network interface, it can be configured
> > at
> > => boot time (I don't  remember if inside the BIOS or pressing
> > cntrl+something
> > => during boot)
> > =>
> >
> > Based on my notes, here's how I configured the DRAC interface on a Dell
> > 1950
> > for use as a fence device:
> >
> >        Configuring the card from Linux depending on the installation of
> > Dell's
> >        OMSA package. Once that's installed, use the following
> > commands:
> >
> >                racadm config -g cfgSerial -o cfgSerialTelnetEnable 1
> >                racadm config -g cfgLanNetworking -o cfgDNSRacName
> > HOSTNAME_FOR_INTERFACE
> >                racadm config -g cfgDNSDomainName DOMAINNAME_FOR_INTERFACE
> >                racadm config -g cfgUserAdmin -o cfgUserAdminPassword -i 2
> > PASSWORD
> >                racadm config -g cfgNicEnable 1
> >                racadm config -g cfgNicIpAddress WWW.XXX.YYY.ZZZ
> >                racadm config -g cfgNicNetmask WWW.XXX.YYY.ZZZ
> >                racadm config -g cfgNicGateway WWW.XXX.YYY.ZZZ
> >                racadm config -g cfgNicUseDhcp 0
> >
> >
> >        I also save a backup of the configuration with:
> >
> >                racadm getconfig -f ~/drac_config
> >
> >
> > Hope this helps,
> >
> > Mark
> >
> > ----
> > Mark Bergman                              voice: 215-662-7310
> > mark.bergman at uphs.upenn.edu                 fax: 215-614-0266
> > System Administrator     Section of Biomedical Image Analysis
> > Department of Radiology            University of Pennsylvania
> >      PGP Key: https://www.rad.upenn.edu/sbia/bergman
> >
> >
> > => Greetings,
> > => Juanra
> > =>
> > => >
> > => > ESG
> > => >
> > => > 2009/8/10 Paras pradhan <pradhanparas at gmail.com>
> > => >
> > => > On Mon, Aug 10, 2009 at 5:24 AM, ESGLinux<esggrupos at gmail.com> wrote:
> > => >> > Hi all,
> > => >> > I was designing a 2 node cluster and I was going to use 2 servers
> > DELL
> > => >> > PowerEdge 1950. I was going to buy a DRAC card to use for fencing
> > but
> > => >> > running several commands in the servers I have noticed that when I
> > run
> > => >> this
> > => >> > command:
> > => >> > #ipmitool lan print
> > => >> > Set in Progress : Set Complete
> > => >> > Auth Type Support : NONE MD2 MD5 PASSWORD
> > => >> > Auth Type Enable : Callback : MD2 MD5
> > => >> >                         : User : MD2 MD5
> > => >> >                         : Operator : MD2 MD5
> > => >> >                         : Admin : MD2 MD5
> > => >> >                         : OEM : MD2 MD5
> > => >> > IP Address Source : Static Address
> > => >> > IP Address : 0.0.0.0
> > => >> > Subnet Mask : 0.0.0.0
> > => >> > MAC Address : 00:1e:c9:ae:6f:7e
> > => >> > SNMP Community String : public
> > => >> > IP Header : TTL=0x40 Flags=0x40 Precedence=0x00 TOS=0x10
> > => >> > Default Gateway IP : 0.0.0.0
> > => >> > Default Gateway MAC : 00:00:00:00:00:00
> > => >> > Backup Gateway IP : 0.0.0.0
> > => >> > Backup Gateway MAC : 00:00:00:00:00:00
> > => >> > 802.1q VLAN ID : Disabled
> > => >> > 802.1q VLAN Priority : 0
> > => >> > RMCP+ Cipher Suites : 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14
> > => >> > Cipher Suite Priv Max : aaaaaaaaaaaaaaa
> > => >> >                         : X=Cipher Suite Unused
> > => >> >                         : c=CALLBACK
> > => >> >                         : u=USER
> > => >> >                         : o=OPERATOR
> > => >> >                         : a=ADMIN
> > => >> >                         : O=OEM
> > => >> > does this mean that I already have an ipmi card (not configured)
> > that
> > => I
> > => >> can
> > => >> > use for fencing? if the anwser is yes, where hell must I configure
> > it?
> > =>  I
> > => >> > don=B4t see wher can I do it.
> > => >> > If I haven=B4t a fencing device which one do you recommed to use?
> > => >> > Thanks in advance
> > => >> > ESG
> > => >> >
> > => >> > --
> > => >> > Linux-cluster mailing list
> > => >> > Linux-cluster at redhat.com
> > => >> > https://www.redhat.com/mailman/listinfo/linux-cluster
> > => >> >
> > => >>
> > => >> Yes you have IPMI and if you are using 1950 Dell, DRAC should be
> > there
> > => >> too. You can see if you have DRAC or not when the server starts and
> > => >> before the loading of the OS.
> > => >>
> > => >> I have 1850s and I am using DRAC for fencing.
> > => >>
> > => >>
> > => >> Paras.
> > => >>
> > => >> --
> > => >> Linux-cluster mailing list
> > => >> Linux-cluster at redhat.com
> > => >> https://www.redhat.com/mailman/listinfo/linux-cluster
> > => >>
> > => >
> > => >
> >
> >
> >
> > --
> > Linux-cluster mailing list
> > Linux-cluster at redhat.com
> > https://www.redhat.com/mailman/listinfo/linux-cluster
> >
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: https://www.redhat.com/archives/linux-cluster/attachments/20090813/a4558d27/attachment.html
> 
> ------------------------------
> 
> Message: 2
> Date: Thu, 13 Aug 2009 14:45:13 +0200
> From: mark benschop <mark.benschop.lists at gmail.com>
> Subject: [Linux-cluster] clusterservice stays in 'recovering' state
> To: linux-cluster at redhat.com
> Message-ID:
> 	<f97c3a70908130545n11ce442ej17d74c9cdc450e45 at mail.gmail.com>
> Content-Type: text/plain; charset="iso-8859-1"
> 
> Hi All,
> 
> I've a problem with a clusterservice. The service was started up while one
> of the resources, an NFS, export was not accessible.
> Therefore the service never started up right but got into the 'recovering'
> state.
> In the mean time the NFS exports are setup properly but to no avail.
> Stopping the clusterservice, using clusvcadm -d <service>, will result in
> the service going down but staying in the 'recovering' state.
> Starting it again doesn't work. The service doesn't start and stays in the
> recovery status.
> I'm suspecting rgmanager lost it somehow.
> 
> Anybody had any ideas on what could be the problem and how to resolve it ?
> 
> Thanks in advance,
> Mark
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: https://www.redhat.com/archives/linux-cluster/attachments/20090813/31731cd3/attachment.html
> 
> ------------------------------
> 
> Message: 3
> Date: Thu, 13 Aug 2009 16:13:12 +0300
> From: Hakan VELIOGLU <veliogluh at itu.edu.tr>
> Subject: Re: [Linux-cluster] Is there any backup heartbeat channel
> To: linux-cluster at redhat.com
> Message-ID: <20090813161312.11546h2sp6psr814 at webmail.itu.edu.tr>
> Content-Type: text/plain; charset=ISO-8859-9; DelSp="Yes";
> 	format="flowed"
> 
> Thanks for all the answers.
> 
> I think there is realy no backup heartbeat channel. Maybe the reason  
> is GFS. DLM works on the heartbeat channel. If you lost your heartbeat  
> you lose your lock consistency so it is better to fence the other  
> node. For this reason I think if you don't have enough network  
> interface on server and switch, loosing the heartbeat network may shut  
> all the cluster members.
> 
> Hakan VEL?O?LU
> 
> 
> ----- robejrm at gmail.com den ileti ---------
>    Tarih: Thu, 13 Aug 2009 10:42:11 +0200
>   Kimden:  Juan Ramon Martin Blanco <robejrm at gmail.com>
> Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
>     Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
>     Kime: linux clustering <linux-cluster at redhat.com>
> 
> 
> > 2009/8/13 Hakan VELIOGLU <veliogluh at itu.edu.tr>
> >
> >> ----- raju.rajsand at gmail.com den ileti ---------
> >>  Tarih: Thu, 13 Aug 2009 08:57:15 +0530
> >>  Kimden:  Rajagopal Swaminathan <raju.rajsand at gmail.com>
> >> Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
> >>   Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
> >>   Kime: linux clustering <linux-cluster at redhat.com>
> >>
> >>
> >>  Greetings,
> >>>
> >>> 2009/8/12 Hakan VELIOGLU <veliogluh at itu.edu.tr>:
> >>>
> >>>> Hi list,
> >>>>
> >>>> I am trying a two node cluster with RH 5.3 on Sun X4150 hardware. I use a
> >>>>
> >>>
> >>> IIRC, Sun x4150 has four ethernet ports. Two can be used for outside
> >>> networking and two can be bonded and used for heartbeat.
> >>>
> >> I think, I couldn't explain my networking. I use two ethernet ports for xen
> >> vm which are trunk and bonded ports. Then there left two. Our network
> >> topology (which is out of my control) available for one port for server
> >> control (SSH).
> >
> > So you can't use a bonded port for both server management  and cluster
> > communications, can you? You can configure an active-passive bonding and
> > then you can have many virtual interfaces on top of that, i.e: bond0:0,
> > bond0:1 and assign them the ip addesses you need.
> >
> >
> > I use the other one with a cross over cable for heartbeat. So there is no
> >> way for bonding these two interfaces. Of course if I buy an extra switch I
> >> may do this.
> >
> > You can connect them to the same switch (though you lost kind of
> > redundancy), or you can use two crossover cables and move the management IP
> > to the same ports you are using for the vm's.
> >
> > Greetings,
> > Juanra
> >
> >>
> >> I don't realy understand why there is no backup heartbeat channel. LVS and
> >> MS cluster has this ability.
> >>
> >>>
> >>> ALOM can be used for fencing and can be on a seperate subnet if required.
> >>>
> >> I used this for fencing_ipmilan.
> >>
> >>>
> >>> Regards
> >>>
> >>> Rajagopal
> >>>
> >>> --
> >>> Linux-cluster mailing list
> >>> Linux-cluster at redhat.com
> >>> https://www.redhat.com/mailman/listinfo/linux-cluster
> >>>
> >>>
> >>
> >> ----- raju.rajsand at gmail.com den iletiyi bitir -----
> >>
> >>
> >>
> >>
> >> --
> >> Linux-cluster mailing list
> >> Linux-cluster at redhat.com
> >> https://www.redhat.com/mailman/listinfo/linux-cluster
> >>
> >
> 
> 
> ----- robejrm at gmail.com den iletiyi bitir -----
> 
> 
> 
> 
> 
> ------------------------------
> 
> Message: 4
> Date: Thu, 13 Aug 2009 15:29:43 +0200
> From: Juan Ramon Martin Blanco <robejrm at gmail.com>
> Subject: Re: [Linux-cluster] Is there any backup heartbeat channel
> To: linux clustering <linux-cluster at redhat.com>
> Message-ID:
> 	<8a5668960908130629n6ec05a88n463a3b03da331dae at mail.gmail.com>
> Content-Type: text/plain; charset="iso-8859-9"
> 
> 2009/8/13 Hakan VELIOGLU <veliogluh at itu.edu.tr>
> 
> > Thanks for all the answers.
> >
> > I think there is realy no backup heartbeat channel. Maybe the reason is
> > GFS. DLM works on the heartbeat channel. If you lost your heartbeat you lose
> > your lock consistency so it is better to fence the other node. For this
> > reason I think if you don't have enough network interface on server and
> > switch, loosing the heartbeat network may shut all the cluster members.
> >
> There is no backup heartbeat channel because you should do the backup at a
> operating system level, i.e: bonding
> That's why you should use a bonded interface for the heartbeat channel with
> at least 2 ethernet slaves; going further (for better redundancy) each of
> the slaves should be on a different network card and you should connect the
> each slave to a different switch.
> But what I am trying to explain, is that you can use that bonded logical
> interface also for things different from hearbeat. ;)
> 
> Greetings,
> Juanra
> 
> 
> > Hakan VEL?O?LU
> >
> >
> > ----- robejrm at gmail.com den ileti ---------
> >  Tarih: Thu, 13 Aug 2009 10:42:11 +0200
> >  Kimden:  Juan Ramon Martin Blanco <robejrm at gmail.com>
> >
> > Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
> >   Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
> >   Kime: linux clustering <linux-cluster at redhat.com>
> >
> >
> >  2009/8/13 Hakan VELIOGLU <veliogluh at itu.edu.tr>
> >>
> >>  ----- raju.rajsand at gmail.com den ileti ---------
> >>>  Tarih: Thu, 13 Aug 2009 08:57:15 +0530
> >>>  Kimden:  Rajagopal Swaminathan <raju.rajsand at gmail.com>
> >>> Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
> >>>  Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
> >>>  Kime: linux clustering <linux-cluster at redhat.com>
> >>>
> >>>
> >>>  Greetings,
> >>>
> >>>>
> >>>> 2009/8/12 Hakan VELIOGLU <veliogluh at itu.edu.tr>:
> >>>>
> >>>>  Hi list,
> >>>>>
> >>>>> I am trying a two node cluster with RH 5.3 on Sun X4150 hardware. I use
> >>>>> a
> >>>>>
> >>>>>
> >>>> IIRC, Sun x4150 has four ethernet ports. Two can be used for outside
> >>>> networking and two can be bonded and used for heartbeat.
> >>>>
> >>>>  I think, I couldn't explain my networking. I use two ethernet ports for
> >>> xen
> >>> vm which are trunk and bonded ports. Then there left two. Our network
> >>> topology (which is out of my control) available for one port for server
> >>> control (SSH).
> >>>
> >>
> >> So you can't use a bonded port for both server management  and cluster
> >> communications, can you? You can configure an active-passive bonding and
> >> then you can have many virtual interfaces on top of that, i.e: bond0:0,
> >> bond0:1 and assign them the ip addesses you need.
> >>
> >>
> >> I use the other one with a cross over cable for heartbeat. So there is no
> >>
> >>> way for bonding these two interfaces. Of course if I buy an extra switch
> >>> I
> >>> may do this.
> >>>
> >>
> >> You can connect them to the same switch (though you lost kind of
> >> redundancy), or you can use two crossover cables and move the management
> >> IP
> >> to the same ports you are using for the vm's.
> >>
> >> Greetings,
> >> Juanra
> >>
> >>
> >>> I don't realy understand why there is no backup heartbeat channel. LVS
> >>> and
> >>> MS cluster has this ability.
> >>>
> >>>
> >>>> ALOM can be used for fencing and can be on a seperate subnet if
> >>>> required.
> >>>>
> >>>>  I used this for fencing_ipmilan.
> >>>
> >>>
> >>>> Regards
> >>>>
> >>>> Rajagopal
> >>>>
> >>>> --
> >>>> Linux-cluster mailing list
> >>>> Linux-cluster at redhat.com
> >>>> https://www.redhat.com/mailman/listinfo/linux-cluster
> >>>>
> >>>>
> >>>>
> >>> ----- raju.rajsand at gmail.com den iletiyi bitir -----
> >>>
> >>>
> >>>
> >>>
> >>> --
> >>> Linux-cluster mailing list
> >>> Linux-cluster at redhat.com
> >>> https://www.redhat.com/mailman/listinfo/linux-cluster
> >>>
> >>>
> >>
> >
> > ----- robejrm at gmail.com den iletiyi bitir -----
> >
> >
> >
> >
> > --
> > Linux-cluster mailing list
> > Linux-cluster at redhat.com
> > https://www.redhat.com/mailman/listinfo/linux-cluster
> >
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: https://www.redhat.com/archives/linux-cluster/attachments/20090813/cd4dc079/attachment.html
> 
> ------------------------------
> 
> Message: 5
> Date: Thu, 13 Aug 2009 10:07:13 -0400
> From: Nehemias Jahcob <nehemiasjahcob at gmail.com>
> Subject: [Linux-cluster] RHCS on KVM
> To: linux clustering <linux-cluster at redhat.com>
> Message-ID:
> 	<5f61ab380908130707q5c936504k7351d0d6b3459090 at mail.gmail.com>
> Content-Type: text/plain; charset="iso-8859-1"
> 
> Hi.
> 
>  How to create a cluster of 2 nodes in rhel5.4 (or Fedora 10) with KVM?
> 
> With XEN follow this guide:
> http://sources.redhat.com/cluster/wiki/VMClusterCookbook?highlight =
> (CategoryHowTo).
> 
> Do you have a guide to implementation of RHCS in KVM?
> 
> Thank you all.
> NJ
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: https://www.redhat.com/archives/linux-cluster/attachments/20090813/f3a69a80/attachment.html
> 
> ------------------------------
> 
> Message: 6
> Date: Thu, 13 Aug 2009 14:16:47 +0000
> From: NTOUGHE GUY-SERGE <ntoughe at hotmail.com>
> Subject: [Linux-cluster] Cman hang
> To: <linux-cluster at redhat.com>
> Message-ID: <BAY119-W410E2F250E8B461752CFC9A5050 at phx.gbl>
> Content-Type: text/plain; charset="iso-8859-1"
> 
> 
> 
> Hi gurus,
> 
> i installed  RHEL 5.3 on 2 servers which participating to a cluster composed of these 2 nodes:
> kernel version:
> kernel-headers-2.6.18-128.el5
> kernel-devel-2.6.18-128.el5
> kernel-2.6.18-128.el5
> cman-devel-2.0.98-1.el5_3.1
> cman-2.0.98-1.el5_3.1
> cluster-cim-0.12.1-2.el5
> lvm2-cluster-2.02.40-7.el5
> cluster-snmp-0.12.1-2.el5
> modcluster-0.12.1-2.el5
> When i want to start cman the following message is sent:
> cman not started: Multicast and node address families differ. /usr/sbin/cman_tool: aisexec daemon didn't start
>                                                            [FAILED]
> 
> I trier to mount gfs2
> and i got theses messages:
> # mount -t gfs2 /dev/VolGroup01/LogVol01 /appli/prod --o lockTablename=arvclust:/appli/prod, Lockproto=lock_dlm
> 
> /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> 
> /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> 
> /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> 
> /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> 
> do you have any clues?
> Please it's an hurry, i waste long time to lok for solution help
> regards
> 
> 
> 
> 
> 
> 
> 
> 
> ntoughe at hotmail.com
> 
> 
> _________________________________________________________________
> With Windows Live, you can organize, edit, and share your photos.
> http://www.microsoft.com/middleeast/windows/windowslive/products/photo-gallery-edit.aspx
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: https://www.redhat.com/archives/linux-cluster/attachments/20090813/0a55101d/attachment.html
> 
> ------------------------------
> 
> Message: 7
> Date: Thu, 13 Aug 2009 09:14:24 -0500
> From: David Teigland <teigland at redhat.com>
> Subject: Re: [Linux-cluster] gfs2 mount hangs
> To: Wengang Wang <wen.gang.wang at oracle.com>
> Cc: linux clustering <linux-cluster at redhat.com>
> Message-ID: <20090813141424.GA8148 at redhat.com>
> Content-Type: text/plain; charset=us-ascii
> 
> On Thu, Aug 13, 2009 at 02:22:11PM +0800, Wengang Wang wrote:
> > <cman two_node="1" expected_votes="2"/>
> 
> That's not a valid combination, two_node="1" requires expected_votes="1".
> 
> You didn't mention which userspace cluster version/release you're using, or
> include any status about the cluster.  Before trying to mount gfs on either
> node, collect from both nodes:
> 
>   cman_tool status
>   cman_tool nodes
>   group_tool
> 
> Then mount on the first node and collect the same information, then try
> mounting on the second node, collect the same information, and look for any
> errors in /var/log/messages.
> 
> Since you're using new kernels, you need to be using the cluster 3.0 userspace
> code.  You're using the old manual fencing config.  There is no more
> fence_manual; the new way to configure manual fencing is to not configure any
> fencing at all.  So, your cluster.conf should look like this:
> 
> <?xml version="1.0"?>
> <cluster name="testgfs2" config_version="1">
> <cman two_node="1" expected_votes="1"/>
> <clusternodes>
> <clusternode name="cool" nodeid="1"/>
> <clusternode name="desk" nodeid="2"/>
> </clusternodes>
> </cluster>
> 
> Dave
> 
> 
> 
> ------------------------------
> 
> Message: 8
> Date: Thu, 13 Aug 2009 10:39:46 -0400
> From: Lon Hohberger <lhh at redhat.com>
> Subject: Re: [Linux-cluster] Qdisk question
> To: linux clustering <linux-cluster at redhat.com>
> Message-ID: <1250174386.23376.1440.camel at localhost.localdomain>
> Content-Type: text/plain
> 
> On Thu, 2009-08-13 at 00:45 +0200, brem belguebli wrote:
> 
> > My understanding of qdisk is that it is used as a tie-breaker, but it
> > looks like it is more a heatbeat vector than a simple tie-breaker.
> 
> Right, it's a secondary membership algorithm.
> 
> 
> > Until here, no real problem indeed, if the site gets apart from the
> > other prod site and also from the third site (hosting the iscsi target
> > qdisk) the 2 nodes from the failing site get evicted from the cluster.
> > 
> > 
> > But, what if my third site gets isolated while the 2 prod ones are
> > fine ?
> 
> Qdisk votes will not be presented to CMAN any more, but the two sites
> should remain online if they still have a "majority" of votes.
> 
> 
> > The real  question is what happens in case all the nodes loose access
> > to the qdisk while they're still able to see each others ?
> 
> Qdisk is just a vote like other voting mechanisms.  If all nodes lose
> access at the same time, it should behave like a node death.  However,
> the default action if _one_ node loses access is to kill that node (even
> if CMAN still sees it).
> 
> 
> > The 4 nodes have each 1 vote and the qdisk 1 vote. The expected quorum
> > is 3.
> 
> 
> > If I loose the qdisk, the number of votes falls to 4, the cluster is
> > quorate (4>3) but it looks like everything goes bad, each node
> > deactivate itself as it can't write its alive status (--> heartbeat
> > vector) to the qdisk even if the network heartbeating is working
> > fine. 
> 
> What happens specifically?  Most of the actions qdiskd performs are
> configurable.  For example, if the nodes are rebooting, you can turn
> that behavior off.
> 
> 
> 
> I wrote a simple 'ping' tiebreaker based the behaviors in RHEL3.  It
> functions in many ways in the same manner as qdiskd with respect to vote
> advertisement to CMAN, but without needing a disk - maybe you would find
> it useful?
> 
> http://people.redhat.com/lhh/qnet.tar.gz
> 
> -- Lon
> 
> 
> 
> ------------------------------
> 
> Message: 9
> Date: Thu, 13 Aug 2009 17:02:15 +0200
> From: Juan Ramon Martin Blanco <robejrm at gmail.com>
> Subject: Re: [Linux-cluster] Cman hang
> To: linux clustering <linux-cluster at redhat.com>
> Message-ID:
> 	<8a5668960908130802p4f5168cbueda86d1e6f1324bb at mail.gmail.com>
> Content-Type: text/plain; charset="iso-8859-1"
> 
> On Thu, Aug 13, 2009 at 4:16 PM, NTOUGHE GUY-SERGE <ntoughe at hotmail.com>wrote:
> 
> >
> > Hi gurus,
> >
> > i installed  RHEL 5.3 on 2 servers which participating to a cluster
> > composed of these 2 nodes:
> > kernel version:
> > kernel-headers-2.6.18-128.el5
> > kernel-devel-2.6.18-128.el5
> > kernel-2.6.18-128.el5
> > cman-devel-2.0.98-1.el5_3.1
> > cman-2.0.98-1.el5_3.1
> > cluster-cim-0.12.1-2.el5
> > lvm2-cluster-2.02.40-7.el5
> > cluster-snmp-0.12.1-2.el5
> > modcluster-0.12.1-2.el5
> > When i want to start cman the following message is sent:
> > cman not started: Multicast and node address families differ.
> > /usr/sbin/cman_tool: aisexec daemon didn't start
> >   [FAILED]
> >
> Please, show us your cluster.conf file so we can help.
> 
> Regards,
> Juanra
> 
> >
> > I trier to mount gfs2
> > and i got theses messages:
> > # mount -t gfs2 /dev/VolGroup01/LogVol01 /appli/prod --o
> > lockTablename=arvclust:/appli/prod, Lockproto=lock_dlm
> >
> > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> >
> > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> >
> > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> >
> > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> >
> > do you have any clues?
> > Please it's an hurry, i waste long time to lok for solution help
> > regards
> >
> >
> >
> >
> >
> >
> >
> >
> > ntoughe at hotmail.com
> >
> >
> >
> > ------------------------------
> > With Windows Live, you can organize, edit, and share your photos.<http://www.microsoft.com/middleeast/windows/windowslive/products/photo-gallery-edit.aspx>
> >
> > --
> > Linux-cluster mailing list
> > Linux-cluster at redhat.com
> > https://www.redhat.com/mailman/listinfo/linux-cluster
> >
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: https://www.redhat.com/archives/linux-cluster/attachments/20090813/9ecbcab1/attachment.html
> 
> ------------------------------
> 
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
> 
> End of Linux-cluster Digest, Vol 64, Issue 16
> *********************************************

_________________________________________________________________
Show them the way! Add maps and directions to your party invites. 
http://www.microsoft.com/windows/windowslive/products/events.aspx
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090813/45029caa/attachment.htm>

From brem.belguebli at gmail.com  Thu Aug 13 15:23:16 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Thu, 13 Aug 2009 17:23:16 +0200
Subject: [Linux-cluster] Qdisk question
In-Reply-To: <1250174386.23376.1440.camel@localhost.localdomain>
References: <29ae894c0908121545r3a92cf8fk59d5c10c162cdfad@mail.gmail.com>
	<1250174386.23376.1440.camel@localhost.localdomain>
Message-ID: <29ae894c0908130823i65667021vdc840ae1f0ded134@mail.gmail.com>

Hi Lon and Thanks for this reply.

In fact, thinking about it, my test wasn't very much representative of what
I was expecting to do.

I blocked the qdisk communications to only one node which, after reading
your reply, kind of confirmed me that I did the wong test. I'm going to re
run it by blocking all the nodes to the qdisk.

I'll also try your ping tie-breaker.

Brem


2009/8/13, Lon Hohberger <lhh at redhat.com>:
>
> On Thu, 2009-08-13 at 00:45 +0200, brem belguebli wrote:
>
> > My understanding of qdisk is that it is used as a tie-breaker, but it
> > looks like it is more a heatbeat vector than a simple tie-breaker.
>
> Right, it's a secondary membership algorithm.
>
>
> > Until here, no real problem indeed, if the site gets apart from the
> > other prod site and also from the third site (hosting the iscsi target
> > qdisk) the 2 nodes from the failing site get evicted from the cluster.
> >
> >
> > But, what if my third site gets isolated while the 2 prod ones are
> > fine ?
>
> Qdisk votes will not be presented to CMAN any more, but the two sites
> should remain online if they still have a "majority" of votes.
>
>
> > The real  question is what happens in case all the nodes loose access
> > to the qdisk while they're still able to see each others ?
>
> Qdisk is just a vote like other voting mechanisms.  If all nodes lose
> access at the same time, it should behave like a node death.  However,
> the default action if _one_ node loses access is to kill that node (even
> if CMAN still sees it).
>
>
> > The 4 nodes have each 1 vote and the qdisk 1 vote. The expected quorum
> > is 3.
>
>
> > If I loose the qdisk, the number of votes falls to 4, the cluster is
> > quorate (4>3) but it looks like everything goes bad, each node
> > deactivate itself as it can't write its alive status (--> heartbeat
> > vector) to the qdisk even if the network heartbeating is working
> > fine.
>
> What happens specifically?  Most of the actions qdiskd performs are
> configurable.  For example, if the nodes are rebooting, you can turn
> that behavior off.
>
>
>
> I wrote a simple 'ping' tiebreaker based the behaviors in RHEL3.  It
> functions in many ways in the same manner as qdiskd with respect to vote
> advertisement to CMAN, but without needing a disk - maybe you would find
> it useful?
>
> http://people.redhat.com/lhh/qnet.tar.gz
>
> -- Lon
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090813/5482c8bf/attachment.htm>

From robejrm at gmail.com  Thu Aug 13 15:23:54 2009
From: robejrm at gmail.com (Juan Ramon Martin Blanco)
Date: Thu, 13 Aug 2009 17:23:54 +0200
Subject: [Linux-cluster] Cman hang
In-Reply-To: <BAY119-W294291EA707ECD00D02139A5050@phx.gbl>
References: <20090813150236.B2D7461A5CE@hormel.redhat.com>
	<BAY119-W294291EA707ECD00D02139A5050@phx.gbl>
Message-ID: <8a5668960908130823u11b46ad9pd1da5af3614ce3d3@mail.gmail.com>

On Thu, Aug 13, 2009 at 5:13 PM, NTOUGHE GUY-SERGE <ntoughe at hotmail.com>wrote:

>  Hi, this is my cluster.conf
> <?xml version="1.0"?>
> <cluster alias="arevclust" config_version="21" name="arevclust">
>   <fence_daemon clean_start="0" post_fail_delay="0" post_join_delay="3"/>
>   <clusternodes>
>   <clusternode name="host1" nodeid="1" votes="1">
>   <fence>
>   <method name="2">
>

>   <device name=""/>
>
You should configure a valid fencing method, and if you don't have any, use
fence_manual until you get it.

>
>   </method>
>   </fence>
>   <multicast addr="" interface=""/>
>
I am not sure, but I think you should erase this <multicast ....> tag

Greetings,
Juanra

>
>   </clusternode>
>

>   <clusternode name="host2" nodeid="2" votes="1">
>   <fence>
>   <method name="1">
>   <device name=""/>
>   </method>
>   <method name=""/>
>   </fence>
>   <multicast addr="" interface=""/>
>   </clusternode>
>   </clusternodes>
>   <cman expected_votes="" two_node="">
>   <multicast addr=""/>
>   </cman>
>   <fencedevices>
>   <fencedevice agent="fence_brocade" ipaddr="" login="" name="" passwd=""/>
>   </fencedevices>
>   <rm>
>   <failoverdomains>
>   </failoverdomains>
>   <resources>
>   </resources>
>   </rm>
> </cluster>
>
> Regards
>
>
>
>
> ntoughe at hotmail.com
>
>
>
>
> > From: linux-cluster-request at redhat.com
> > Subject: Linux-cluster Digest, Vol 64, Issue 16
> > To: linux-cluster at redhat.com
> > Date: Thu, 13 Aug 2009 11:02:36 -0400
> >
> > Send Linux-cluster mailing list submissions to
> > linux-cluster at redhat.com
> >
> > To subscribe or unsubscribe via the World Wide Web, visit
> > https://www.redhat.com/mailman/listinfo/linux-cluster
> > or, via email, send a message with subject or body 'help' to
> > linux-cluster-request at redhat.com
> >
> > You can reach the person managing the list at
> > linux-cluster-owner at redhat.com
> >
> > When replying, please edit your Subject line so it is more specific
> > than "Re: Contents of Linux-cluster digest..."
> >
> >
> > Today's Topics:
> >
> > 1. Re: do I have a fence DRAC device? (ESGLinux)
> > 2. clusterservice stays in 'recovering' state (mark benschop)
> > 3. Re: Is there any backup heartbeat channel (Hakan VELIOGLU)
> > 4. Re: Is there any backup heartbeat channel
> > (Juan Ramon Martin Blanco)
> > 5. RHCS on KVM (Nehemias Jahcob)
> > 6. Cman hang (NTOUGHE GUY-SERGE)
> > 7. Re: gfs2 mount hangs (David Teigland)
> > 8. Re: Qdisk question (Lon Hohberger)
> > 9. Re: Cman hang (Juan Ramon Martin Blanco)
> >
> >
> > ----------------------------------------------------------------------
> >
> > Message: 1
> > Date: Thu, 13 Aug 2009 13:27:16 +0200
> > From: ESGLinux <esggrupos at gmail.com>
> > Subject: Re: [Linux-cluster] do I have a fence DRAC device?
> > To: linux clustering <linux-cluster at redhat.com>
> > Message-ID:
> > <3128ba140908130427i6ab85406ye6da34073e6a6e97 at mail.gmail.com>
> > Content-Type: text/plain; charset="iso-8859-1"
> >
> > Hi,
> > I couldn?t reboot my system yet but I have installed the openmanage
> > packages:
> >
> > srvadmin-omacore-5.4.0-260
> > srvadmin-iws-5.4.0-260
> > srvadmin-syscheck-5.4.0-260
> > srvadmin-rac5-components-5.4.0-260
> > srvadmin-deng-5.4.0-260
> > srvadmin-ipmi-5.4.0-260.DUP
> > srvadmin-racadm5-5.4.0-260
> > srvadmin-omauth-5.4.0-260.rhel5
> > srvadmin-hapi-5.4.0-260
> > srvadmin-cm-5.4.0-260
> > srvadmin-racdrsc5-5.4.0-260
> > srvadmin-omilcore-5.4.0-260
> > srvadmin-isvc-5.4.0-260
> > srvadmin-storage-5.4.0-260
> > srvadmin-jre-5.4.0-260
> > srvadmin-omhip-5.4.0-260
> >
> > Now I have the command racadm but when I try to execut it I get this:
> >
> > racadm config -g cfgSerial -o cfgSerialTelnetEnable 1
> > ERROR: RACADM is unable to process the requested subcommand because there
> is
> > no
> > local RAC configuration to communicate with.
> >
> > Local RACADM subcommand execution requires the following:
> >
> > 1. A Remote Access Controller (RAC) must be present on the managed server
> > 2. Appropriate managed node software must be installed and running on the
> > server
> >
> >
> > What do I need to install/start? or until I configure the bios I can?t
> get
> > this work?
> >
> > Greetings
> >
> > ESG
> >
> >
> > 2009/8/11 <bergman at merctech.com>
> >
> > >
> > >
> > > In the message dated: Tue, 11 Aug 2009 14:14:03 +0200,
> > > The pithy ruminations from Juan Ramon Martin Blanco on
> > > <Re: [Linux-cluster] do I have a fence DRAC device?> were:
> > > => --===============1917368601==
> > > => Content-Type: multipart/alternative;
> > > boundary=0016364c7c07663f600470dca3b8
> > > =>
> > > => --0016364c7c07663f600470dca3b8
> > > => Content-Type: text/plain; charset=ISO-8859-1
> > > => Content-Transfer-Encoding: quoted-printable
> > > =>
> > > => On Tue, Aug 11, 2009 at 2:03 PM, ESGLinux <esggrupos at gmail.com>
> wrote:
> > > =>
> > > => > Thanks
> > > => > I=B4ll check it when I could reboot the server.
> > > => >
> > > => > greetings,
> > > => >
> > > => You have a BMC ipmi in the first network interface, it can be
> configured
> > > at
> > > => boot time (I don't remember if inside the BIOS or pressing
> > > cntrl+something
> > > => during boot)
> > > =>
> > >
> > > Based on my notes, here's how I configured the DRAC interface on a Dell
> > > 1950
> > > for use as a fence device:
> > >
> > > Configuring the card from Linux depending on the installation of
> > > Dell's
> > > OMSA package. Once that's installed, use the following
> > > commands:
> > >
> > > racadm config -g cfgSerial -o cfgSerialTelnetEnable 1
> > > racadm config -g cfgLanNetworking -o cfgDNSRacName
> > > HOSTNAME_FOR_INTERFACE
> > > racadm config -g cfgDNSDomainName DOMAINNAME_FOR_INTERFACE
> > > racadm config -g cfgUserAdmin -o cfgUserAdminPassword -i 2
> > > PASSWORD
> > > racadm config -g cfgNicEnable 1
> > > racadm config -g cfgNicIpAddress WWW.XXX.YYY.ZZZ
> > > racadm config -g cfgNicNetmask WWW.XXX.YYY.ZZZ
> > > racadm config -g cfgNicGateway WWW.XXX.YYY.ZZZ
> > > racadm config -g cfgNicUseDhcp 0
> > >
> > >
> > > I also save a backup of the configuration with:
> > >
> > > racadm getconfig -f ~/drac_config
> > >
> > >
> > > Hope this helps,
> > >
> > > Mark
> > >
> > > ----
> > > Mark Bergman voice: 215-662-7310
> > > mark.bergman at uphs.upenn.edu fax: 215-614-0266
> > > System Administrator Section of Biomedical Image Analysis
> > > Department of Radiology University of Pennsylvania
> > > PGP Key: https://www.rad.upenn.edu/sbia/bergman
> > >
> > >
> > > => Greetings,
> > > => Juanra
> > > =>
> > > => >
> > > => > ESG
> > > => >
> > > => > 2009/8/10 Paras pradhan <pradhanparas at gmail.com>
> > > => >
> > > => > On Mon, Aug 10, 2009 at 5:24 AM, ESGLinux<esggrupos at gmail.com>
> wrote:
> > > => >> > Hi all,
> > > => >> > I was designing a 2 node cluster and I was going to use 2
> servers
> > > DELL
> > > => >> > PowerEdge 1950. I was going to buy a DRAC card to use for
> fencing
> > > but
> > > => >> > running several commands in the servers I have noticed that
> when I
> > > run
> > > => >> this
> > > => >> > command:
> > > => >> > #ipmitool lan print
> > > => >> > Set in Progress : Set Complete
> > > => >> > Auth Type Support : NONE MD2 MD5 PASSWORD
> > > => >> > Auth Type Enable : Callback : MD2 MD5
> > > => >> > : User : MD2 MD5
> > > => >> > : Operator : MD2 MD5
> > > => >> > : Admin : MD2 MD5
> > > => >> > : OEM : MD2 MD5
> > > => >> > IP Address Source : Static Address
> > > => >> > IP Address : 0.0.0.0
> > > => >> > Subnet Mask : 0.0.0.0
> > > => >> > MAC Address : 00:1e:c9:ae:6f:7e
> > > => >> > SNMP Community String : public
> > > => >> > IP Header : TTL=0x40 Flags=0x40 Precedence=0x00 TOS=0x10
> > > => >> > Default Gateway IP : 0.0.0.0
> > > => >> > Default Gateway MAC : 00:00:00:00:00:00
> > > => >> > Backup Gateway IP : 0.0.0.0
> > > => >> > Backup Gateway MAC : 00:00:00:00:00:00
> > > => >> > 802.1q VLAN ID : Disabled
> > > => >> > 802.1q VLAN Priority : 0
> > > => >> > RMCP+ Cipher Suites : 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14
> > > => >> > Cipher Suite Priv Max : aaaaaaaaaaaaaaa
> > > => >> > : X=Cipher Suite Unused
> > > => >> > : c=CALLBACK
> > > => >> > : u=USER
> > > => >> > : o=OPERATOR
> > > => >> > : a=ADMIN
> > > => >> > : O=OEM
> > > => >> > does this mean that I already have an ipmi card (not
> configured)
> > > that
> > > => I
> > > => >> can
> > > => >> > use for fencing? if the anwser is yes, where hell must I
> configure
> > > it?
> > > => I
> > > => >> > don=B4t see wher can I do it.
> > > => >> > If I haven=B4t a fencing device which one do you recommed to
> use?
> > > => >> > Thanks in advance
> > > => >> > ESG
> > > => >> >
> > > => >> > --
> > > => >> > Linux-cluster mailing list
> > > => >> > Linux-cluster at redhat.com
> > > => >> > https://www.redhat.com/mailman/listinfo/linux-cluster
> > > => >> >
> > > => >>
> > > => >> Yes you have IPMI and if you are using 1950 Dell, DRAC should be
> > > there
> > > => >> too. You can see if you have DRAC or not when the server starts
> and
> > > => >> before the loading of the OS.
> > > => >>
> > > => >> I have 1850s and I am using DRAC for fencing.
> > > => >>
> > > => >>
> > > => >> Paras.
> > > => >>
> > > => >> --
> > > => >> Linux-cluster mailing list
> > > => >> Linux-cluster at redhat.com
> > > => >> https://www.redhat.com/mailman/listinfo/linux-cluster
> > > => >>
> > > => >
> > > => >
> > >
> > >
> > >
> > > --
> > > Linux-cluster mailing list
> > > Linux-cluster at redhat.com
> > > https://www.redhat.com/mailman/listinfo/linux-cluster
> > >
> > -------------- next part --------------
> > An HTML attachment was scrubbed...
> > URL:
> https://www.redhat.com/archives/linux-cluster/attachments/20090813/a4558d27/attachment.html
> >
> > ------------------------------
> >
> > Message: 2
> > Date: Thu, 13 Aug 2009 14:45:13 +0200
> > From: mark benschop <mark.benschop.lists at gmail.com>
> > Subject: [Linux-cluster] clusterservice stays in 'recovering' state
> > To: linux-cluster at redhat.com
> > Message-ID:
> > <f97c3a70908130545n11ce442ej17d74c9cdc450e45 at mail.gmail.com>
> > Content-Type: text/plain; charset="iso-8859-1"
> >
> > Hi All,
> >
> > I've a problem with a clusterservice. The service was started up while
> one
> > of the resources, an NFS, export was not accessible.
> > Therefore the service never started up right but got into the
> 'recovering'
> > state.
> > In the mean time the NFS exports are setup properly but to no avail.
> > Stopping the clusterservice, using clusvcadm -d <service>, will result in
> > the service going down but staying in the 'recovering' state.
> > Starting it again doesn't work. The service doesn't start and stays in
> the
> > recovery status.
> > I'm suspecting rgmanager lost it somehow.
> >
> > Anybody had any ideas on what could be the problem and how to resolve it
> ?
> >
> > Thanks in advance,
> > Mark
> > -------------- next part --------------
> > An HTML attachment was scrubbed...
> > URL:
> https://www.redhat.com/archives/linux-cluster/attachments/20090813/31731cd3/attachment.html
> >
> > ------------------------------
> >
> > Message: 3
> > Date: Thu, 13 Aug 2009 16:13:12 +0300
> > From: Hakan VELIOGLU <veliogluh at itu.edu.tr>
> > Subject: Re: [Linux-cluster] Is there any backup heartbeat channel
> > To: linux-cluster at redhat.com
> > Message-ID: <20090813161312.11546h2sp6psr814 at webmail.itu.edu.tr>
> > Content-Type: text/plain; charset=ISO-8859-9; DelSp="Yes";
> > format="flowed"
> >
> > Thanks for all the answers.
> >
> > I think there is realy no backup heartbeat channel. Maybe the reason
> > is GFS. DLM works on the heartbeat channel. If you lost your heartbeat
> > you lose your lock consistency so it is better to fence the other
> > node. For this reason I think if you don't have enough network
> > interface on server and switch, loosing the heartbeat network may shut
> > all the cluster members.
> >
> > Hakan VEL?O?LU
> >
> >
> > ----- robejrm at gmail.com den ileti ---------
> > Tarih: Thu, 13 Aug 2009 10:42:11 +0200
> > Kimden: Juan Ramon Martin Blanco <robejrm at gmail.com>
> > Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
> > Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
> > Kime: linux clustering <linux-cluster at redhat.com>
> >
> >
> > > 2009/8/13 Hakan VELIOGLU <veliogluh at itu.edu.tr>
> > >
> > >> ----- raju.rajsand at gmail.com den ileti ---------
> > >> Tarih: Thu, 13 Aug 2009 08:57:15 +0530
> > >> Kimden: Rajagopal Swaminathan <raju.rajsand at gmail.com>
> > >> Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
> > >> Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
> > >> Kime: linux clustering <linux-cluster at redhat.com>
> > >>
> > >>
> > >> Greetings,
> > >>>
> > >>> 2009/8/12 Hakan VELIOGLU <veliogluh at itu.edu.tr>:
> > >>>
> > >>>> Hi list,
> > >>>>
> > >>>> I am trying a two node cluster with RH 5.3 on Sun X4150 hardware. I
> use a
> > >>>>
> > >>>
> > >>> IIRC, Sun x4150 has four ethernet ports. Two can be used for outside
> > >>> networking and two can be bonded and used for heartbeat.
> > >>>
> > >> I think, I couldn't explain my networking. I use two ethernet ports
> for xen
> > >> vm which are trunk and bonded ports. Then there left two. Our network
> > >> topology (which is out of my control) available for one port for
> server
> > >> control (SSH).
> > >
> > > So you can't use a bonded port for both server management and cluster
> > > communications, can you? You can configure an active-passive bonding
> and
> > > then you can have many virtual interfaces on top of that, i.e: bond0:0,
> > > bond0:1 and assign them the ip addesses you need.
> > >
> > >
> > > I use the other one with a cross over cable for heartbeat. So there is
> no
> > >> way for bonding these two interfaces. Of course if I buy an extra
> switch I
> > >> may do this.
> > >
> > > You can connect them to the same switch (though you lost kind of
> > > redundancy), or you can use two crossover cables and move the
> management IP
> > > to the same ports you are using for the vm's.
> > >
> > > Greetings,
> > > Juanra
> > >
> > >>
> > >> I don't realy understand why there is no backup heartbeat channel. LVS
> and
> > >> MS cluster has this ability.
> > >>
> > >>>
> > >>> ALOM can be used for fencing and can be on a seperate subnet if
> required.
> > >>>
> > >> I used this for fencing_ipmilan.
> > >>
> > >>>
> > >>> Regards
> > >>>
> > >>> Rajagopal
> > >>>
> > >>> --
> > >>> Linux-cluster mailing list
> > >>> Linux-cluster at redhat.com
> > >>> https://www.redhat.com/mailman/listinfo/linux-cluster
> > >>>
> > >>>
> > >>
> > >> ----- raju.rajsand at gmail.com den iletiyi bitir -----
> > >>
> > >>
> > >>
> > >>
> > >> --
> > >> Linux-cluster mailing list
> > >> Linux-cluster at redhat.com
> > >> https://www.redhat.com/mailman/listinfo/linux-cluster
> > >>
> > >
> >
> >
> > ----- robejrm at gmail.com den iletiyi bitir -----
> >
> >
> >
> >
> >
> > ------------------------------
> >
> > Message: 4
> > Date: Thu, 13 Aug 2009 15:29:43 +0200
> > From: Juan Ramon Martin Blanco <robejrm at gmail.com>
> > Subject: Re: [Linux-cluster] Is there any backup heartbeat channel
> > To: linux clustering <linux-cluster at redhat.com>
> > Message-ID:
> > <8a5668960908130629n6ec05a88n463a3b03da331dae at mail.gmail.com>
> > Content-Type: text/plain; charset="iso-8859-9"
> >
> > 2009/8/13 Hakan VELIOGLU <veliogluh at itu.edu.tr>
> >
> > > Thanks for all the answers.
> > >
> > > I think there is realy no backup heartbeat channel. Maybe the reason is
> > > GFS. DLM works on the heartbeat channel. If you lost your heartbeat you
> lose
> > > your lock consistency so it is better to fence the other node. For this
> > > reason I think if you don't have enough network interface on server and
> > > switch, loosing the heartbeat network may shut all the cluster members.
> > >
> > There is no backup heartbeat channel because you should do the backup at
> a
> > operating system level, i.e: bonding
> > That's why you should use a bonded interface for the heartbeat channel
> with
> > at least 2 ethernet slaves; going further (for better redundancy) each of
> > the slaves should be on a different network card and you should connect
> the
> > each slave to a different switch.
> > But what I am trying to explain, is that you can use that bonded logical
> > interface also for things different from hearbeat. ;)
> >
> > Greetings,
> > Juanra
> >
> >
> > > Hakan VEL?O?LU
> > >
> > >
> > > ----- robejrm at gmail.com den ileti ---------
> > > Tarih: Thu, 13 Aug 2009 10:42:11 +0200
> > > Kimden: Juan Ramon Martin Blanco <robejrm at gmail.com>
> > >
> > > Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
> > > Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
> > > Kime: linux clustering <linux-cluster at redhat.com>
> > >
> > >
> > > 2009/8/13 Hakan VELIOGLU <veliogluh at itu.edu.tr>
> > >>
> > >> ----- raju.rajsand at gmail.com den ileti ---------
> > >>> Tarih: Thu, 13 Aug 2009 08:57:15 +0530
> > >>> Kimden: Rajagopal Swaminathan <raju.rajsand at gmail.com>
> > >>> Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
> > >>> Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
> > >>> Kime: linux clustering <linux-cluster at redhat.com>
> > >>>
> > >>>
> > >>> Greetings,
> > >>>
> > >>>>
> > >>>> 2009/8/12 Hakan VELIOGLU <veliogluh at itu.edu.tr>:
> > >>>>
> > >>>> Hi list,
> > >>>>>
> > >>>>> I am trying a two node cluster with RH 5.3 on Sun X4150 hardware. I
> use
> > >>>>> a
> > >>>>>
> > >>>>>
> > >>>> IIRC, Sun x4150 has four ethernet ports. Two can be used for outside
> > >>>> networking and two can be bonded and used for heartbeat.
> > >>>>
> > >>>> I think, I couldn't explain my networking. I use two ethernet ports
> for
> > >>> xen
> > >>> vm which are trunk and bonded ports. Then there left two. Our network
> > >>> topology (which is out of my control) available for one port for
> server
> > >>> control (SSH).
> > >>>
> > >>
> > >> So you can't use a bonded port for both server management and cluster
> > >> communications, can you? You can configure an active-passive bonding
> and
> > >> then you can have many virtual interfaces on top of that, i.e:
> bond0:0,
> > >> bond0:1 and assign them the ip addesses you need.
> > >>
> > >>
> > >> I use the other one with a cross over cable for heartbeat. So there is
> no
> > >>
> > >>> way for bonding these two interfaces. Of course if I buy an extra
> switch
> > >>> I
> > >>> may do this.
> > >>>
> > >>
> > >> You can connect them to the same switch (though you lost kind of
> > >> redundancy), or you can use two crossover cables and move the
> management
> > >> IP
> > >> to the same ports you are using for the vm's.
> > >>
> > >> Greetings,
> > >> Juanra
> > >>
> > >>
> > >>> I don't realy understand why there is no backup heartbeat channel.
> LVS
> > >>> and
> > >>> MS cluster has this ability.
> > >>>
> > >>>
> > >>>> ALOM can be used for fencing and can be on a seperate subnet if
> > >>>> required.
> > >>>>
> > >>>> I used this for fencing_ipmilan.
> > >>>
> > >>>
> > >>>> Regards
> > >>>>
> > >>>> Rajagopal
> > >>>>
> > >>>> --
> > >>>> Linux-cluster mailing list
> > >>>> Linux-cluster at redhat.com
> > >>>> https://www.redhat.com/mailman/listinfo/linux-cluster
> > >>>>
> > >>>>
> > >>>>
> > >>> ----- raju.rajsand at gmail.com den iletiyi bitir -----
> > >>>
> > >>>
> > >>>
> > >>>
> > >>> --
> > >>> Linux-cluster mailing list
> > >>> Linux-cluster at redhat.com
> > >>> https://www.redhat.com/mailman/listinfo/linux-cluster
> > >>>
> > >>>
> > >>
> > >
> > > ----- robejrm at gmail.com den iletiyi bitir -----
> > >
> > >
> > >
> > >
> > > --
> > > Linux-cluster mailing list
> > > Linux-cluster at redhat.com
> > > https://www.redhat.com/mailman/listinfo/linux-cluster
> > >
> > -------------- next part --------------
> > An HTML attachment was scrubbed...
> > URL:
> https://www.redhat.com/archives/linux-cluster/attachments/20090813/cd4dc079/attachment.html
> >
> > ------------------------------
> >
> > Message: 5
> > Date: Thu, 13 Aug 2009 10:07:13 -0400
> > From: Nehemias Jahcob <nehemiasjahcob at gmail.com>
> > Subject: [Linux-cluster] RHCS on KVM
> > To: linux clustering <linux-cluster at redhat.com>
> > Message-ID:
> > <5f61ab380908130707q5c936504k7351d0d6b3459090 at mail.gmail.com>
> > Content-Type: text/plain; charset="iso-8859-1"
> >
> > Hi.
> >
> > How to create a cluster of 2 nodes in rhel5.4 (or Fedora 10) with KVM?
> >
> > With XEN follow this guide:
> > http://sources.redhat.com/cluster/wiki/VMClusterCookbook?highlight =
> > (CategoryHowTo).
> >
> > Do you have a guide to implementation of RHCS in KVM?
> >
> > Thank you all.
> > NJ
> > -------------- next part --------------
> > An HTML attachment was scrubbed...
> > URL:
> https://www.redhat.com/archives/linux-cluster/attachments/20090813/f3a69a80/attachment.html
> >
> > ------------------------------
> >
> > Message: 6
> > Date: Thu, 13 Aug 2009 14:16:47 +0000
> > From: NTOUGHE GUY-SERGE <ntoughe at hotmail.com>
> > Subject: [Linux-cluster] Cman hang
> > To: <linux-cluster at redhat.com>
> > Message-ID: <BAY119-W410E2F250E8B461752CFC9A5050 at phx.gbl>
> > Content-Type: text/plain; charset="iso-8859-1"
> >
> >
> >
> > Hi gurus,
> >
> > i installed RHEL 5.3 on 2 servers which participating to a cluster
> composed of these 2 nodes:
> > kernel version:
> > kernel-headers-2.6.18-128.el5
> > kernel-devel-2.6.18-128.el5
> > kernel-2.6.18-128.el5
> > cman-devel-2.0.98-1.el5_3.1
> > cman-2.0.98-1.el5_3.1
> > cluster-cim-0.12.1-2.el5
> > lvm2-cluster-2.02.40-7.el5
> > cluster-snmp-0.12.1-2.el5
> > modcluster-0.12.1-2.el5
> > When i want to start cman the following message is sent:
> > cman not started: Multicast and node address families differ.
> /usr/sbin/cman_tool: aisexec daemon didn't start
> > [FAILED]
> >
> > I trier to mount gfs2
> > and i got theses messages:
> > # mount -t gfs2 /dev/VolGroup01/LogVol01 /appli/prod --o
> lockTablename=arvclust:/appli/prod, Lockproto=lock_dlm
> >
> > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> >
> > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> >
> > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> >
> > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> >
> > do you have any clues?
> > Please it's an hurry, i waste long time to lok for solution help
> > regards
> >
> >
> >
> >
> >
> >
> >
> >
> > ntoughe at hotmail.com
> >
> >
> > _________________________________________________________________
> > With Windows Live, you can organize, edit, and share your photos.
> >
> http://www.microsoft.com/middleeast/windows/windowslive/products/photo-gallery-edit.aspx
> > -------------- next part --------------
> > An HTML attachment was scrubbed...
> > URL:
> https://www.redhat.com/archives/linux-cluster/attachments/20090813/0a55101d/attachment.html
> >
> > ------------------------------
> >
> > Message: 7
> > Date: Thu, 13 Aug 2009 09:14:24 -0500
> > From: David Teigland <teigland at redhat.com>
> > Subject: Re: [Linux-cluster] gfs2 mount hangs
> > To: Wengang Wang <wen.gang.wang at oracle.com>
> > Cc: linux clustering <linux-cluster at redhat.com>
> > Message-ID: <20090813141424.GA8148 at redhat.com>
> > Content-Type: text/plain; charset=us-ascii
> >
> > On Thu, Aug 13, 2009 at 02:22:11PM +0800, Wengang Wang wrote:
> > > <cman two_node="1" expected_votes="2"/>
> >
> > That's not a valid combination, two_node="1" requires expected_votes="1".
> >
> > You didn't mention which userspace cluster version/release you're using,
> or
> > include any status about the cluster. Before trying to mount gfs on
> either
> > node, collect from both nodes:
> >
> > cman_tool status
> > cman_tool nodes
> > group_tool
> >
> > Then mount on the first node and collect the same information, then try
> > mounting on the second node, collect the same information, and look for
> any
> > errors in /var/log/messages.
> >
> > Since you're using new kernels, you need to be using the cluster 3.0
> userspace
> > code. You're using the old manual fencing config. There is no more
> > fence_manual; the new way to configure manual fencing is to not configure
> any
> > fencing at all. So, your cluster.conf should look like this:
> >
> > <?xml version="1.0"?>
> > <cluster name="testgfs2" config_version="1">
> > <cman two_node="1" expected_votes="1"/>
> > <clusternodes>
> > <clusternode name="cool" nodeid="1"/>
> > <clusternode name="desk" nodeid="2"/>
> > </clusternodes>
> > </cluster>
> >
> > Dave
> >
> >
> >
> > ------------------------------
> >
> > Message: 8
> > Date: Thu, 13 Aug 2009 10:39:46 -0400
> > From: Lon Hohberger <lhh at redhat.com>
> > Subject: Re: [Linux-cluster] Qdisk question
> > To: linux clustering <linux-cluster at redhat.com>
> > Message-ID: <1250174386.23376.1440.camel at localhost.localdomain>
> > Content-Type: text/plain
> >
> > On Thu, 2009-08-13 at 00:45 +0200, brem belguebli wrote:
> >
> > > My understanding of qdisk is that it is used as a tie-breaker, but it
> > > looks like it is more a heatbeat vector than a simple tie-breaker.
> >
> > Right, it's a secondary membership algorithm.
> >
> >
> > > Until here, no real problem indeed, if the site gets apart from the
> > > other prod site and also from the third site (hosting the iscsi target
> > > qdisk) the 2 nodes from the failing site get evicted from the cluster.
> > >
> > >
> > > But, what if my third site gets isolated while the 2 prod ones are
> > > fine ?
> >
> > Qdisk votes will not be presented to CMAN any more, but the two sites
> > should remain online if they still have a "majority" of votes.
> >
> >
> > > The real question is what happens in case all the nodes loose access
> > > to the qdisk while they're still able to see each others ?
> >
> > Qdisk is just a vote like other voting mechanisms. If all nodes lose
> > access at the same time, it should behave like a node death. However,
> > the default action if _one_ node loses access is to kill that node (even
> > if CMAN still sees it).
> >
> >
> > > The 4 nodes have each 1 vote and the qdisk 1 vote. The expected quorum
> > > is 3.
> >
> >
> > > If I loose the qdisk, the number of votes falls to 4, the cluster is
> > > quorate (4>3) but it looks like everything goes bad, each node
> > > deactivate itself as it can't write its alive status (--> heartbeat
> > > vector) to the qdisk even if the network heartbeating is working
> > > fine.
> >
> > What happens specifically? Most of the actions qdiskd performs are
> > configurable. For example, if the nodes are rebooting, you can turn
> > that behavior off.
> >
> >
> >
> > I wrote a simple 'ping' tiebreaker based the behaviors in RHEL3. It
> > functions in many ways in the same manner as qdiskd with respect to vote
> > advertisement to CMAN, but without needing a disk - maybe you would find
> > it useful?
> >
> > http://people.redhat.com/lhh/qnet.tar.gz
> >
> > -- Lon
> >
> >
> >
> > ------------------------------
> >
> > Message: 9
> > Date: Thu, 13 Aug 2009 17:02:15 +0200
> > From: Juan Ramon Martin Blanco <robejrm at gmail.com>
> > Subject: Re: [Linux-cluster] Cman hang
> > To: linux clustering <linux-cluster at redhat.com>
> > Message-ID:
> > <8a5668960908130802p4f5168cbueda86d1e6f1324bb at mail.gmail.com>
> > Content-Type: text/plain; charset="iso-8859-1"
> >
> > On Thu, Aug 13, 2009 at 4:16 PM, NTOUGHE GUY-SERGE <ntoughe at hotmail.com
> >wrote:
> >
> > >
> > > Hi gurus,
> > >
> > > i installed RHEL 5.3 on 2 servers which participating to a cluster
> > > composed of these 2 nodes:
> > > kernel version:
> > > kernel-headers-2.6.18-128.el5
> > > kernel-devel-2.6.18-128.el5
> > > kernel-2.6.18-128.el5
> > > cman-devel-2.0.98-1.el5_3.1
> > > cman-2.0.98-1.el5_3.1
> > > cluster-cim-0.12.1-2.el5
> > > lvm2-cluster-2.02.40-7.el5
> > > cluster-snmp-0.12.1-2.el5
> > > modcluster-0.12.1-2.el5
> > > When i want to start cman the following message is sent:
> > > cman not started: Multicast and node address families differ.
> > > /usr/sbin/cman_tool: aisexec daemon didn't start
> > > [FAILED]
> > >
> > Please, show us your cluster.conf file so we can help.
> >
> > Regards,
> > Juanra
> >
> > >
> > > I trier to mount gfs2
> > > and i got theses messages:
> > > # mount -t gfs2 /dev/VolGroup01/LogVol01 /appli/prod --o
> > > lockTablename=arvclust:/appli/prod, Lockproto=lock_dlm
> > >
> > > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> > >
> > > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> > >
> > > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> > >
> > > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> > >
> > > do you have any clues?
> > > Please it's an hurry, i waste long time to lok for solution help
> > > regards
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > > ntoughe at hotmail.com
> > >
> > >
> > >
> > > ------------------------------
> > > With Windows Live, you can organize, edit, and share your photos.<
> http://www.microsoft.com/middleeast/windows/windowslive/products/photo-gallery-edit.aspx
> >
> > >
> > > --
> > > Linux-cluster mailing list
> > > Linux-cluster at redhat.com
> > > https://www.redhat.com/mailman/listinfo/linux-cluster
> > >
> > -------------- next part --------------
> > An HTML attachment was scrubbed...
> > URL:
> https://www.redhat.com/archives/linux-cluster/attachments/20090813/9ecbcab1/attachment.html
> >
> > ------------------------------
> >
> > --
> > Linux-cluster mailing list
> > Linux-cluster at redhat.com
> > https://www.redhat.com/mailman/listinfo/linux-cluster
> >
> > End of Linux-cluster Digest, Vol 64, Issue 16
> > *********************************************
>
> ------------------------------
> See all the ways you can stay connected to friends and family<http://www.microsoft.com/windows/windowslive/default.aspx>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090813/f9557411/attachment.htm>

From pradhanparas at gmail.com  Thu Aug 13 15:30:47 2009
From: pradhanparas at gmail.com (Paras pradhan)
Date: Thu, 13 Aug 2009 10:30:47 -0500
Subject: [Linux-cluster] Min Quorum
Message-ID: <8b711df40908130830u6923f642td319b97a1e259861@mail.gmail.com>

I have a three nodes cluster with Total Cluster Votes: 3 and Minimum
Required Quorum: 2

Can I change Minimum Required Quorum  to 1?


Thanks
Paras.



From ccaulfie at redhat.com  Thu Aug 13 15:37:06 2009
From: ccaulfie at redhat.com (Christine Caulfield)
Date: Thu, 13 Aug 2009 16:37:06 +0100
Subject: [Linux-cluster] Min Quorum
In-Reply-To: <8b711df40908130830u6923f642td319b97a1e259861@mail.gmail.com>
References: <8b711df40908130830u6923f642td319b97a1e259861@mail.gmail.com>
Message-ID: <4A843322.1050304@redhat.com>

On 13/08/09 16:30, Paras pradhan wrote:
> I have a three nodes cluster with Total Cluster Votes: 3 and Minimum
> Required Quorum: 2
>
> Can I change Minimum Required Quorum  to 1?
>

You can change it what whatever you like, but if you change it to 
something too low (ie anything less that cman decides for you) you run 
the risk of split brain and potential data corruption.

If you want the cluster to keep running with two nodes missing then you 
ought to look into qdiskd. The man page is very detailed :-)

Chrissie



From pradhanparas at gmail.com  Thu Aug 13 15:42:40 2009
From: pradhanparas at gmail.com (Paras pradhan)
Date: Thu, 13 Aug 2009 10:42:40 -0500
Subject: [Linux-cluster] Min Quorum
In-Reply-To: <4A843322.1050304@redhat.com>
References: <8b711df40908130830u6923f642td319b97a1e259861@mail.gmail.com>
	<4A843322.1050304@redhat.com>
Message-ID: <8b711df40908130842p7379016coaad5666d24bd49cc@mail.gmail.com>

On Thu, Aug 13, 2009 at 10:37 AM, Christine
Caulfield<ccaulfie at redhat.com> wrote:
> On 13/08/09 16:30, Paras pradhan wrote:
>>
>> I have a three nodes cluster with Total Cluster Votes: 3 and Minimum
>> Required Quorum: 2
>>
>> Can I change Minimum Required Quorum ?to 1?
>>
>
> You can change it what whatever you like, but if you change it to something
> too low (ie anything less that cman decides for you) you run the risk of
> split brain and potential data corruption.
>
> If you want the cluster to keep running with two nodes missing then you
> ought to look into qdiskd. The man page is very detailed :-)

I have a quorum partition running with 2 votes. But when 2 of my nodes
failed, Quorum is dissolved.

I have 3 nodes + 2 votes to quorum parition. So altogether the total
votes should be 5 (?). I can see total votes to be only 3 .


Paras.


>
> Chrissie
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>



From ntoughe at hotmail.com  Thu Aug 13 15:45:16 2009
From: ntoughe at hotmail.com (NTOUGHE GUY-SERGE)
Date: Thu, 13 Aug 2009 15:45:16 +0000
Subject: [Linux-cluster] Cman hang
In-Reply-To: <20090813152421.A4A86619076@hormel.redhat.com>
References: <20090813152421.A4A86619076@hormel.redhat.com>
Message-ID: <BAY119-W4089E3353F6D3368CBADE0A5050@phx.gbl>


Thank's Junra,
I did it , and i tried to restart the cman service  without more success, thise are the messages i got:

my new cluster.conf:
<?xml version="1.0"?>
<cluster alias="arevclust" config_version="1" name="arevclust">
        <clusternodes>
        <cman expected_votes="1" two_node="1">
        </cman>
                <clusternode name="gs21spli003.occ.lan" nodeid="1" votes="1">
                </clusternode>
                <clusternode name="gs21spli004.occ.lan" nodeid="2" votes="1">
                </clusternode>
        </clusternodes>
</cluster>
~                                                                      
j

cman not started: Multicast and node address families differ. /usr/sbin/cman_tool: aisexec daemon didn't start
when i mounted the gfs FS i got this:
# mount -t gfs2 /dev/mapper/VolGroup01-LogVol01  /appli/prod
/sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
/sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
/sbin/mount.gfs2: can't connect to gfs_controld: Connection refused

I 'm not sure but i have a doubt on the lockTableName  during  the FS GFS creation:
mkfs.gfs2 -p lock_dlm -t arevclust:appli/prod -j 1 /dev/mapper/VolGroup00-LogVol01,  the name of the cluster is: arevclust
The gfs File system is made by /dev/mapper/volGroup00-LogVol01 and the mounting point is /appli/prod, so if i have to precise my gfs file i put : appli/prod is it correct?
regards



ntoughe at hotmail.com



> From: linux-cluster-request at redhat.com
> Subject: Linux-cluster Digest, Vol 64, Issue 18
> To: linux-cluster at redhat.com
> Date: Thu, 13 Aug 2009 11:24:21 -0400
> 
> Send Linux-cluster mailing list submissions to
> 	linux-cluster at redhat.com
> 
> To subscribe or unsubscribe via the World Wide Web, visit
> 	https://www.redhat.com/mailman/listinfo/linux-cluster
> or, via email, send a message with subject or body 'help' to
> 	linux-cluster-request at redhat.com
> 
> You can reach the person managing the list at
> 	linux-cluster-owner at redhat.com
> 
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Linux-cluster digest..."
> 
> 
> Today's Topics:
> 
>    1. Re: Qdisk question (brem belguebli)
>    2. Re: Cman hang (Juan Ramon Martin Blanco)
> 
> 
> ----------------------------------------------------------------------
> 
> Message: 1
> Date: Thu, 13 Aug 2009 17:23:16 +0200
> From: brem belguebli <brem.belguebli at gmail.com>
> Subject: Re: [Linux-cluster] Qdisk question
> To: linux clustering <linux-cluster at redhat.com>
> Message-ID:
> 	<29ae894c0908130823i65667021vdc840ae1f0ded134 at mail.gmail.com>
> Content-Type: text/plain; charset="iso-8859-1"
> 
> Hi Lon and Thanks for this reply.
> 
> In fact, thinking about it, my test wasn't very much representative of what
> I was expecting to do.
> 
> I blocked the qdisk communications to only one node which, after reading
> your reply, kind of confirmed me that I did the wong test. I'm going to re
> run it by blocking all the nodes to the qdisk.
> 
> I'll also try your ping tie-breaker.
> 
> Brem
> 
> 
> 2009/8/13, Lon Hohberger <lhh at redhat.com>:
> >
> > On Thu, 2009-08-13 at 00:45 +0200, brem belguebli wrote:
> >
> > > My understanding of qdisk is that it is used as a tie-breaker, but it
> > > looks like it is more a heatbeat vector than a simple tie-breaker.
> >
> > Right, it's a secondary membership algorithm.
> >
> >
> > > Until here, no real problem indeed, if the site gets apart from the
> > > other prod site and also from the third site (hosting the iscsi target
> > > qdisk) the 2 nodes from the failing site get evicted from the cluster.
> > >
> > >
> > > But, what if my third site gets isolated while the 2 prod ones are
> > > fine ?
> >
> > Qdisk votes will not be presented to CMAN any more, but the two sites
> > should remain online if they still have a "majority" of votes.
> >
> >
> > > The real  question is what happens in case all the nodes loose access
> > > to the qdisk while they're still able to see each others ?
> >
> > Qdisk is just a vote like other voting mechanisms.  If all nodes lose
> > access at the same time, it should behave like a node death.  However,
> > the default action if _one_ node loses access is to kill that node (even
> > if CMAN still sees it).
> >
> >
> > > The 4 nodes have each 1 vote and the qdisk 1 vote. The expected quorum
> > > is 3.
> >
> >
> > > If I loose the qdisk, the number of votes falls to 4, the cluster is
> > > quorate (4>3) but it looks like everything goes bad, each node
> > > deactivate itself as it can't write its alive status (--> heartbeat
> > > vector) to the qdisk even if the network heartbeating is working
> > > fine.
> >
> > What happens specifically?  Most of the actions qdiskd performs are
> > configurable.  For example, if the nodes are rebooting, you can turn
> > that behavior off.
> >
> >
> >
> > I wrote a simple 'ping' tiebreaker based the behaviors in RHEL3.  It
> > functions in many ways in the same manner as qdiskd with respect to vote
> > advertisement to CMAN, but without needing a disk - maybe you would find
> > it useful?
> >
> > http://people.redhat.com/lhh/qnet.tar.gz
> >
> > -- Lon
> >
> > --
> > Linux-cluster mailing list
> > Linux-cluster at redhat.com
> > https://www.redhat.com/mailman/listinfo/linux-cluster
> >
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: https://www.redhat.com/archives/linux-cluster/attachments/20090813/5482c8bf/attachment.html
> 
> ------------------------------
> 
> Message: 2
> Date: Thu, 13 Aug 2009 17:23:54 +0200
> From: Juan Ramon Martin Blanco <robejrm at gmail.com>
> Subject: Re: [Linux-cluster] Cman hang
> To: linux clustering <linux-cluster at redhat.com>
> Message-ID:
> 	<8a5668960908130823u11b46ad9pd1da5af3614ce3d3 at mail.gmail.com>
> Content-Type: text/plain; charset="iso-8859-1"
> 
> On Thu, Aug 13, 2009 at 5:13 PM, NTOUGHE GUY-SERGE <ntoughe at hotmail.com>wrote:
> 
> >  Hi, this is my cluster.conf
> > <?xml version="1.0"?>
> > <cluster alias="arevclust" config_version="21" name="arevclust">
> >   <fence_daemon clean_start="0" post_fail_delay="0" post_join_delay="3"/>
> >   <clusternodes>
> >   <clusternode name="host1" nodeid="1" votes="1">
> >   <fence>
> >   <method name="2">
> >
> 
> >   <device name=""/>
> >
> You should configure a valid fencing method, and if you don't have any, use
> fence_manual until you get it.
> 
> >
> >   </method>
> >   </fence>
> >   <multicast addr="" interface=""/>
> >
> I am not sure, but I think you should erase this <multicast ....> tag
> 
> Greetings,
> Juanra
> 
> >
> >   </clusternode>
> >
> 
> >   <clusternode name="host2" nodeid="2" votes="1">
> >   <fence>
> >   <method name="1">
> >   <device name=""/>
> >   </method>
> >   <method name=""/>
> >   </fence>
> >   <multicast addr="" interface=""/>
> >   </clusternode>
> >   </clusternodes>
> >   <cman expected_votes="" two_node="">
> >   <multicast addr=""/>
> >   </cman>
> >   <fencedevices>
> >   <fencedevice agent="fence_brocade" ipaddr="" login="" name="" passwd=""/>
> >   </fencedevices>
> >   <rm>
> >   <failoverdomains>
> >   </failoverdomains>
> >   <resources>
> >   </resources>
> >   </rm>
> > </cluster>
> >
> > Regards
> >
> >
> >
> >
> > ntoughe at hotmail.com
> >
> >
> >
> >
> > > From: linux-cluster-request at redhat.com
> > > Subject: Linux-cluster Digest, Vol 64, Issue 16
> > > To: linux-cluster at redhat.com
> > > Date: Thu, 13 Aug 2009 11:02:36 -0400
> > >
> > > Send Linux-cluster mailing list submissions to
> > > linux-cluster at redhat.com
> > >
> > > To subscribe or unsubscribe via the World Wide Web, visit
> > > https://www.redhat.com/mailman/listinfo/linux-cluster
> > > or, via email, send a message with subject or body 'help' to
> > > linux-cluster-request at redhat.com
> > >
> > > You can reach the person managing the list at
> > > linux-cluster-owner at redhat.com
> > >
> > > When replying, please edit your Subject line so it is more specific
> > > than "Re: Contents of Linux-cluster digest..."
> > >
> > >
> > > Today's Topics:
> > >
> > > 1. Re: do I have a fence DRAC device? (ESGLinux)
> > > 2. clusterservice stays in 'recovering' state (mark benschop)
> > > 3. Re: Is there any backup heartbeat channel (Hakan VELIOGLU)
> > > 4. Re: Is there any backup heartbeat channel
> > > (Juan Ramon Martin Blanco)
> > > 5. RHCS on KVM (Nehemias Jahcob)
> > > 6. Cman hang (NTOUGHE GUY-SERGE)
> > > 7. Re: gfs2 mount hangs (David Teigland)
> > > 8. Re: Qdisk question (Lon Hohberger)
> > > 9. Re: Cman hang (Juan Ramon Martin Blanco)
> > >
> > >
> > > ----------------------------------------------------------------------
> > >
> > > Message: 1
> > > Date: Thu, 13 Aug 2009 13:27:16 +0200
> > > From: ESGLinux <esggrupos at gmail.com>
> > > Subject: Re: [Linux-cluster] do I have a fence DRAC device?
> > > To: linux clustering <linux-cluster at redhat.com>
> > > Message-ID:
> > > <3128ba140908130427i6ab85406ye6da34073e6a6e97 at mail.gmail.com>
> > > Content-Type: text/plain; charset="iso-8859-1"
> > >
> > > Hi,
> > > I couldn?t reboot my system yet but I have installed the openmanage
> > > packages:
> > >
> > > srvadmin-omacore-5.4.0-260
> > > srvadmin-iws-5.4.0-260
> > > srvadmin-syscheck-5.4.0-260
> > > srvadmin-rac5-components-5.4.0-260
> > > srvadmin-deng-5.4.0-260
> > > srvadmin-ipmi-5.4.0-260.DUP
> > > srvadmin-racadm5-5.4.0-260
> > > srvadmin-omauth-5.4.0-260.rhel5
> > > srvadmin-hapi-5.4.0-260
> > > srvadmin-cm-5.4.0-260
> > > srvadmin-racdrsc5-5.4.0-260
> > > srvadmin-omilcore-5.4.0-260
> > > srvadmin-isvc-5.4.0-260
> > > srvadmin-storage-5.4.0-260
> > > srvadmin-jre-5.4.0-260
> > > srvadmin-omhip-5.4.0-260
> > >
> > > Now I have the command racadm but when I try to execut it I get this:
> > >
> > > racadm config -g cfgSerial -o cfgSerialTelnetEnable 1
> > > ERROR: RACADM is unable to process the requested subcommand because there
> > is
> > > no
> > > local RAC configuration to communicate with.
> > >
> > > Local RACADM subcommand execution requires the following:
> > >
> > > 1. A Remote Access Controller (RAC) must be present on the managed server
> > > 2. Appropriate managed node software must be installed and running on the
> > > server
> > >
> > >
> > > What do I need to install/start? or until I configure the bios I can?t
> > get
> > > this work?
> > >
> > > Greetings
> > >
> > > ESG
> > >
> > >
> > > 2009/8/11 <bergman at merctech.com>
> > >
> > > >
> > > >
> > > > In the message dated: Tue, 11 Aug 2009 14:14:03 +0200,
> > > > The pithy ruminations from Juan Ramon Martin Blanco on
> > > > <Re: [Linux-cluster] do I have a fence DRAC device?> were:
> > > > => --===============1917368601==
> > > > => Content-Type: multipart/alternative;
> > > > boundary=0016364c7c07663f600470dca3b8
> > > > =>
> > > > => --0016364c7c07663f600470dca3b8
> > > > => Content-Type: text/plain; charset=ISO-8859-1
> > > > => Content-Transfer-Encoding: quoted-printable
> > > > =>
> > > > => On Tue, Aug 11, 2009 at 2:03 PM, ESGLinux <esggrupos at gmail.com>
> > wrote:
> > > > =>
> > > > => > Thanks
> > > > => > I=B4ll check it when I could reboot the server.
> > > > => >
> > > > => > greetings,
> > > > => >
> > > > => You have a BMC ipmi in the first network interface, it can be
> > configured
> > > > at
> > > > => boot time (I don't remember if inside the BIOS or pressing
> > > > cntrl+something
> > > > => during boot)
> > > > =>
> > > >
> > > > Based on my notes, here's how I configured the DRAC interface on a Dell
> > > > 1950
> > > > for use as a fence device:
> > > >
> > > > Configuring the card from Linux depending on the installation of
> > > > Dell's
> > > > OMSA package. Once that's installed, use the following
> > > > commands:
> > > >
> > > > racadm config -g cfgSerial -o cfgSerialTelnetEnable 1
> > > > racadm config -g cfgLanNetworking -o cfgDNSRacName
> > > > HOSTNAME_FOR_INTERFACE
> > > > racadm config -g cfgDNSDomainName DOMAINNAME_FOR_INTERFACE
> > > > racadm config -g cfgUserAdmin -o cfgUserAdminPassword -i 2
> > > > PASSWORD
> > > > racadm config -g cfgNicEnable 1
> > > > racadm config -g cfgNicIpAddress WWW.XXX.YYY.ZZZ
> > > > racadm config -g cfgNicNetmask WWW.XXX.YYY.ZZZ
> > > > racadm config -g cfgNicGateway WWW.XXX.YYY.ZZZ
> > > > racadm config -g cfgNicUseDhcp 0
> > > >
> > > >
> > > > I also save a backup of the configuration with:
> > > >
> > > > racadm getconfig -f ~/drac_config
> > > >
> > > >
> > > > Hope this helps,
> > > >
> > > > Mark
> > > >
> > > > ----
> > > > Mark Bergman voice: 215-662-7310
> > > > mark.bergman at uphs.upenn.edu fax: 215-614-0266
> > > > System Administrator Section of Biomedical Image Analysis
> > > > Department of Radiology University of Pennsylvania
> > > > PGP Key: https://www.rad.upenn.edu/sbia/bergman
> > > >
> > > >
> > > > => Greetings,
> > > > => Juanra
> > > > =>
> > > > => >
> > > > => > ESG
> > > > => >
> > > > => > 2009/8/10 Paras pradhan <pradhanparas at gmail.com>
> > > > => >
> > > > => > On Mon, Aug 10, 2009 at 5:24 AM, ESGLinux<esggrupos at gmail.com>
> > wrote:
> > > > => >> > Hi all,
> > > > => >> > I was designing a 2 node cluster and I was going to use 2
> > servers
> > > > DELL
> > > > => >> > PowerEdge 1950. I was going to buy a DRAC card to use for
> > fencing
> > > > but
> > > > => >> > running several commands in the servers I have noticed that
> > when I
> > > > run
> > > > => >> this
> > > > => >> > command:
> > > > => >> > #ipmitool lan print
> > > > => >> > Set in Progress : Set Complete
> > > > => >> > Auth Type Support : NONE MD2 MD5 PASSWORD
> > > > => >> > Auth Type Enable : Callback : MD2 MD5
> > > > => >> > : User : MD2 MD5
> > > > => >> > : Operator : MD2 MD5
> > > > => >> > : Admin : MD2 MD5
> > > > => >> > : OEM : MD2 MD5
> > > > => >> > IP Address Source : Static Address
> > > > => >> > IP Address : 0.0.0.0
> > > > => >> > Subnet Mask : 0.0.0.0
> > > > => >> > MAC Address : 00:1e:c9:ae:6f:7e
> > > > => >> > SNMP Community String : public
> > > > => >> > IP Header : TTL=0x40 Flags=0x40 Precedence=0x00 TOS=0x10
> > > > => >> > Default Gateway IP : 0.0.0.0
> > > > => >> > Default Gateway MAC : 00:00:00:00:00:00
> > > > => >> > Backup Gateway IP : 0.0.0.0
> > > > => >> > Backup Gateway MAC : 00:00:00:00:00:00
> > > > => >> > 802.1q VLAN ID : Disabled
> > > > => >> > 802.1q VLAN Priority : 0
> > > > => >> > RMCP+ Cipher Suites : 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14
> > > > => >> > Cipher Suite Priv Max : aaaaaaaaaaaaaaa
> > > > => >> > : X=Cipher Suite Unused
> > > > => >> > : c=CALLBACK
> > > > => >> > : u=USER
> > > > => >> > : o=OPERATOR
> > > > => >> > : a=ADMIN
> > > > => >> > : O=OEM
> > > > => >> > does this mean that I already have an ipmi card (not
> > configured)
> > > > that
> > > > => I
> > > > => >> can
> > > > => >> > use for fencing? if the anwser is yes, where hell must I
> > configure
> > > > it?
> > > > => I
> > > > => >> > don=B4t see wher can I do it.
> > > > => >> > If I haven=B4t a fencing device which one do you recommed to
> > use?
> > > > => >> > Thanks in advance
> > > > => >> > ESG
> > > > => >> >
> > > > => >> > --
> > > > => >> > Linux-cluster mailing list
> > > > => >> > Linux-cluster at redhat.com
> > > > => >> > https://www.redhat.com/mailman/listinfo/linux-cluster
> > > > => >> >
> > > > => >>
> > > > => >> Yes you have IPMI and if you are using 1950 Dell, DRAC should be
> > > > there
> > > > => >> too. You can see if you have DRAC or not when the server starts
> > and
> > > > => >> before the loading of the OS.
> > > > => >>
> > > > => >> I have 1850s and I am using DRAC for fencing.
> > > > => >>
> > > > => >>
> > > > => >> Paras.
> > > > => >>
> > > > => >> --
> > > > => >> Linux-cluster mailing list
> > > > => >> Linux-cluster at redhat.com
> > > > => >> https://www.redhat.com/mailman/listinfo/linux-cluster
> > > > => >>
> > > > => >
> > > > => >
> > > >
> > > >
> > > >
> > > > --
> > > > Linux-cluster mailing list
> > > > Linux-cluster at redhat.com
> > > > https://www.redhat.com/mailman/listinfo/linux-cluster
> > > >
> > > -------------- next part --------------
> > > An HTML attachment was scrubbed...
> > > URL:
> > https://www.redhat.com/archives/linux-cluster/attachments/20090813/a4558d27/attachment.html
> > >
> > > ------------------------------
> > >
> > > Message: 2
> > > Date: Thu, 13 Aug 2009 14:45:13 +0200
> > > From: mark benschop <mark.benschop.lists at gmail.com>
> > > Subject: [Linux-cluster] clusterservice stays in 'recovering' state
> > > To: linux-cluster at redhat.com
> > > Message-ID:
> > > <f97c3a70908130545n11ce442ej17d74c9cdc450e45 at mail.gmail.com>
> > > Content-Type: text/plain; charset="iso-8859-1"
> > >
> > > Hi All,
> > >
> > > I've a problem with a clusterservice. The service was started up while
> > one
> > > of the resources, an NFS, export was not accessible.
> > > Therefore the service never started up right but got into the
> > 'recovering'
> > > state.
> > > In the mean time the NFS exports are setup properly but to no avail.
> > > Stopping the clusterservice, using clusvcadm -d <service>, will result in
> > > the service going down but staying in the 'recovering' state.
> > > Starting it again doesn't work. The service doesn't start and stays in
> > the
> > > recovery status.
> > > I'm suspecting rgmanager lost it somehow.
> > >
> > > Anybody had any ideas on what could be the problem and how to resolve it
> > ?
> > >
> > > Thanks in advance,
> > > Mark
> > > -------------- next part --------------
> > > An HTML attachment was scrubbed...
> > > URL:
> > https://www.redhat.com/archives/linux-cluster/attachments/20090813/31731cd3/attachment.html
> > >
> > > ------------------------------
> > >
> > > Message: 3
> > > Date: Thu, 13 Aug 2009 16:13:12 +0300
> > > From: Hakan VELIOGLU <veliogluh at itu.edu.tr>
> > > Subject: Re: [Linux-cluster] Is there any backup heartbeat channel
> > > To: linux-cluster at redhat.com
> > > Message-ID: <20090813161312.11546h2sp6psr814 at webmail.itu.edu.tr>
> > > Content-Type: text/plain; charset=ISO-8859-9; DelSp="Yes";
> > > format="flowed"
> > >
> > > Thanks for all the answers.
> > >
> > > I think there is realy no backup heartbeat channel. Maybe the reason
> > > is GFS. DLM works on the heartbeat channel. If you lost your heartbeat
> > > you lose your lock consistency so it is better to fence the other
> > > node. For this reason I think if you don't have enough network
> > > interface on server and switch, loosing the heartbeat network may shut
> > > all the cluster members.
> > >
> > > Hakan VEL?O?LU
> > >
> > >
> > > ----- robejrm at gmail.com den ileti ---------
> > > Tarih: Thu, 13 Aug 2009 10:42:11 +0200
> > > Kimden: Juan Ramon Martin Blanco <robejrm at gmail.com>
> > > Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
> > > Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
> > > Kime: linux clustering <linux-cluster at redhat.com>
> > >
> > >
> > > > 2009/8/13 Hakan VELIOGLU <veliogluh at itu.edu.tr>
> > > >
> > > >> ----- raju.rajsand at gmail.com den ileti ---------
> > > >> Tarih: Thu, 13 Aug 2009 08:57:15 +0530
> > > >> Kimden: Rajagopal Swaminathan <raju.rajsand at gmail.com>
> > > >> Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
> > > >> Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
> > > >> Kime: linux clustering <linux-cluster at redhat.com>
> > > >>
> > > >>
> > > >> Greetings,
> > > >>>
> > > >>> 2009/8/12 Hakan VELIOGLU <veliogluh at itu.edu.tr>:
> > > >>>
> > > >>>> Hi list,
> > > >>>>
> > > >>>> I am trying a two node cluster with RH 5.3 on Sun X4150 hardware. I
> > use a
> > > >>>>
> > > >>>
> > > >>> IIRC, Sun x4150 has four ethernet ports. Two can be used for outside
> > > >>> networking and two can be bonded and used for heartbeat.
> > > >>>
> > > >> I think, I couldn't explain my networking. I use two ethernet ports
> > for xen
> > > >> vm which are trunk and bonded ports. Then there left two. Our network
> > > >> topology (which is out of my control) available for one port for
> > server
> > > >> control (SSH).
> > > >
> > > > So you can't use a bonded port for both server management and cluster
> > > > communications, can you? You can configure an active-passive bonding
> > and
> > > > then you can have many virtual interfaces on top of that, i.e: bond0:0,
> > > > bond0:1 and assign them the ip addesses you need.
> > > >
> > > >
> > > > I use the other one with a cross over cable for heartbeat. So there is
> > no
> > > >> way for bonding these two interfaces. Of course if I buy an extra
> > switch I
> > > >> may do this.
> > > >
> > > > You can connect them to the same switch (though you lost kind of
> > > > redundancy), or you can use two crossover cables and move the
> > management IP
> > > > to the same ports you are using for the vm's.
> > > >
> > > > Greetings,
> > > > Juanra
> > > >
> > > >>
> > > >> I don't realy understand why there is no backup heartbeat channel. LVS
> > and
> > > >> MS cluster has this ability.
> > > >>
> > > >>>
> > > >>> ALOM can be used for fencing and can be on a seperate subnet if
> > required.
> > > >>>
> > > >> I used this for fencing_ipmilan.
> > > >>
> > > >>>
> > > >>> Regards
> > > >>>
> > > >>> Rajagopal
> > > >>>
> > > >>> --
> > > >>> Linux-cluster mailing list
> > > >>> Linux-cluster at redhat.com
> > > >>> https://www.redhat.com/mailman/listinfo/linux-cluster
> > > >>>
> > > >>>
> > > >>
> > > >> ----- raju.rajsand at gmail.com den iletiyi bitir -----
> > > >>
> > > >>
> > > >>
> > > >>
> > > >> --
> > > >> Linux-cluster mailing list
> > > >> Linux-cluster at redhat.com
> > > >> https://www.redhat.com/mailman/listinfo/linux-cluster
> > > >>
> > > >
> > >
> > >
> > > ----- robejrm at gmail.com den iletiyi bitir -----
> > >
> > >
> > >
> > >
> > >
> > > ------------------------------
> > >
> > > Message: 4
> > > Date: Thu, 13 Aug 2009 15:29:43 +0200
> > > From: Juan Ramon Martin Blanco <robejrm at gmail.com>
> > > Subject: Re: [Linux-cluster] Is there any backup heartbeat channel
> > > To: linux clustering <linux-cluster at redhat.com>
> > > Message-ID:
> > > <8a5668960908130629n6ec05a88n463a3b03da331dae at mail.gmail.com>
> > > Content-Type: text/plain; charset="iso-8859-9"
> > >
> > > 2009/8/13 Hakan VELIOGLU <veliogluh at itu.edu.tr>
> > >
> > > > Thanks for all the answers.
> > > >
> > > > I think there is realy no backup heartbeat channel. Maybe the reason is
> > > > GFS. DLM works on the heartbeat channel. If you lost your heartbeat you
> > lose
> > > > your lock consistency so it is better to fence the other node. For this
> > > > reason I think if you don't have enough network interface on server and
> > > > switch, loosing the heartbeat network may shut all the cluster members.
> > > >
> > > There is no backup heartbeat channel because you should do the backup at
> > a
> > > operating system level, i.e: bonding
> > > That's why you should use a bonded interface for the heartbeat channel
> > with
> > > at least 2 ethernet slaves; going further (for better redundancy) each of
> > > the slaves should be on a different network card and you should connect
> > the
> > > each slave to a different switch.
> > > But what I am trying to explain, is that you can use that bonded logical
> > > interface also for things different from hearbeat. ;)
> > >
> > > Greetings,
> > > Juanra
> > >
> > >
> > > > Hakan VEL?O?LU
> > > >
> > > >
> > > > ----- robejrm at gmail.com den ileti ---------
> > > > Tarih: Thu, 13 Aug 2009 10:42:11 +0200
> > > > Kimden: Juan Ramon Martin Blanco <robejrm at gmail.com>
> > > >
> > > > Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
> > > > Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
> > > > Kime: linux clustering <linux-cluster at redhat.com>
> > > >
> > > >
> > > > 2009/8/13 Hakan VELIOGLU <veliogluh at itu.edu.tr>
> > > >>
> > > >> ----- raju.rajsand at gmail.com den ileti ---------
> > > >>> Tarih: Thu, 13 Aug 2009 08:57:15 +0530
> > > >>> Kimden: Rajagopal Swaminathan <raju.rajsand at gmail.com>
> > > >>> Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
> > > >>> Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
> > > >>> Kime: linux clustering <linux-cluster at redhat.com>
> > > >>>
> > > >>>
> > > >>> Greetings,
> > > >>>
> > > >>>>
> > > >>>> 2009/8/12 Hakan VELIOGLU <veliogluh at itu.edu.tr>:
> > > >>>>
> > > >>>> Hi list,
> > > >>>>>
> > > >>>>> I am trying a two node cluster with RH 5.3 on Sun X4150 hardware. I
> > use
> > > >>>>> a
> > > >>>>>
> > > >>>>>
> > > >>>> IIRC, Sun x4150 has four ethernet ports. Two can be used for outside
> > > >>>> networking and two can be bonded and used for heartbeat.
> > > >>>>
> > > >>>> I think, I couldn't explain my networking. I use two ethernet ports
> > for
> > > >>> xen
> > > >>> vm which are trunk and bonded ports. Then there left two. Our network
> > > >>> topology (which is out of my control) available for one port for
> > server
> > > >>> control (SSH).
> > > >>>
> > > >>
> > > >> So you can't use a bonded port for both server management and cluster
> > > >> communications, can you? You can configure an active-passive bonding
> > and
> > > >> then you can have many virtual interfaces on top of that, i.e:
> > bond0:0,
> > > >> bond0:1 and assign them the ip addesses you need.
> > > >>
> > > >>
> > > >> I use the other one with a cross over cable for heartbeat. So there is
> > no
> > > >>
> > > >>> way for bonding these two interfaces. Of course if I buy an extra
> > switch
> > > >>> I
> > > >>> may do this.
> > > >>>
> > > >>
> > > >> You can connect them to the same switch (though you lost kind of
> > > >> redundancy), or you can use two crossover cables and move the
> > management
> > > >> IP
> > > >> to the same ports you are using for the vm's.
> > > >>
> > > >> Greetings,
> > > >> Juanra
> > > >>
> > > >>
> > > >>> I don't realy understand why there is no backup heartbeat channel.
> > LVS
> > > >>> and
> > > >>> MS cluster has this ability.
> > > >>>
> > > >>>
> > > >>>> ALOM can be used for fencing and can be on a seperate subnet if
> > > >>>> required.
> > > >>>>
> > > >>>> I used this for fencing_ipmilan.
> > > >>>
> > > >>>
> > > >>>> Regards
> > > >>>>
> > > >>>> Rajagopal
> > > >>>>
> > > >>>> --
> > > >>>> Linux-cluster mailing list
> > > >>>> Linux-cluster at redhat.com
> > > >>>> https://www.redhat.com/mailman/listinfo/linux-cluster
> > > >>>>
> > > >>>>
> > > >>>>
> > > >>> ----- raju.rajsand at gmail.com den iletiyi bitir -----
> > > >>>
> > > >>>
> > > >>>
> > > >>>
> > > >>> --
> > > >>> Linux-cluster mailing list
> > > >>> Linux-cluster at redhat.com
> > > >>> https://www.redhat.com/mailman/listinfo/linux-cluster
> > > >>>
> > > >>>
> > > >>
> > > >
> > > > ----- robejrm at gmail.com den iletiyi bitir -----
> > > >
> > > >
> > > >
> > > >
> > > > --
> > > > Linux-cluster mailing list
> > > > Linux-cluster at redhat.com
> > > > https://www.redhat.com/mailman/listinfo/linux-cluster
> > > >
> > > -------------- next part --------------
> > > An HTML attachment was scrubbed...
> > > URL:
> > https://www.redhat.com/archives/linux-cluster/attachments/20090813/cd4dc079/attachment.html
> > >
> > > ------------------------------
> > >
> > > Message: 5
> > > Date: Thu, 13 Aug 2009 10:07:13 -0400
> > > From: Nehemias Jahcob <nehemiasjahcob at gmail.com>
> > > Subject: [Linux-cluster] RHCS on KVM
> > > To: linux clustering <linux-cluster at redhat.com>
> > > Message-ID:
> > > <5f61ab380908130707q5c936504k7351d0d6b3459090 at mail.gmail.com>
> > > Content-Type: text/plain; charset="iso-8859-1"
> > >
> > > Hi.
> > >
> > > How to create a cluster of 2 nodes in rhel5.4 (or Fedora 10) with KVM?
> > >
> > > With XEN follow this guide:
> > > http://sources.redhat.com/cluster/wiki/VMClusterCookbook?highlight =
> > > (CategoryHowTo).
> > >
> > > Do you have a guide to implementation of RHCS in KVM?
> > >
> > > Thank you all.
> > > NJ
> > > -------------- next part --------------
> > > An HTML attachment was scrubbed...
> > > URL:
> > https://www.redhat.com/archives/linux-cluster/attachments/20090813/f3a69a80/attachment.html
> > >
> > > ------------------------------
> > >
> > > Message: 6
> > > Date: Thu, 13 Aug 2009 14:16:47 +0000
> > > From: NTOUGHE GUY-SERGE <ntoughe at hotmail.com>
> > > Subject: [Linux-cluster] Cman hang
> > > To: <linux-cluster at redhat.com>
> > > Message-ID: <BAY119-W410E2F250E8B461752CFC9A5050 at phx.gbl>
> > > Content-Type: text/plain; charset="iso-8859-1"
> > >
> > >
> > >
> > > Hi gurus,
> > >
> > > i installed RHEL 5.3 on 2 servers which participating to a cluster
> > composed of these 2 nodes:
> > > kernel version:
> > > kernel-headers-2.6.18-128.el5
> > > kernel-devel-2.6.18-128.el5
> > > kernel-2.6.18-128.el5
> > > cman-devel-2.0.98-1.el5_3.1
> > > cman-2.0.98-1.el5_3.1
> > > cluster-cim-0.12.1-2.el5
> > > lvm2-cluster-2.02.40-7.el5
> > > cluster-snmp-0.12.1-2.el5
> > > modcluster-0.12.1-2.el5
> > > When i want to start cman the following message is sent:
> > > cman not started: Multicast and node address families differ.
> > /usr/sbin/cman_tool: aisexec daemon didn't start
> > > [FAILED]
> > >
> > > I trier to mount gfs2
> > > and i got theses messages:
> > > # mount -t gfs2 /dev/VolGroup01/LogVol01 /appli/prod --o
> > lockTablename=arvclust:/appli/prod, Lockproto=lock_dlm
> > >
> > > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> > >
> > > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> > >
> > > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> > >
> > > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> > >
> > > do you have any clues?
> > > Please it's an hurry, i waste long time to lok for solution help
> > > regards
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > > ntoughe at hotmail.com
> > >
> > >
> > > _________________________________________________________________
> > > With Windows Live, you can organize, edit, and share your photos.
> > >
> > http://www.microsoft.com/middleeast/windows/windowslive/products/photo-gallery-edit.aspx
> > > -------------- next part --------------
> > > An HTML attachment was scrubbed...
> > > URL:
> > https://www.redhat.com/archives/linux-cluster/attachments/20090813/0a55101d/attachment.html
> > >
> > > ------------------------------
> > >
> > > Message: 7
> > > Date: Thu, 13 Aug 2009 09:14:24 -0500
> > > From: David Teigland <teigland at redhat.com>
> > > Subject: Re: [Linux-cluster] gfs2 mount hangs
> > > To: Wengang Wang <wen.gang.wang at oracle.com>
> > > Cc: linux clustering <linux-cluster at redhat.com>
> > > Message-ID: <20090813141424.GA8148 at redhat.com>
> > > Content-Type: text/plain; charset=us-ascii
> > >
> > > On Thu, Aug 13, 2009 at 02:22:11PM +0800, Wengang Wang wrote:
> > > > <cman two_node="1" expected_votes="2"/>
> > >
> > > That's not a valid combination, two_node="1" requires expected_votes="1".
> > >
> > > You didn't mention which userspace cluster version/release you're using,
> > or
> > > include any status about the cluster. Before trying to mount gfs on
> > either
> > > node, collect from both nodes:
> > >
> > > cman_tool status
> > > cman_tool nodes
> > > group_tool
> > >
> > > Then mount on the first node and collect the same information, then try
> > > mounting on the second node, collect the same information, and look for
> > any
> > > errors in /var/log/messages.
> > >
> > > Since you're using new kernels, you need to be using the cluster 3.0
> > userspace
> > > code. You're using the old manual fencing config. There is no more
> > > fence_manual; the new way to configure manual fencing is to not configure
> > any
> > > fencing at all. So, your cluster.conf should look like this:
> > >
> > > <?xml version="1.0"?>
> > > <cluster name="testgfs2" config_version="1">
> > > <cman two_node="1" expected_votes="1"/>
> > > <clusternodes>
> > > <clusternode name="cool" nodeid="1"/>
> > > <clusternode name="desk" nodeid="2"/>
> > > </clusternodes>
> > > </cluster>
> > >
> > > Dave
> > >
> > >
> > >
> > > ------------------------------
> > >
> > > Message: 8
> > > Date: Thu, 13 Aug 2009 10:39:46 -0400
> > > From: Lon Hohberger <lhh at redhat.com>
> > > Subject: Re: [Linux-cluster] Qdisk question
> > > To: linux clustering <linux-cluster at redhat.com>
> > > Message-ID: <1250174386.23376.1440.camel at localhost.localdomain>
> > > Content-Type: text/plain
> > >
> > > On Thu, 2009-08-13 at 00:45 +0200, brem belguebli wrote:
> > >
> > > > My understanding of qdisk is that it is used as a tie-breaker, but it
> > > > looks like it is more a heatbeat vector than a simple tie-breaker.
> > >
> > > Right, it's a secondary membership algorithm.
> > >
> > >
> > > > Until here, no real problem indeed, if the site gets apart from the
> > > > other prod site and also from the third site (hosting the iscsi target
> > > > qdisk) the 2 nodes from the failing site get evicted from the cluster.
> > > >
> > > >
> > > > But, what if my third site gets isolated while the 2 prod ones are
> > > > fine ?
> > >
> > > Qdisk votes will not be presented to CMAN any more, but the two sites
> > > should remain online if they still have a "majority" of votes.
> > >
> > >
> > > > The real question is what happens in case all the nodes loose access
> > > > to the qdisk while they're still able to see each others ?
> > >
> > > Qdisk is just a vote like other voting mechanisms. If all nodes lose
> > > access at the same time, it should behave like a node death. However,
> > > the default action if _one_ node loses access is to kill that node (even
> > > if CMAN still sees it).
> > >
> > >
> > > > The 4 nodes have each 1 vote and the qdisk 1 vote. The expected quorum
> > > > is 3.
> > >
> > >
> > > > If I loose the qdisk, the number of votes falls to 4, the cluster is
> > > > quorate (4>3) but it looks like everything goes bad, each node
> > > > deactivate itself as it can't write its alive status (--> heartbeat
> > > > vector) to the qdisk even if the network heartbeating is working
> > > > fine.
> > >
> > > What happens specifically? Most of the actions qdiskd performs are
> > > configurable. For example, if the nodes are rebooting, you can turn
> > > that behavior off.
> > >
> > >
> > >
> > > I wrote a simple 'ping' tiebreaker based the behaviors in RHEL3. It
> > > functions in many ways in the same manner as qdiskd with respect to vote
> > > advertisement to CMAN, but without needing a disk - maybe you would find
> > > it useful?
> > >
> > > http://people.redhat.com/lhh/qnet.tar.gz
> > >
> > > -- Lon
> > >
> > >
> > >
> > > ------------------------------
> > >
> > > Message: 9
> > > Date: Thu, 13 Aug 2009 17:02:15 +0200
> > > From: Juan Ramon Martin Blanco <robejrm at gmail.com>
> > > Subject: Re: [Linux-cluster] Cman hang
> > > To: linux clustering <linux-cluster at redhat.com>
> > > Message-ID:
> > > <8a5668960908130802p4f5168cbueda86d1e6f1324bb at mail.gmail.com>
> > > Content-Type: text/plain; charset="iso-8859-1"
> > >
> > > On Thu, Aug 13, 2009 at 4:16 PM, NTOUGHE GUY-SERGE <ntoughe at hotmail.com
> > >wrote:
> > >
> > > >
> > > > Hi gurus,
> > > >
> > > > i installed RHEL 5.3 on 2 servers which participating to a cluster
> > > > composed of these 2 nodes:
> > > > kernel version:
> > > > kernel-headers-2.6.18-128.el5
> > > > kernel-devel-2.6.18-128.el5
> > > > kernel-2.6.18-128.el5
> > > > cman-devel-2.0.98-1.el5_3.1
> > > > cman-2.0.98-1.el5_3.1
> > > > cluster-cim-0.12.1-2.el5
> > > > lvm2-cluster-2.02.40-7.el5
> > > > cluster-snmp-0.12.1-2.el5
> > > > modcluster-0.12.1-2.el5
> > > > When i want to start cman the following message is sent:
> > > > cman not started: Multicast and node address families differ.
> > > > /usr/sbin/cman_tool: aisexec daemon didn't start
> > > > [FAILED]
> > > >
> > > Please, show us your cluster.conf file so we can help.
> > >
> > > Regards,
> > > Juanra
> > >
> > > >
> > > > I trier to mount gfs2
> > > > and i got theses messages:
> > > > # mount -t gfs2 /dev/VolGroup01/LogVol01 /appli/prod --o
> > > > lockTablename=arvclust:/appli/prod, Lockproto=lock_dlm
> > > >
> > > > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> > > >
> > > > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> > > >
> > > > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> > > >
> > > > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> > > >
> > > > do you have any clues?
> > > > Please it's an hurry, i waste long time to lok for solution help
> > > > regards
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > > ntoughe at hotmail.com
> > > >
> > > >
> > > >
> > > > ------------------------------
> > > > With Windows Live, you can organize, edit, and share your photos.<
> > http://www.microsoft.com/middleeast/windows/windowslive/products/photo-gallery-edit.aspx
> > >
> > > >
> > > > --
> > > > Linux-cluster mailing list
> > > > Linux-cluster at redhat.com
> > > > https://www.redhat.com/mailman/listinfo/linux-cluster
> > > >
> > > -------------- next part --------------
> > > An HTML attachment was scrubbed...
> > > URL:
> > https://www.redhat.com/archives/linux-cluster/attachments/20090813/9ecbcab1/attachment.html
> > >
> > > ------------------------------
> > >
> > > --
> > > Linux-cluster mailing list
> > > Linux-cluster at redhat.com
> > > https://www.redhat.com/mailman/listinfo/linux-cluster
> > >
> > > End of Linux-cluster Digest, Vol 64, Issue 16
> > > *********************************************
> >
> > ------------------------------
> > See all the ways you can stay connected to friends and family<http://www.microsoft.com/windows/windowslive/default.aspx>
> >
> > --
> > Linux-cluster mailing list
> > Linux-cluster at redhat.com
> > https://www.redhat.com/mailman/listinfo/linux-cluster
> >
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: https://www.redhat.com/archives/linux-cluster/attachments/20090813/f9557411/attachment.html
> 
> ------------------------------
> 
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
> 
> End of Linux-cluster Digest, Vol 64, Issue 18
> *********************************************

_________________________________________________________________
Share your memories online with anyone you want.
http://www.microsoft.com/middleeast/windows/windowslive/products/photos-share.aspx?tab=1
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090813/b25ff822/attachment.htm>

From robejrm at gmail.com  Thu Aug 13 16:10:58 2009
From: robejrm at gmail.com (Juan Ramon Martin Blanco)
Date: Thu, 13 Aug 2009 18:10:58 +0200
Subject: [Linux-cluster] Cman hang
In-Reply-To: <BAY119-W4089E3353F6D3368CBADE0A5050@phx.gbl>
References: <20090813152421.A4A86619076@hormel.redhat.com>
	<BAY119-W4089E3353F6D3368CBADE0A5050@phx.gbl>
Message-ID: <8a5668960908130910x7f83c293q44a015e44fe3570a@mail.gmail.com>

On Thu, Aug 13, 2009 at 5:45 PM, NTOUGHE GUY-SERGE <ntoughe at hotmail.com>wrote:

>  Thank's Junra,
> I did it , and i tried to restart the cman service  without more success,
> thise are the messages i got:
>
> my new cluster.conf:
> <?xml version="1.0"?>
> <cluster alias="arevclust" config_version="1" name="arevclust">
>   <clusternodes>
>   <cman expected_votes="1" two_node="1">
>   </cman>
>   <clusternode name="gs21spli003.occ.lan" nodeid="1" votes="1">
>   </clusternode>
>   <clusternode name="gs21spli004.occ.lan" nodeid="2" votes="1">
>
You should put the node names in /etc/hosts to be resolved to the IP
addresses configured on the network interfaces you want to be used for
cluster comms.

Greetings,
Juanra

>
>   </clusternode>
>   </clusternodes>
> </cluster>
> ~
> j
>
> cman not started: Multicast and node address families differ.
> /usr/sbin/cman_tool: aisexec daemon didn't start
> when i mounted the gfs FS i got this:
> # mount -t gfs2 /dev/mapper/VolGroup01-LogVol01 /appli/prod
> /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
>
> I 'm not sure but i have a doubt on the lockTableName  during  the FS GFS
> creation:
> mkfs.gfs2 -p lock_dlm -t arevclust:appli/prod -j 1
> /dev/mapper/VolGroup00-LogVol01,  the name of the cluster is: arevclust
> The gfs File system is made by /dev/mapper/volGroup00-LogVol01 and the
> mounting point is /appli/prod, so if i have to precise my gfs file i put :
> appli/prod is it correct?
> regards
>
>
>
> ntoughe at hotmail.com
>
>
>
>
> > From: linux-cluster-request at redhat.com
> > Subject: Linux-cluster Digest, Vol 64, Issue 18
> > To: linux-cluster at redhat.com
> > Date: Thu, 13 Aug 2009 11:24:21 -0400
> >
> > Send Linux-cluster mailing list submissions to
> > linux-cluster at redhat.com
> >
> > To subscribe or unsubscribe via the World Wide Web, visit
> > https://www.redhat.com/mailman/listinfo/linux-cluster
> > or, via email, send a message with subject or body 'help' to
> > linux-cluster-request at redhat.com
> >
> > You can reach the person managing the list at
> > linux-cluster-owner at redhat.com
> >
> > When replying, please edit your Subject line so it is more specific
> > than "Re: Contents of Linux-cluster digest..."
> >
> >
> > Today's Topics:
> >
> > 1. Re: Qdisk question (brem belguebli)
> > 2. Re: Cman hang (Juan Ramon Martin Blanco)
> >
> >
> > ----------------------------------------------------------------------
> >
> > Message: 1
> > Date: Thu, 13 Aug 2009 17:23:16 +0200
> > From: brem belguebli <brem.belguebli at gmail.com>
> > Subject: Re: [Linux-cluster] Qdisk question
> > To: linux clustering <linux-cluster at redhat.com>
> > Message-ID:
> > <29ae894c0908130823i65667021vdc840ae1f0ded134 at mail.gmail.com>
> > Content-Type: text/plain; charset="iso-8859-1"
> >
> > Hi Lon and Thanks for this reply.
> >
> > In fact, thinking about it, my test wasn't very much representative of
> what
> > I was expecting to do.
> >
> > I blocked the qdisk communications to only one node which, after reading
> > your reply, kind of confirmed me that I did the wong test. I'm going to
> re
> > run it by blocking all the nodes to the qdisk.
> >
> > I'll also try your ping tie-breaker.
> >
> > Brem
> >
> >
> > 2009/8/13, Lon Hohberger <lhh at redhat.com>:
> > >
> > > On Thu, 2009-08-13 at 00:45 +0200, brem belguebli wrote:
> > >
> > > > My understanding of qdisk is that it is used as a tie-breaker, but it
> > > > looks like it is more a heatbeat vector than a simple tie-breaker.
> > >
> > > Right, it's a secondary membership algorithm.
> > >
> > >
> > > > Until here, no real problem indeed, if the site gets apart from the
> > > > other prod site and also from the third site (hosting the iscsi
> target
> > > > qdisk) the 2 nodes from the failing site get evicted from the
> cluster.
> > > >
> > > >
> > > > But, what if my third site gets isolated while the 2 prod ones are
> > > > fine ?
> > >
> > > Qdisk votes will not be presented to CMAN any more, but the two sites
> > > should remain online if they still have a "majority" of votes.
> > >
> > >
> > > > The real question is what happens in case all the nodes loose access
> > > > to the qdisk while they're still able to see each others ?
> > >
> > > Qdisk is just a vote like other voting mechanisms. If all nodes lose
> > > access at the same time, it should behave like a node death. However,
> > > the default action if _one_ node loses access is to kill that node
> (even
> > > if CMAN still sees it).
> > >
> > >
> > > > The 4 nodes have each 1 vote and the qdisk 1 vote. The expected
> quorum
> > > > is 3.
> > >
> > >
> > > > If I loose the qdisk, the number of votes falls to 4, the cluster is
> > > > quorate (4>3) but it looks like everything goes bad, each node
> > > > deactivate itself as it can't write its alive status (--> heartbeat
> > > > vector) to the qdisk even if the network heartbeating is working
> > > > fine.
> > >
> > > What happens specifically? Most of the actions qdiskd performs are
> > > configurable. For example, if the nodes are rebooting, you can turn
> > > that behavior off.
> > >
> > >
> > >
> > > I wrote a simple 'ping' tiebreaker based the behaviors in RHEL3. It
> > > functions in many ways in the same manner as qdiskd with respect to
> vote
> > > advertisement to CMAN, but without needing a disk - maybe you would
> find
> > > it useful?
> > >
> > > http://people.redhat.com/lhh/qnet.tar.gz
> > >
> > > -- Lon
> > >
> > > --
> > > Linux-cluster mailing list
> > > Linux-cluster at redhat.com
> > > https://www.redhat.com/mailman/listinfo/linux-cluster
> > >
> > -------------- next part --------------
> > An HTML attachment was scrubbed...
> > URL:
> https://www.redhat.com/archives/linux-cluster/attachments/20090813/5482c8bf/attachment.html
> >
> > ------------------------------
> >
> > Message: 2
> > Date: Thu, 13 Aug 2009 17:23:54 +0200
> > From: Juan Ramon Martin Blanco <robejrm at gmail.com>
> > Subject: Re: [Linux-cluster] Cman hang
> > To: linux clustering <linux-cluster at redhat.com>
> > Message-ID:
> > <8a5668960908130823u11b46ad9pd1da5af3614ce3d3 at mail.gmail.com>
> > Content-Type: text/plain; charset="iso-8859-1"
> >
> > On Thu, Aug 13, 2009 at 5:13 PM, NTOUGHE GUY-SERGE <ntoughe at hotmail.com
> >wrote:
> >
> > > Hi, this is my cluster.conf
> > > <?xml version="1.0"?>
> > > <cluster alias="arevclust" config_version="21" name="arevclust">
> > > <fence_daemon clean_start="0" post_fail_delay="0" post_join_delay="3"/>
> > > <clusternodes>
> > > <clusternode name="host1" nodeid="1" votes="1">
> > > <fence>
> > > <method name="2">
> > >
> >
> > > <device name=""/>
> > >
> > You should configure a valid fencing method, and if you don't have any,
> use
> > fence_manual until you get it.
> >
> > >
> > > </method>
> > > </fence>
> > > <multicast addr="" interface=""/>
> > >
> > I am not sure, but I think you should erase this <multicast ....> tag
> >
> > Greetings,
> > Juanra
> >
> > >
> > > </clusternode>
> > >
> >
> > > <clusternode name="host2" nodeid="2" votes="1">
> > > <fence>
> > > <method name="1">
> > > <device name=""/>
> > > </method>
> > > <method name=""/>
> > > </fence>
> > > <multicast addr="" interface=""/>
> > > </clusternode>
> > > </clusternodes>
> > > <cman expected_votes="" two_node="">
> > > <multicast addr=""/>
> > > </cman>
> > > <fencedevices>
> > > <fencedevice agent="fence_brocade" ipaddr="" login="" name=""
> passwd=""/>
> > > </fencedevices>
> > > <rm>
> > > <failoverdomains>
> > > </failoverdomains>
> > > <resources>
> > > </resources>
> > > </rm>
> > > </cluster>
> > >
> > > Regards
> > >
> > >
> > >
> > >
> > > ntoughe at hotmail.com
> > >
> > >
> > >
> > >
> > > > From: linux-cluster-request at redhat.com
> > > > Subject: Linux-cluster Digest, Vol 64, Issue 16
> > > > To: linux-cluster at redhat.com
> > > > Date: Thu, 13 Aug 2009 11:02:36 -0400
> > > >
> > > > Send Linux-cluster mailing list submissions to
> > > > linux-cluster at redhat.com
> > > >
> > > > To subscribe or unsubscribe via the World Wide Web, visit
> > > > https://www.redhat.com/mailman/listinfo/linux-cluster
> > > > or, via email, send a message with subject or body 'help' to
> > > > linux-cluster-request at redhat.com
> > > >
> > > > You can reach the person managing the list at
> > > > linux-cluster-owner at redhat.com
> > > >
> > > > When replying, please edit your Subject line so it is more specific
> > > > than "Re: Contents of Linux-cluster digest..."
> > > >
> > > >
> > > > Today's Topics:
> > > >
> > > > 1. Re: do I have a fence DRAC device? (ESGLinux)
> > > > 2. clusterservice stays in 'recovering' state (mark benschop)
> > > > 3. Re: Is there any backup heartbeat channel (Hakan VELIOGLU)
> > > > 4. Re: Is there any backup heartbeat channel
> > > > (Juan Ramon Martin Blanco)
> > > > 5. RHCS on KVM (Nehemias Jahcob)
> > > > 6. Cman hang (NTOUGHE GUY-SERGE)
> > > > 7. Re: gfs2 mount hangs (David Teigland)
> > > > 8. Re: Qdisk question (Lon Hohberger)
> > > > 9. Re: Cman hang (Juan Ramon Martin Blanco)
> > > >
> > > >
> > > >
> ----------------------------------------------------------------------
> > > >
> > > > Message: 1
> > > > Date: Thu, 13 Aug 2009 13:27:16 +0200
> > > > From: ESGLinux <esggrupos at gmail.com>
> > > > Subject: Re: [Linux-cluster] do I have a fence DRAC device?
> > > > To: linux clustering <linux-cluster at redhat.com>
> > > > Message-ID:
> > > > <3128ba140908130427i6ab85406ye6da34073e6a6e97 at mail.gmail.com>
> > > > Content-Type: text/plain; charset="iso-8859-1"
> > > >
> > > > Hi,
> > > > I couldn?t reboot my system yet but I have installed the openmanage
> > > > packages:
> > > >
> > > > srvadmin-omacore-5.4.0-260
> > > > srvadmin-iws-5.4.0-260
> > > > srvadmin-syscheck-5.4.0-260
> > > > srvadmin-rac5-components-5.4.0-260
> > > > srvadmin-deng-5.4.0-260
> > > > srvadmin-ipmi-5.4.0-260.DUP
> > > > srvadmin-racadm5-5.4.0-260
> > > > srvadmin-omauth-5.4.0-260.rhel5
> > > > srvadmin-hapi-5.4.0-260
> > > > srvadmin-cm-5.4.0-260
> > > > srvadmin-racdrsc5-5.4.0-260
> > > > srvadmin-omilcore-5.4.0-260
> > > > srvadmin-isvc-5.4.0-260
> > > > srvadmin-storage-5.4.0-260
> > > > srvadmin-jre-5.4.0-260
> > > > srvadmin-omhip-5.4.0-260
> > > >
> > > > Now I have the command racadm but when I try to execut it I get this:
> > > >
> > > > racadm config -g cfgSerial -o cfgSerialTelnetEnable 1
> > > > ERROR: RACADM is unable to process the requested subcommand because
> there
> > > is
> > > > no
> > > > local RAC configuration to communicate with.
> > > >
> > > > Local RACADM subcommand execution requires the following:
> > > >
> > > > 1. A Remote Access Controller (RAC) must be present on the managed
> server
> > > > 2. Appropriate managed node software must be installed and running on
> the
> > > > server
> > > >
> > > >
> > > > What do I need to install/start? or until I configure the bios I
> can?t
> > > get
> > > > this work?
> > > >
> > > > Greetings
> > > >
> > > > ESG
> > > >
> > > >
> > > > 2009/8/11 <bergman at merctech.com>
> > > >
> > > > >
> > > > >
> > > > > In the message dated: Tue, 11 Aug 2009 14:14:03 +0200,
> > > > > The pithy ruminations from Juan Ramon Martin Blanco on
> > > > > <Re: [Linux-cluster] do I have a fence DRAC device?> were:
> > > > > => --===============1917368601==
> > > > > => Content-Type: multipart/alternative;
> > > > > boundary=0016364c7c07663f600470dca3b8
> > > > > =>
> > > > > => --0016364c7c07663f600470dca3b8
> > > > > => Content-Type: text/plain; charset=ISO-8859-1
> > > > > => Content-Transfer-Encoding: quoted-printable
> > > > > =>
> > > > > => On Tue, Aug 11, 2009 at 2:03 PM, ESGLinux <esggrupos at gmail.com>
> > > wrote:
> > > > > =>
> > > > > => > Thanks
> > > > > => > I=B4ll check it when I could reboot the server.
> > > > > => >
> > > > > => > greetings,
> > > > > => >
> > > > > => You have a BMC ipmi in the first network interface, it can be
> > > configured
> > > > > at
> > > > > => boot time (I don't remember if inside the BIOS or pressing
> > > > > cntrl+something
> > > > > => during boot)
> > > > > =>
> > > > >
> > > > > Based on my notes, here's how I configured the DRAC interface on a
> Dell
> > > > > 1950
> > > > > for use as a fence device:
> > > > >
> > > > > Configuring the card from Linux depending on the installation of
> > > > > Dell's
> > > > > OMSA package. Once that's installed, use the following
> > > > > commands:
> > > > >
> > > > > racadm config -g cfgSerial -o cfgSerialTelnetEnable 1
> > > > > racadm config -g cfgLanNetworking -o cfgDNSRacName
> > > > > HOSTNAME_FOR_INTERFACE
> > > > > racadm config -g cfgDNSDomainName DOMAINNAME_FOR_INTERFACE
> > > > > racadm config -g cfgUserAdmin -o cfgUserAdminPassword -i 2
> > > > > PASSWORD
> > > > > racadm config -g cfgNicEnable 1
> > > > > racadm config -g cfgNicIpAddress WWW.XXX.YYY.ZZZ
> > > > > racadm config -g cfgNicNetmask WWW.XXX.YYY.ZZZ
> > > > > racadm config -g cfgNicGateway WWW.XXX.YYY.ZZZ
> > > > > racadm config -g cfgNicUseDhcp 0
> > > > >
> > > > >
> > > > > I also save a backup of the configuration with:
> > > > >
> > > > > racadm getconfig -f ~/drac_config
> > > > >
> > > > >
> > > > > Hope this helps,
> > > > >
> > > > > Mark
> > > > >
> > > > > ----
> > > > > Mark Bergman voice: 215-662-7310
> > > > > mark.bergman at uphs.upenn.edu fax: 215-614-0266
> > > > > System Administrator Section of Biomedical Image Analysis
> > > > > Department of Radiology University of Pennsylvania
> > > > > PGP Key: https://www.rad.upenn.edu/sbia/bergman
> > > > >
> > > > >
> > > > > => Greetings,
> > > > > => Juanra
> > > > > =>
> > > > > => >
> > > > > => > ESG
> > > > > => >
> > > > > => > 2009/8/10 Paras pradhan <pradhanparas at gmail.com>
> > > > > => >
> > > > > => > On Mon, Aug 10, 2009 at 5:24 AM, ESGLinux<esggrupos at gmail.com
> >
> > > wrote:
> > > > > => >> > Hi all,
> > > > > => >> > I was designing a 2 node cluster and I was going to use 2
> > > servers
> > > > > DELL
> > > > > => >> > PowerEdge 1950. I was going to buy a DRAC card to use for
> > > fencing
> > > > > but
> > > > > => >> > running several commands in the servers I have noticed that
> > > when I
> > > > > run
> > > > > => >> this
> > > > > => >> > command:
> > > > > => >> > #ipmitool lan print
> > > > > => >> > Set in Progress : Set Complete
> > > > > => >> > Auth Type Support : NONE MD2 MD5 PASSWORD
> > > > > => >> > Auth Type Enable : Callback : MD2 MD5
> > > > > => >> > : User : MD2 MD5
> > > > > => >> > : Operator : MD2 MD5
> > > > > => >> > : Admin : MD2 MD5
> > > > > => >> > : OEM : MD2 MD5
> > > > > => >> > IP Address Source : Static Address
> > > > > => >> > IP Address : 0.0.0.0
> > > > > => >> > Subnet Mask : 0.0.0.0
> > > > > => >> > MAC Address : 00:1e:c9:ae:6f:7e
> > > > > => >> > SNMP Community String : public
> > > > > => >> > IP Header : TTL=0x40 Flags=0x40 Precedence=0x00 TOS=0x10
> > > > > => >> > Default Gateway IP : 0.0.0.0
> > > > > => >> > Default Gateway MAC : 00:00:00:00:00:00
> > > > > => >> > Backup Gateway IP : 0.0.0.0
> > > > > => >> > Backup Gateway MAC : 00:00:00:00:00:00
> > > > > => >> > 802.1q VLAN ID : Disabled
> > > > > => >> > 802.1q VLAN Priority : 0
> > > > > => >> > RMCP+ Cipher Suites : 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14
> > > > > => >> > Cipher Suite Priv Max : aaaaaaaaaaaaaaa
> > > > > => >> > : X=Cipher Suite Unused
> > > > > => >> > : c=CALLBACK
> > > > > => >> > : u=USER
> > > > > => >> > : o=OPERATOR
> > > > > => >> > : a=ADMIN
> > > > > => >> > : O=OEM
> > > > > => >> > does this mean that I already have an ipmi card (not
> > > configured)
> > > > > that
> > > > > => I
> > > > > => >> can
> > > > > => >> > use for fencing? if the anwser is yes, where hell must I
> > > configure
> > > > > it?
> > > > > => I
> > > > > => >> > don=B4t see wher can I do it.
> > > > > => >> > If I haven=B4t a fencing device which one do you recommed
> to
> > > use?
> > > > > => >> > Thanks in advance
> > > > > => >> > ESG
> > > > > => >> >
> > > > > => >> > --
> > > > > => >> > Linux-cluster mailing list
> > > > > => >> > Linux-cluster at redhat.com
> > > > > => >> > https://www.redhat.com/mailman/listinfo/linux-cluster
> > > > > => >> >
> > > > > => >>
> > > > > => >> Yes you have IPMI and if you are using 1950 Dell, DRAC should
> be
> > > > > there
> > > > > => >> too. You can see if you have DRAC or not when the server
> starts
> > > and
> > > > > => >> before the loading of the OS.
> > > > > => >>
> > > > > => >> I have 1850s and I am using DRAC for fencing.
> > > > > => >>
> > > > > => >>
> > > > > => >> Paras.
> > > > > => >>
> > > > > => >> --
> > > > > => >> Linux-cluster mailing list
> > > > > => >> Linux-cluster at redhat.com
> > > > > => >> https://www.redhat.com/mailman/listinfo/linux-cluster
> > > > > => >>
> > > > > => >
> > > > > => >
> > > > >
> > > > >
> > > > >
> > > > > --
> > > > > Linux-cluster mailing list
> > > > > Linux-cluster at redhat.com
> > > > > https://www.redhat.com/mailman/listinfo/linux-cluster
> > > > >
> > > > -------------- next part --------------
> > > > An HTML attachment was scrubbed...
> > > > URL:
> > >
> https://www.redhat.com/archives/linux-cluster/attachments/20090813/a4558d27/attachment.html
> > > >
> > > > ------------------------------
> > > >
> > > > Message: 2
> > > > Date: Thu, 13 Aug 2009 14:45:13 +0200
> > > > From: mark benschop <mark.benschop.lists at gmail.com>
> > > > Subject: [Linux-cluster] clusterservice stays in 'recovering' state
> > > > To: linux-cluster at redhat.com
> > > > Message-ID:
> > > > <f97c3a70908130545n11ce442ej17d74c9cdc450e45 at mail.gmail.com>
> > > > Content-Type: text/plain; charset="iso-8859-1"
> > > >
> > > > Hi All,
> > > >
> > > > I've a problem with a clusterservice. The service was started up
> while
> > > one
> > > > of the resources, an NFS, export was not accessible.
> > > > Therefore the service never started up right but got into the
> > > 'recovering'
> > > > state.
> > > > In the mean time the NFS exports are setup properly but to no avail.
> > > > Stopping the clusterservice, using clusvcadm -d <service>, will
> result in
> > > > the service going down but staying in the 'recovering' state.
> > > > Starting it again doesn't work. The service doesn't start and stays
> in
> > > the
> > > > recovery status.
> > > > I'm suspecting rgmanager lost it somehow.
> > > >
> > > > Anybody had any ideas on what could be the problem and how to resolve
> it
> > > ?
> > > >
> > > > Thanks in advance,
> > > > Mark
> > > > -------------- next part --------------
> > > > An HTML attachment was scrubbed...
> > > > URL:
> > >
> https://www.redhat.com/archives/linux-cluster/attachments/20090813/31731cd3/attachment.html
> > > >
> > > > ------------------------------
> > > >
> > > > Message: 3
> > > > Date: Thu, 13 Aug 2009 16:13:12 +0300
> > > > From: Hakan VELIOGLU <veliogluh at itu.edu.tr>
> > > > Subject: Re: [Linux-cluster] Is there any backup heartbeat channel
> > > > To: linux-cluster at redhat.com
> > > > Message-ID: <20090813161312.11546h2sp6psr814 at webmail.itu.edu.tr>
> > > > Content-Type: text/plain; charset=ISO-8859-9; DelSp="Yes";
> > > > format="flowed"
> > > >
> > > > Thanks for all the answers.
> > > >
> > > > I think there is realy no backup heartbeat channel. Maybe the reason
> > > > is GFS. DLM works on the heartbeat channel. If you lost your
> heartbeat
> > > > you lose your lock consistency so it is better to fence the other
> > > > node. For this reason I think if you don't have enough network
> > > > interface on server and switch, loosing the heartbeat network may
> shut
> > > > all the cluster members.
> > > >
> > > > Hakan VEL?O?LU
> > > >
> > > >
> > > > ----- robejrm at gmail.com den ileti ---------
> > > > Tarih: Thu, 13 Aug 2009 10:42:11 +0200
> > > > Kimden: Juan Ramon Martin Blanco <robejrm at gmail.com>
> > > > Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
> > > > Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
> > > > Kime: linux clustering <linux-cluster at redhat.com>
> > > >
> > > >
> > > > > 2009/8/13 Hakan VELIOGLU <veliogluh at itu.edu.tr>
> > > > >
> > > > >> ----- raju.rajsand at gmail.com den ileti ---------
> > > > >> Tarih: Thu, 13 Aug 2009 08:57:15 +0530
> > > > >> Kimden: Rajagopal Swaminathan <raju.rajsand at gmail.com>
> > > > >> Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
> > > > >> Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
> > > > >> Kime: linux clustering <linux-cluster at redhat.com>
> > > > >>
> > > > >>
> > > > >> Greetings,
> > > > >>>
> > > > >>> 2009/8/12 Hakan VELIOGLU <veliogluh at itu.edu.tr>:
> > > > >>>
> > > > >>>> Hi list,
> > > > >>>>
> > > > >>>> I am trying a two node cluster with RH 5.3 on Sun X4150
> hardware. I
> > > use a
> > > > >>>>
> > > > >>>
> > > > >>> IIRC, Sun x4150 has four ethernet ports. Two can be used for
> outside
> > > > >>> networking and two can be bonded and used for heartbeat.
> > > > >>>
> > > > >> I think, I couldn't explain my networking. I use two ethernet
> ports
> > > for xen
> > > > >> vm which are trunk and bonded ports. Then there left two. Our
> network
> > > > >> topology (which is out of my control) available for one port for
> > > server
> > > > >> control (SSH).
> > > > >
> > > > > So you can't use a bonded port for both server management and
> cluster
> > > > > communications, can you? You can configure an active-passive
> bonding
> > > and
> > > > > then you can have many virtual interfaces on top of that, i.e:
> bond0:0,
> > > > > bond0:1 and assign them the ip addesses you need.
> > > > >
> > > > >
> > > > > I use the other one with a cross over cable for heartbeat. So there
> is
> > > no
> > > > >> way for bonding these two interfaces. Of course if I buy an extra
> > > switch I
> > > > >> may do this.
> > > > >
> > > > > You can connect them to the same switch (though you lost kind of
> > > > > redundancy), or you can use two crossover cables and move the
> > > management IP
> > > > > to the same ports you are using for the vm's.
> > > > >
> > > > > Greetings,
> > > > > Juanra
> > > > >
> > > > >>
> > > > >> I don't realy understand why there is no backup heartbeat channel.
> LVS
> > > and
> > > > >> MS cluster has this ability.
> > > > >>
> > > > >>>
> > > > >>> ALOM can be used for fencing and can be on a seperate subnet if
> > > required.
> > > > >>>
> > > > >> I used this for fencing_ipmilan.
> > > > >>
> > > > >>>
> > > > >>> Regards
> > > > >>>
> > > > >>> Rajagopal
> > > > >>>
> > > > >>> --
> > > > >>> Linux-cluster mailing list
> > > > >>> Linux-cluster at redhat.com
> > > > >>> https://www.redhat.com/mailman/listinfo/linux-cluster
> > > > >>>
> > > > >>>
> > > > >>
> > > > >> ----- raju.rajsand at gmail.com den iletiyi bitir -----
> > > > >>
> > > > >>
> > > > >>
> > > > >>
> > > > >> --
> > > > >> Linux-cluster mailing list
> > > > >> Linux-cluster at redhat.com
> > > > >> https://www.redhat.com/mailman/listinfo/linux-cluster
> > > > >>
> > > > >
> > > >
> > > >
> > > > ----- robejrm at gmail.com den iletiyi bitir -----
> > > >
> > > >
> > > >
> > > >
> > > >
> > > > ------------------------------
> > > >
> > > > Message: 4
> > > > Date: Thu, 13 Aug 2009 15:29:43 +0200
> > > > From: Juan Ramon Martin Blanco <robejrm at gmail.com>
> > > > Subject: Re: [Linux-cluster] Is there any backup heartbeat channel
> > > > To: linux clustering <linux-cluster at redhat.com>
> > > > Message-ID:
> > > > <8a5668960908130629n6ec05a88n463a3b03da331dae at mail.gmail.com>
> > > > Content-Type: text/plain; charset="iso-8859-9"
> > > >
> > > > 2009/8/13 Hakan VELIOGLU <veliogluh at itu.edu.tr>
> > > >
> > > > > Thanks for all the answers.
> > > > >
> > > > > I think there is realy no backup heartbeat channel. Maybe the
> reason is
> > > > > GFS. DLM works on the heartbeat channel. If you lost your heartbeat
> you
> > > lose
> > > > > your lock consistency so it is better to fence the other node. For
> this
> > > > > reason I think if you don't have enough network interface on server
> and
> > > > > switch, loosing the heartbeat network may shut all the cluster
> members.
> > > > >
> > > > There is no backup heartbeat channel because you should do the backup
> at
> > > a
> > > > operating system level, i.e: bonding
> > > > That's why you should use a bonded interface for the heartbeat
> channel
> > > with
> > > > at least 2 ethernet slaves; going further (for better redundancy)
> each of
> > > > the slaves should be on a different network card and you should
> connect
> > > the
> > > > each slave to a different switch.
> > > > But what I am trying to explain, is that you can use that bonded
> logical
> > > > interface also for things different from hearbeat. ;)
> > > >
> > > > Greetings,
> > > > Juanra
> > > >
> > > >
> > > > > Hakan VEL?O?LU
> > > > >
> > > > >
> > > > > ----- robejrm at gmail.com den ileti ---------
> > > > > Tarih: Thu, 13 Aug 2009 10:42:11 +0200
> > > > > Kimden: Juan Ramon Martin Blanco <robejrm at gmail.com>
> > > > >
> > > > > Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
> > > > > Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
> > > > > Kime: linux clustering <linux-cluster at redhat.com>
> > > > >
> > > > >
> > > > > 2009/8/13 Hakan VELIOGLU <veliogluh at itu.edu.tr>
> > > > >>
> > > > >> ----- raju.rajsand at gmail.com den ileti ---------
> > > > >>> Tarih: Thu, 13 Aug 2009 08:57:15 +0530
> > > > >>> Kimden: Rajagopal Swaminathan <raju.rajsand at gmail.com>
> > > > >>> Yan?t Adresi:linux clustering <linux-cluster at redhat.com>
> > > > >>> Konu: Re: [Linux-cluster] Is there any backup heartbeat channel
> > > > >>> Kime: linux clustering <linux-cluster at redhat.com>
> > > > >>>
> > > > >>>
> > > > >>> Greetings,
> > > > >>>
> > > > >>>>
> > > > >>>> 2009/8/12 Hakan VELIOGLU <veliogluh at itu.edu.tr>:
> > > > >>>>
> > > > >>>> Hi list,
> > > > >>>>>
> > > > >>>>> I am trying a two node cluster with RH 5.3 on Sun X4150
> hardware. I
> > > use
> > > > >>>>> a
> > > > >>>>>
> > > > >>>>>
> > > > >>>> IIRC, Sun x4150 has four ethernet ports. Two can be used for
> outside
> > > > >>>> networking and two can be bonded and used for heartbeat.
> > > > >>>>
> > > > >>>> I think, I couldn't explain my networking. I use two ethernet
> ports
> > > for
> > > > >>> xen
> > > > >>> vm which are trunk and bonded ports. Then there left two. Our
> network
> > > > >>> topology (which is out of my control) available for one port for
> > > server
> > > > >>> control (SSH).
> > > > >>>
> > > > >>
> > > > >> So you can't use a bonded port for both server management and
> cluster
> > > > >> communications, can you? You can configure an active-passive
> bonding
> > > and
> > > > >> then you can have many virtual interfaces on top of that, i.e:
> > > bond0:0,
> > > > >> bond0:1 and assign them the ip addesses you need.
> > > > >>
> > > > >>
> > > > >> I use the other one with a cross over cable for heartbeat. So
> there is
> > > no
> > > > >>
> > > > >>> way for bonding these two interfaces. Of course if I buy an extra
> > > switch
> > > > >>> I
> > > > >>> may do this.
> > > > >>>
> > > > >>
> > > > >> You can connect them to the same switch (though you lost kind of
> > > > >> redundancy), or you can use two crossover cables and move the
> > > management
> > > > >> IP
> > > > >> to the same ports you are using for the vm's.
> > > > >>
> > > > >> Greetings,
> > > > >> Juanra
> > > > >>
> > > > >>
> > > > >>> I don't realy understand why there is no backup heartbeat
> channel.
> > > LVS
> > > > >>> and
> > > > >>> MS cluster has this ability.
> > > > >>>
> > > > >>>
> > > > >>>> ALOM can be used for fencing and can be on a seperate subnet if
> > > > >>>> required.
> > > > >>>>
> > > > >>>> I used this for fencing_ipmilan.
> > > > >>>
> > > > >>>
> > > > >>>> Regards
> > > > >>>>
> > > > >>>> Rajagopal
> > > > >>>>
> > > > >>>> --
> > > > >>>> Linux-cluster mailing list
> > > > >>>> Linux-cluster at redhat.com
> > > > >>>> https://www.redhat.com/mailman/listinfo/linux-cluster
> > > > >>>>
> > > > >>>>
> > > > >>>>
> > > > >>> ----- raju.rajsand at gmail.com den iletiyi bitir -----
> > > > >>>
> > > > >>>
> > > > >>>
> > > > >>>
> > > > >>> --
> > > > >>> Linux-cluster mailing list
> > > > >>> Linux-cluster at redhat.com
> > > > >>> https://www.redhat.com/mailman/listinfo/linux-cluster
> > > > >>>
> > > > >>>
> > > > >>
> > > > >
> > > > > ----- robejrm at gmail.com den iletiyi bitir -----
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > --
> > > > > Linux-cluster mailing list
> > > > > Linux-cluster at redhat.com
> > > > > https://www.redhat.com/mailman/listinfo/linux-cluster
> > > > >
> > > > -------------- next part --------------
> > > > An HTML attachment was scrubbed...
> > > > URL:
> > >
> https://www.redhat.com/archives/linux-cluster/attachments/20090813/cd4dc079/attachment.html
> > > >
> > > > ------------------------------
> > > >
> > > > Message: 5
> > > > Date: Thu, 13 Aug 2009 10:07:13 -0400
> > > > From: Nehemias Jahcob <nehemiasjahcob at gmail.com>
> > > > Subject: [Linux-cluster] RHCS on KVM
> > > > To: linux clustering <linux-cluster at redhat.com>
> > > > Message-ID:
> > > > <5f61ab380908130707q5c936504k7351d0d6b3459090 at mail.gmail.com>
> > > > Content-Type: text/plain; charset="iso-8859-1"
> > > >
> > > > Hi.
> > > >
> > > > How to create a cluster of 2 nodes in rhel5.4 (or Fedora 10) with
> KVM?
> > > >
> > > > With XEN follow this guide:
> > > > http://sources.redhat.com/cluster/wiki/VMClusterCookbook?highlight =
> > > > (CategoryHowTo).
> > > >
> > > > Do you have a guide to implementation of RHCS in KVM?
> > > >
> > > > Thank you all.
> > > > NJ
> > > > -------------- next part --------------
> > > > An HTML attachment was scrubbed...
> > > > URL:
> > >
> https://www.redhat.com/archives/linux-cluster/attachments/20090813/f3a69a80/attachment.html
> > > >
> > > > ------------------------------
> > > >
> > > > Message: 6
> > > > Date: Thu, 13 Aug 2009 14:16:47 +0000
> > > > From: NTOUGHE GUY-SERGE <ntoughe at hotmail.com>
> > > > Subject: [Linux-cluster] Cman hang
> > > > To: <linux-cluster at redhat.com>
> > > > Message-ID: <BAY119-W410E2F250E8B461752CFC9A5050 at phx.gbl>
> > > > Content-Type: text/plain; charset="iso-8859-1"
> > > >
> > > >
> > > >
> > > > Hi gurus,
> > > >
> > > > i installed RHEL 5.3 on 2 servers which participating to a cluster
> > > composed of these 2 nodes:
> > > > kernel version:
> > > > kernel-headers-2.6.18-128.el5
> > > > kernel-devel-2.6.18-128.el5
> > > > kernel-2.6.18-128.el5
> > > > cman-devel-2.0.98-1.el5_3.1
> > > > cman-2.0.98-1.el5_3.1
> > > > cluster-cim-0.12.1-2.el5
> > > > lvm2-cluster-2.02.40-7.el5
> > > > cluster-snmp-0.12.1-2.el5
> > > > modcluster-0.12.1-2.el5
> > > > When i want to start cman the following message is sent:
> > > > cman not started: Multicast and node address families differ.
> > > /usr/sbin/cman_tool: aisexec daemon didn't start
> > > > [FAILED]
> > > >
> > > > I trier to mount gfs2
> > > > and i got theses messages:
> > > > # mount -t gfs2 /dev/VolGroup01/LogVol01 /appli/prod --o
> > > lockTablename=arvclust:/appli/prod, Lockproto=lock_dlm
> > > >
> > > > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> > > >
> > > > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> > > >
> > > > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> > > >
> > > > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> > > >
> > > > do you have any clues?
> > > > Please it's an hurry, i waste long time to lok for solution help
> > > > regards
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > > ntoughe at hotmail.com
> > > >
> > > >
> > > > _________________________________________________________________
> > > > With Windows Live, you can organize, edit, and share your photos.
> > > >
> > >
> http://www.microsoft.com/middleeast/windows/windowslive/products/photo-gallery-edit.aspx
> > > > -------------- next part --------------
> > > > An HTML attachment was scrubbed...
> > > > URL:
> > >
> https://www.redhat.com/archives/linux-cluster/attachments/20090813/0a55101d/attachment.html
> > > >
> > > > ------------------------------
> > > >
> > > > Message: 7
> > > > Date: Thu, 13 Aug 2009 09:14:24 -0500
> > > > From: David Teigland <teigland at redhat.com>
> > > > Subject: Re: [Linux-cluster] gfs2 mount hangs
> > > > To: Wengang Wang <wen.gang.wang at oracle.com>
> > > > Cc: linux clustering <linux-cluster at redhat.com>
> > > > Message-ID: <20090813141424.GA8148 at redhat.com>
> > > > Content-Type: text/plain; charset=us-ascii
> > > >
> > > > On Thu, Aug 13, 2009 at 02:22:11PM +0800, Wengang Wang wrote:
> > > > > <cman two_node="1" expected_votes="2"/>
> > > >
> > > > That's not a valid combination, two_node="1" requires
> expected_votes="1".
> > > >
> > > > You didn't mention which userspace cluster version/release you're
> using,
> > > or
> > > > include any status about the cluster. Before trying to mount gfs on
> > > either
> > > > node, collect from both nodes:
> > > >
> > > > cman_tool status
> > > > cman_tool nodes
> > > > group_tool
> > > >
> > > > Then mount on the first node and collect the same information, then
> try
> > > > mounting on the second node, collect the same information, and look
> for
> > > any
> > > > errors in /var/log/messages.
> > > >
> > > > Since you're using new kernels, you need to be using the cluster 3.0
> > > userspace
> > > > code. You're using the old manual fencing config. There is no more
> > > > fence_manual; the new way to configure manual fencing is to not
> configure
> > > any
> > > > fencing at all. So, your cluster.conf should look like this:
> > > >
> > > > <?xml version="1.0"?>
> > > > <cluster name="testgfs2" config_version="1">
> > > > <cman two_node="1" expected_votes="1"/>
> > > > <clusternodes>
> > > > <clusternode name="cool" nodeid="1"/>
> > > > <clusternode name="desk" nodeid="2"/>
> > > > </clusternodes>
> > > > </cluster>
> > > >
> > > > Dave
> > > >
> > > >
> > > >
> > > > ------------------------------
> > > >
> > > > Message: 8
> > > > Date: Thu, 13 Aug 2009 10:39:46 -0400
> > > > From: Lon Hohberger <lhh at redhat.com>
> > > > Subject: Re: [Linux-cluster] Qdisk question
> > > > To: linux clustering <linux-cluster at redhat.com>
> > > > Message-ID: <1250174386.23376.1440.camel at localhost.localdomain>
> > > > Content-Type: text/plain
> > > >
> > > > On Thu, 2009-08-13 at 00:45 +0200, brem belguebli wrote:
> > > >
> > > > > My understanding of qdisk is that it is used as a tie-breaker, but
> it
> > > > > looks like it is more a heatbeat vector than a simple tie-breaker.
> > > >
> > > > Right, it's a secondary membership algorithm.
> > > >
> > > >
> > > > > Until here, no real problem indeed, if the site gets apart from the
> > > > > other prod site and also from the third site (hosting the iscsi
> target
> > > > > qdisk) the 2 nodes from the failing site get evicted from the
> cluster.
> > > > >
> > > > >
> > > > > But, what if my third site gets isolated while the 2 prod ones are
> > > > > fine ?
> > > >
> > > > Qdisk votes will not be presented to CMAN any more, but the two sites
> > > > should remain online if they still have a "majority" of votes.
> > > >
> > > >
> > > > > The real question is what happens in case all the nodes loose
> access
> > > > > to the qdisk while they're still able to see each others ?
> > > >
> > > > Qdisk is just a vote like other voting mechanisms. If all nodes lose
> > > > access at the same time, it should behave like a node death. However,
> > > > the default action if _one_ node loses access is to kill that node
> (even
> > > > if CMAN still sees it).
> > > >
> > > >
> > > > > The 4 nodes have each 1 vote and the qdisk 1 vote. The expected
> quorum
> > > > > is 3.
> > > >
> > > >
> > > > > If I loose the qdisk, the number of votes falls to 4, the cluster
> is
> > > > > quorate (4>3) but it looks like everything goes bad, each node
> > > > > deactivate itself as it can't write its alive status (--> heartbeat
> > > > > vector) to the qdisk even if the network heartbeating is working
> > > > > fine.
> > > >
> > > > What happens specifically? Most of the actions qdiskd performs are
> > > > configurable. For example, if the nodes are rebooting, you can turn
> > > > that behavior off.
> > > >
> > > >
> > > >
> > > > I wrote a simple 'ping' tiebreaker based the behaviors in RHEL3. It
> > > > functions in many ways in the same manner as qdiskd with respect to
> vote
> > > > advertisement to CMAN, but without needing a disk - maybe you would
> find
> > > > it useful?
> > > >
> > > > http://people.redhat.com/lhh/qnet.tar.gz
> > > >
> > > > -- Lon
> > > >
> > > >
> > > >
> > > > ------------------------------
> > > >
> > > > Message: 9
> > > > Date: Thu, 13 Aug 2009 17:02:15 +0200
> > > > From: Juan Ramon Martin Blanco <robejrm at gmail.com>
> > > > Subject: Re: [Linux-cluster] Cman hang
> > > > To: linux clustering <linux-cluster at redhat.com>
> > > > Message-ID:
> > > > <8a5668960908130802p4f5168cbueda86d1e6f1324bb at mail.gmail.com>
> > > > Content-Type: text/plain; charset="iso-8859-1"
> > > >
> > > > On Thu, Aug 13, 2009 at 4:16 PM, NTOUGHE GUY-SERGE <
> ntoughe at hotmail.com
> > > >wrote:
> > > >
> > > > >
> > > > > Hi gurus,
> > > > >
> > > > > i installed RHEL 5.3 on 2 servers which participating to a cluster
> > > > > composed of these 2 nodes:
> > > > > kernel version:
> > > > > kernel-headers-2.6.18-128.el5
> > > > > kernel-devel-2.6.18-128.el5
> > > > > kernel-2.6.18-128.el5
> > > > > cman-devel-2.0.98-1.el5_3.1
> > > > > cman-2.0.98-1.el5_3.1
> > > > > cluster-cim-0.12.1-2.el5
> > > > > lvm2-cluster-2.02.40-7.el5
> > > > > cluster-snmp-0.12.1-2.el5
> > > > > modcluster-0.12.1-2.el5
> > > > > When i want to start cman the following message is sent:
> > > > > cman not started: Multicast and node address families differ.
> > > > > /usr/sbin/cman_tool: aisexec daemon didn't start
> > > > > [FAILED]
> > > > >
> > > > Please, show us your cluster.conf file so we can help.
> > > >
> > > > Regards,
> > > > Juanra
> > > >
> > > > >
> > > > > I trier to mount gfs2
> > > > > and i got theses messages:
> > > > > # mount -t gfs2 /dev/VolGroup01/LogVol01 /appli/prod --o
> > > > > lockTablename=arvclust:/appli/prod, Lockproto=lock_dlm
> > > > >
> > > > > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> > > > >
> > > > > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> > > > >
> > > > > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> > > > >
> > > > > /sbin/mount.gfs2: can't connect to gfs_controld: Connection refused
> > > > >
> > > > > do you have any clues?
> > > > > Please it's an hurry, i waste long time to lok for solution help
> > > > > regards
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > ntoughe at hotmail.com
> > > > >
> > > > >
> > > > >
> > > > > ------------------------------
> > > > > With Windows Live, you can organize, edit, and share your photos.<
> > >
> http://www.microsoft.com/middleeast/windows/windowslive/products/photo-gallery-edit.aspx
> > > >
> > > > >
> > > > > --
> > > > > Linux-cluster mailing list
> > > > > Linux-cluster at redhat.com
> > > > > https://www.redhat.com/mailman/listinfo/linux-cluster
> > > > >
> > > > -------------- next part --------------
> > > > An HTML attachment was scrubbed...
> > > > URL:
> > >
> https://www.redhat.com/archives/linux-cluster/attachments/20090813/9ecbcab1/attachment.html
> > > >
> > > > ------------------------------
> > > >
> > > > --
> > > > Linux-cluster mailing list
> > > > Linux-cluster at redhat.com
> > > > https://www.redhat.com/mailman/listinfo/linux-cluster
> > > >
> > > > End of Linux-cluster Digest, Vol 64, Issue 16
> > > > *********************************************
> > >
> > > ------------------------------
> > > See all the ways you can stay connected to friends and family<
> http://www.microsoft.com/windows/windowslive/default.aspx>
> > >
> > > --
> > > Linux-cluster mailing list
> > > Linux-cluster at redhat.com
> > > https://www.redhat.com/mailman/listinfo/linux-cluster
> > >
> > -------------- next part --------------
> > An HTML attachment was scrubbed...
> > URL:
> https://www.redhat.com/archives/linux-cluster/attachments/20090813/f9557411/attachment.html
> >
> > ------------------------------
> >
> > --
> > Linux-cluster mailing list
> > Linux-cluster at redhat.com
> > https://www.redhat.com/mailman/listinfo/linux-cluster
> >
> > End of Linux-cluster Digest, Vol 64, Issue 18
> > *********************************************
>
> ------------------------------
> Share your memories online with anyone you want anyone you want.<http://www.microsoft.com/middleeast/windows/windowslive/products/photos-share.aspx?tab=1>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090813/4e49e3c1/attachment.htm>

From wen.gang.wang at oracle.com  Fri Aug 14 02:30:17 2009
From: wen.gang.wang at oracle.com (Wengang Wang)
Date: Fri, 14 Aug 2009 10:30:17 +0800
Subject: [Linux-cluster] gfs2 mount hangs
In-Reply-To: <20090813141424.GA8148@redhat.com>
References: <4A83B113.60609@oracle.com> <20090813141424.GA8148@redhat.com>
Message-ID: <4A84CC39.3080809@oracle.com>

Hi Dave,

Thanks for your rely!

I changed the cluster.conf as you suggested on both nodes. and 
re-formatted the block device by
'mkfs.gfs2 -t testgfs2:1 -j 2 /dev/sdb'.
the following are the info needed.

I) user space utils version:
[root at cool ~]# rpm -qi cman
Name        : cman                         Relocations: (not relocatable)
Version     : 3.0.0                             Vendor: Fedora Project
Release     : 20.fc11                       Build Date: Wed 29 Jul 2009 
01:27:54 AM CST
Install Date: Tue 04 Aug 2009 05:43:32 PM CST      Build Host: 
xenbuilder4.fedora.phx.redhat.com
Group       : System Environment/Base       Source RPM: 
cluster-3.0.0-20.fc11.src.rpm
Size        : 1146269                          License: GPLv2+ and LGPLv2+
Signature   : RSA/8, Wed 29 Jul 2009 08:35:33 PM CST, Key ID 
1dc5c758d22e77f2
Packager    : Fedora Project
URL         : http://sources.redhat.com/cluster/wiki/
Summary     : Red Hat Cluster Manager
Description :
Red Hat Cluster Manager

[root at cool ~]# rpm -qi gfs2-utils
Name        : gfs2-utils                   Relocations: (not relocatable)
Version     : 3.0.0                             Vendor: Fedora Project
Release     : 20.fc11                       Build Date: Wed 29 Jul 2009 
01:27:54 AM CST
Install Date: Tue 04 Aug 2009 05:46:17 PM CST      Build Host: 
xenbuilder4.fedora.phx.redhat.com
Group       : System Environment/Kernel     Source RPM: 
cluster-3.0.0-20.fc11.src.rpm
Size        : 682088                           License: GPLv2+ and LGPLv2+
Signature   : RSA/8, Wed 29 Jul 2009 08:36:42 PM CST, Key ID 
1dc5c758d22e77f2
Packager    : Fedora Project
URL         : http://sources.redhat.com/cluster/wiki/
Summary     : Utilities for managing the global filesystem (GFS2)
Description :
The gfs2-utils package contains a number of utilities for creating,
checking, modifying, and correcting any inconsistencies in GFS2
filesystems

anything else needed?

II) status outputs before and after mounts
'cman_tool status' before the mounts:
[root at desk ~]# cman_tool status
Version: 6.2.0
Config Version: 1
Cluster Name: testgfs2
Cluster Id: 28360
Cluster Member: Yes
Cluster Generation: 88
Membership state: Cluster-Member
Nodes: 2
Expected votes: 1
Total votes: 2
Node votes: 1
Quorum: 1
Active subsystems: 7
Flags: 2node HaveState
Ports Bound: 0
Node name: desk
Node ID: 2
Multicast addresses: 239.192.110.55
Node addresses: 192.168.1.104

#don't know where 239.192.110.55 comes from. does it matter?

[root at cool ~]# cman_tool status
Version: 6.2.0
Config Version: 1
Cluster Name: testgfs2
Cluster Id: 28360
Cluster Member: Yes
Cluster Generation: 88
Membership state: Cluster-Member
Nodes: 2
Expected votes: 1
Total votes: 2
Node votes: 1
Quorum: 1
Active subsystems: 7
Flags: 2node HaveState
Ports Bound: 0
Node name: cool
Node ID: 1
Multicast addresses: 239.192.110.55
Node addresses: 192.168.1.101

'cman_tool nodes' before the mounts:
[root at desk ~]# cman_tool nodes
Node  Sts   Inc   Joined               Name
    1   M     88   2009-08-14 17:44:00  cool
    2   M     76   2009-08-14 17:43:52  desk

[root at cool ~]# cman_tool nodes
Node  Sts   Inc   Joined               Name
    1   M     84   2009-08-14 09:46:06  cool
    2   M     88   2009-08-14 09:46:06  desk

'group_tool' before the mounts:
[root at desk ~]# group_tool
groupd not running
fence domain
member count  2
victim count  0
victim now    0
master nodeid 2
wait state    none
members       1 2

[root at cool ~]# group_tool
groupd not running
fence domain
member count  2
victim count  0
victim now    0
master nodeid 2
wait state    none
members       1 2
----

'cman_tool status' after the first mount(on node desk):
[root at desk ~]#  cman_tool status
Version: 6.2.0
Config Version: 1
Cluster Name: testgfs2
Cluster Id: 28360
Cluster Member: Yes
Cluster Generation: 88
Membership state: Cluster-Member
Nodes: 2
Expected votes: 1
Total votes: 2
Node votes: 1
Quorum: 1
Active subsystems: 7
Flags: 2node HaveState
Ports Bound: 0
Node name: desk
Node ID: 2
Multicast addresses: 239.192.110.55
Node addresses: 192.168.1.104

[root at cool ~]# cman_tool status
Version: 6.2.0
Config Version: 1
Cluster Name: testgfs2
Cluster Id: 28360
Cluster Member: Yes
Cluster Generation: 88
Membership state: Cluster-Member
Nodes: 2
Expected votes: 1
Total votes: 2
Node votes: 1
Quorum: 1
Active subsystems: 7
Flags: 2node HaveState
Ports Bound: 0
Node name: cool
Node ID: 1
Multicast addresses: 239.192.110.55
Node addresses: 192.168.1.101

'cman_tool nodes' after the first mount:
[root at desk ~]# cman_tool nodes
Node  Sts   Inc   Joined               Name
    1   M     88   2009-08-14 17:44:00  cool
    2   M     76   2009-08-14 17:43:52  desk

[root at cool ~]# cman_tool nodes
Node  Sts   Inc   Joined               Name
    1   M     84   2009-08-14 09:46:06  cool
    2   M     88   2009-08-14 09:46:06  desk

'group_tool' after the first mount:
[root at desk ~]# group_tool
groupd not running
fence domain
member count  2
victim count  0
victim now    0
master nodeid 2
wait state    none
members       1 2

dlm lockspaces
name          1
id            0xe5ab8ad6
flags         0x00000000
change        member 1 joined 1 remove 0 failed 0 seq 1,1
members       2

[root at cool ~]# group_tool
groupd not running
fence domain
member count  2
victim count  0
victim now    0
master nodeid 2
wait state    none
members       1 2
---

'cman_tool status' after the second mount(no node cool, hang)
[root at desk ~]#  cman_tool status
Version: 6.2.0
Config Version: 1
Cluster Name: testgfs2
Cluster Id: 28360
Cluster Member: Yes
Cluster Generation: 88
Membership state: Cluster-Member
Nodes: 2
Expected votes: 1
Total votes: 2
Node votes: 1
Quorum: 1
Active subsystems: 7
Flags: 2node HaveState
Ports Bound: 0
Node name: desk
Node ID: 2
Multicast addresses: 239.192.110.55
Node addresses: 192.168.1.104

[root at cool ~]# cman_tool status
Version: 6.2.0
Config Version: 1
Cluster Name: testgfs2
Cluster Id: 28360
Cluster Member: Yes
Cluster Generation: 88
Membership state: Cluster-Member
Nodes: 2
Expected votes: 1
Total votes: 2
Node votes: 1
Quorum: 1
Active subsystems: 7
Flags: 2node HaveState
Ports Bound: 0
Node name: cool
Node ID: 1
Multicast addresses: 239.192.110.55
Node addresses: 192.168.1.101

'cman_tool nodes' after the second mount
[root at desk ~]# cman_tool nodes
Node  Sts   Inc   Joined               Name
    1   M     88   2009-08-14 17:44:00  cool
    2   M     76   2009-08-14 17:43:52  desk

[root at cool ~]#  cman_tool nodes
Node  Sts   Inc   Joined               Name
    1   M     84   2009-08-14 09:46:06  cool
    2   M     88   2009-08-14 09:46:06  desk

'group_tool' after the second mount
[root at desk ~]# group_tool
groupd not running
fence domain
member count  2
victim count  0
victim now    0
master nodeid 2
wait state    none
members       1 2

dlm lockspaces
name          1
id            0xe5ab8ad6
flags         0x00000000
change        member 2 joined 1 remove 0 failed 0 seq 2,2
members       1 2

[root at cool ~]# group_tool
groupd compatibility mode 0
fence domain
member count  2
victim count  0
victim now    0
master nodeid 2
wait state    none
members       1 2

dlm lockspaces
name          1
id            0xe5ab8ad6
flags         0x00000008 fs_reg
change        member 2 joined 1 remove 0 failed 0 seq 1,1
members       1 2

gfs mountgroups
name          1
id            0x791ee743
flags         0x00000020 need_first
change        member 1 joined 1 remove 0 failed 0 seq 1,1
members       1
---

#checking for difference, seems only the group_tool has different 
output. problem is in groupd? it starts automatically? I didn't start it 
by hand. what I do is "service cman start" on both nodes and then "mount 
...." on both nodes.

messages files shows:
node desk:
Aug 14 18:07:44 desk kernel: GFS2: fsid=: Trying to join cluster 
"lock_dlm", "testgfs2:1"
Aug 14 18:07:44 desk kernel: dlm: Using TCP for communications
Aug 14 18:07:44 desk kernel: GFS2: fsid=testgfs2:1.0: Joined cluster. 
Now mounting FS...
Aug 14 18:07:44 desk kernel: GFS2: fsid=testgfs2:1.0: jid=0, already 
locked for use
Aug 14 18:07:44 desk kernel: GFS2: fsid=testgfs2:1.0: jid=0: Looking at 
journal...
Aug 14 18:07:44 desk gfs_controld[2206]: recovery_uevent mg not found 1
Aug 14 18:07:44 desk kernel: GFS2: fsid=testgfs2:1.0: jid=0: Done
Aug 14 18:07:44 desk kernel: GFS2: fsid=testgfs2:1.0: jid=1: Trying to 
acquire journal lock...
Aug 14 18:07:44 desk kernel: GFS2: fsid=testgfs2:1.0: jid=1: Looking at 
journal...
Aug 14 18:07:44 desk gfs_controld[2206]: recovery_uevent mg not found 1
Aug 14 18:07:44 desk gfs_controld[2206]: recovery_uevent mg not found 1
Aug 14 18:07:44 desk kernel: GFS2: fsid=testgfs2:1.0: jid=1: Done
Aug 14 18:07:54 desk kernel: dlm: connecting to 1

node cool:
Aug 14 10:10:00 cool kernel: GFS2: fsid=: Trying to join cluster 
"lock_dlm", "testgfs2:1"
Aug 14 10:10:00 cool kernel: dlm: Using TCP for communications
Aug 14 10:10:00 cool kernel: dlm: got connection from 2
Aug 14 10:10:00 cool kernel: GFS2: fsid=testgfs2:1.0: Joined cluster. 
Now mounting FS...
Aug 14 10:14:00 cool kernel: INFO: task mount.gfs2:2458 blocked for more 
than 120 seconds.
Aug 14 10:14:00 cool kernel: "echo 0 > 
/proc/sys/kernel/hung_task_timeout_secs" disables this message.
Aug 14 10:14:00 cool kernel: mount.gfs2    D 000001db     0  2458   2457 
0x00000080
Aug 14 10:14:00 cool kernel: dbd13c6c 00000086 31dc2c4d 000001db 
f6575c44 c09a5260 c09a9b60 f6575c44
Aug 14 10:14:00 cool kernel: dbd13c3c c09a9b60 c09a9b60 f4c3b6c0 
dbcf4000 dbd13c58 dbd13c44 00000001
Aug 14 10:14:00 cool kernel: 31dbe4d2 000001db f65759b0 dbd13c5c 
dbd13c5c 00000246 c1f98554 c1f98558
Aug 14 10:14:00 cool kernel: Call Trace:
Aug 14 10:14:00 cool kernel: [<f8d5a6e6>] 
gfs2_glock_holder_wait+0xd/0x11 [gfs2]
Aug 14 10:14:00 cool kernel: [<c0742897>] __wait_on_bit+0x39/0x60
Aug 14 10:14:00 cool kernel: [<f8d5a6d9>] ? 
gfs2_glock_holder_wait+0x0/0x11 [gfs2]
Aug 14 10:14:00 cool kernel: [<f8d5a6d9>] ? 
gfs2_glock_holder_wait+0x0/0x11 [gfs2]
Aug 14 10:14:00 cool kernel: [<c074295e>] out_of_line_wait_on_bit+0xa0/0xa8
Aug 14 10:14:00 cool kernel: [<c044779d>] ? wake_bit_function+0x0/0x3c
Aug 14 10:14:00 cool kernel: [<f8d5d487>] wait_on_bit.clone.1+0x1c/0x28 
[gfs2]
Aug 14 10:14:00 cool kernel: [<f8d5d4fe>] gfs2_glock_wait+0x31/0x37 [gfs2]
Aug 14 10:14:00 cool kernel: [<f8d5d774>] gfs2_glock_nq+0x270/0x278 [gfs2]
Aug 14 10:14:00 cool kernel: [<f8d5d8a0>] gfs2_glock_nq_num+0x4c/0x6c [gfs2]
Aug 14 10:14:00 cool kernel: [<f8d66567>] init_journal+0x2b2/0x675 [gfs2]
Aug 14 10:14:00 cool kernel: [<c04b1e75>] ? __slab_alloc+0x40a/0x421
Aug 14 10:14:00 cool kernel: [<c04b21b8>] ? kmem_cache_alloc+0x6d/0x105
Aug 14 10:14:00 cool kernel: [<c04c6fe1>] ? d_alloc+0x23/0x15e
Aug 14 10:14:00 cool kernel: [<c04dd3a0>] ? inotify_d_instantiate+0x17/0x3a
Aug 14 10:14:00 cool kernel: [<f8d65e5f>] ? gfs2_glock_nq_init+0x13/0x31 
[gfs2]
Aug 14 10:14:00 cool kernel: [<f8d6694f>] init_inodes+0x25/0x152 [gfs2]
Aug 14 10:14:00 cool kernel: [<f8d6750d>] fill_super+0xa91/0xc10 [gfs2]
Aug 14 10:14:00 cool kernel: [<f8d5d899>] ? gfs2_glock_nq_num+0x45/0x6c 
[gfs2]
Aug 14 10:14:00 cool kernel: [<c04bade0>] get_sb_bdev+0xdc/0x119
Aug 14 10:14:00 cool kernel: [<c04b4aea>] ? pcpu_alloc+0x352/0x38b
Aug 14 10:14:00 cool kernel: [<f8d65ad8>] gfs2_get_sb+0x18/0x1a [gfs2]
Aug 14 10:14:00 cool kernel: [<f8d66a7c>] ? fill_super+0x0/0xc10 [gfs2]
Aug 14 10:14:00 cool kernel: [<c04baacc>] vfs_kern_mount+0x82/0xf0
Aug 14 10:14:00 cool kernel: [<c04bab89>] do_kern_mount+0x38/0xc3
Aug 14 10:14:00 cool kernel: [<c04cc03f>] do_mount+0x68c/0x6e4
Aug 14 10:14:00 cool kernel: [<c0492dc1>] ? __get_free_pages+0x24/0x26
Aug 14 10:14:00 cool kernel: [<c04cc0fd>] sys_mount+0x66/0x98
Aug 14 10:14:00 cool kernel: [<c0402a28>] sysenter_do_call+0x12/0x27

there will be more like the "INFO: task mount.gfs2:2458 blocked for more 
than 120 seconds." if I wait longer.

regards,
wengang.

David Teigland wrote:
> On Thu, Aug 13, 2009 at 02:22:11PM +0800, Wengang Wang wrote:
>> <cman two_node="1" expected_votes="2"/>
> 
> That's not a valid combination, two_node="1" requires expected_votes="1".
> 
> You didn't mention which userspace cluster version/release you're using, or
> include any status about the cluster.  Before trying to mount gfs on either
> node, collect from both nodes:
> 
>   cman_tool status
>   cman_tool nodes
>   group_tool
> 
> Then mount on the first node and collect the same information, then try
> mounting on the second node, collect the same information, and look for any
> errors in /var/log/messages.
> 
> Since you're using new kernels, you need to be using the cluster 3.0 userspace
> code.  You're using the old manual fencing config.  There is no more
> fence_manual; the new way to configure manual fencing is to not configure any
> fencing at all.  So, your cluster.conf should look like this:
> 
> <?xml version="1.0"?>
> <cluster name="testgfs2" config_version="1">
> <cman two_node="1" expected_votes="1"/>
> <clusternodes>
> <clusternode name="cool" nodeid="1"/>
> <clusternode name="desk" nodeid="2"/>
> </clusternodes>
> </cluster>
> 
> Dave
> 



From ntoughe at hotmail.com  Fri Aug 14 08:24:46 2009
From: ntoughe at hotmail.com (NTOUGHE GUY-SERGE)
Date: Fri, 14 Aug 2009 08:24:46 +0000
Subject: [Linux-cluster] Cman Hang
Message-ID: <BAY119-W12634207EA807E5B78021CA5020@phx.gbl>


Hi, 

I added the  hostnames and ip addresses in /etc/hosts like this:
127.0.0.1		 localhost
10.81.10.07	host1.occ.lan
10.81.10.06	host2.occ.lan
I also  executed the following command:
 gfs2_tool sb /dev/VolGroup01/LogVol01 table arev_clust:LogVol01
You shouldn't change any of these values if the filesystem is mounted.

Are you sure? [y/n] y

current lock table name = "arev_clust:LogVol01"
new lock table name = "arev_clust:LogVol01"
Done

mkfs.gfs2 -p lock_dlm -t arev_clust:LogVol01 -j 1 /dev/VolGroup01/LogVol01
This will destroy any data on /dev/VolGroup01/LogVol01.
  It appears to contain a gfs2 filesystem.

Are you sure you want to proceed? [y/n] y

Device:                    /dev/VolGroup01/LogVol01
Blocksize:                 4096
Device Size                240.00 GB (62913536 blocks)
Filesystem Size:           240.00 GB (62913535 blocks)
Journals:                  1
Resource Groups:           960
Locking Protocol:          "lock_dlm"
Lock Table:                "arev_clust:LogVol01"

BUT when i want to mount th e gfs FS i got:
# mount -t gfs2 /dev/VolGroup01/LogVol01 /appli/prod
/sbin/mount.gfs2: can't connect to gfs_controld: Connection refused

And the starting of the service  gives the same error:
cman not started: Multicast and node address families differ. /usr/sbin/cman_tool: aisexec daemon didn't start
                                                           [FAILED]
When i creating the cluster via Conga this is below the error i got:

The following errors occurred:


    * Unable to add the key for node gs21spli004.occ.lan to the trusted keys list.
    * Unable to connect to host2.occ.lan: Unable to establish an SSL connection to host2.occ.lan:11111: ClientSocket(hostname, port, timeout): connect() failed
    * host1.occ.lan reports it is a member of cluster "arev_clust"
Of course the port 1111 is opened:

tcp        0      0 0.0.0.0:11111               0.0.0.0:*                   LISTEN      0          573641     9435/ricci
SO what can you advise me i'm really lost.
greetings



ntoughe at hotmail.com


_________________________________________________________________
Show them the way! Add maps and directions to your party invites. 
http://www.microsoft.com/windows/windowslive/products/events.aspx
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090814/4fccaf9b/attachment.htm>

From teigland at redhat.com  Fri Aug 14 14:57:38 2009
From: teigland at redhat.com (David Teigland)
Date: Fri, 14 Aug 2009 09:57:38 -0500
Subject: [Linux-cluster] gfs2 mount hangs
In-Reply-To: <4A84CC39.3080809@oracle.com>
References: <4A83B113.60609@oracle.com> <20090813141424.GA8148@redhat.com>
	<4A84CC39.3080809@oracle.com>
Message-ID: <20090814145738.GA7094@redhat.com>

On Fri, Aug 14, 2009 at 10:30:17AM +0800, Wengang Wang wrote:
> anything else needed?

versions are good

> #don't know where 239.192.110.55 comes from. does it matter?

That's good, the multicast address is generated by cman.

> [root at desk ~]# cman_tool nodes
> Node  Sts   Inc   Joined               Name
>    1   M     88   2009-08-14 17:44:00  cool
>    2   M     76   2009-08-14 17:43:52  desk
> 
> [root at cool ~]# cman_tool nodes
> Node  Sts   Inc   Joined               Name
>    1   M     84   2009-08-14 09:46:06  cool
>    2   M     88   2009-08-14 09:46:06  desk
> 

> [root at desk ~]# group_tool
> groupd not running
> fence domain
> member count  2
> victim count  0
> victim now    0
> master nodeid 2
> wait state    none
> members       1 2
> 
> [root at cool ~]# group_tool
> groupd not running
> fence domain
> member count  2
> victim count  0
> victim now    0
> master nodeid 2
> wait state    none
> members       1 2

the cluster is fine

> #checking for difference, seems only the group_tool has different 
> output. problem is in groupd? it starts automatically? I didn't start it 
> by hand. what I do is "service cman start" on both nodes and then "mount 
> ...." on both nodes.

groupd is not needed

> node desk:
> Aug 14 18:07:44 desk gfs_controld[2206]: recovery_uevent mg not found 1
> Aug 14 18:07:44 desk gfs_controld[2206]: recovery_uevent mg not found 1
> Aug 14 18:07:44 desk gfs_controld[2206]: recovery_uevent mg not found 1

There's a problem here, but it's not clear what has gone wrong.  Could you try
this again and after these messages appear send the output of
"gfs_control dump" from both nodes?

> Aug 14 10:14:00 cool kernel: INFO: task mount.gfs2:2458 blocked for more 
> than 120 seconds.

The second mount is stuck probably because the first went bad.

Dave



From wen.gang.wang at oracle.com  Fri Aug 14 15:05:22 2009
From: wen.gang.wang at oracle.com (Wengang Wang)
Date: Fri, 14 Aug 2009 23:05:22 +0800
Subject: [Linux-cluster] [PATCH] gfs2: free disk inode which is deleted by
	remote node
Message-ID: <200908141511.n7EFBPhZ030555@acsinet15.oracle.com>

this patch is for the same problem that Benjamin Marzinski fixes at commit
b94a170e96dc416828af9d350ae2e34b70ae7347

quotation of the original problem:

---cut here---
When a file is deleted from a gfs2 filesystem on one node, a dcache
entry for it may still exist on other nodes in the cluster. If this
happens, gfs2 will be unable to free this file on disk. Because of this,
it's possible to have a gfs2 filesystem with no files on it and no free
space. With this patch, when a node receives a callback notifying it
that the file is being deleted on another node, it schedules a new
workqueue thread to remove the file's dcache entry.
---end cut---

after applying Benjamin's patch, I think there is still a case in which the disk
inode remains even when "no space" is hit. the case is that when running
d_prune_aliases() against the inode, there are one or more dentries(aliases)
which have reference count number > 0. in this case the dentries won't be pruned.
and even later, the reference cound becomes to 0, the dentries can still be
cached in memory. unfortunately, no callback come again, things come back to
the state before the callback runs. thus the on disk inode remains there until
in memory inode is removed for some other reason(shrinking inode cache or unmount
the gfs2 volume..).

this patch is to remove those dentries when their reference count becomes to 0 and
the inode is deleted by remote node. for implementation, gfs2_dentry_delete() is
added as dentry_operations.d_delete. the function returns true when the inode is
deleted by remote node. in dput(), gfs2_dentry_delete() is called and since it
returns true, the dentry is unhashed from dcache and then removed. when all dentries
are removed, the in memory inode get removed so that the on disk inode can be freed.

Signed-off-by: Wengang Wang <wen.gang.wang at oracle.com>
---
 fs/gfs2/dentry.c |   11 +++++++++++
 1 files changed, 11 insertions(+), 0 deletions(-)

diff --git a/fs/gfs2/dentry.c b/fs/gfs2/dentry.c
index 022c66c..434f204 100644
--- a/fs/gfs2/dentry.c
+++ b/fs/gfs2/dentry.c
@@ -107,8 +107,19 @@ static int gfs2_dhash(struct dentry *dentry, struct qstr *str)
 	return 0;
 }
 
+static int gfs2_dentry_delete(struct dentry *dentry)
+{
+	struct gfs2_inode *ginode = GFS2_I(dentry->d_inode);
+
+	if (test_bit(GLF_DEMOTE, &ginode->i_iopen_gh.gh_gl->gl_flags))
+		return 1;
+
+	return 0;
+}
+
 const struct dentry_operations gfs2_dops = {
 	.d_revalidate = gfs2_drevalidate,
 	.d_hash = gfs2_dhash,
+	.d_delete = gfs2_dentry_delete,
 };
 
-- 
1.6.2.5



From thomas at sjolshagen.net  Fri Aug 14 20:45:50 2009
From: thomas at sjolshagen.net (Thomas Sjolshagen)
Date: Fri, 14 Aug 2009 16:45:50 -0400
Subject: [Linux-cluster] vm.sh resource script
Message-ID: <20090814164550.16372xa21kmg4xz2@www.sjolshagen.net>

Hi,

I'm trying to use; "node1> clusvcadm -M <service> -m node2" in Fedora  
11 w/latest cluster updates installed and get the following error  
message that I'm having a hard time deciphering/debugging:

Trying to migrate service:svc1 to node2...S/Lang Script Error

In /var/log/messages:

Aug 14 16:44:08 node2 rgmanager[3210]: [S/Lang] Variable Uninitialized Error
Aug 14 16:44:08 node2 rgmanager[3210]: [S/Lang] Script Execution Failure

Not sure if there's something wrong with my cluster.conf or that error  
is indicative of something else?

// Thomas


My /etc/cluster.conf:

<?xml version="1.0" ?>
<cluster alias="virt-cluster" config_version="130" name="virt-cluster">

	<cman expected_votes="1" two_node="1">
		<multicast addr="225.0.0.12"/>
	</cman>
	<dlm plock_ownership="1" plock_rate_limit="0"/>
	<gfs_controld plock_rate_limit="0"/>
	<fence_daemon clean_start="1" post_fail_delay="0" post_join_delay="3"/>
	<clusternodes>
		<clusternode name="virt1-backup.sjolshagen.net" nodeid="1" votes="1">
			<multicast addr="225.0.0.12" interface="kvmbr1"/>
			<fence>
				<method name="1">
					<device name="power" port="1"/>
				</method>
				<method name="2">
					<device name="human" nodename="virt1-backup.sjolshagen.net"/>
				</method>
			</fence>
		</clusternode>
		<clusternode name="virt0-backup.sjolshagen.net" nodeid="3" votes="1">
			<multicast addr="225.0.0.12" interface="kvmbr1"/>
			<fence>
				<method name="1">
					<device name="power" port="8"/>
				</method>
				<method name="2">
					<device name="human" nodename="virt0-backup.sjolshagen.net"/>
				</method>
			</fence>
		</clusternode>
	</clusternodes>
	<fencedevices>
		<!-- Definition of fence device named "power", referenced previously -->
		<fencedevice agent="fence_wti" name="power" ipaddr="192.168.1.10"  
password="thomas"/>
		<fencedevice agent="fence_manual" name="human"/>
	</fencedevices>
	<rm log_level="7" central_processing="1">
		<!-- Configure events to do service relationships -->
   		<events>
			<event class="service" name="service-ers">
				notice("Event service triggered!");
				evalfile("/usr/share/cluster/follow-service.sl");
				follow_service("service:dhcpd", "service:rhc_mail_db",  
"service:rhc_mail_old", "service:rhc_mail_ui");
			</event>
			<event class="node" name="node-ers">
				notice("Event node triggered!");
				evalfile("/usr/share/cluster/follow-service.sl");
				follow_service("service:dhcpd", "service:rhc_mail_db",  
"service:rhc_mail_old", "service:rhc_mail_ui");
			</event>
                                 follow_service("service:dhcpd",  
"service:rhc_mail_db", "service:rhc_mail_old", "service:rhc_mail_ui");
                         </event>
                 </events>
                 <failoverdomains>
                         <failoverdomain name="ALL">
                                 <failoverdomainnode name="node1"/>
                                 <failoverdomainnode name="node2"/>
                         </failoverdomain>
                         <failoverdomain name="prefer-node1"  
restricted="1" ordered="1">
                                 <failoverdomainnode name="node1"  
priority="10" />
                                 <failoverdomainnode name="node2"  
priority="20" />
                         </failoverdomain>
                 </failoverdomains>
                         <!-- Internet services definition -->

                 <service autostart="1" exclusive="0" domain="ALL"  
name="rhc_mail_db" recovery="relocate">
                         <vm name="kvm01-hvm" migrate="live"  
snapshot="/cluster/kvm-guests/snapshots" use_virsh="1"  
hypervisor="qemu"/>
                 </service>

                 <service autostart="1" exclusive="0" domain="ALL"  
name="rhc_mail_old" recovery="relocate" depend="service:rhc_mail_db"  
depend_mode="soft">
                         <vm name="kvm02-hvm" migrate="live"  
snapshot="/cluster/kvm-guests/snapshots" use_virsh="1"  
hypervisor="qemu">
                                         <vm name="kvm03-hvm"  
migrate="live" snapshot="/cluster/kvm-guests/snapshots" use_virsh="1"  
hypervisor="qemu">
                                         </vm>
                         </vm>
                 </service>

                 <service autostart="1" exclusive="0" domain="ALL"  
name="windows_guests" recovery="relocate" depend="service:dhcpd"  
depend_mode="soft">
                         <vm name="kvm20-hvm" migrate="live"  
snapshot="/cluster/kvm-guests/snapshots" use_virsh="1"  
hypervisor="qemu"/>
                 </service>

                 <service autostart="1" exclusive="0" domain="ALL"  
name="rhc_mail_ui" recovery="relocate" depend="service:rhc_mail_old"  
depend_mode="soft">
                         <vm name="kvm04-hvm" migrate="live"  
snapshot="/cluster/kvm-guests/snapshots" use_virsh="1"  
hypervisor="qemu"/>
                 </service>

                 <service autostart="1" exclusive="0" domain="ALL"  
name="rhc_samba" recovery="relocate">
                         <vm name="kvm06-hvm" recovery="restart"  
snapshot="/cluster/kvm-guests/snapshots" use_virsh="1"  
hypervisor="qemu" />
                 </service>
                 <service autostart="0" exclusive="0"  
name="rhc_mail_new" domain="prefer-node1" recovery="restart">
                         <vm name="kvm30-hvm" migrate="live"  
snapshot="/cluster/kvm-guests/snapshots" use_virsh="1"  
hypervisor="qemu"/>
                         <vm name="kvm31-hvm" migrate="live"  
snapshot="/cluster/kvm-guests/snapshots" use_virsh="1"  
hypervisor="qemu"/>
                 </service>
                 <service autostart="1" exclusive="0" domain="ALL"  
name="dhcpd" recovery="relocate">
                         <script name="dhcpd" domain="virt-cluster"  
file="/cluster/network/configs/dhcp/dhcpd" max_restarts="2"/>
                 </service>
         </rm>
</cluster>


----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.




From cryptogrid at gmail.com  Fri Aug 14 22:12:34 2009
From: cryptogrid at gmail.com (crypto grid)
Date: Fri, 14 Aug 2009 19:12:34 -0300
Subject: [Linux-cluster] sanity check
In-Reply-To: <20090811031639.GA728@monsterjam.org>
References: <20090811031639.GA728@monsterjam.org>
Message-ID: <a9f464b80908141512s41661d0cu4b9b7bf91ce16604@mail.gmail.com>

Your failover domain includes server1 and server2, and you have two diferent
services using the same ip address in the services section. I wonder what
happen if you have to realocate just one of the services? I think you
couldn't.
I think you have to define two different ip addresses in the resource
section, one for apache and one for postfix.
I'm not saying that this is causing the problem, I'd try disabling one
service and enabling just one.

On Tue, Aug 11, 2009 at 12:16 AM, <jason at monsterjam.org> wrote:

> can someone please look at the attached config and let me know if anything
> looks
> blatantly wrong.. This cluster config was working earlier for my httpd and
> postfix
> services , but now the virtual cluster ip address (.7) wont come up any
> more and the
> httpd and postfix will therefore not start. Im not sure why.
>
> Jason
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090814/6d369426/attachment.htm>

From tpeirce at etrafficsolutions.com  Fri Aug 14 22:14:31 2009
From: tpeirce at etrafficsolutions.com (Trevor Peirce)
Date: Fri, 14 Aug 2009 15:14:31 -0700
Subject: [Linux-cluster] cluster-3.0.0 problems starting up
Message-ID: <4A85E1C7.1070502@etrafficsolutions.com>

Hello,

I've attempted to compile and install the cluster-3.0.0 suite and have 
these problems when trying to start cman:

Aug 14 12:10:01 localhost corosync[27739]: parse error in config: parse 
error in config: .
Aug 14 12:10:01 localhost corosync[27739]: [MAIN ] Corosync Cluster 
Engine ('trunk'): started and ready to provide service.
Aug 14 12:10:01 localhost corosync[27739]: [MAIN ] Successfully read 
config from /etc/cluster/cluster.conf
Aug 14 12:10:01 localhost corosync[27739]: [MAIN ] Successfully parsed 
cman config
Aug 14 12:10:01 localhost corosync[27739]: [MAIN ] Successfully 
configured openais services to load
Aug 14 12:10:01 localhost corosync[27739]: [MAIN ] parse error in 
config: parse error in config: .
Aug 14 12:10:01 localhost corosync[27739]: [MAIN ] Corosync Cluster 
Engine exiting with status -9 at main.c:879.

Here is cluster.conf, as generated by ccs_tool:

<?xml version="1.0"?>
<cluster name="TestCluster" config_version="3">
<clusternodes><clusternode name="guinneapig" votes="1" 
nodeid="1"><fence><method name="single"><device 
name="manual"/></method></fence></clusternode></clusternodes>
<fencedevices><fencedevice name="manual" agent="manual"/></fencedevices>
<rm>
<failoverdomains/>
<resources/>
</rm>
</cluster>


Any advice would be greatly appreciated!

-- 
Trevor Peirce
Systems Administrator
Etraffic Solutions Inc.

Suite 100 ? 710 Redbrick Street
Victoria, B.C. V8T 5J3
phone: 250.658.8238 x49
fax: 250.658.5936 




From dhoffutt at gmail.com  Fri Aug 14 22:23:19 2009
From: dhoffutt at gmail.com (Dusty)
Date: Fri, 14 Aug 2009 17:23:19 -0500
Subject: [Linux-cluster] cluster-3.0.0 problems starting up
In-Reply-To: <4A85E1C7.1070502@etrafficsolutions.com>
References: <4A85E1C7.1070502@etrafficsolutions.com>
Message-ID: <d740f8db0908141523y7a0d4a7eta656459bf16bc145@mail.gmail.com>

I seem to recall that in v3 the new "manual_fence" is simply using no fence
at all. Take out the fencing tags.

If that doesn't solve it, you might try switching SElinux to permissive to
test if that solve the issue.

On Fri, Aug 14, 2009 at 5:14 PM, Trevor Peirce <
tpeirce at etrafficsolutions.com> wrote:

> Hello,
>
> I've attempted to compile and install the cluster-3.0.0 suite and have
> these problems when trying to start cman:
>
> Aug 14 12:10:01 localhost corosync[27739]: parse error in config: parse
> error in config: .
> Aug 14 12:10:01 localhost corosync[27739]: [MAIN ] Corosync Cluster Engine
> ('trunk'): started and ready to provide service.
> Aug 14 12:10:01 localhost corosync[27739]: [MAIN ] Successfully read config
> from /etc/cluster/cluster.conf
> Aug 14 12:10:01 localhost corosync[27739]: [MAIN ] Successfully parsed cman
> config
> Aug 14 12:10:01 localhost corosync[27739]: [MAIN ] Successfully configured
> openais services to load
> Aug 14 12:10:01 localhost corosync[27739]: [MAIN ] parse error in config:
> parse error in config: .
> Aug 14 12:10:01 localhost corosync[27739]: [MAIN ] Corosync Cluster Engine
> exiting with status -9 at main.c:879.
>
> Here is cluster.conf, as generated by ccs_tool:
>
> <?xml version="1.0"?>
> <cluster name="TestCluster" config_version="3">
> <clusternodes><clusternode name="guinneapig" votes="1"
> nodeid="1"><fence><method name="single"><device
> name="manual"/></method></fence></clusternode></clusternodes>
> <fencedevices><fencedevice name="manual" agent="manual"/></fencedevices>
> <rm>
> <failoverdomains/>
> <resources/>
> </rm>
> </cluster>
>
>
> Any advice would be greatly appreciated!
>
> --
> Trevor Peirce
> Systems Administrator
> Etraffic Solutions Inc.
>
> Suite 100 ? 710 Redbrick Street
> Victoria, B.C. V8T 5J3
> phone: 250.658.8238 x49
> fax: 250.658.5936
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090814/b37c4fdf/attachment.htm>

From tpeirce at etrafficsolutions.com  Fri Aug 14 22:27:59 2009
From: tpeirce at etrafficsolutions.com (Trevor Peirce)
Date: Fri, 14 Aug 2009 15:27:59 -0700
Subject: [Linux-cluster] cluster-3.0.0 problems starting up
In-Reply-To: <d740f8db0908141523y7a0d4a7eta656459bf16bc145@mail.gmail.com>
References: <4A85E1C7.1070502@etrafficsolutions.com>
	<d740f8db0908141523y7a0d4a7eta656459bf16bc145@mail.gmail.com>
Message-ID: <4A85E4EF.4020501@etrafficsolutions.com>

Dusty wrote:
> I seem to recall that in v3 the new "manual_fence" is simply using no 
> fence at all. Take out the fencing tags.
>
> If that doesn't solve it, you might try switching SElinux to 
> permissive to test if that solve the issue.

Nay...  I have already removed SElinux form the equation, so that isn't it.

Aug 14 15:24:58 localhost corosync[28324]: parse error in config: parse 
error in config: .
Aug 14 15:24:58 localhost corosync[28324]:   [MAIN  ] Corosync Cluster 
Engine ('trunk'): started and ready to provide service.
Aug 14 15:24:58 localhost corosync[28324]:   [MAIN  ] Successfully read 
config from /etc/cluster/cluster.conf
Aug 14 15:24:58 localhost corosync[28324]:   [MAIN  ] Successfully 
parsed cman config
Aug 14 15:24:58 localhost corosync[28324]:   [MAIN  ] Successfully 
configured openais services to load
Aug 14 15:24:58 localhost corosync[28324]:   [MAIN  ] parse error in 
config: parse error in config: .
Aug 14 15:24:58 localhost corosync[28324]:   [MAIN  ] Corosync Cluster 
Engine exiting with status -9 at main.c:879.
Aug 14 15:24:59 localhost groupd[28332]: groupd 3.0.0
Aug 14 15:24:59 localhost fenced[28342]: fenced 3.0.0 started
Aug 14 15:25:00 localhost dlm_controld[28358]: dlm_controld 3.0.0 started
Aug 14 15:25:00 localhost gfs_controld[28369]: gfs_controld 3.0.0 started
Aug 14 15:25:01 localhost fenced[28342]: cman_admin_init error 111
Aug 14 15:25:02 localhost dlm_controld[28358]: cman_admin_init error 111
Aug 14 15:25:02 localhost gfs_controld[28369]: cman_admin_init error 111

using:

<?xml version="1.0"?>
<cluster name="TestCluster" config_version="3">
  <clusternodes><clusternode name="guinneapig" votes="1" 
nodeid="1"></clusternode></clusternodes>
  <rm>
    <failoverdomains/>
    <resources/>
  </rm>
</cluster>

-- 
Trevor Peirce
Systems Administrator
Etraffic Solutions Inc.

Suite 100 ? 710 Redbrick Street
Victoria, B.C. V8T 5J3
phone: 250.658.8238 x49
fax: 250.658.5936 




From jason at monsterjam.org  Sat Aug 15 01:23:39 2009
From: jason at monsterjam.org (jason at monsterjam.org)
Date: Fri, 14 Aug 2009 21:23:39 -0400
Subject: [Linux-cluster] sanity check
In-Reply-To: <a9f464b80908141512s41661d0cu4b9b7bf91ce16604@mail.gmail.com>
References: <20090811031639.GA728@monsterjam.org>
	<a9f464b80908141512s41661d0cu4b9b7bf91ce16604@mail.gmail.com>
Message-ID: <20090815012339.GB6468@monsterjam.org>

On Fri, Aug 14, 2009 at 07:12:34PM -0300, crypto grid wrote:
>    Your failover domain includes server1 and server2, and you have two
>    diferent services using the same ip address in the services section. I
>    wonder what happen if you have to realocate just one of the services? I
>    think you couldn't.
> 
>    I think you have to define two different ip addresses in the resource
>    section, one for apache and one for postfix.
> 
>    I'm not saying that this is causing the problem, I'd try disabling one
>    service and enabling just one.
> 

ok, Ive just modified it to be the following config, and it looks like it works fine. 
But please tell me if theres anything  else Ive missed.

thanks for the help!

Jason


[jason at server1 ~]$ sudo cat /etc/cluster/cluster.conf
<?xml version="1.0"?>
<cluster config_version="26" name="mycluster">
        <fence_daemon clean_start="0" post_fail_delay="0" post_join_delay="3"/>
        <clusternodes>
                <clusternode name="server1" votes="1">
                        <fence> 
                                <method name="1">
                                        <device name="apc_power_switch" option="off" port="1" switch="1"/>
                                        <device name="apc_power_switch" option="off" port="2" switch="1"/>
                                        <device name="apc_power_switch" option="on" port="1" switch="1"/>
                                        <device name="apc_power_switch" option="on" port="2" switch="1"/>
                                </method>
                        </fence>
                </clusternode>
                <clusternode name="server2" votes="1">
                        <fence> 
                                <method name="1">
                                        <device name="apc_power_switch" option="off" port="3" switch="1"/>
                                        <device name="apc_power_switch" option="off" port="4" switch="1"/>
                                        <device name="apc_power_switch" option="on" port="3" switch="1"/>
                                        <device name="apc_power_switch" option="on" port="4" switch="1"/>
                                </method>
                        </fence>
                </clusternode>
        </clusternodes>
        <cman expected_votes="1" two_node="1"/>
        <fencedevices>
                <fencedevice agent="fence_apc" ipaddr="192.168.1.8" login="xxxxx" name="apc_power_switch" passwd="xxxx2xxx"/>
        </fencedevices>
        <rm>    
                <failoverdomains>
                        <failoverdomain name="httpd" ordered="1" restricted="1">
                                <failoverdomainnode name="server1" priority="1"/>
                                <failoverdomainnode name="server2" priority="2"/>
                        </failoverdomain>
                </failoverdomains>
                <resources>
                        <script file="/etc/init.d/httpd" name="cluster_apache"/>
                        <ip address="192.168.1.9" monitor_link="1"/>
                        <script file="/etc/init.d/postfix" name="cluster_posstfix"/>
                        <ip address="192.168.1.7" monitor_link="1"/>
                </resources>
                <service autostart="1" domain="httpd" name="Apache Service">
                        <script ref="cluster_apache"/>
                        <ip ref="192.168.1.9"/>
                </service>
                <service autostart="1" domain="httpd" name="postfix service">
                        <script ref="cluster_posstfix"/>
                        <ip ref="192.168.1.7"/>
                </service>
        </rm>
</cluster>



From thomas at sjolshagen.net  Sat Aug 15 02:13:16 2009
From: thomas at sjolshagen.net (Thomas Sjolshagen)
Date: Fri, 14 Aug 2009 22:13:16 -0400
Subject: [Linux-cluster] fence_node fails for fence_wti agent - cluster 3.0
Message-ID: <20090814221316.89106gn410mavk58@www.sjolshagen.net>

Hi,

I've got a WTI NPS power switch that I've configured as a fence device  
for my 2-node cluster. If I call 'fence_wti' directly with  
action=status or action=reboot, it does what I expect it to do; fence  
the node I'm telling it to fence.

However, if I issue "fence_node -vv <node>", the command fails with  
the following error:

# fence_node -vv node1
fence node1 dev 0.0 agent fence_wti result: error from agent
agent args: plug=1 agent=fence_wti ipaddr=1.1.1.1  
debug-file=/tmp/wti-fence.debug inet4-only=1 verbose=1 ipport=23  
action=reboot
fence node1 failed

Also, there is no /tmp/wti-fence* file and the only entry in the error  
logs is in /var/log/messages:

fence_node[19172]: fence virt1-backup.sjolshagen.net failed

Suggestions on how I can further troubleshoot this or fix the problem?

Thanks in advance,

// Thomas

My cluster.conf file contains the following fencedevice entry:

<fencedevices>
    <!-- Definition of fence device named "power", referenced previously -->
    <fencedevice agent="fence_wti" name="power" ipaddr="1.1.1.1"  
debug-file="/tmp/wti-fence.debug" inet4-only="1" verbose="1"  
ipport="23" action="reboot"/>
</fencedevices>
`
And for the clusternodes definitions:

<clusternodes>
         <clusternode name="node2" nodeid="1" votes="1">
                 <fence>
                         <method name="1">
                                 <device name="power" plug="1"/>
                         </method>
                 </fence>
         </clusternode>
         <clusternode name="node2" nodeid="3" votes="1">
                 <fence>
                         <method name="1">
                                 <device name="power" plug="8"/>
                         </method>
                 </fence>
         </clusternode>
</clusternodes>



----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.




From thomas at sjolshagen.net  Sat Aug 15 02:16:22 2009
From: thomas at sjolshagen.net (Thomas Sjolshagen)
Date: Fri, 14 Aug 2009 22:16:22 -0400
Subject: [Linux-cluster] fence_node fails for fence_wti agent - cluster 3.0
In-Reply-To: <20090814221316.89106gn410mavk58@www.sjolshagen.net>
References: <20090814221316.89106gn410mavk58@www.sjolshagen.net>
Message-ID: <20090814221622.154918bmw0zb9euu@www.sjolshagen.net>


Quoting Thomas Sjolshagen <thomas at sjolshagen.net>:
...
>         <clusternode name="node2" nodeid="1" votes="1">
...
>         <clusternode name="node2" nodeid="3" votes="1">
                              ^^^^^
Obviously a typo while sanitizing the config file, sorry :)

// Thomas


----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.




From esanchezvela.redhatcluster at gmail.com  Sat Aug 15 06:11:52 2009
From: esanchezvela.redhatcluster at gmail.com (Enrique Sanchez)
Date: Sat, 15 Aug 2009 02:11:52 -0400
Subject: [Linux-cluster] problems with gfs and openais cluster.
Message-ID: <1247a04a0908142311o5ff51b2am3f245a736eda45b8@mail.gmail.com>

folks,
    I have little experience with RH and even less with RH clusters,
but kind of know my way around, I am concerned about an RH cluster
where I am getting nonstop errors like these...

Aug 15 05:45:42 XXXXX   openais[5283]: [MAIN ] Received message has
invalid digest... ignoring.
Aug 15 05:45:42 XXXXX   openais[5283]: [MAIN ] Invalid packet data


We are using RedHat's GFS and I believe the error should be fixed
following this post, however  I have a copule of questions.....

http://www.mail-archive.com/linux-cluster at redhat.com/msg06172.html

a) here is my /etc/ais/openais.conf file (same config for both hosts
in the cluster)...
======================================================
totem {
        version: 2
        secauth: off
        threads: 0
        interface {
                ringnumber: 0
                bindnetaddr: 192.168.2.0
                mcastaddr: 226.94.1.1
                mcastport: 5405
        }
}
logging {
        debug: off
        timestamp: on
}
amf {
        mode: disabled
}
===================================================
but here is a list of my IP addresses:

============================
ifconfig -a | grep inet | grep Mask
          inet addr:10.205.226.91  Bcast:10.205.239.255  Mask:255.255.240.0
          inet addr:10.205.39.26  Bcast:10.205.39.255  Mask:255.255.255.0
          inet addr:127.0.0.1  Mask:255.0.0.0
          inet addr:192.168.122.1  Bcast:192.168.122.255
Mask:255.255.255.0 ((virbr0))
============================
first of all, I believe the  bindnetaddr parameter should have been
set to 192.168.122.0 but I am not sure.....

next, there is a 2nd cluster with the exact same configuration but
different IP addresses except the debugging setup...

here it is...
========================================
totem {
        version: 2
        secauth: off
        threads: 0
        interface {
                ringnumber: 0
                bindnetaddr: 192.168.2.0
                mcastaddr: 226.94.1.1
                mcastport: 5405
        }
}
logging {
        debug: on
        logfile: /var/log/openais.log
        timestamp: on
}

amf {
        mode: disabled
}
===================================================
and here are the IP addresses found in the server:
============================================
#ifconfig -a | grep inet | grep Mask
          inet addr:10.205.26.38  Bcast:10.205.26.255  Mask:255.255.255.0
          inet addr:10.205.39.28  Bcast:10.205.39.255  Mask:255.255.255.0
          inet addr:127.0.0.1  Mask:255.0.0.0
          inet addr:192.168.122.1  Bcast:192.168.122.255
Mask:255.255.255.0 (virbr0)
============================================


given that 10.205.26.38 and 10.205.226.91 are on different VLANs I
doubt that is the problem, but it could be in the adapters using the
10.205.39.x IP address range since they are in the same VLAN and given
the interface binding is incorrect the cluster picked up an interface
and clashed with the existing server.... is that correct?


2nd question: I see openais service is running when chkconfig --list
openais ; shows it should be down ....

chkconfig --list openais
openais         0:off   1:off   2:off   3:off   4:off   5:off   6:off

who's starting it???

thank you,
enrique sanchez.





-- 
Enrique Sanchez Vela
------------------------------------------
"What you have been obliged to discover
by yourself leaves a path in your mind
which you can use again when the need
arises."    --G. C. Lichtenberg
http://themathcircle.org/



From sdake at redhat.com  Sat Aug 15 06:23:27 2009
From: sdake at redhat.com (Steven Dake)
Date: Fri, 14 Aug 2009 23:23:27 -0700
Subject: [Linux-cluster] problems with gfs and openais cluster.
In-Reply-To: <1247a04a0908142311o5ff51b2am3f245a736eda45b8@mail.gmail.com>
References: <1247a04a0908142311o5ff51b2am3f245a736eda45b8@mail.gmail.com>
Message-ID: <1250317407.2649.16.camel@localhost.localdomain>

On Sat, 2009-08-15 at 02:11 -0400, Enrique Sanchez wrote:
> folks,
>     I have little experience with RH and even less with RH clusters,
> but kind of know my way around, I am concerned about an RH cluster
> where I am getting nonstop errors like these...
> 
> Aug 15 05:45:42 XXXXX   openais[5283]: [MAIN ] Received message has
> invalid digest... ignoring.
> Aug 15 05:45:42 XXXXX   openais[5283]: [MAIN ] Invalid packet data
> 
> 
> We are using RedHat's GFS and I believe the error should be fixed
> following this post, however  I have a copule of questions.....
> 
> http://www.mail-archive.com/linux-cluster at redhat.com/msg06172.html
> 
> a) here is my /etc/ais/openais.conf file (same config for both hosts
> in the cluster)...
> ======================================================
> totem {
>         version: 2
>         secauth: off
>         threads: 0
>         interface {
>                 ringnumber: 0
>                 bindnetaddr: 192.168.2.0
>                 mcastaddr: 226.94.1.1
>                 mcastport: 5405
>         }
> }
> logging {
>         debug: off
>         timestamp: on
> }
> amf {
>         mode: disabled
> }
> ===================================================
> but here is a list of my IP addresses:
> 
> ============================
> ifconfig -a | grep inet | grep Mask
>           inet addr:10.205.226.91  Bcast:10.205.239.255  Mask:255.255.240.0
>           inet addr:10.205.39.26  Bcast:10.205.39.255  Mask:255.255.255.0
>           inet addr:127.0.0.1  Mask:255.0.0.0
>           inet addr:192.168.122.1  Bcast:192.168.122.255
> Mask:255.255.255.0 ((virbr0))
> ============================
> first of all, I believe the  bindnetaddr parameter should have been
> set to 192.168.122.0 but I am not sure.....
> 
> next, there is a 2nd cluster with the exact same configuration but
> different IP addresses except the debugging setup...
> 
> here it is...
> ========================================
> totem {
>         version: 2
>         secauth: off
>         threads: 0
>         interface {
>                 ringnumber: 0
>                 bindnetaddr: 192.168.2.0
>                 mcastaddr: 226.94.1.1
>                 mcastport: 5405
>         }
> }
> logging {
>         debug: on
>         logfile: /var/log/openais.log
>         timestamp: on
> }
> 
> amf {
>         mode: disabled
> }
> ===================================================
> and here are the IP addresses found in the server:
> ============================================
> #ifconfig -a | grep inet | grep Mask
>           inet addr:10.205.26.38  Bcast:10.205.26.255  Mask:255.255.255.0
>           inet addr:10.205.39.28  Bcast:10.205.39.255  Mask:255.255.255.0
>           inet addr:127.0.0.1  Mask:255.0.0.0
>           inet addr:192.168.122.1  Bcast:192.168.122.255
> Mask:255.255.255.0 (virbr0)
> ============================================
> 
> 
> given that 10.205.26.38 and 10.205.226.91 are on different VLANs I
> doubt that is the problem, but it could be in the adapters using the
> 10.205.39.x IP address range since they are in the same VLAN and given
> the interface binding is incorrect the cluster picked up an interface
> and clashed with the existing server.... is that correct?
> 
> 
> 2nd question: I see openais service is running when chkconfig --list
> openais ; shows it should be down ....
> 
> chkconfig --list openais
> openais         0:off   1:off   2:off   3:off   4:off   5:off   6:off
> 
> who's starting it???
> 
> thank you,
> enrique sanchez.
> 
> 
> 

The openais.conf file is not used when starting with the cman service.
Instead cman uses its own configuration file
in /etc/cluster/cluster.conf.  openais should not be chkconfig'ed on
unless running without cman.

One reason you could have the error is that you have multiple clusters
on the same multicast address as specified in /etc/cluster/cluster.conf.

Regards
-steve

> 
> 



From esanchezvela.redhatcluster at gmail.com  Sat Aug 15 12:12:56 2009
From: esanchezvela.redhatcluster at gmail.com (Enrique Sanchez)
Date: Sat, 15 Aug 2009 08:12:56 -0400
Subject: [Linux-cluster] problems with gfs and openais cluster.
In-Reply-To: <1250317407.2649.16.camel@localhost.localdomain>
References: <1247a04a0908142311o5ff51b2am3f245a736eda45b8@mail.gmail.com>
	<1250317407.2649.16.camel@localhost.localdomain>
Message-ID: <1247a04a0908150512g2966f1cnefffb0c8dc5d033e@mail.gmail.com>

On Sat, Aug 15, 2009 at 2:23 AM, Steven Dake<sdake at redhat.com> wrote:
>
> The openais.conf file is not used when starting with the cman service.
> Instead cman uses its own configuration file
> in /etc/cluster/cluster.conf.  openais should not be chkconfig'ed on
> unless running without cman.
>
> One reason you could have the error is that you have multiple clusters
> on the same multicast address as specified in /etc/cluster/cluster.conf.
>
> Regards
> -steve
>

steve,

thanks for the quick reply: the mcast address is the same and it
looks like it is using the same VLAN here are the details....

this is one cluster (cluster one):

        <cman expected_votes="1" two_node="1">
                <multicast addr="224.0.0.1"/>
        </cman>

and this is in the 2nd cluster (cluster two)......

        <cman expected_votes="1" two_node="1">
                <multicast addr="224.0.0.1" interface="eth1"/>
        </cman>


unfortunately eth1 is on the same VLAN for both clusters...is cluster
1 using the mcast address on both adapters?

thnx...
enrique.


-- 
Enrique Sanchez Vela
------------------------------------------
"What you have been obliged to discover
by yourself leaves a path in your mind
which you can use again when the need
arises."    --G. C. Lichtenberg
http://themathcircle.org/



From wen.gang.wang at oracle.com  Sun Aug 16 10:45:14 2009
From: wen.gang.wang at oracle.com (Wengang Wang)
Date: Sun, 16 Aug 2009 18:45:14 +0800
Subject: [Linux-cluster] [PATCH] gfs2: free disk inode which is deleted
	by remote node
In-Reply-To: <200908141511.n7EFBPhZ030555@acsinet15.oracle.com>
References: <200908141511.n7EFBPhZ030555@acsinet15.oracle.com>
Message-ID: <4A87E33A.4020109@oracle.com>

sorry, I should send to cluster-devel at redhat.com. please ignore this email.

regards,
wengang.

Wengang Wang wrote:
> this patch is for the same problem that Benjamin Marzinski fixes at commit
> b94a170e96dc416828af9d350ae2e34b70ae7347

-- 
--just begin to learn, you are never too late...



From brem.belguebli at gmail.com  Sun Aug 16 11:59:08 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Sun, 16 Aug 2009 13:59:08 +0200
Subject: [Linux-cluster] Fwd: CLVM exclusive mode
In-Reply-To: <4A7A8240.7000206@redhat.com>
References: <4A7AA7DE02000027000536E3@lucius.provo.novell.com>
	<4A7A8240.7000206@redhat.com>
Message-ID: <29ae894c0908160459m34dd02dcp8601e63d7bf62c3a@mail.gmail.com>

Hi,

I don't think adding more security can be considered as pointless,
especially when this has no impact on performance or behaviour.

The question is, what's the point in  allowing the clustered active
exclusive lock to be bypassed ?

In comparison to other volume management solutions (on various unices) where
these barriers are already implemented, the lack of them on Linux can  be
seen as a weakness.

Regards


2009/8/6, Christine Caulfield <ccaulfie at redhat.com>:
>
> On 06/08/09 02:52, Jia Ju Zhang wrote:
>
>> Just RFC:
>> I noticed that 'vgchange -ay' can convert the lock which locked by
>> 'vgchange -aey'
>> from EX to CR. Is that acceptable to change the logic into always
>> allocating a new lock
>> rather than converting an existing lock?
>> In that case, 'vgchange -ay' won't change the result of 'vgchange -aey'.
>> But if we really
>> want to convert the lock, we can firstly invoke 'vgchange -aen' to release
>> the EX lock,
>> then invoke the 'vgchange -ay'.
>>
>> Does this make sense? Or what side effect it may introduce?
>>
>
>
> I think it makes no sense at all, and have already said so on this list. As
> far as I know there is no bugzilla for this problem and therefore it isn't
> being worked on.
>
> So ... if you care about this ... you know what to do ;-)
>
> Chrissie
>
>   On 8/6/2009 at  9:39 AM, in message<4A7A346B.A94 : 39 : 18251>, Jia Ju
>>>>> Zhang
>>>>>
>>>> wrote:
>>
>>> On Fri, 2009-07-31 at 21:29 +0200, brem belguebli wrote:
>>>
>>>> Hi,
>>>>
>>>> Same behaviour as the one from Rafael.
>>>>
>>>> Everything is coherent as long as you use the exclusive flag from the
>>>> rogue node, the locking does the job. Deactivating an already opened
>>>> VG (mounted lvol) is not possible either. How could this behave in
>>>> case one used raw devices instead of FS ?
>>>>
>>>> But when you come to ignore the exclusive flag on the rogue node
>>>> (vgchange -a y vgXX) the locking is completely bypassed. It's
>>>> definitely here that the watchdog has to be (within the tools
>>>> lvchange, vgchange, or at dlm level).
>>>>
>>> Is there an open bugzilla # for this? Would like to follow this issue.
>>>
>>>
>>
>>
>> --
>> Linux-cluster mailing list
>> Linux-cluster at redhat.com
>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090816/45c00d65/attachment.htm>

From jfenal at redhat.com  Sun Aug 16 23:14:08 2009
From: jfenal at redhat.com (Jerome Fenal)
Date: Mon, 17 Aug 2009 01:14:08 +0200
Subject: [Linux-cluster] fence device
In-Reply-To: <EE5A9AA37E4A40D5BD7B22FED6A61FE9@siim94.local>
References: <EE5A9AA37E4A40D5BD7B22FED6A61FE9@siim94.local>
Message-ID: <1250464448.3325.2.camel@localhost.localdomain>

Le mercredi 12 ao?t 2009 ? 09:08 +0200, LEROUX Ludovic a ?crit :
> i have a linux redhat cluster with two redhat 5.3 nodes.
> I have a shared iscsi disk with gfs.
> And i created an oracle service which mounts the gfs file system and
> start an oracle instance in fail over mode.
> I have two questions:
> 1?) I don't have a fence device. Is it possible to create a cluster
> without a fence device? 

Having a fencing device is mandatory for supportability.
You can open a support ticket to get directions in order to correct this
with your actual setup.

> 2?) is it possible to create a high available cluster with two or more
> nodes which works together (for example http server).

You mean multiple processes across multiple nodes for one service ?
No.

Regards,

J.

-- 
J?r?me Fenal, RHCE                                      Tel.: +33 1 41 91 23 37
Solution Architect                                      Mob.: +33 6 88 06 51 15
Consultant Avant-ventes                                 Fax.: +33 1 41 91 23 32
http://www.fr.redhat.com/                                     jfenal at redhat.com
Red Hat France SARL                                  Siret n? 421 199 464 00064
Le Linea, 1 rue du G?n?ral Leclerc                 92047 Paris La D?fense Cedex
JBoss Open Choice, la strat?gie Java   http://www.jboss.com/solutions/javaapps/



From jfenal at redhat.com  Sun Aug 16 23:18:04 2009
From: jfenal at redhat.com (Jerome Fenal)
Date: Mon, 17 Aug 2009 01:18:04 +0200
Subject: [Linux-cluster] Cman hang
In-Reply-To: <BAY119-W410E2F250E8B461752CFC9A5050@phx.gbl>
References: <BAY119-W410E2F250E8B461752CFC9A5050@phx.gbl>
Message-ID: <1250464684.3325.4.camel@localhost.localdomain>

Le jeudi 13 ao?t 2009 ? 14:16 +0000, NTOUGHE GUY-SERGE a ?crit :
> 
> Hi gurus,
> 
> i installed  RHEL 5.3 on 2 servers which participating to a cluster
> composed of these 2 nodes:

Hi Guy-Serge !

please open a ticket at Red Hat support to get a timely answer.

Regards,

J.
-- 
J?r?me Fenal, RHCE                                      Tel.: +33 1 41 91 23 37
Solution Architect                                      Mob.: +33 6 88 06 51 15
Consultant Avant-ventes                                 Fax.: +33 1 41 91 23 32
http://www.fr.redhat.com/                                     jfenal at redhat.com
Red Hat France SARL                                  Siret n? 421 199 464 00064
Le Linea, 1 rue du G?n?ral Leclerc                 92047 Paris La D?fense Cedex
JBoss Open Choice, la strat?gie Java   http://www.jboss.com/solutions/javaapps/



From sdake at redhat.com  Mon Aug 17 00:31:46 2009
From: sdake at redhat.com (Steven Dake)
Date: Sun, 16 Aug 2009 17:31:46 -0700
Subject: [Linux-cluster] fence device
In-Reply-To: <1250464448.3325.2.camel@localhost.localdomain>
References: <EE5A9AA37E4A40D5BD7B22FED6A61FE9@siim94.local>
	<1250464448.3325.2.camel@localhost.localdomain>
Message-ID: <1250469106.19265.16.camel@localhost.localdomain>

On Mon, 2009-08-17 at 01:14 +0200, Jerome Fenal wrote:
> Le mercredi 12 ao?t 2009 ? 09:08 +0200, LEROUX Ludovic a ?crit :
> > i have a linux redhat cluster with two redhat 5.3 nodes.
> > I have a shared iscsi disk with gfs.
> > And i created an oracle service which mounts the gfs file system and
> > start an oracle instance in fail over mode.
> > I have two questions:
> > 1?) I don't have a fence device. Is it possible to create a cluster
> > without a fence device? 
> 
> Having a fencing device is mandatory for supportability.
> You can open a support ticket to get directions in order to correct this
> with your actual setup.
> 
> > 2?) is it possible to create a high available cluster with two or more
> > nodes which works together (for example http server).
> 
> You mean multiple processes across multiple nodes for one service ?
> No.
> 
It is possible using the cpg C programming interface but requires custom
development of the server application.  If you are developing your own
server application, you may find this method of HA (replicated state)
very useful.

man evs_overview to see the programming model.  man -k cpg will show a
list of the apis.

Looks like we need a cpg_overview man page.

regards
-steve

> Regards,
> 
> J.
> 



From brem.belguebli at gmail.com  Mon Aug 17 10:22:17 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Mon, 17 Aug 2009 12:22:17 +0200
Subject: [Linux-cluster] Submitting a mdadm resource script
Message-ID: <29ae894c0908170322u447ff473s5b0512e6b37267af@mail.gmail.com>

Hi,

I have written a resource script (mdmonitor.sh) to manage mdadm mirrors
across a cluster.

This is for testing purposes and still needs some additionnal features and
some deeper debugging.

If someone is interested in hacking it to improve it or to correct it, I'll
be glad.

A readme file is provided that explains the goal of this resource script.

Thanks to Rafael Mico Miranda for his lvm-cluster resource script used on
top of mdmonitor.

 Brem
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090817/8e4a0b79/attachment.htm>
-------------- next part --------------
<?xml version="1.0"?>
<resource-agent name="mdmonitor" version="rgmanager 2.0">
    <version>1.0</version>

    <longdesc lang="en">
	Resource for activating, deactivating and monitoring MD devices	
    </longdesc>
    <shortdesc lang="en">
	Resource for activating, deactivating and monitoring MD devices	
    </shortdesc>

    <parameters>

        <parameter name="raidconf" required="1">
            <longdesc lang="en">
                Name of the file containing the description of the array(s) 
		Format is strict, must be "servicename"-raid.conf
		It will be  placed in /etc/cluster/"servicename"
            </longdesc>
            <shortdesc lang="en">
                raid.conf  filename
            </shortdesc>
	    <content type="string"/>
	</parameter>
	
	<parameter name="policy" required="0">
	   <longdesc lang="en">
		Policy to adopt at start, strict will prevent 
		assembling any array partially thus causing the 
		resource to fail.
		Quorum will allow partial start. If not defined,
		defaults to quorum
	   </longdesc>
	   <shortdesc lang="en">
                Assembly Policy    
            </shortdesc>
	    <content type="string"/>
        </parameter>

    </parameters>

    <actions>
        <action name="start" timeout="5"/>
	<action name="stop" timeout="5"/>

	<action name="status" interval="10m" timeout="5" />
	<action name="monitor" interval="10m" timeout="5" />

	<action name="meta-data"  timeout="5"/>
    </actions>

    <special tag="rgmanager">
    </special>

</resource-agent>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: mdmonitor.README
Type: application/octet-stream
Size: 11193 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090817/8e4a0b79/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: mdmonitor.sh
Type: application/x-sh
Size: 33420 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090817/8e4a0b79/attachment.sh>

From esggrupos at gmail.com  Mon Aug 17 10:23:05 2009
From: esggrupos at gmail.com (ESGLinux)
Date: Mon, 17 Aug 2009 12:23:05 +0200
Subject: [Linux-cluster] do I have a fence DRAC device?
In-Reply-To: <8a5668960908110514u4703a951qa49938375d0a7e52@mail.gmail.com>
References: <3128ba140908100324l6cdb4c34ra5f5edb39c6903e9@mail.gmail.com>
	<8b711df40908101134t69b8e12cof6cc551809421e45@mail.gmail.com>
	<3128ba140908110503m27532c62rce2844564c35da06@mail.gmail.com>
	<8a5668960908110514u4703a951qa49938375d0a7e52@mail.gmail.com>
Message-ID: <3128ba140908170323u48e878c3s40337ad2e3bf3ec5@mail.gmail.com>

2009/8/11 Juan Ramon Martin Blanco <robejrm at gmail.com>

>
>
> On Tue, Aug 11, 2009 at 2:03 PM, ESGLinux <esggrupos at gmail.com> wrote:
>
>> Thanks
>> I?ll check it when I could reboot the server.
>>
>> greetings,
>>
> You have a BMC ipmi in the first network interface, it can be configured at
> boot time (I don't  remember if inside the BIOS or pressing cntrl+something
> during boot)
>


Hi,

I have finally rebooted my system and entered inside the BIOS with CTRL+E
command.
I have configured the card and now when I run the command ipmitool lan print
I get this information:

ipmitool lan print
Set in Progress         : Set Complete
Auth Type Support       : NONE MD2 MD5 PASSWORD
Auth Type Enable        : Callback : MD2 MD5
                        : User     : MD2 MD5
                        : Operator : MD2 MD5
                        : Admin    : MD2 MD5
                        : OEM      : MD2 MD5
IP Address Source       : Static Address
IP Address              : 192.168.1.250
Subnet Mask             : 255.255.255.0
MAC Address             : 00:1e:c9:ae:6f:7e
SNMP Community String   : public
IP Header               : TTL=0x40 Flags=0x40 Precedence=0x00 TOS=0x10
Default Gateway IP      : 192.168.1.1
Default Gateway MAC     : 00:00:00:00:00:00
Backup Gateway IP       : 0.0.0.0
Backup Gateway MAC      : 00:00:00:00:00:00
802.1q VLAN ID          : Disabled
802.1q VLAN Priority    : 0
RMCP+ Cipher Suites     : 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14
Cipher Suite Priv Max   : aaaaaaaaaaaaaaa
                        :     X=Cipher Suite Unused
                        :     c=CALLBACK
                        :     u=USER
                        :     o=OPERATOR
                        :     a=ADMIN
                        :     O=OEM


If I make ping 192.168.1.250 I get response but I can?t do anymore. If I run
the racadm commands I still get the errors:

ERROR: RACADM is unable to process the requested subcommand because there is
no
local RAC configuration to communicate with.

Local RACADM subcommand execution requires the following:

 1. A Remote Access Controller (RAC) must be present on the managed server
 2. Appropriate managed node software must be installed and running on the
    server


I have tried to access that ip, but I can?t do nothing with it. I have done
a port scan and I nothing is listening there,

What else do I have to do to get this working?

Thanks,

ESG






>
> Greetings,
> Juanra
>
>>
>> ESG
>>
>> 2009/8/10 Paras pradhan <pradhanparas at gmail.com>
>>
>> On Mon, Aug 10, 2009 at 5:24 AM, ESGLinux<esggrupos at gmail.com> wrote:
>>> > Hi all,
>>> > I was designing a 2 node cluster and I was going to use 2 servers DELL
>>> > PowerEdge 1950. I was going to buy a DRAC card to use for fencing but
>>> > running several commands in the servers I have noticed that when I run
>>> this
>>> > command:
>>> > #ipmitool lan print
>>> > Set in Progress : Set Complete
>>> > Auth Type Support : NONE MD2 MD5 PASSWORD
>>> > Auth Type Enable : Callback : MD2 MD5
>>> >                         : User : MD2 MD5
>>> >                         : Operator : MD2 MD5
>>> >                         : Admin : MD2 MD5
>>> >                         : OEM : MD2 MD5
>>> > IP Address Source : Static Address
>>> > IP Address : 0.0.0.0
>>> > Subnet Mask : 0.0.0.0
>>> > MAC Address : 00:1e:c9:ae:6f:7e
>>> > SNMP Community String : public
>>> > IP Header : TTL=0x40 Flags=0x40 Precedence=0x00 TOS=0x10
>>> > Default Gateway IP : 0.0.0.0
>>> > Default Gateway MAC : 00:00:00:00:00:00
>>> > Backup Gateway IP : 0.0.0.0
>>> > Backup Gateway MAC : 00:00:00:00:00:00
>>> > 802.1q VLAN ID : Disabled
>>> > 802.1q VLAN Priority : 0
>>> > RMCP+ Cipher Suites : 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14
>>> > Cipher Suite Priv Max : aaaaaaaaaaaaaaa
>>> >                         : X=Cipher Suite Unused
>>> >                         : c=CALLBACK
>>> >                         : u=USER
>>> >                         : o=OPERATOR
>>> >                         : a=ADMIN
>>> >                         : O=OEM
>>> > does this mean that I already have an ipmi card (not configured) that I
>>> can
>>> > use for fencing? if the anwser is yes, where hell must I configure it?
>>> I
>>> > don?t see wher can I do it.
>>> > If I haven?t a fencing device which one do you recommed to use?
>>> > Thanks in advance
>>> > ESG
>>> >
>>> > --
>>> > Linux-cluster mailing list
>>> > Linux-cluster at redhat.com
>>> > https://www.redhat.com/mailman/listinfo/linux-cluster
>>> >
>>>
>>> Yes you have IPMI and if you are using 1950 Dell, DRAC should be there
>>> too. You can see if you have DRAC or not when the server starts and
>>> before the loading of the OS.
>>>
>>> I have 1850s and I am using DRAC for fencing.
>>>
>>>
>>> Paras.
>>>
>>> --
>>> Linux-cluster mailing list
>>> Linux-cluster at redhat.com
>>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>>
>>
>>
>> --
>> Linux-cluster mailing list
>> Linux-cluster at redhat.com
>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090817/9f5dfa6f/attachment.htm>

From esggrupos at gmail.com  Mon Aug 17 10:37:46 2009
From: esggrupos at gmail.com (ESGLinux)
Date: Mon, 17 Aug 2009 12:37:46 +0200
Subject: [Linux-cluster] do I have a fence DRAC device?
In-Reply-To: <3128ba140908130427i6ab85406ye6da34073e6a6e97@mail.gmail.com>
References: <3128ba140908100324l6cdb4c34ra5f5edb39c6903e9@mail.gmail.com>
	<8b711df40908101134t69b8e12cof6cc551809421e45@mail.gmail.com>
	<3128ba140908110503m27532c62rce2844564c35da06@mail.gmail.com>
	<8a5668960908110514u4703a951qa49938375d0a7e52@mail.gmail.com>
	<27649.1250005179@mirchi>
	<3128ba140908130427i6ab85406ye6da34073e6a6e97@mail.gmail.com>
Message-ID: <3128ba140908170337w7d448186h60f8316ac6037326@mail.gmail.com>

2009/8/13 ESGLinux <esggrupos at gmail.com>

> Hi,
> I couldn?t reboot my system yet but I have installed the openmanage
> packages:
>
> srvadmin-omacore-5.4.0-260
> srvadmin-iws-5.4.0-260
> srvadmin-syscheck-5.4.0-260
> srvadmin-rac5-components-5.4.0-260
> srvadmin-deng-5.4.0-260
> srvadmin-ipmi-5.4.0-260.DUP
> srvadmin-racadm5-5.4.0-260
> srvadmin-omauth-5.4.0-260.rhel5
> srvadmin-hapi-5.4.0-260
> srvadmin-cm-5.4.0-260
> srvadmin-racdrsc5-5.4.0-260
> srvadmin-omilcore-5.4.0-260
> srvadmin-isvc-5.4.0-260
> srvadmin-storage-5.4.0-260
> srvadmin-jre-5.4.0-260
> srvadmin-omhip-5.4.0-260
>
> Now I have the command racadm but when I try to execut it I get this:
>
> racadm config -g cfgSerial -o cfgSerialTelnetEnable 1
> ERROR: RACADM is unable to process the requested subcommand because there
> is no
> local RAC configuration to communicate with.
>
> Local RACADM subcommand execution requires the following:
>
>  1. A Remote Access Controller (RAC) must be present on the managed server
>  2. Appropriate managed node software must be installed and running on the
>     server
>
>
> What do I need to install/start? or until I configure the bios I can?t get
> this work?
>
> Greetings
>
> ESG
>
>
> 2009/8/11 <bergman at merctech.com>
>
>
>>
>> In the message dated: Tue, 11 Aug 2009 14:14:03 +0200,
>> The pithy ruminations from Juan Ramon Martin Blanco on
>> <Re: [Linux-cluster] do I have a fence DRAC device?> were:
>> => --===============1917368601==
>> => Content-Type: multipart/alternative;
>> boundary=0016364c7c07663f600470dca3b8
>> =>
>> => --0016364c7c07663f600470dca3b8
>> => Content-Type: text/plain; charset=ISO-8859-1
>> => Content-Transfer-Encoding: quoted-printable
>> =>
>> => On Tue, Aug 11, 2009 at 2:03 PM, ESGLinux <esggrupos at gmail.com> wrote:
>> =>
>> => > Thanks
>> => > I=B4ll check it when I could reboot the server.
>> => >
>> => > greetings,
>> => >
>> => You have a BMC ipmi in the first network interface, it can be
>> configured at
>> => boot time (I don't  remember if inside the BIOS or pressing
>> cntrl+something
>> => during boot)
>> =>
>>
>> Based on my notes, here's how I configured the DRAC interface on a Dell
>> 1950
>> for use as a fence device:
>>
>>        Configuring the card from Linux depending on the installation of
>> Dell's
>>        OMSA package. Once that's installed, use the following
>> commands:
>>
>>                racadm config -g cfgSerial -o cfgSerialTelnetEnable 1
>>                racadm config -g cfgLanNetworking -o cfgDNSRacName
>> HOSTNAME_FOR_INTERFACE
>>                racadm config -g cfgDNSDomainName DOMAINNAME_FOR_INTERFACE
>>                racadm config -g cfgUserAdmin -o cfgUserAdminPassword -i 2
>> PASSWORD
>>                racadm config -g cfgNicEnable 1
>>                racadm config -g cfgNicIpAddress WWW.XXX.YYY.ZZZ
>>                racadm config -g cfgNicNetmask WWW.XXX.YYY.ZZZ
>>                racadm config -g cfgNicGateway WWW.XXX.YYY.ZZZ
>>                racadm config -g cfgNicUseDhcp 0
>>
>>
>>        I also save a backup of the configuration with:
>>
>>                racadm getconfig -f ~/drac_config
>>
>>
>> Hope this helps,
>>
>> Mark
>>
>> ----
>> Mark Bergman                              voice: 215-662-7310
>> mark.bergman at uphs.upenn.edu                 fax: 215-614-0266
>> System Administrator     Section of Biomedical Image Analysis
>> Department of Radiology            University of Pennsylvania
>>      PGP Key: https://www.rad.upenn.edu/sbia/bergman
>>
>>
>> => Greetings,
>> => Juanra
>> =>
>> => >
>> => > ESG
>> => >
>> => > 2009/8/10 Paras pradhan <pradhanparas at gmail.com>
>> => >
>> => > On Mon, Aug 10, 2009 at 5:24 AM, ESGLinux<esggrupos at gmail.com>
>> wrote:
>> => >> > Hi all,
>> => >> > I was designing a 2 node cluster and I was going to use 2 servers
>> DELL
>> => >> > PowerEdge 1950. I was going to buy a DRAC card to use for fencing
>> but
>> => >> > running several commands in the servers I have noticed that when I
>> run
>> => >> this
>> => >> > command:
>> => >> > #ipmitool lan print
>> => >> > Set in Progress : Set Complete
>> => >> > Auth Type Support : NONE MD2 MD5 PASSWORD
>> => >> > Auth Type Enable : Callback : MD2 MD5
>> => >> >                         : User : MD2 MD5
>> => >> >                         : Operator : MD2 MD5
>> => >> >                         : Admin : MD2 MD5
>> => >> >                         : OEM : MD2 MD5
>> => >> > IP Address Source : Static Address
>> => >> > IP Address : 0.0.0.0
>> => >> > Subnet Mask : 0.0.0.0
>> => >> > MAC Address : 00:1e:c9:ae:6f:7e
>> => >> > SNMP Community String : public
>> => >> > IP Header : TTL=0x40 Flags=0x40 Precedence=0x00 TOS=0x10
>> => >> > Default Gateway IP : 0.0.0.0
>> => >> > Default Gateway MAC : 00:00:00:00:00:00
>> => >> > Backup Gateway IP : 0.0.0.0
>> => >> > Backup Gateway MAC : 00:00:00:00:00:00
>> => >> > 802.1q VLAN ID : Disabled
>> => >> > 802.1q VLAN Priority : 0
>> => >> > RMCP+ Cipher Suites : 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14
>> => >> > Cipher Suite Priv Max : aaaaaaaaaaaaaaa
>> => >> >                         : X=Cipher Suite Unused
>> => >> >                         : c=CALLBACK
>> => >> >                         : u=USER
>> => >> >                         : o=OPERATOR
>> => >> >                         : a=ADMIN
>> => >> >                         : O=OEM
>> => >> > does this mean that I already have an ipmi card (not configured)
>> that
>> => I
>> => >> can
>> => >> > use for fencing? if the anwser is yes, where hell must I configure
>> it?
>> =>  I
>> => >> > don=B4t see wher can I do it.
>> => >> > If I haven=B4t a fencing device which one do you recommed to use?
>> => >> > Thanks in advance
>> => >> > ESG
>> => >> >
>> => >> > --
>> => >> > Linux-cluster mailing list
>> => >> > Linux-cluster at redhat.com
>> => >> > https://www.redhat.com/mailman/listinfo/linux-cluster
>> => >> >
>> => >>
>> => >> Yes you have IPMI and if you are using 1950 Dell, DRAC should be
>> there
>> => >> too. You can see if you have DRAC or not when the server starts and
>> => >> before the loading of the OS.
>> => >>
>> => >> I have 1850s and I am using DRAC for fencing.
>> => >>
>> => >>
>> => >> Paras.
>> => >>
>> => >> --
>> => >> Linux-cluster mailing list
>> => >> Linux-cluster at redhat.com
>> => >> https://www.redhat.com/mailman/listinfo/linux-cluster
>> => >>
>> => >
>> => >
>>
>>
>>
>> --
>> Linux-cluster mailing list
>> Linux-cluster at redhat.com
>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090817/165aa00f/attachment.htm>

From esggrupos at gmail.com  Mon Aug 17 10:50:59 2009
From: esggrupos at gmail.com (ESGLinux)
Date: Mon, 17 Aug 2009 12:50:59 +0200
Subject: [Linux-cluster] do I have a fence DRAC device?
In-Reply-To: <8b711df40908101134t69b8e12cof6cc551809421e45@mail.gmail.com>
References: <3128ba140908100324l6cdb4c34ra5f5edb39c6903e9@mail.gmail.com>
	<8b711df40908101134t69b8e12cof6cc551809421e45@mail.gmail.com>
Message-ID: <3128ba140908170350ge619930w5c17368ff0d3cf42@mail.gmail.com>

Hi,

>
> Yes you have IPMI and if you are using 1950 Dell, DRAC should be there
> too.


What do I have to see at start up? I haven?t seen anything about DRAC. I
have seen that with CTRL+E I entered in an utility to configure de ip of the
card, but I don?t know if it?s a DRAC card. How can I check it?

Thanks in advance

ESG




> You can see if you have DRAC or not when the server starts and
> before the loading of the OS.
>
> I have 1850s and I am using DRAC for fencing.
>
>
> Paras.
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090817/ac4428a4/attachment.htm>

From hxinwei at gmail.com  Mon Aug 17 11:17:13 2009
From: hxinwei at gmail.com (Xinwei Hu)
Date: Mon, 17 Aug 2009 19:17:13 +0800
Subject: [Linux-cluster] Fwd: CLVM exclusive mode
In-Reply-To: <29ae894c0908160459m34dd02dcp8601e63d7bf62c3a@mail.gmail.com>
References: <4A7AA7DE02000027000536E3@lucius.provo.novell.com>
	<4A7A8240.7000206@redhat.com>
	<29ae894c0908160459m34dd02dcp8601e63d7bf62c3a@mail.gmail.com>
Message-ID: <1cafab770908170417n16919354h38e22bcb97d31bf7@mail.gmail.com>

Hi,

  Attached a very naive try to solve the issue you have.

  Would you give it a test ?

  Thanks.

2009/8/16 brem belguebli <brem.belguebli at gmail.com>:
> Hi,
>
> I don't think adding more security can be considered as pointless,
> especially when this has no impact on performance or behaviour.
> The question is, what's the point in ?allowing the clustered active
> exclusive lock to be bypassed ?
>
> In comparison to other volume management solutions (on various unices) where
> these barriers are already implemented, the lack of them on Linux can ?be
> seen as a weakness.
> Regards
>
>
> 2009/8/6, Christine Caulfield <ccaulfie at redhat.com>:
>>
>> On 06/08/09 02:52, Jia Ju Zhang wrote:
>>>
>>> Just RFC:
>>> I noticed that 'vgchange -ay' can convert the lock which locked by
>>> 'vgchange -aey'
>>> from EX to CR. Is that acceptable to change the logic into always
>>> allocating a new lock
>>> rather than converting an existing lock?
>>> In that case, 'vgchange -ay' won't change the result of 'vgchange -aey'.
>>> But if we really
>>> want to convert the lock, we can firstly invoke 'vgchange -aen' to
>>> release the EX lock,
>>> then invoke the 'vgchange -ay'.
>>>
>>> Does this make sense? Or what side effect it may introduce?
>>
>>
>> I think it makes no sense at all, and have already said so on this list.
>> As far as I know there is no bugzilla for this problem and therefore it
>> isn't being worked on.
>>
>> So ... if you care about this ... you know what to do ;-)
>>
>> Chrissie
>>
>>>>>> On 8/6/2009 at ?9:39 AM, in message<4A7A346B.A94 : 39 : 18251>, Jia Ju
>>>>>> Zhang
>>>
>>> wrote:
>>>>
>>>> On Fri, 2009-07-31 at 21:29 +0200, brem belguebli wrote:
>>>>>
>>>>> Hi,
>>>>>
>>>>> Same behaviour as the one from Rafael.
>>>>>
>>>>> Everything is coherent as long as you use the exclusive flag from the
>>>>> rogue node, the locking does the job. Deactivating an already opened
>>>>> VG (mounted lvol) is not possible either. How could this behave in
>>>>> case one used raw devices instead of FS ?
>>>>>
>>>>> But when you come to ignore the exclusive flag on the rogue node
>>>>> (vgchange -a y vgXX) the locking is completely bypassed. It's
>>>>> definitely here that the watchdog has to be (within the tools
>>>>> lvchange, vgchange, or at dlm level).
>>>>
>>>> Is there an open bugzilla # for this? Would like to follow this issue.
>>>>
>>>
>>>
>>>
>>> --
>>> Linux-cluster mailing list
>>> Linux-cluster at redhat.com
>>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>
>> --
>> Linux-cluster mailing list
>> Linux-cluster at redhat.com
>> https://www.redhat.com/mailman/listinfo/linux-cluster
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test.diff
Type: text/x-patch
Size: 664 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090817/d9aeb2b4/attachment.bin>

From brem.belguebli at gmail.com  Mon Aug 17 14:07:45 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Mon, 17 Aug 2009 16:07:45 +0200
Subject: [Linux-cluster] Re: Submitting a mdadm resource script
In-Reply-To: <29ae894c0908170322u447ff473s5b0512e6b37267af@mail.gmail.com>
References: <29ae894c0908170322u447ff473s5b0512e6b37267af@mail.gmail.com>
Message-ID: <29ae894c0908170707p4f42cd78j24592b02a6e2983@mail.gmail.com>

Hi,

there was  a bug in my script that occurs when debugging was set to off.

Please find attached  the corrected version.

Regards


2009/8/17, brem belguebli <brem.belguebli at gmail.com>:
>
> Hi,
>
> I have written a resource script (mdmonitor.sh) to manage mdadm mirrors
> across a cluster.
>
> This is for testing purposes and still needs some additionnal features and
> some deeper debugging.
>
> If someone is interested in hacking it to improve it or to correct it, I'll
> be glad.
>
> A readme file is provided that explains the goal of this resource script.
>
> Thanks to Rafael Mico Miranda for his lvm-cluster resource script used on
> top of mdmonitor.
>
>  Brem
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090817/4ce57201/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: mdmonitor.sh
Type: application/x-sh
Size: 33490 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090817/4ce57201/attachment.sh>

From brem.belguebli at gmail.com  Mon Aug 17 14:10:01 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Mon, 17 Aug 2009 16:10:01 +0200
Subject: [Linux-cluster] Fwd: CLVM exclusive mode
In-Reply-To: <1cafab770908170417n16919354h38e22bcb97d31bf7@mail.gmail.com>
References: <4A7AA7DE02000027000536E3@lucius.provo.novell.com>
	<4A7A8240.7000206@redhat.com>
	<29ae894c0908160459m34dd02dcp8601e63d7bf62c3a@mail.gmail.com>
	<1cafab770908170417n16919354h38e22bcb97d31bf7@mail.gmail.com>
Message-ID: <29ae894c0908170710t14071b3ai59f793b9390b9b38@mail.gmail.com>

Hi,

Thanks a lot.

I'll try it, but would be enjoyed if RH could implement it.

Regards


2009/8/17, Xinwei Hu <hxinwei at gmail.com>:
>
> Hi,
>
> Attached a very naive try to solve the issue you have.
>
> Would you give it a test ?
>
> Thanks.
>
> 2009/8/16 brem belguebli <brem.belguebli at gmail.com>:
> > Hi,
> >
> > I don't think adding more security can be considered as pointless,
> > especially when this has no impact on performance or behaviour.
> > The question is, what's the point in  allowing the clustered active
> > exclusive lock to be bypassed ?
> >
> > In comparison to other volume management solutions (on various unices)
> where
> > these barriers are already implemented, the lack of them on Linux can  be
> > seen as a weakness.
> > Regards
> >
> >
> > 2009/8/6, Christine Caulfield <ccaulfie at redhat.com>:
> >>
> >> On 06/08/09 02:52, Jia Ju Zhang wrote:
> >>>
> >>> Just RFC:
> >>> I noticed that 'vgchange -ay' can convert the lock which locked by
> >>> 'vgchange -aey'
> >>> from EX to CR. Is that acceptable to change the logic into always
> >>> allocating a new lock
> >>> rather than converting an existing lock?
> >>> In that case, 'vgchange -ay' won't change the result of 'vgchange
> -aey'.
> >>> But if we really
> >>> want to convert the lock, we can firstly invoke 'vgchange -aen' to
> >>> release the EX lock,
> >>> then invoke the 'vgchange -ay'.
> >>>
> >>> Does this make sense? Or what side effect it may introduce?
> >>
> >>
> >> I think it makes no sense at all, and have already said so on this list.
> >> As far as I know there is no bugzilla for this problem and therefore it
> >> isn't being worked on.
> >>
> >> So ... if you care about this ... you know what to do ;-)
> >>
> >> Chrissie
> >>
> >>>>>> On 8/6/2009 at  9:39 AM, in message<4A7A346B.A94 : 39 : 18251>, Jia
> Ju
> >>>>>> Zhang
> >>>
> >>> wrote:
> >>>>
> >>>> On Fri, 2009-07-31 at 21:29 +0200, brem belguebli wrote:
> >>>>>
> >>>>> Hi,
> >>>>>
> >>>>> Same behaviour as the one from Rafael.
> >>>>>
> >>>>> Everything is coherent as long as you use the exclusive flag from the
> >>>>> rogue node, the locking does the job. Deactivating an already opened
> >>>>> VG (mounted lvol) is not possible either. How could this behave in
> >>>>> case one used raw devices instead of FS ?
> >>>>>
> >>>>> But when you come to ignore the exclusive flag on the rogue node
> >>>>> (vgchange -a y vgXX) the locking is completely bypassed. It's
> >>>>> definitely here that the watchdog has to be (within the tools
> >>>>> lvchange, vgchange, or at dlm level).
> >>>>
> >>>> Is there an open bugzilla # for this? Would like to follow this issue.
> >>>>
> >>>
> >>>
> >>>
> >>> --
> >>> Linux-cluster mailing list
> >>> Linux-cluster at redhat.com
> >>> https://www.redhat.com/mailman/listinfo/linux-cluster
> >>
> >> --
> >> Linux-cluster mailing list
> >> Linux-cluster at redhat.com
> >> https://www.redhat.com/mailman/listinfo/linux-cluster
> >
> >
> > --
> > Linux-cluster mailing list
> > Linux-cluster at redhat.com
> > https://www.redhat.com/mailman/listinfo/linux-cluster
> >
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090817/e5359846/attachment.htm>

From pasik at iki.fi  Mon Aug 17 18:12:05 2009
From: pasik at iki.fi (Pasi =?iso-8859-1?Q?K=E4rkk=E4inen?=)
Date: Mon, 17 Aug 2009 21:12:05 +0300
Subject: [Linux-cluster] Fwd: CLVM exclusive mode
In-Reply-To: <29ae894c0908170710t14071b3ai59f793b9390b9b38@mail.gmail.com>
References: <4A7AA7DE02000027000536E3@lucius.provo.novell.com>
	<4A7A8240.7000206@redhat.com>
	<29ae894c0908160459m34dd02dcp8601e63d7bf62c3a@mail.gmail.com>
	<1cafab770908170417n16919354h38e22bcb97d31bf7@mail.gmail.com>
	<29ae894c0908170710t14071b3ai59f793b9390b9b38@mail.gmail.com>
Message-ID: <20090817181205.GK24960@edu.joroinen.fi>

On Mon, Aug 17, 2009 at 04:10:01PM +0200, brem belguebli wrote:
> Hi,
> 
> Thanks a lot.
> 
> I'll try it, but would be enjoyed if RH could implement it.
> 

Did you already open bugzilla entry about it? 

Quote from this same thread:

"I think it makes no sense at all, and have already said so on this list.
As far as I know there is no bugzilla for this problem and therefore it
isn't being worked on.

So ... if you care about this ... you know what to do ;-)

Chrissie"

-- Pasi

> Regards
> 
> 
> 2009/8/17, Xinwei Hu <hxinwei at gmail.com>:
> >
> > Hi,
> >
> > Attached a very naive try to solve the issue you have.
> >
> > Would you give it a test ?
> >
> > Thanks.
> >
> > 2009/8/16 brem belguebli <brem.belguebli at gmail.com>:
> > > Hi,
> > >
> > > I don't think adding more security can be considered as pointless,
> > > especially when this has no impact on performance or behaviour.
> > > The question is, what's the point in  allowing the clustered active
> > > exclusive lock to be bypassed ?
> > >
> > > In comparison to other volume management solutions (on various unices)
> > where
> > > these barriers are already implemented, the lack of them on Linux can  be
> > > seen as a weakness.
> > > Regards
> > >
> > >
> > > 2009/8/6, Christine Caulfield <ccaulfie at redhat.com>:
> > >>
> > >> On 06/08/09 02:52, Jia Ju Zhang wrote:
> > >>>
> > >>> Just RFC:
> > >>> I noticed that 'vgchange -ay' can convert the lock which locked by
> > >>> 'vgchange -aey'
> > >>> from EX to CR. Is that acceptable to change the logic into always
> > >>> allocating a new lock
> > >>> rather than converting an existing lock?
> > >>> In that case, 'vgchange -ay' won't change the result of 'vgchange
> > -aey'.
> > >>> But if we really
> > >>> want to convert the lock, we can firstly invoke 'vgchange -aen' to
> > >>> release the EX lock,
> > >>> then invoke the 'vgchange -ay'.
> > >>>
> > >>> Does this make sense? Or what side effect it may introduce?
> > >>
> > >>
> > >> I think it makes no sense at all, and have already said so on this list.
> > >> As far as I know there is no bugzilla for this problem and therefore it
> > >> isn't being worked on.
> > >>
> > >> So ... if you care about this ... you know what to do ;-)
> > >>
> > >> Chrissie
> > >>
> > >>>>>> On 8/6/2009 at  9:39 AM, in message<4A7A346B.A94 : 39 : 18251>, Jia
> > Ju
> > >>>>>> Zhang
> > >>>
> > >>> wrote:
> > >>>>
> > >>>> On Fri, 2009-07-31 at 21:29 +0200, brem belguebli wrote:
> > >>>>>
> > >>>>> Hi,
> > >>>>>
> > >>>>> Same behaviour as the one from Rafael.
> > >>>>>
> > >>>>> Everything is coherent as long as you use the exclusive flag from the
> > >>>>> rogue node, the locking does the job. Deactivating an already opened
> > >>>>> VG (mounted lvol) is not possible either. How could this behave in
> > >>>>> case one used raw devices instead of FS ?
> > >>>>>
> > >>>>> But when you come to ignore the exclusive flag on the rogue node
> > >>>>> (vgchange -a y vgXX) the locking is completely bypassed. It's
> > >>>>> definitely here that the watchdog has to be (within the tools
> > >>>>> lvchange, vgchange, or at dlm level).
> > >>>>
> > >>>> Is there an open bugzilla # for this? Would like to follow this issue.
> > >>>>
> > >>>
> > >>>
> > >>>
> > >>> --
> > >>> Linux-cluster mailing list
> > >>> Linux-cluster at redhat.com
> > >>> https://www.redhat.com/mailman/listinfo/linux-cluster
> > >>
> > >> --
> > >> Linux-cluster mailing list
> > >> Linux-cluster at redhat.com
> > >> https://www.redhat.com/mailman/listinfo/linux-cluster
> > >
> > >
> > > --
> > > Linux-cluster mailing list
> > > Linux-cluster at redhat.com
> > > https://www.redhat.com/mailman/listinfo/linux-cluster
> > >
> >
> > --
> > Linux-cluster mailing list
> > Linux-cluster at redhat.com
> > https://www.redhat.com/mailman/listinfo/linux-cluster
> >
> >

> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster



From pradhanparas at gmail.com  Mon Aug 17 18:28:10 2009
From: pradhanparas at gmail.com (Paras pradhan)
Date: Mon, 17 Aug 2009 13:28:10 -0500
Subject: [Linux-cluster] do I have a fence DRAC device?
In-Reply-To: <3128ba140908170350ge619930w5c17368ff0d3cf42@mail.gmail.com>
References: <3128ba140908100324l6cdb4c34ra5f5edb39c6903e9@mail.gmail.com>
	<8b711df40908101134t69b8e12cof6cc551809421e45@mail.gmail.com>
	<3128ba140908170350ge619930w5c17368ff0d3cf42@mail.gmail.com>
Message-ID: <8b711df40908171128j1fc18525nfd7df01d7604cda0@mail.gmail.com>

On Mon, Aug 17, 2009 at 5:50 AM, ESGLinux<esggrupos at gmail.com> wrote:
> Hi,
>>
>> Yes you have IPMI and if you are using 1950 Dell, DRAC should be there
>> too.
>
> What do I have to see at start up? I haven?t seen anything about DRAC. I
> have seen that with CTRL+E I entered in an utility to configure de ip of the
> card, but I don?t know if it?s a DRAC card. How can I check it?
> Thanks in advance
> ESG
>

When you press the power button, it comes as "Remote Access Card".
Then after few other screens you can login to enable and configure
DRAC.

racadm did't work for me  as well. But you do not need it if you
configure it boot time.


Paras.

>
>>
>> You can see if you have DRAC or not when the server starts and
>> before the loading of the OS.
>>
>> I have 1850s and I am using DRAC for fencing.
>>
>>
>> Paras.
>>
>> --
>> Linux-cluster mailing list
>> Linux-cluster at redhat.com
>> https://www.redhat.com/mailman/listinfo/linux-cluster
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>



From brem.belguebli at gmail.com  Mon Aug 17 18:46:12 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Mon, 17 Aug 2009 20:46:12 +0200
Subject: [Linux-cluster] Fwd: CLVM exclusive mode
In-Reply-To: <20090817181205.GK24960@edu.joroinen.fi>
References: <4A7AA7DE02000027000536E3@lucius.provo.novell.com>
	<4A7A8240.7000206@redhat.com>
	<29ae894c0908160459m34dd02dcp8601e63d7bf62c3a@mail.gmail.com>
	<1cafab770908170417n16919354h38e22bcb97d31bf7@mail.gmail.com>
	<29ae894c0908170710t14071b3ai59f793b9390b9b38@mail.gmail.com>
	<20090817181205.GK24960@edu.joroinen.fi>
Message-ID: <29ae894c0908171146u5130bb71t1e8a573eda5f3311@mail.gmail.com>

done,

Bug 517900 <https://bugzilla.redhat.com/show_bug.cgi?id=517900>


2009/8/17, Pasi K?rkk?inen <pasik at iki.fi>:
>
> On Mon, Aug 17, 2009 at 04:10:01PM +0200, brem belguebli wrote:
> > Hi,
> >
> > Thanks a lot.
> >
> > I'll try it, but would be enjoyed if RH could implement it.
> >
>
> Did you already open bugzilla entry about it?
>
> Quote from this same thread:
>
> "I think it makes no sense at all, and have already said so on this list.
> As far as I know there is no bugzilla for this problem and therefore it
> isn't being worked on.
>
> So ... if you care about this ... you know what to do ;-)
>
> Chrissie"
>
> -- Pasi
>
> > Regards
> >
> >
> > 2009/8/17, Xinwei Hu <hxinwei at gmail.com>:
> > >
> > > Hi,
> > >
> > > Attached a very naive try to solve the issue you have.
> > >
> > > Would you give it a test ?
> > >
> > > Thanks.
> > >
> > > 2009/8/16 brem belguebli <brem.belguebli at gmail.com>:
> > > > Hi,
> > > >
> > > > I don't think adding more security can be considered as pointless,
> > > > especially when this has no impact on performance or behaviour.
> > > > The question is, what's the point in  allowing the clustered active
> > > > exclusive lock to be bypassed ?
> > > >
> > > > In comparison to other volume management solutions (on various
> unices)
> > > where
> > > > these barriers are already implemented, the lack of them on Linux
> can  be
> > > > seen as a weakness.
> > > > Regards
> > > >
> > > >
> > > > 2009/8/6, Christine Caulfield <ccaulfie at redhat.com>:
> > > >>
> > > >> On 06/08/09 02:52, Jia Ju Zhang wrote:
> > > >>>
> > > >>> Just RFC:
> > > >>> I noticed that 'vgchange -ay' can convert the lock which locked by
> > > >>> 'vgchange -aey'
> > > >>> from EX to CR. Is that acceptable to change the logic into always
> > > >>> allocating a new lock
> > > >>> rather than converting an existing lock?
> > > >>> In that case, 'vgchange -ay' won't change the result of 'vgchange
> > > -aey'.
> > > >>> But if we really
> > > >>> want to convert the lock, we can firstly invoke 'vgchange -aen' to
> > > >>> release the EX lock,
> > > >>> then invoke the 'vgchange -ay'.
> > > >>>
> > > >>> Does this make sense? Or what side effect it may introduce?
> > > >>
> > > >>
> > > >> I think it makes no sense at all, and have already said so on this
> list.
> > > >> As far as I know there is no bugzilla for this problem and therefore
> it
> > > >> isn't being worked on.
> > > >>
> > > >> So ... if you care about this ... you know what to do ;-)
> > > >>
> > > >> Chrissie
> > > >>
> > > >>>>>> On 8/6/2009 at  9:39 AM, in message<4A7A346B.A94 : 39 : 18251>,
> Jia
> > > Ju
> > > >>>>>> Zhang
> > > >>>
> > > >>> wrote:
> > > >>>>
> > > >>>> On Fri, 2009-07-31 at 21:29 +0200, brem belguebli wrote:
> > > >>>>>
> > > >>>>> Hi,
> > > >>>>>
> > > >>>>> Same behaviour as the one from Rafael.
> > > >>>>>
> > > >>>>> Everything is coherent as long as you use the exclusive flag from
> the
> > > >>>>> rogue node, the locking does the job. Deactivating an already
> opened
> > > >>>>> VG (mounted lvol) is not possible either. How could this behave
> in
> > > >>>>> case one used raw devices instead of FS ?
> > > >>>>>
> > > >>>>> But when you come to ignore the exclusive flag on the rogue node
> > > >>>>> (vgchange -a y vgXX) the locking is completely bypassed. It's
> > > >>>>> definitely here that the watchdog has to be (within the tools
> > > >>>>> lvchange, vgchange, or at dlm level).
> > > >>>>
> > > >>>> Is there an open bugzilla # for this? Would like to follow this
> issue.
> > > >>>>
> > > >>>
> > > >>>
> > > >>>
> > > >>> --
> > > >>> Linux-cluster mailing list
> > > >>> Linux-cluster at redhat.com
> > > >>> https://www.redhat.com/mailman/listinfo/linux-cluster
> > > >>
> > > >> --
> > > >> Linux-cluster mailing list
> > > >> Linux-cluster at redhat.com
> > > >> https://www.redhat.com/mailman/listinfo/linux-cluster
> > > >
> > > >
> > > > --
> > > > Linux-cluster mailing list
> > > > Linux-cluster at redhat.com
> > > > https://www.redhat.com/mailman/listinfo/linux-cluster
> > > >
> > >
> > > --
> > > Linux-cluster mailing list
> > > Linux-cluster at redhat.com
> > > https://www.redhat.com/mailman/listinfo/linux-cluster
> > >
> > >
>
> > --
> > Linux-cluster mailing list
> > Linux-cluster at redhat.com
> > https://www.redhat.com/mailman/listinfo/linux-cluster
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090817/6fa7bf38/attachment.htm>

From alan.zg at gmail.com  Mon Aug 17 21:38:10 2009
From: alan.zg at gmail.com (Alan A)
Date: Mon, 17 Aug 2009 16:38:10 -0500
Subject: [Linux-cluster] What is the kernel recommendation for 32bit RHEL5
	GFS cluster
Message-ID: <fac531740908171438j36cfec67qf28cb55522e8810c@mail.gmail.com>

We are moving from Development to Production. Can you please specify what
are the recommended kernel versions to run on 32bit system with 32GB+
memory.

Would you recommend a PAE Kernel on GFS cluster?
Would you recommend 64bit over 32bit version, if so why?

Our Experience so far:
1) We have new HP Proliant 385 servers with 32bit RHEL5 where non PAE kernel
sees all the memory (32GB).
2) In our test environment recent update to kernel 2.6.18-128.4.1.el5 we
were unable to register all the memory and PAE kernel gave us problems when
mounting GFS.

[root at dev02 log]# dmidecode | grep -i "Range Size" ; uname -a; free -m
    Range Size: 12 GB
    Range Size: 3932161 kB
    Range Size: 3932161 kB
    Range Size: 4 GB
    Range Size: 4 GB
    Range Size: 4 GB
    Range Size: 4 GB
    Range Size: 1 kB
    Range Size: 1 kB
Linux dev02.foo.com 2.6.18-128.4.1.el5 #1 SMP Thu Jul 23 19:59:17 EDT 2009
i686 i686 i386 GNU/Linux
             total       used       free     shared    buffers     cached
Mem:          3674       3534        140          0        229       2818
-/+ buffers/cache:        486       3188
Swap:        12286          0      12285
[root at dev02 log]#


-- 
Alan A.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090817/5d90d106/attachment.htm>

From pradhanparas at gmail.com  Mon Aug 17 21:59:09 2009
From: pradhanparas at gmail.com (Paras pradhan)
Date: Mon, 17 Aug 2009 16:59:09 -0500
Subject: [Linux-cluster] Cluster behavior
Message-ID: <8b711df40908171459q1afecda7k7b0dcc4773153b84@mail.gmail.com>

I have a 3 nodes cluster.

Node A - Vote 1

Node B - Vote 1

Node C - Votes 2

Qdisk - Votes 3


Altogether the cluster has 7 votes. The required min quorum to run the
cluster would be in this case 4. Now if I poweroff node 1, I can see
Quorum Dissolved in the terminals of Node 2 and Node3 . This cluster
has xen virtual machines.

Whats wrong and how to do debug the problem?

Thanks
Paras.



From JInserra at zebra.com  Mon Aug 17 22:03:21 2009
From: JInserra at zebra.com (Inserra, Joseph)
Date: Mon, 17 Aug 2009 17:03:21 -0500
Subject: [Linux-cluster] (no subject)
In-Reply-To: <8b711df40908171459q1afecda7k7b0dcc4773153b84@mail.gmail.com>
References: <8b711df40908171459q1afecda7k7b0dcc4773153b84@mail.gmail.com>
Message-ID: <61A9E641F597D648B02F65D737DA6CA0170833B5@03S08EXMB01.zebra.lan>



- CONFIDENTIAL-

This email and any files transmitted with it are confidential, and may also be legally privileged.  If you are not the intended recipient, you may not review, use, copy, or distribute this message. If you receive this email in error, please notify the sender immediately by reply email and then delete this email.



From cryptogrid at gmail.com  Mon Aug 17 22:11:01 2009
From: cryptogrid at gmail.com (crypto grid)
Date: Mon, 17 Aug 2009 19:11:01 -0300
Subject: [Linux-cluster] sanity check
In-Reply-To: <20090815012339.GB6468@monsterjam.org>
References: <20090811031639.GA728@monsterjam.org>
	<a9f464b80908141512s41661d0cu4b9b7bf91ce16604@mail.gmail.com>
	<20090815012339.GB6468@monsterjam.org>
Message-ID: <a9f464b80908171511p61b950f8pa84309e8d5daf706@mail.gmail.com>

On Fri, Aug 14, 2009 at 10:23 PM, <jason at monsterjam.org> wrote:

> On Fri, Aug 14, 2009 at 07:12:34PM -0300, crypto grid wrote:
> >    Your failover domain includes server1 and server2, and you have two
> >    diferent services using the same ip address in the services section. I
> >    wonder what happen if you have to realocate just one of the services?
> I
> >    think you couldn't.
> >
> >    I think you have to define two different ip addresses in the resource
> >    section, one for apache and one for postfix.
> >
> >    I'm not saying that this is causing the problem, I'd try disabling one
> >    service and enabling just one.
> >
>
> ok, Ive just modified it to be the following config, and it looks like it
> works fine.
> But please tell me if theres anything  else Ive missed.


I don't see anything wrong regarding to the resources and service sections.
After the ip address adition, are the cluster and services working
correctly?


>
>
> thanks for the help!
>
> Jason
>
>
> [jason at server1 ~]$ sudo cat /etc/cluster/cluster.conf
> <?xml version="1.0"?>
> <cluster config_version="26" name="mycluster">
>        <fence_daemon clean_start="0" post_fail_delay="0"
> post_join_delay="3"/>
>        <clusternodes>
>                <clusternode name="server1" votes="1">
>                        <fence>
>                                <method name="1">
>                                        <device name="apc_power_switch"
> option="off" port="1" switch="1"/>
>                                        <device name="apc_power_switch"
> option="off" port="2" switch="1"/>
>                                        <device name="apc_power_switch"
> option="on" port="1" switch="1"/>
>                                        <device name="apc_power_switch"
> option="on" port="2" switch="1"/>
>                                </method>
>                        </fence>
>                </clusternode>
>                <clusternode name="server2" votes="1">
>                        <fence>
>                                <method name="1">
>                                        <device name="apc_power_switch"
> option="off" port="3" switch="1"/>
>                                        <device name="apc_power_switch"
> option="off" port="4" switch="1"/>
>                                        <device name="apc_power_switch"
> option="on" port="3" switch="1"/>
>                                        <device name="apc_power_switch"
> option="on" port="4" switch="1"/>
>                                </method>
>                        </fence>
>                </clusternode>
>        </clusternodes>
>        <cman expected_votes="1" two_node="1"/>
>        <fencedevices>
>                <fencedevice agent="fence_apc" ipaddr="192.168.1.8"
> login="xxxxx" name="apc_power_switch" passwd="xxxx2xxx"/>
>        </fencedevices>
>        <rm>
>                <failoverdomains>
>                        <failoverdomain name="httpd" ordered="1"
> restricted="1">
>                                <failoverdomainnode name="server1"
> priority="1"/>
>                                <failoverdomainnode name="server2"
> priority="2"/>
>                        </failoverdomain>
>                </failoverdomains>
>                <resources>
>                        <script file="/etc/init.d/httpd"
> name="cluster_apache"/>
>                        <ip address="192.168.1.9" monitor_link="1"/>
>                        <script file="/etc/init.d/postfix"
> name="cluster_posstfix"/>
>                        <ip address="192.168.1.7" monitor_link="1"/>
>                </resources>
>                <service autostart="1" domain="httpd" name="Apache Service">
>                        <script ref="cluster_apache"/>
>                        <ip ref="192.168.1.9"/>
>                </service>
>                <service autostart="1" domain="httpd" name="postfix
> service">
>                        <script ref="cluster_posstfix"/>
>                        <ip ref="192.168.1.7"/>
>                </service>
>        </rm>
> </cluster>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090817/70751f6e/attachment.htm>

From jason at monsterjam.org  Mon Aug 17 22:39:14 2009
From: jason at monsterjam.org (jason at monsterjam.org)
Date: Mon, 17 Aug 2009 18:39:14 -0400
Subject: [Linux-cluster] sanity check
In-Reply-To: <a9f464b80908171511p61b950f8pa84309e8d5daf706@mail.gmail.com>
References: <20090811031639.GA728@monsterjam.org>
	<a9f464b80908141512s41661d0cu4b9b7bf91ce16604@mail.gmail.com>
	<20090815012339.GB6468@monsterjam.org>
	<a9f464b80908171511p61b950f8pa84309e8d5daf706@mail.gmail.com>
Message-ID: <20090817223914.GA21646@monsterjam.org>

> 
>    I don't see anything wrong regarding to the resources and service
>    sections.
>    After the ip address adition, are the cluster and services working
>    correctly?
> 
ooh my aplogies for not reporting back.. YES they are working correctly now as far as I can tell.

thanks/regards,
Jason



From sdake at redhat.com  Mon Aug 17 23:06:23 2009
From: sdake at redhat.com (Steven Dake)
Date: Mon, 17 Aug 2009 16:06:23 -0700
Subject: [Linux-cluster] problems with gfs and openais cluster.
In-Reply-To: <1247a04a0908150512g2966f1cnefffb0c8dc5d033e@mail.gmail.com>
References: <1247a04a0908142311o5ff51b2am3f245a736eda45b8@mail.gmail.com>
	<1250317407.2649.16.camel@localhost.localdomain>
	<1247a04a0908150512g2966f1cnefffb0c8dc5d033e@mail.gmail.com>
Message-ID: <1250550383.19265.86.camel@localhost.localdomain>

On Sat, 2009-08-15 at 08:12 -0400, Enrique Sanchez wrote:
> On Sat, Aug 15, 2009 at 2:23 AM, Steven Dake<sdake at redhat.com> wrote:
> >
> > The openais.conf file is not used when starting with the cman service.
> > Instead cman uses its own configuration file
> > in /etc/cluster/cluster.conf.  openais should not be chkconfig'ed on
> > unless running without cman.
> >
> > One reason you could have the error is that you have multiple clusters
> > on the same multicast address as specified in /etc/cluster/cluster.conf.
> >
> > Regards
> > -steve
> >
> 
> steve,
> 
> thanks for the quick reply: the mcast address is the same and it
> looks like it is using the same VLAN here are the details....
> 
> this is one cluster (cluster one):
> 
>         <cman expected_votes="1" two_node="1">
>                 <multicast addr="224.0.0.1"/>
>         </cman>
> 
> and this is in the 2nd cluster (cluster two)......
> 
>         <cman expected_votes="1" two_node="1">
>                 <multicast addr="224.0.0.1" interface="eth1"/>
>         </cman>
> 
> 
> unfortunately eth1 is on the same VLAN for both clusters...is cluster
> 1 using the mcast address on both adapters?
> 
> thnx...
> enrique.

clusters are uniquely identified by their multicast address:port pair.
I suggest changing one of the cluster multicast addresses to 224.0.0.2.

Regards
-steve
> 
> 



From muffaleta at gmail.com  Mon Aug 17 23:51:51 2009
From: muffaleta at gmail.com (Christopher Chen)
Date: Mon, 17 Aug 2009 16:51:51 -0700
Subject: [Linux-cluster] Rgmanager and Xen VMs
Message-ID: <7bc80d500908171651m48927838v21644fc03cd21fca@mail.gmail.com>

Hi there--

I'm wondering about the behavior of rgmanager with regards to Xen VMs
and memory planning.

I see there's a memory parameter defined in vm.sh. Does cluster
manager take this into account when trying to start up guests? Right
now it's undefined so my rgmanager will try to start up guests in the
order defined for the failover domain--of course, at some point, VMs
fail to start when the first Dom0's free memory gets too low, then
they get started on the next machine, and so on...

I hoping the answer is yes!

cc

-- 
Chris Chen <muffaleta at gmail.com>
"The fact that yours is better than anyone else's
is not a guarantee that it's any good."
-- Seen on a wall



From wen.gang.wang at oracle.com  Tue Aug 18 02:32:54 2009
From: wen.gang.wang at oracle.com (Wengang Wang)
Date: Tue, 18 Aug 2009 10:32:54 +0800
Subject: [Linux-cluster] gfs2 mount hangs
In-Reply-To: <20090814145738.GA7094@redhat.com>
References: <4A83B113.60609@oracle.com> <20090813141424.GA8148@redhat.com>
	<4A84CC39.3080809@oracle.com> <20090814145738.GA7094@redhat.com>
Message-ID: <4A8A12D6.5050907@oracle.com>

Hi Dave,

> 
>> node desk:
>> Aug 14 18:07:44 desk gfs_controld[2206]: recovery_uevent mg not found 1
>> Aug 14 18:07:44 desk gfs_controld[2206]: recovery_uevent mg not found 1
>> Aug 14 18:07:44 desk gfs_controld[2206]: recovery_uevent mg not found 1
> 
> There's a problem here, but it's not clear what has gone wrong.  Could you try
> this again and after these messages appear send the output of
> "gfs_control dump" from both nodes?
> 
I got the outputs:
1) for node cool(the second mount is on this node):

[root at cool ~]# gfs_control dump
1250560388 logging mode 3 syslog f 160 p 6 logfile p 6 
/var/log/cluster/gfs_controld.log
1250560388 gfs_controld 3.0.0 started
1250560389 logging mode 3 syslog f 160 p 6 logfile p 6 
/var/log/cluster/gfs_controld.log
1250560389 group_mode 3 compat 0
1250560389 setup_cpg 13
1250560389 set_protocol member_count 1 propose daemon 1.1.1 kernel 1.1.1
1250560389 run protocol from nodeid 1
1250560389 daemon run 1.1.1 max 1.1.1 kernel run 1.1.1 max 1.1.1
1250561975 client connection 5 fd 15
1250561975 join: /gfs2 gfs2 lock_dlm testgfs2:1 rw /dev/sdb
1250561975 1 join: cluster name matches: testgfs2
1250561975 1 process_dlmcontrol register 0
1250561975 1 add_change cg 1 joined nodeid 1
1250561975 1 add_change cg 1 we joined
1250561975 1 add_change cg 1 counts member 1 joined 1 remove 0 failed 0
1250561975 1 wait_conditions skip for zero started_count
1250561975 1 send_start cg 1 id_count 1 om 0 nm 1 oj 0 nj 0
1250561975 1 receive_start 1:1 len 92
1250561975 1 match_change 1:1 matches cg 1
1250561975 1 wait_messages cg 1 got all 1
1250561975 1 pick_first_recovery_master low 1 old 0
1250561975 1 sync_state all_nodes_new first_recovery_needed master 1
1250561975 1 create_old_nodes all new
1250561975 1 create_new_nodes 1 ro 0 spect 0
1250561975 1 create_failed_journals all new
1250561975 1 create_new_journals 1 gets jid 0
1250561975 1 apply_recovery first start_kernel
1250561975 1 start_kernel cg 1 member_count 1
1250561975 1 set /sys/fs/gfs2/testgfs2:1/lock_module/block to 0
1250561975 1 set open /sys/fs/gfs2/testgfs2:1/lock_module/block error -1 2
1250561975 1 client_reply_join_full ci 5 result 0 
hostdata=jid=0:id=2032068419:first=1
1250561975 client_reply_join 1 ci 5 result 0
1250561975 uevent add gfs2 /fs/gfs2/testgfs2:1
1250561975 1 ping_kernel_mount

there is a line with "...error -1 2", but seems the line is out during 
the second(hang) mount.

2) for node desk(the first mount is on this node).
[root at desk ~]# gfs_control dump
1250589068 logging mode 3 syslog f 160 p 6 logfile p 6 
/var/log/cluster/gfs_controld.log
1250589068 gfs_controld 3.0.0 started
1250589068 logging mode 3 syslog f 160 p 6 logfile p 6 
/var/log/cluster/gfs_controld.log
1250589068 group_mode 3 compat 0
1250589068 setup_cpg 13
1250589068 run protocol from nodeid 1
1250589068 daemon run 1.1.1 max 1.1.1 kernel run 1.1.1 max 1.1.1
1250590628 uevent add gfs2 /fs/gfs2/testgfs2:1
1250590628 uevent change gfs2 /fs/gfs2/testgfs2:1
1250590628 recovery_uevent mg not found 1
1250590628 uevent change gfs2 /fs/gfs2/testgfs2:1
1250590628 recovery_uevent mg not found 1
1250590628 uevent change gfs2 /fs/gfs2/testgfs2:1
1250590628 recovery_uevent mg not found 1
1250590628 uevent online gfs2 /fs/gfs2/testgfs2:1

the log looks be with no much info, can it help us?

regards,
wengang.



From alfredo.moralejo at roche.com  Tue Aug 18 08:34:59 2009
From: alfredo.moralejo at roche.com (Moralejo, Alfredo)
Date: Tue, 18 Aug 2009 10:34:59 +0200
Subject: [Linux-cluster] CLVMD without GFS
In-Reply-To: <29ae894c0907311725k4759e631m332ffd38857a6f99@mail.gmail.com>
References: <MJ2mY0lYjZGc@NBjvyXzB> <1249079337.6494.10.camel@mecatol>
	<29ae894c0907311725k4759e631m332ffd38857a6f99@mail.gmail.com>
Message-ID: <FD716ADFEE797543AA8B2AB0B3F9148E01A09C46B9@rkamsem701.emea.roche.com>

Hi,

I have the same concern with the activation of non-GFS file systems and I agree with the idea of using exclusive activation (keeping in mind that the strange behavior of exclusive tag being bypassed should be fixed). However, from my point of view, the best way is not to create a new resource but adding a new "exclusive" option to the regular fs resource or something like that, as I think it's the desired behavior most times the fs resource is used.

Best regards,

Alfredo



________________________________
From: linux-cluster-bounces at redhat.com [mailto:linux-cluster-bounces at redhat.com] On Behalf Of brem belguebli
Sent: Saturday, August 01, 2009 2:25 AM
To: linux clustering
Subject: Re: [Linux-cluster] CLVMD without GFS

Hi Rafael,
Of course, that is what I was telling you the other day, your script works fine, but I just wanted to twist a little clvm.
The prerequisites are already setup (locking type, dm-mp, lvm, etc...).
The only thing that disturbes me is that you can bypass under certain conditions the locking.
Thanks for the help
Brem

2009/8/1 Rafael Mic? Miranda <rmicmirregs at gmail.com<mailto:rmicmirregs at gmail.com>>
Hi Brem,

El vie, 31-07-2009 a las 06:09 -0400, crosa at redhat.com<mailto:crosa at redhat.com> escribi?:
>
> --- mensagem original ---
> De: brem belguebli <brem.belguebli at gmail.com<mailto:brem.belguebli at gmail.com>>
> Assunto: Re: [Linux-cluster] CLVMD without GFS
> Data: 29 de Julho de 2009
> Hora: 10:1:29
>
> Hi Rafael,
>
> Just posted the basic tests I'm doing on both linux-cluster and linux-lvm.
>
> I can't get exclusive activation to work properly, I may be missing some
> step in my process.
>
> Brem
>
>

>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com<mailto:Linux-cluster at redhat.com>
> https://www.redhat.com/mailman/listinfo/linux-cluster
Sorry, i'm not sure if I missed this mail.

Maybe i need to explain a couple of things first, just to make them
clear.

As part of the use of the lvm-cluster.sh resource script, you need to:

1.- Configure properly the lock type into lvm.conf to type 3 (example
into this link:

https://www.redhat.com/archives/linux-cluster/2009-July/msg00253.html

)

2.- Configure your multipathing software

3.- Create your Physical Volumes

4.- Configure your Volume Groups as clustered Volume Groups

5.- Create your Logical Volumens into the clustered Volume Groups

6.- And, and the not so obvious, de-activate all the Logical Volumes you
plan to use as exclusive Logical Volumes (using exclusive flag).

De-activation can be done with "vgchange -an volgrp01/logvol01" or
similar command.

If you don't do step 6, you will receive an error message when executing
"lvchange -aey volgrp01/logvol01". This command is executed (with the
proper volume group and logical volume names) internally in the resource
script.

I designed the lvm-cluster.sh resource script to be verbose, maybe you
can copy here your logs (default to /var/log/messages on RHEL systems).

Cheers and thanks,

Rafael

--
Rafael Mic? Miranda

--
Linux-cluster mailing list
Linux-cluster at redhat.com<mailto:Linux-cluster at redhat.com>
https://www.redhat.com/mailman/listinfo/linux-cluster

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090818/010a2126/attachment.htm>

From brem.belguebli at gmail.com  Tue Aug 18 09:01:05 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Tue, 18 Aug 2009 11:01:05 +0200
Subject: [Linux-cluster] Questions about rgmanager
Message-ID: <29ae894c0908180201l14ab18aexddefd0f04bf2ba29@mail.gmail.com>

Hello,

I have a few questions about rgmanager:

1) How to make it take into account resource metadata changes without
restarting it. service rgmanager reload (or kill -HUP) doesn't.
    In case one needs to change some timers in a resource metadata, the only
way to make it is to stop and start rgmanager which brings down the
resources.

2) Is there a way to log what "rules" rgmanager succeded to load at start
time (or the ones it failed to), as having a "24 rules loaded" message isn't
very explicit.

The last point concerns the potential replacement of rgmanager with
pacemaker (as stated at http://sources.redhat.com/cluster/wiki/RGManager).
Will this happen in a near future ?

Brem
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090818/e7d6105a/attachment.htm>

From brem.belguebli at gmail.com  Tue Aug 18 09:09:53 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Tue, 18 Aug 2009 11:09:53 +0200
Subject: [Linux-cluster] CLVMD without GFS
In-Reply-To: <FD716ADFEE797543AA8B2AB0B3F9148E01A09C46B9@rkamsem701.emea.roche.com>
References: <MJ2mY0lYjZGc@NBjvyXzB> <1249079337.6494.10.camel@mecatol>
	<29ae894c0907311725k4759e631m332ffd38857a6f99@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C46B9@rkamsem701.emea.roche.com>
Message-ID: <29ae894c0908180209i46e23ed1o8fc2d372c8371e12@mail.gmail.com>

Hello Alfredo,

For the exclusive bypass, I have opened a bugzilla ticket. Hopefully it'll
be fixed (enhanced)0

For your second point, I'm not sure to ubndestand what you're saying.

My concern is to avoid manual activation on the wrong node, not an
accidental rgmanager activation.

Even if you configure your FS resource as exclusive, if it is possible, this
won't be of any effect with ext3 outside of rgmanager control as it is not
cluster aware.

Brem
2009/8/18, Moralejo, Alfredo <alfredo.moralejo at roche.com>:
>
>  Hi,
>
>
>
> I have the same concern with the activation of non-GFS file systems and I
> agree with the idea of using exclusive activation (keeping in mind that the
> strange behavior of exclusive tag being bypassed should be fixed). However,
> from my point of view, the best way is not to create a new resource but
> adding a new ?exclusive? option to the regular fs resource or something like
> that, as I think it?s the desired behavior most times the fs resource is
> used.
>
>
>
> Best regards,
>
>
>
> Alfredo
>
>
>
>
>
>
>  ------------------------------
>
> *From:* linux-cluster-bounces at redhat.com [mailto:
> linux-cluster-bounces at redhat.com] *On Behalf Of *brem belguebli
> *Sent:* Saturday, August 01, 2009 2:25 AM
> *To:* linux clustering
> *Subject:* Re: [Linux-cluster] CLVMD without GFS
>
>
>
> Hi Rafael,
>
> Of course, that is what I was telling you the other day, your script works
> fine, but I just wanted to twist a little clvm.
>
> The prerequisites are already setup (locking type, dm-mp, lvm, etc...).
>
> The only thing that disturbes me is that you can bypass under certain
> conditions the locking.
>
> Thanks for the help
>
> Brem
>
>
>
> 2009/8/1 Rafael Mic? Miranda <rmicmirregs at gmail.com>
>
> Hi Brem,
>
> El vie, 31-07-2009 a las 06:09 -0400, crosa at redhat.com escribi?:
>
> >
> > --- mensagem original ---
> > De: brem belguebli <brem.belguebli at gmail.com>
> > Assunto: Re: [Linux-cluster] CLVMD without GFS
> > Data: 29 de Julho de 2009
> > Hora: 10:1:29
> >
> > Hi Rafael,
> >
> > Just posted the basic tests I'm doing on both linux-cluster and
> linux-lvm.
> >
> > I can't get exclusive activation to work properly, I may be missing some
> > step in my process.
> >
> > Brem
> >
> >
>
> >
>
> > --
> > Linux-cluster mailing list
> > Linux-cluster at redhat.com
> > https://www.redhat.com/mailman/listinfo/linux-cluster
>
> Sorry, i'm not sure if I missed this mail.
>
> Maybe i need to explain a couple of things first, just to make them
> clear.
>
> As part of the use of the lvm-cluster.sh resource script, you need to:
>
> 1.- Configure properly the lock type into lvm.conf to type 3 (example
> into this link:
>
> https://www.redhat.com/archives/linux-cluster/2009-July/msg00253.html
>
> )
>
> 2.- Configure your multipathing software
>
> 3.- Create your Physical Volumes
>
> 4.- Configure your Volume Groups as clustered Volume Groups
>
> 5.- Create your Logical Volumens into the clustered Volume Groups
>
> 6.- And, and the not so obvious, de-activate all the Logical Volumes you
> plan to use as exclusive Logical Volumes (using exclusive flag).
>
> De-activation can be done with "vgchange -an volgrp01/logvol01" or
> similar command.
>
> If you don't do step 6, you will receive an error message when executing
> "lvchange -aey volgrp01/logvol01". This command is executed (with the
> proper volume group and logical volume names) internally in the resource
> script.
>
> I designed the lvm-cluster.sh resource script to be verbose, maybe you
> can copy here your logs (default to /var/log/messages on RHEL systems).
>
> Cheers and thanks,
>
> Rafael
>
> --
> Rafael Mic? Miranda
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090818/d6b70e26/attachment.htm>

From alfredo.moralejo at roche.com  Tue Aug 18 09:16:39 2009
From: alfredo.moralejo at roche.com (Moralejo, Alfredo)
Date: Tue, 18 Aug 2009 11:16:39 +0200
Subject: [Linux-cluster] CLVMD without GFS
In-Reply-To: <29ae894c0908180209i46e23ed1o8fc2d372c8371e12@mail.gmail.com>
References: <MJ2mY0lYjZGc@NBjvyXzB> <1249079337.6494.10.camel@mecatol>
	<29ae894c0907311725k4759e631m332ffd38857a6f99@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C46B9@rkamsem701.emea.roche.com>
	<29ae894c0908180209i46e23ed1o8fc2d372c8371e12@mail.gmail.com>
Message-ID: <FD716ADFEE797543AA8B2AB0B3F9148E01A09C4700@rkamsem701.emea.roche.com>

Hi Brem,

Maybe I did not explain it properly.

What I mean is that fs resource should activate the filesystems in exclusive mode as the resource created by Rafael does (lvm-cluster.sh) as I far as I understand to ensure the filesystem can not be mounted in other nodes accidentally.

Best regards,

Alfredo


________________________________
From: linux-cluster-bounces at redhat.com [mailto:linux-cluster-bounces at redhat.com] On Behalf Of brem belguebli
Sent: Tuesday, August 18, 2009 11:10 AM
To: linux clustering
Subject: Re: [Linux-cluster] CLVMD without GFS

Hello Alfredo,

For the exclusive bypass, I have opened a bugzilla ticket. Hopefully it'll be fixed (enhanced)0

For your second point, I'm not sure to ubndestand what you're saying.

My concern is to avoid manual activation on the wrong node, not an accidental rgmanager activation.

Even if you configure your FS resource as exclusive, if it is possible, this won't be of any effect with ext3 outside of rgmanager control as it is not cluster aware.

Brem
2009/8/18, Moralejo, Alfredo <alfredo.moralejo at roche.com<mailto:alfredo.moralejo at roche.com>>:

Hi,



I have the same concern with the activation of non-GFS file systems and I agree with the idea of using exclusive activation (keeping in mind that the strange behavior of exclusive tag being bypassed should be fixed). However, from my point of view, the best way is not to create a new resource but adding a new "exclusive" option to the regular fs resource or something like that, as I think it's the desired behavior most times the fs resource is used.



Best regards,



Alfredo







________________________________

From: linux-cluster-bounces at redhat.com<mailto:linux-cluster-bounces at redhat.com> [mailto:linux-cluster-bounces at redhat.com<mailto:linux-cluster-bounces at redhat.com>] On Behalf Of brem belguebli
Sent: Saturday, August 01, 2009 2:25 AM
To: linux clustering
Subject: Re: [Linux-cluster] CLVMD without GFS



Hi Rafael,

Of course, that is what I was telling you the other day, your script works fine, but I just wanted to twist a little clvm.

The prerequisites are already setup (locking type, dm-mp, lvm, etc...).

The only thing that disturbes me is that you can bypass under certain conditions the locking.

Thanks for the help

Brem



2009/8/1 Rafael Mic? Miranda <rmicmirregs at gmail.com<mailto:rmicmirregs at gmail.com>>

Hi Brem,

El vie, 31-07-2009 a las 06:09 -0400, crosa at redhat.com<mailto:crosa at redhat.com> escribi?:

>
> --- mensagem original ---
> De: brem belguebli <brem.belguebli at gmail.com<mailto:brem.belguebli at gmail.com>>
> Assunto: Re: [Linux-cluster] CLVMD without GFS
> Data: 29 de Julho de 2009
> Hora: 10:1:29
>
> Hi Rafael,
>
> Just posted the basic tests I'm doing on both linux-cluster and linux-lvm.
>
> I can't get exclusive activation to work properly, I may be missing some
> step in my process.
>
> Brem
>
>

>

> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com<mailto:Linux-cluster at redhat.com>
> https://www.redhat.com/mailman/listinfo/linux-cluster

Sorry, i'm not sure if I missed this mail.

Maybe i need to explain a couple of things first, just to make them
clear.

As part of the use of the lvm-cluster.sh resource script, you need to:

1.- Configure properly the lock type into lvm.conf to type 3 (example
into this link:

https://www.redhat.com/archives/linux-cluster/2009-July/msg00253.html

)

2.- Configure your multipathing software

3.- Create your Physical Volumes

4.- Configure your Volume Groups as clustered Volume Groups

5.- Create your Logical Volumens into the clustered Volume Groups

6.- And, and the not so obvious, de-activate all the Logical Volumes you
plan to use as exclusive Logical Volumes (using exclusive flag).

De-activation can be done with "vgchange -an volgrp01/logvol01" or
similar command.

If you don't do step 6, you will receive an error message when executing
"lvchange -aey volgrp01/logvol01". This command is executed (with the
proper volume group and logical volume names) internally in the resource
script.

I designed the lvm-cluster.sh resource script to be verbose, maybe you
can copy here your logs (default to /var/log/messages on RHEL systems).

Cheers and thanks,

Rafael

--
Rafael Mic? Miranda

--
Linux-cluster mailing list
Linux-cluster at redhat.com<mailto:Linux-cluster at redhat.com>
https://www.redhat.com/mailman/listinfo/linux-cluster



--
Linux-cluster mailing list
Linux-cluster at redhat.com<mailto:Linux-cluster at redhat.com>
https://www.redhat.com/mailman/listinfo/linux-cluster

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090818/58a95037/attachment.htm>

From brem.belguebli at gmail.com  Tue Aug 18 09:35:46 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Tue, 18 Aug 2009 11:35:46 +0200
Subject: [Linux-cluster] CLVMD without GFS
In-Reply-To: <FD716ADFEE797543AA8B2AB0B3F9148E01A09C4700@rkamsem701.emea.roche.com>
References: <MJ2mY0lYjZGc@NBjvyXzB> <1249079337.6494.10.camel@mecatol>
	<29ae894c0907311725k4759e631m332ffd38857a6f99@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C46B9@rkamsem701.emea.roche.com>
	<29ae894c0908180209i46e23ed1o8fc2d372c8371e12@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C4700@rkamsem701.emea.roche.com>
Message-ID: <29ae894c0908180235g756ce310nc678a922ce5f718e@mail.gmail.com>

Alfredo,

The resource Rafael created operates at LV level not FS.

As far as I understand, a FS exclusive tag would be efficient only under
rgmanager control, but it wouldn't if an admin decided to mount the FS by
hand on another cluster node.

Brem


2009/8/18, Moralejo, Alfredo <alfredo.moralejo at roche.com>:
>
>  Hi Brem,
>
>
>
> Maybe I did not explain it properly.
>
>
>
> What I mean is that fs resource should activate the filesystems in *
> exclusive* mode as the resource created by Rafael does (lvm-cluster.sh) as
> I far as I understand to ensure the filesystem can not be mounted in other
> nodes accidentally.
>
>
>
> Best regards,
>
>
>
> Alfredo
>
>
>
>
>  ------------------------------
>
> *From:* linux-cluster-bounces at redhat.com [mailto:
> linux-cluster-bounces at redhat.com] *On Behalf Of *brem belguebli
> *Sent:* Tuesday, August 18, 2009 11:10 AM
> *To:* linux clustering
> *Subject:* Re: [Linux-cluster] CLVMD without GFS
>
>
>
> Hello Alfredo,
>
>
>
> For the exclusive bypass, I have opened a bugzilla ticket. Hopefully it'll
> be fixed (enhanced)0
>
>
>
> For your second point, I'm not sure to ubndestand what you're saying.
>
>
>
> My concern is to avoid manual activation on the wrong node, not an
> accidental rgmanager activation.
>
>
>
> Even if you configure your FS resource as exclusive, if it is possible,
> this won't be of any effect with ext3 outside of rgmanager control as it is
> not cluster aware.
>
> Brem
>
> 2009/8/18, Moralejo, Alfredo <alfredo.moralejo at roche.com>:
>
> Hi,
>
>
>
> I have the same concern with the activation of non-GFS file systems and I
> agree with the idea of using exclusive activation (keeping in mind that the
> strange behavior of exclusive tag being bypassed should be fixed). However,
> from my point of view, the best way is not to create a new resource but
> adding a new ?exclusive? option to the regular fs resource or something like
> that, as I think it?s the desired behavior most times the fs resource is
> used.
>
>
>
> Best regards,
>
>
>
> Alfredo
>
>
>
>
>
>
>  ------------------------------
>
> *From:* linux-cluster-bounces at redhat.com [mailto:
> linux-cluster-bounces at redhat.com] *On Behalf Of *brem belguebli
> *Sent:* Saturday, August 01, 2009 2:25 AM
> *To:* linux clustering
> *Subject:* Re: [Linux-cluster] CLVMD without GFS
>
>
>
> Hi Rafael,
>
> Of course, that is what I was telling you the other day, your script works
> fine, but I just wanted to twist a little clvm.
>
> The prerequisites are already setup (locking type, dm-mp, lvm, etc...).
>
> The only thing that disturbes me is that you can bypass under certain
> conditions the locking.
>
> Thanks for the help
>
> Brem
>
>
>
> 2009/8/1 Rafael Mic? Miranda <rmicmirregs at gmail.com>
>
> Hi Brem,
>
> El vie, 31-07-2009 a las 06:09 -0400, crosa at redhat.com escribi?:
>
> >
> > --- mensagem original ---
> > De: brem belguebli <brem.belguebli at gmail.com>
> > Assunto: Re: [Linux-cluster] CLVMD without GFS
> > Data: 29 de Julho de 2009
> > Hora: 10:1:29
> >
> > Hi Rafael,
> >
> > Just posted the basic tests I'm doing on both linux-cluster and
> linux-lvm.
> >
> > I can't get exclusive activation to work properly, I may be missing some
> > step in my process.
> >
> > Brem
> >
> >
>
> >
>
> > --
> > Linux-cluster mailing list
> > Linux-cluster at redhat.com
> > https://www.redhat.com/mailman/listinfo/linux-cluster
>
> Sorry, i'm not sure if I missed this mail.
>
> Maybe i need to explain a couple of things first, just to make them
> clear.
>
> As part of the use of the lvm-cluster.sh resource script, you need to:
>
> 1.- Configure properly the lock type into lvm.conf to type 3 (example
> into this link:
>
> https://www.redhat.com/archives/linux-cluster/2009-July/msg00253.html
>
> )
>
> 2.- Configure your multipathing software
>
> 3.- Create your Physical Volumes
>
> 4.- Configure your Volume Groups as clustered Volume Groups
>
> 5.- Create your Logical Volumens into the clustered Volume Groups
>
> 6.- And, and the not so obvious, de-activate all the Logical Volumes you
> plan to use as exclusive Logical Volumes (using exclusive flag).
>
> De-activation can be done with "vgchange -an volgrp01/logvol01" or
> similar command.
>
> If you don't do step 6, you will receive an error message when executing
> "lvchange -aey volgrp01/logvol01". This command is executed (with the
> proper volume group and logical volume names) internally in the resource
> script.
>
> I designed the lvm-cluster.sh resource script to be verbose, maybe you
> can copy here your logs (default to /var/log/messages on RHEL systems).
>
> Cheers and thanks,
>
> Rafael
>
> --
> Rafael Mic? Miranda
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
>
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090818/5fd8ffe7/attachment.htm>

From harri.paivaniemi at tieto.com  Tue Aug 18 10:07:08 2009
From: harri.paivaniemi at tieto.com (=?iso-8859-1?q?H=2EP=E4iv=E4niemi?=)
Date: Tue, 18 Aug 2009 13:07:08 +0300
Subject: [Linux-cluster] About adding a new parameter to failover domain
In-Reply-To: <FD716ADFEE797543AA8B2AB0B3F9148E01A09C46B9@rkamsem701.emea.roche.com>
References: <MJ2mY0lYjZGc@NBjvyXzB>
	<29ae894c0907311725k4759e631m332ffd38857a6f99@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C46B9@rkamsem701.emea.roche.com>
Message-ID: <200908181307.08257.harri.paivaniemi@tieto.com>

Hi all,

I noticed that there is a new parameter "nofailback" that can be added to failover domain like this:

<failoverdomain name="FOOFD" ordered="1" restricted="0" nofailback="1">

Do you know if this is in use right after propagation of the cluster.conf or should I restart the whole cluster 
to activate that?

Thanks,

-hjp



From esggrupos at gmail.com  Tue Aug 18 12:35:25 2009
From: esggrupos at gmail.com (ESGLinux)
Date: Tue, 18 Aug 2009 14:35:25 +0200
Subject: [Linux-cluster] do I have a fence DRAC device?
In-Reply-To: <8b711df40908171128j1fc18525nfd7df01d7604cda0@mail.gmail.com>
References: <3128ba140908100324l6cdb4c34ra5f5edb39c6903e9@mail.gmail.com>
	<8b711df40908101134t69b8e12cof6cc551809421e45@mail.gmail.com>
	<3128ba140908170350ge619930w5c17368ff0d3cf42@mail.gmail.com>
	<8b711df40908171128j1fc18525nfd7df01d7604cda0@mail.gmail.com>
Message-ID: <3128ba140908180535o4f62b011vc41e5ec6517ac388@mail.gmail.com>

>
>
>
> When you press the power button, it comes as "Remote Access Card".
> Then after few other screens you can login to enable and configure
> DRAC.
>
> racadm did't work for me  as well. But you do not need it if you
> configure it boot time.
>
>
> Paras.
>

Thanks Paras,

I?ll try when they let me to reboot the system again :-(

Grettings

ESG
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090818/9e006b73/attachment.htm>

From alfredo.moralejo at roche.com  Tue Aug 18 13:19:19 2009
From: alfredo.moralejo at roche.com (Moralejo, Alfredo)
Date: Tue, 18 Aug 2009 15:19:19 +0200
Subject: [Linux-cluster] CLVMD without GFS
In-Reply-To: <29ae894c0908180235g756ce310nc678a922ce5f718e@mail.gmail.com>
References: <MJ2mY0lYjZGc@NBjvyXzB> <1249079337.6494.10.camel@mecatol>
	<29ae894c0907311725k4759e631m332ffd38857a6f99@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C46B9@rkamsem701.emea.roche.com>
	<29ae894c0908180209i46e23ed1o8fc2d372c8371e12@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C4700@rkamsem701.emea.roche.com>
	<29ae894c0908180235g756ce310nc678a922ce5f718e@mail.gmail.com>
Message-ID: <FD716ADFEE797543AA8B2AB0B3F9148E01A09C4841@rkamsem701.emea.roche.com>

Hi,

Yes, I understand that.

However, what I meant is that, from a configuration simplicity point of view, it could be convenient to add the functionality into a only resource that take care of the entire cycle, logical volume activation in exclusive mode and mounting it. I think this is the most common usage of the fs resource. In this way, I don't need to create two resources for each filesystem (one for the logical volume and one other for the fs) and I can be sure a filesystem can not be mounted by hand in another node (as it has been activate exclusively in the node running the service).

Best regards,

Alfredo


________________________________
From: linux-cluster-bounces at redhat.com [mailto:linux-cluster-bounces at redhat.com] On Behalf Of brem belguebli
Sent: Tuesday, August 18, 2009 11:36 AM
To: linux clustering
Subject: Re: [Linux-cluster] CLVMD without GFS

Alfredo,

The resource Rafael created operates at LV level not FS.

As far as I understand, a FS exclusive tag would be efficient only under rgmanager control, but it wouldn't if an admin decided to mount the FS by hand on another cluster node.

Brem


2009/8/18, Moralejo, Alfredo <alfredo.moralejo at roche.com<mailto:alfredo.moralejo at roche.com>>:

Hi Brem,



Maybe I did not explain it properly.



What I mean is that fs resource should activate the filesystems in exclusive mode as the resource created by Rafael does (lvm-cluster.sh) as I far as I understand to ensure the filesystem can not be mounted in other nodes accidentally.



Best regards,



Alfredo





________________________________

From: linux-cluster-bounces at redhat.com<mailto:linux-cluster-bounces at redhat.com> [mailto:linux-cluster-bounces at redhat.com<mailto:linux-cluster-bounces at redhat.com>] On Behalf Of brem belguebli
Sent: Tuesday, August 18, 2009 11:10 AM

To: linux clustering
Subject: Re: [Linux-cluster] CLVMD without GFS



Hello Alfredo,



For the exclusive bypass, I have opened a bugzilla ticket. Hopefully it'll be fixed (enhanced)0



For your second point, I'm not sure to ubndestand what you're saying.



My concern is to avoid manual activation on the wrong node, not an accidental rgmanager activation.



Even if you configure your FS resource as exclusive, if it is possible, this won't be of any effect with ext3 outside of rgmanager control as it is not cluster aware.

Brem

2009/8/18, Moralejo, Alfredo <alfredo.moralejo at roche.com<mailto:alfredo.moralejo at roche.com>>:

Hi,



I have the same concern with the activation of non-GFS file systems and I agree with the idea of using exclusive activation (keeping in mind that the strange behavior of exclusive tag being bypassed should be fixed). However, from my point of view, the best way is not to create a new resource but adding a new "exclusive" option to the regular fs resource or something like that, as I think it's the desired behavior most times the fs resource is used.



Best regards,



Alfredo







________________________________

From: linux-cluster-bounces at redhat.com<mailto:linux-cluster-bounces at redhat.com> [mailto:linux-cluster-bounces at redhat.com<mailto:linux-cluster-bounces at redhat.com>] On Behalf Of brem belguebli
Sent: Saturday, August 01, 2009 2:25 AM
To: linux clustering
Subject: Re: [Linux-cluster] CLVMD without GFS



Hi Rafael,

Of course, that is what I was telling you the other day, your script works fine, but I just wanted to twist a little clvm.

The prerequisites are already setup (locking type, dm-mp, lvm, etc...).

The only thing that disturbes me is that you can bypass under certain conditions the locking.

Thanks for the help

Brem



2009/8/1 Rafael Mic? Miranda <rmicmirregs at gmail.com<mailto:rmicmirregs at gmail.com>>

Hi Brem,

El vie, 31-07-2009 a las 06:09 -0400, crosa at redhat.com<mailto:crosa at redhat.com> escribi?:

>
> --- mensagem original ---
> De: brem belguebli <brem.belguebli at gmail.com<mailto:brem.belguebli at gmail.com>>
> Assunto: Re: [Linux-cluster] CLVMD without GFS
> Data: 29 de Julho de 2009
> Hora: 10:1:29
>
> Hi Rafael,
>
> Just posted the basic tests I'm doing on both linux-cluster and linux-lvm.
>
> I can't get exclusive activation to work properly, I may be missing some
> step in my process.
>
> Brem
>
>

>

> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com<mailto:Linux-cluster at redhat.com>
> https://www.redhat.com/mailman/listinfo/linux-cluster

Sorry, i'm not sure if I missed this mail.

Maybe i need to explain a couple of things first, just to make them
clear.

As part of the use of the lvm-cluster.sh resource script, you need to:

1.- Configure properly the lock type into lvm.conf to type 3 (example
into this link:

https://www.redhat.com/archives/linux-cluster/2009-July/msg00253.html

)

2.- Configure your multipathing software

3.- Create your Physical Volumes

4.- Configure your Volume Groups as clustered Volume Groups

5.- Create your Logical Volumens into the clustered Volume Groups

6.- And, and the not so obvious, de-activate all the Logical Volumes you
plan to use as exclusive Logical Volumes (using exclusive flag).

De-activation can be done with "vgchange -an volgrp01/logvol01" or
similar command.

If you don't do step 6, you will receive an error message when executing
"lvchange -aey volgrp01/logvol01". This command is executed (with the
proper volume group and logical volume names) internally in the resource
script.

I designed the lvm-cluster.sh resource script to be verbose, maybe you
can copy here your logs (default to /var/log/messages on RHEL systems).

Cheers and thanks,

Rafael

--
Rafael Mic? Miranda

--
Linux-cluster mailing list
Linux-cluster at redhat.com<mailto:Linux-cluster at redhat.com>
https://www.redhat.com/mailman/listinfo/linux-cluster



--
Linux-cluster mailing list
Linux-cluster at redhat.com<mailto:Linux-cluster at redhat.com>
https://www.redhat.com/mailman/listinfo/linux-cluster



--
Linux-cluster mailing list
Linux-cluster at redhat.com<mailto:Linux-cluster at redhat.com>
https://www.redhat.com/mailman/listinfo/linux-cluster

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090818/d7cdf3dc/attachment.htm>

From alfredo.moralejo at roche.com  Tue Aug 18 13:26:20 2009
From: alfredo.moralejo at roche.com (Moralejo, Alfredo)
Date: Tue, 18 Aug 2009 15:26:20 +0200
Subject: [Linux-cluster] Cluster behavior
In-Reply-To: <8b711df40908171459q1afecda7k7b0dcc4773153b84@mail.gmail.com>
References: <8b711df40908171459q1afecda7k7b0dcc4773153b84@mail.gmail.com>
Message-ID: <FD716ADFEE797543AA8B2AB0B3F9148E01A09C484B@rkamsem701.emea.roche.com>

Could you send logs in messages file including the part where the cluster dissolved messages is logged?

Regards,

Alfredo 

-----Original Message-----
From: linux-cluster-bounces at redhat.com [mailto:linux-cluster-bounces at redhat.com] On Behalf Of Paras pradhan
Sent: Monday, August 17, 2009 11:59 PM
To: linux clustering
Subject: [Linux-cluster] Cluster behavior

I have a 3 nodes cluster.

Node A - Vote 1

Node B - Vote 1

Node C - Votes 2

Qdisk - Votes 3


Altogether the cluster has 7 votes. The required min quorum to run the
cluster would be in this case 4. Now if I poweroff node 1, I can see
Quorum Dissolved in the terminals of Node 2 and Node3 . This cluster
has xen virtual machines.

Whats wrong and how to do debug the problem?

Thanks
Paras.

--
Linux-cluster mailing list
Linux-cluster at redhat.com
https://www.redhat.com/mailman/listinfo/linux-cluster



From Harri.Paivaniemi at tieto.com  Tue Aug 18 13:51:09 2009
From: Harri.Paivaniemi at tieto.com (Harri.Paivaniemi at tieto.com)
Date: Tue, 18 Aug 2009 16:51:09 +0300
Subject: [Linux-cluster] Cluster behavior
In-Reply-To: <FD716ADFEE797543AA8B2AB0B3F9148E01A09C484B@rkamsem701.emea.roche.com>
References: <8b711df40908171459q1afecda7k7b0dcc4773153b84@mail.gmail.com>,
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C484B@rkamsem701.emea.roche.com>
Message-ID: <F5A4C3A7AC59A74FB0A0D35F1E8BDB460352A205E1@EXMB03.eu.tieto.com>

Are you sure,

that your qdisk-configuration is ok and it's working properly? Just based on maths, it seems that you have only 4 votes up in your cluster and it goes down if it loses one vote...

So is cman_tool really showing exp.votes=7? And qdisk brings it's votes ok?

-hjp

________________________________________
From: linux-cluster-bounces at redhat.com [linux-cluster-bounces at redhat.com] On Behalf Of Moralejo, Alfredo [alfredo.moralejo at roche.com]
Sent: Tuesday, August 18, 2009 4:26 PM
To: linux clustering
Subject: RE: [Linux-cluster] Cluster behavior

Could you send logs in messages file including the part where the cluster dissolved messages is logged?

Regards,

Alfredo

-----Original Message-----
From: linux-cluster-bounces at redhat.com [mailto:linux-cluster-bounces at redhat.com] On Behalf Of Paras pradhan
Sent: Monday, August 17, 2009 11:59 PM
To: linux clustering
Subject: [Linux-cluster] Cluster behavior

I have a 3 nodes cluster.

Node A - Vote 1

Node B - Vote 1

Node C - Votes 2

Qdisk - Votes 3


Altogether the cluster has 7 votes. The required min quorum to run the
cluster would be in this case 4. Now if I poweroff node 1, I can see
Quorum Dissolved in the terminals of Node 2 and Node3 . This cluster
has xen virtual machines.

Whats wrong and how to do debug the problem?

Thanks
Paras.

--
Linux-cluster mailing list
Linux-cluster at redhat.com
https://www.redhat.com/mailman/listinfo/linux-cluster

--
Linux-cluster mailing list
Linux-cluster at redhat.com
https://www.redhat.com/mailman/listinfo/linux-cluster



From brem.belguebli at gmail.com  Tue Aug 18 14:19:28 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Tue, 18 Aug 2009 16:19:28 +0200
Subject: [Linux-cluster] CLVMD without GFS
In-Reply-To: <FD716ADFEE797543AA8B2AB0B3F9148E01A09C4841@rkamsem701.emea.roche.com>
References: <MJ2mY0lYjZGc@NBjvyXzB> <1249079337.6494.10.camel@mecatol>
	<29ae894c0907311725k4759e631m332ffd38857a6f99@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C46B9@rkamsem701.emea.roche.com>
	<29ae894c0908180209i46e23ed1o8fc2d372c8371e12@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C4700@rkamsem701.emea.roche.com>
	<29ae894c0908180235g756ce310nc678a922ce5f718e@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C4841@rkamsem701.emea.roche.com>
Message-ID: <29ae894c0908180719tf00e87di485d691fd8d84679@mail.gmail.com>

Hi,

I see what you mean.

On the other hand, it would may be lessen some flexibility. The typical case
being the fact that you do not want to use FS on top of your LV's but raw
devices.

Brem


2009/8/18, Moralejo, Alfredo <alfredo.moralejo at roche.com>:
>
>  Hi,
>
>
>
> Yes, I understand that.
>
>
>
> However, what I meant is that, from a configuration simplicity point of
> view, it could be convenient to add the functionality into a only resource
> that take care of the entire cycle, logical volume activation in exclusive
> mode and mounting it. I think this is the most common usage of the fs
> resource. In this way, I don?t need to create two resources for each
> filesystem (one for the logical volume and one other for the fs) and I can
> be sure a filesystem can not be mounted by hand in another node (as it has
> been activate exclusively in the node running the service).
>
>
>
> Best regards,
>
>
>
> Alfredo
>
>
>
>
>  ------------------------------
>
> *From:* linux-cluster-bounces at redhat.com [mailto:
> linux-cluster-bounces at redhat.com] *On Behalf Of *brem belguebli
> *Sent:* Tuesday, August 18, 2009 11:36 AM
> *To:* linux clustering
> *Subject:* Re: [Linux-cluster] CLVMD without GFS
>
>
>
> Alfredo,
>
>
>
> The resource Rafael created operates at LV level not FS.
>
>
>
> As far as I understand, a FS exclusive tag would be efficient only under
> rgmanager control, but it wouldn't if an admin decided to mount the FS by
> hand on another cluster node.
>
>
>
> Brem
>
>
>
> 2009/8/18, Moralejo, Alfredo <alfredo.moralejo at roche.com>:
>
> Hi Brem,
>
>
>
> Maybe I did not explain it properly.
>
>
>
> What I mean is that fs resource should activate the filesystems in *
> exclusive* mode as the resource created by Rafael does (lvm-cluster.sh) as
> I far as I understand to ensure the filesystem can not be mounted in other
> nodes accidentally.
>
>
>
> Best regards,
>
>
>
> Alfredo
>
>
>
>
>  ------------------------------
>
> *From:* linux-cluster-bounces at redhat.com [mailto:
> linux-cluster-bounces at redhat.com] *On Behalf Of *brem belguebli
> *Sent:* Tuesday, August 18, 2009 11:10 AM
>
>
> *To:* linux clustering
> *Subject:* Re: [Linux-cluster] CLVMD without GFS
>
>
>
> Hello Alfredo,
>
>
>
> For the exclusive bypass, I have opened a bugzilla ticket. Hopefully it'll
> be fixed (enhanced)0
>
>
>
> For your second point, I'm not sure to ubndestand what you're saying.
>
>
>
> My concern is to avoid manual activation on the wrong node, not an
> accidental rgmanager activation.
>
>
>
> Even if you configure your FS resource as exclusive, if it is possible,
> this won't be of any effect with ext3 outside of rgmanager control as it is
> not cluster aware.
>
> Brem
>
> 2009/8/18, Moralejo, Alfredo <alfredo.moralejo at roche.com>:
>
> Hi,
>
>
>
> I have the same concern with the activation of non-GFS file systems and I
> agree with the idea of using exclusive activation (keeping in mind that the
> strange behavior of exclusive tag being bypassed should be fixed). However,
> from my point of view, the best way is not to create a new resource but
> adding a new ?exclusive? option to the regular fs resource or something like
> that, as I think it?s the desired behavior most times the fs resource is
> used.
>
>
>
> Best regards,
>
>
>
> Alfredo
>
>
>
>
>
>
>  ------------------------------
>
> *From:* linux-cluster-bounces at redhat.com [mailto:
> linux-cluster-bounces at redhat.com] *On Behalf Of *brem belguebli
> *Sent:* Saturday, August 01, 2009 2:25 AM
> *To:* linux clustering
> *Subject:* Re: [Linux-cluster] CLVMD without GFS
>
>
>
> Hi Rafael,
>
> Of course, that is what I was telling you the other day, your script works
> fine, but I just wanted to twist a little clvm.
>
> The prerequisites are already setup (locking type, dm-mp, lvm, etc...).
>
> The only thing that disturbes me is that you can bypass under certain
> conditions the locking.
>
> Thanks for the help
>
> Brem
>
>
>
> 2009/8/1 Rafael Mic? Miranda <rmicmirregs at gmail.com>
>
> Hi Brem,
>
> El vie, 31-07-2009 a las 06:09 -0400, crosa at redhat.com escribi?:
>
> >
> > --- mensagem original ---
> > De: brem belguebli <brem.belguebli at gmail.com>
> > Assunto: Re: [Linux-cluster] CLVMD without GFS
> > Data: 29 de Julho de 2009
> > Hora: 10:1:29
> >
> > Hi Rafael,
> >
> > Just posted the basic tests I'm doing on both linux-cluster and
> linux-lvm.
> >
> > I can't get exclusive activation to work properly, I may be missing some
> > step in my process.
> >
> > Brem
> >
> >
>
> >
>
> > --
> > Linux-cluster mailing list
> > Linux-cluster at redhat.com
> > https://www.redhat.com/mailman/listinfo/linux-cluster
>
> Sorry, i'm not sure if I missed this mail.
>
> Maybe i need to explain a couple of things first, just to make them
> clear.
>
> As part of the use of the lvm-cluster.sh resource script, you need to:
>
> 1.- Configure properly the lock type into lvm.conf to type 3 (example
> into this link:
>
> https://www.redhat.com/archives/linux-cluster/2009-July/msg00253.html
>
> )
>
> 2.- Configure your multipathing software
>
> 3.- Create your Physical Volumes
>
> 4.- Configure your Volume Groups as clustered Volume Groups
>
> 5.- Create your Logical Volumens into the clustered Volume Groups
>
> 6.- And, and the not so obvious, de-activate all the Logical Volumes you
> plan to use as exclusive Logical Volumes (using exclusive flag).
>
> De-activation can be done with "vgchange -an volgrp01/logvol01" or
> similar command.
>
> If you don't do step 6, you will receive an error message when executing
> "lvchange -aey volgrp01/logvol01". This command is executed (with the
> proper volume group and logical volume names) internally in the resource
> script.
>
> I designed the lvm-cluster.sh resource script to be verbose, maybe you
> can copy here your logs (default to /var/log/messages on RHEL systems).
>
> Cheers and thanks,
>
> Rafael
>
> --
> Rafael Mic? Miranda
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
>
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
>
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090818/1c45d2fa/attachment.htm>

From teigland at redhat.com  Tue Aug 18 14:14:42 2009
From: teigland at redhat.com (David Teigland)
Date: Tue, 18 Aug 2009 09:14:42 -0500
Subject: [Linux-cluster] gfs2 mount hangs
In-Reply-To: <4A8A12D6.5050907@oracle.com>
References: <4A83B113.60609@oracle.com> <20090813141424.GA8148@redhat.com>
	<4A84CC39.3080809@oracle.com> <20090814145738.GA7094@redhat.com>
	<4A8A12D6.5050907@oracle.com>
Message-ID: <20090818141442.GA6303@redhat.com>

On Tue, Aug 18, 2009 at 10:32:54AM +0800, Wengang Wang wrote:
> there is a line with "...error -1 2", but seems the line is out during 
> the second(hang) mount.
> 
> 2) for node desk(the first mount is on this node).
> [root at desk ~]# gfs_control dump
> 1250589068 logging mode 3 syslog f 160 p 6 logfile p 6 
> /var/log/cluster/gfs_controld.log
> 1250589068 gfs_controld 3.0.0 started
> 1250589068 logging mode 3 syslog f 160 p 6 logfile p 6 
> /var/log/cluster/gfs_controld.log
> 1250589068 group_mode 3 compat 0
> 1250589068 setup_cpg 13
> 1250589068 run protocol from nodeid 1
> 1250589068 daemon run 1.1.1 max 1.1.1 kernel run 1.1.1 max 1.1.1
> 1250590628 uevent add gfs2 /fs/gfs2/testgfs2:1
> 1250590628 uevent change gfs2 /fs/gfs2/testgfs2:1
> 1250590628 recovery_uevent mg not found 1
> 1250590628 uevent change gfs2 /fs/gfs2/testgfs2:1
> 1250590628 recovery_uevent mg not found 1
> 1250590628 uevent change gfs2 /fs/gfs2/testgfs2:1
> 1250590628 recovery_uevent mg not found 1
> 1250590628 uevent online gfs2 /fs/gfs2/testgfs2:1

That's the problem, there are a bunch of steps missing that come from
mount.gfs2.  My guess is that you're missing /sbin/mount.gfs2 on this node.

It also seems that gfs2 no longer properly handles this error condition
(missing/empty hostdata string).  In the past you would have gotten a nice
message in /var/log/messages explaining that you were missing mount.gfs2.

Dave



From pradhanparas at gmail.com  Tue Aug 18 16:22:34 2009
From: pradhanparas at gmail.com (Paras pradhan)
Date: Tue, 18 Aug 2009 11:22:34 -0500
Subject: [Linux-cluster] Cluster behavior
In-Reply-To: <FD716ADFEE797543AA8B2AB0B3F9148E01A09C484B@rkamsem701.emea.roche.com>
References: <8b711df40908171459q1afecda7k7b0dcc4773153b84@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C484B@rkamsem701.emea.roche.com>
Message-ID: <8b711df40908180922u5da38a5bw206deeb94729896b@mail.gmail.com>

On Tue, Aug 18, 2009 at 8:26 AM, Moralejo,
Alfredo<alfredo.moralejo at roche.com> wrote:
> Could you send logs in messages file including the part where the cluster dissolved messages is logged?

Ok.. here are the logs . I shutdown node 3 which have 2 votes.  Below
are the logs from node1 and node 2.

X.X.X.165 is node 1
X.X.X.172 is node 2
X.X.X.173 is node 3

Log from node1:

ug 18 10:35:28 cvtst1 ntpd[3862]: synchronized to X.X.X.103, stratum 2
Aug 18 11:01:44 cvtst1 clurgmgrd[4309]: <notice> Member 3 shutting down
Aug 18 11:01:57 cvtst1 kernel: peth0: received packet with  own
address as source address
Aug 18 11:01:57 cvtst1 qdiskd[3361]: <info> Node 3 shutdown
Aug 18 11:02:02 cvtst1 kernel: peth0: received packet with  own
address as source address
Aug 18 11:02:07 cvtst1 openais[3318]: [TOTEM] The token was lost in
the OPERATIONAL state.
Aug 18 11:02:07 cvtst1 openais[3318]: [TOTEM] Receive multicast socket
recv buffer size (288000 bytes).
Aug 18 11:02:07 cvtst1 openais[3318]: [TOTEM] Transmit multicast
socket send buffer size (262142 bytes).
Aug 18 11:02:07 cvtst1 openais[3318]: [TOTEM] entering GATHER state from 2.
Aug 18 11:02:12 cvtst1 openais[3318]: [TOTEM] entering GATHER state from 11.
Aug 18 11:02:12 cvtst1 openais[3318]: [TOTEM] Creating commit token
because I am the rep.
Aug 18 11:02:12 cvtst1 openais[3318]: [TOTEM] Saving state aru 9f high
seq received 9f
Aug 18 11:02:12 cvtst1 openais[3318]: [TOTEM] Storing new sequence id
for ring 758
Aug 18 11:02:12 cvtst1 openais[3318]: [TOTEM] entering COMMIT state.
Aug 18 11:02:12 cvtst1 openais[3318]: [TOTEM] entering RECOVERY state.
Aug 18 11:02:12 cvtst1 openais[3318]: [TOTEM] position [0] member X.X.X.165:
Aug 18 11:02:12 cvtst1 openais[3318]: [TOTEM] previous ring seq 1876
rep X.X.X.165
Aug 18 11:02:12 cvtst1 openais[3318]: [TOTEM] aru 9f high delivered 9f
received flag 1
Aug 18 11:02:12 cvtst1 openais[3318]: [TOTEM] position [1] member X.X.X.172:
Aug 18 11:02:12 cvtst1 openais[3318]: [TOTEM] previous ring seq 1876
rep X.X.X.165
Aug 18 11:02:12 cvtst1 openais[3318]: [TOTEM] aru 9f high delivered 9f
received flag 1
Aug 18 11:02:12 cvtst1 openais[3318]: [TOTEM] Did not need to
originate any messages in recovery.
Aug 18 11:02:12 cvtst1 openais[3318]: [TOTEM] Sending initial ORF token
Aug 18 11:02:12 cvtst1 kernel: dlm: closing connection to node 3
Aug 18 11:02:12 cvtst1 openais[3318]: [CLM  ] CLM CONFIGURATION CHANGE
Aug 18 11:02:12 cvtst1 openais[3318]: [CLM  ] New Configuration:
Aug 18 11:02:12 cvtst1 openais[3318]: [CLM  ] 	r(0) ip(X.X.X.165)
Aug 18 11:02:12 cvtst1 openais[3318]: [CLM  ] 	r(0) ip(X.X.X.172)
Aug 18 11:02:12 cvtst1 openais[3318]: [CLM  ] Members Left:
Aug 18 11:02:12 cvtst1 openais[3318]: [CLM  ] 	r(0) ip(X.X.X.173)
Aug 18 11:02:12 cvtst1 clurgmgrd[4309]: <emerg> #1: Quorum Dissolved
Aug 18 11:02:12 cvtst1 openais[3318]: [CLM  ] Members Joined:
Aug 18 11:02:12 cvtst1 openais[3318]: [CLM  ] CLM CONFIGURATION CHANGE
Aug 18 11:02:12 cvtst1 openais[3318]: [CLM  ] New Configuration:
Aug 18 11:02:12 cvtst1 openais[3318]: [CLM  ] 	r(0) ip(X.X.X.165)
Aug 18 11:02:12 cvtst1 openais[3318]: [CLM  ] 	r(0) ip(X.X.X.172)
Aug 18 11:02:12 cvtst1 openais[3318]: [CLM  ] Members Left:
Aug 18 11:02:12 cvtst1 openais[3318]: [CLM  ] Members Joined:
Aug 18 11:02:12 cvtst1 openais[3318]: [SYNC ] This node is within the
primary component and will provide service.
Aug 18 11:02:12 cvtst1 openais[3318]: [TOTEM] entering OPERATIONAL state.
Aug 18 11:02:12 cvtst1 openais[3318]: [CLM  ] got nodejoin message X.X.X.165
Aug 18 11:02:12 cvtst1 openais[3318]: [CLM  ] got nodejoin message X.X.X.172
Aug 18 11:02:13 cvtst1 openais[3318]: [CPG  ] got joinlist message from node 2
Aug 18 11:02:13 cvtst1 openais[3318]: [CPG  ] got joinlist message from node 1
Aug 18 11:02:13 cvtst1 openais[3318]: [CMAN ] lost contact with quorum device
Aug 18 11:02:13 cvtst1 openais[3318]: [CMAN ] quorum lost, blocking activity
Aug 18 11:02:13 cvtst1 ccsd[3274]: Cluster is not quorate.  Refusing
connection.
Aug 18 11:02:13 cvtst1 ccsd[3274]: Error while processing connect:
Connection refused
Aug 18 11:02:22 cvtst1 ccsd[3274]: Cluster is not quorate.  Refusing
connection.
Aug 18 11:02:22 cvtst1 ccsd[3274]: Error while processing connect:
Connection refused
Aug 18 11:02:27 cvtst1 qdiskd[3361]: <info> Node 1 is the master
Aug 18 11:02:30 cvtst1 openais[3318]: [CMAN ] quorum regained,
resuming activity
Aug 18 11:02:36 cvtst1 kernel: xenbr0: port 3(vif1.0) entering disabled state
Aug 18 11:02:36 cvtst1 kernel: device vif1.0 left promiscuous mode
Aug 18 11:02:36 cvtst1 kernel: xenbr0: port 3(vif1.0) entering disabled state
Aug 18 11:02:39 cvtst1 clurgmgrd[4309]: <notice> Quorum Regained
Aug 18 11:02:41 cvtst1 clurgmgrd[4309]: <notice> Starting stopped
service vm:guest1
Aug 18 11:02:42 cvtst1 kernel: tap tap-2-51712: 2 getting info
Aug 18 11:02:43 cvtst1 kernel: device vif2.0 entered promiscuous mode
Aug 18 11:02:43 cvtst1 kernel: ADDRCONF(NETDEV_UP): vif2.0: link is not ready
Aug 18 11:02:43 cvtst1 clurgmgrd[4309]: <notice> Service vm:guest1 started
Aug 18 11:02:47 cvtst1 kernel: blktap: ring-ref 8, event-channel 6,
protocol 1 (x86_64-abi)
Aug 18 11:02:56 cvtst1 kernel: xenbr0: topology change detected, propagating
Aug 18 11:02:56 cvtst1 kernel: xenbr0: port 3(vif2.0) entering forwarding state
Aug 18 11:02:56 cvtst1 kernel: ADDRCONF(NETDEV_CHANGE): vif2.0: link
becomes ready

>From node2:

Aug 18 11:01:44 cvtst2 clurgmgrd[4365]: <notice> Member 3 shutting down
Aug 18 11:01:55 cvtst2 qdiskd[3403]: <info> Node 3 shutdown
Aug 18 11:01:57 cvtst2 kernel: peth0: received packet with  own
address as source address
Aug 18 11:02:02 cvtst2 kernel: peth0: received packet with  own
address as source address
Aug 18 11:02:07 cvtst2 openais[3359]: [TOTEM] entering GATHER state from 12.
Aug 18 11:02:09 cvtst2 openais[3359]: [CMAN ] lost contact with quorum device
Aug 18 11:02:09 cvtst2 openais[3359]: [CMAN ] quorum lost, blocking activity
Aug 18 11:02:09 cvtst2 clurgmgrd[4365]: <emerg> #1: Quorum Dissolved
Aug 18 11:02:09 cvtst2 kernel: dlm: closing connection to node 3
Aug 18 11:02:10 cvtst2 ccsd[3316]: Cluster is not quorate.  Refusing
connection.
Aug 18 11:02:10 cvtst2 ccsd[3316]: Error while processing connect:
Connection refused
Aug 18 11:02:12 cvtst2 openais[3359]: [TOTEM] entering GATHER state from 0.
Aug 18 11:02:12 cvtst2 openais[3359]: [TOTEM] Saving state aru 9f high
seq received 9f
Aug 18 11:02:12 cvtst2 openais[3359]: [TOTEM] Storing new sequence id
for ring 758
Aug 18 11:02:12 cvtst2 openais[3359]: [TOTEM] entering COMMIT state.
Aug 18 11:02:12 cvtst2 openais[3359]: [TOTEM] entering RECOVERY state.
Aug 18 11:02:12 cvtst2 openais[3359]: [TOTEM] position [0] member X.X.X.165:
Aug 18 11:02:12 cvtst2 openais[3359]: [TOTEM] previous ring seq 1876
rep X.X.X.165
Aug 18 11:02:12 cvtst2 openais[3359]: [TOTEM] aru 9f high delivered 9f
received flag 1
Aug 18 11:02:12 cvtst2 openais[3359]: [TOTEM] position [1] member X.X.X.172:
Aug 18 11:02:12 cvtst2 openais[3359]: [TOTEM] previous ring seq 1876
rep X.X.X.165
Aug 18 11:02:12 cvtst2 openais[3359]: [TOTEM] aru 9f high delivered 9f
received flag 1
Aug 18 11:02:12 cvtst2 openais[3359]: [TOTEM] Did not need to
originate any messages in recovery.
Aug 18 11:02:12 cvtst2 openais[3359]: [CLM  ] CLM CONFIGURATION CHANGE
Aug 18 11:02:12 cvtst2 openais[3359]: [CLM  ] New Configuration:
Aug 18 11:02:12 cvtst2 openais[3359]: [CLM  ] 	r(0) ip(X.X.X.165)
Aug 18 11:02:12 cvtst2 openais[3359]: [CLM  ] 	r(0) ip(X.X.X.172)
Aug 18 11:02:12 cvtst2 openais[3359]: [CLM  ] Members Left:
Aug 18 11:02:12 cvtst2 openais[3359]: [CLM  ] 	r(0) ip(X.X.X.173)
Aug 18 11:02:12 cvtst2 openais[3359]: [CLM  ] Members Joined:
Aug 18 11:02:12 cvtst2 openais[3359]: [CMAN ] quorum regained,
resuming activity
Aug 18 11:02:12 cvtst2 openais[3359]: [CLM  ] CLM CONFIGURATION CHANGE
Aug 18 11:02:12 cvtst2 openais[3359]: [CLM  ] New Configuration:
Aug 18 11:02:12 cvtst2 openais[3359]: [CLM  ] 	r(0) ip(X.X.X.165)
Aug 18 11:02:12 cvtst2 openais[3359]: [CLM  ] 	r(0) ip(X.X.X.172)
Aug 18 11:02:12 cvtst2 openais[3359]: [CLM  ] Members Left:
Aug 18 11:02:12 cvtst2 openais[3359]: [CLM  ] Members Joined:
Aug 18 11:02:12 cvtst2 openais[3359]: [SYNC ] This node is within the
primary component and will provide service.
Aug 18 11:02:12 cvtst2 openais[3359]: [TOTEM] entering OPERATIONAL state.
Aug 18 11:02:12 cvtst2 openais[3359]: [CLM  ] got nodejoin message X.X.X.165
Aug 18 11:02:12 cvtst2 openais[3359]: [CLM  ] got nodejoin message X.X.X.172
Aug 18 11:02:12 cvtst2 openais[3359]: [CPG  ] got joinlist message from node 2
Aug 18 11:02:12 cvtst2 openais[3359]: [CPG  ] got joinlist message from node 1
Aug 18 11:02:25 cvtst2 qdiskd[3403]: <info> Assuming master role
Aug 18 11:02:33 cvtst2 kernel: xenbr0: port 3(vif1.0) entering disabled state
Aug 18 11:02:33 cvtst2 kernel: device vif1.0 left promiscuous mode
Aug 18 11:02:33 cvtst2 kernel: xenbr0: port 3(vif1.0) entering disabled state
Aug 18 11:02:36 cvtst2 clurgmgrd[4365]: <notice> Quorum Regained
Aug 18 11:02:39 cvtst2 clurgmgrd[4365]: <notice> Starting stopped
service vm:guest2
Aug 18 11:02:41 cvtst2 kernel: tap tap-2-51712: 2 getting info
Aug 18 11:02:41 cvtst2 kernel: device vif2.0 entered promiscuous mode
Aug 18 11:02:41 cvtst2 kernel: ADDRCONF(NETDEV_UP): vif2.0: link is not ready
Aug 18 11:02:41 cvtst2 clurgmgrd[4365]: <notice> Service vm:guest2 started
Aug 18 11:02:45 cvtst2 kernel: blktap: ring-ref 8, event-channel 6,
protocol 1 (x86_64-abi)
Aug 18 11:02:53 cvtst2 kernel: xenbr0: topology change detected, propagating
Aug 18 11:02:53 cvtst2 kernel: xenbr0: port 3(vif2.0) entering forwarding state
Aug 18 11:02:53 cvtst2 kernel: ADDRCONF(NETDEV_CHANGE): vif2.0: link
becomes read



>
> Regards,
>
> Alfredo
>
> -----Original Message-----
> From: linux-cluster-bounces at redhat.com [mailto:linux-cluster-bounces at redhat.com] On Behalf Of Paras pradhan
> Sent: Monday, August 17, 2009 11:59 PM
> To: linux clustering
> Subject: [Linux-cluster] Cluster behavior
>
> I have a 3 nodes cluster.
>
> Node A - Vote 1
>
> Node B - Vote 1
>
> Node C - Votes 2
>
> Qdisk - Votes 3
>
>
> Altogether the cluster has 7 votes. The required min quorum to run the
> cluster would be in this case 4. Now if I poweroff node 1, I can see
> Quorum Dissolved in the terminals of Node 2 and Node3 . This cluster
> has xen virtual machines.
>
> Whats wrong and how to do debug the problem?
>
> Thanks
> Paras.
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>

Thanks
Paras.



From pradhanparas at gmail.com  Tue Aug 18 16:26:17 2009
From: pradhanparas at gmail.com (Paras pradhan)
Date: Tue, 18 Aug 2009 11:26:17 -0500
Subject: [Linux-cluster] Cluster behavior
In-Reply-To: <F5A4C3A7AC59A74FB0A0D35F1E8BDB460352A205E1@EXMB03.eu.tieto.com>
References: <8b711df40908171459q1afecda7k7b0dcc4773153b84@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C484B@rkamsem701.emea.roche.com>
	<F5A4C3A7AC59A74FB0A0D35F1E8BDB460352A205E1@EXMB03.eu.tieto.com>
Message-ID: <8b711df40908180926v5b7e7131rffae94430c103eb7@mail.gmail.com>

On Tue, Aug 18, 2009 at 8:51 AM, <Harri.Paivaniemi at tieto.com> wrote:
> Are you sure,
>
> that your qdisk-configuration is ok and it's working properly? Just based on maths, it seems that you have only 4 votes up in

How do I check if my qdisk config is working properly?

I can see qdiskd -Q running in the bg.

I can see

dev/disk/by-id/scsi-360060e8004770d000000770d000001a7-part1    0
Online, Quorum Disk

from clustat.



your cluster and it goes down if it loses one vote...
>
> So is cman_tool really showing exp.votes=7? And qdisk brings it's votes ok?
>

Yes it is showing 7.

> -hjp
>
> ________________________________________
> From: linux-cluster-bounces at redhat.com [linux-cluster-bounces at redhat.com] On Behalf Of Moralejo, Alfredo [alfredo.moralejo at roche.com]
> Sent: Tuesday, August 18, 2009 4:26 PM
> To: linux clustering
> Subject: RE: [Linux-cluster] Cluster behavior
>
> Could you send logs in messages file including the part where the cluster dissolved messages is logged?
>
> Regards,
>
> Alfredo
>
> -----Original Message-----
> From: linux-cluster-bounces at redhat.com [mailto:linux-cluster-bounces at redhat.com] On Behalf Of Paras pradhan
> Sent: Monday, August 17, 2009 11:59 PM
> To: linux clustering
> Subject: [Linux-cluster] Cluster behavior
>
> I have a 3 nodes cluster.
>
> Node A - Vote 1
>
> Node B - Vote 1
>
> Node C - Votes 2
>
> Qdisk - Votes 3
>
>
> Altogether the cluster has 7 votes. The required min quorum to run the
> cluster would be in this case 4. Now if I poweroff node 1, I can see
> Quorum Dissolved in the terminals of Node 2 and Node3 . This cluster
> has xen virtual machines.
>
> Whats wrong and how to do debug the problem?
>
> Thanks
> Paras.
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>

Paras.



From lhh at redhat.com  Tue Aug 18 20:27:56 2009
From: lhh at redhat.com (Lon Hohberger)
Date: Tue, 18 Aug 2009 16:27:56 -0400
Subject: [Linux-cluster] clusterservice stays in 'recovering' state
In-Reply-To: <f97c3a70908130545n11ce442ej17d74c9cdc450e45@mail.gmail.com>
References: <f97c3a70908130545n11ce442ej17d74c9cdc450e45@mail.gmail.com>
Message-ID: <1250627276.9711.168.camel@localhost.localdomain>

On Thu, 2009-08-13 at 14:45 +0200, mark benschop wrote:
> Hi All,
> 
> I've a problem with a clusterservice. The service was started up while
> one of the resources, an NFS, export was not accessible.
> Therefore the service never started up right but got into the
> 'recovering' state.
> In the mean time the NFS exports are setup properly but to no avail.
> Stopping the clusterservice, using clusvcadm -d <service>, will result
> in the service going down but staying in the 'recovering' state.
> Starting it again doesn't work. The service doesn't start and stays in
> the recovery status.
> I'm suspecting rgmanager lost it somehow.
> 
> Anybody had any ideas on what could be the problem and how to resolve
> it ?

Assuming it's the most recent version, file a bugzilla.

-- Lon




From tpeirce at etrafficsolutions.com  Tue Aug 18 20:44:04 2009
From: tpeirce at etrafficsolutions.com (Trevor Peirce)
Date: Tue, 18 Aug 2009 13:44:04 -0700
Subject: [Linux-cluster] cluster-3.0.0 problems starting up
In-Reply-To: <4A85E4EF.4020501@etrafficsolutions.com>
References: <4A85E1C7.1070502@etrafficsolutions.com>	<d740f8db0908141523y7a0d4a7eta656459bf16bc145@mail.gmail.com>
	<4A85E4EF.4020501@etrafficsolutions.com>
Message-ID: <4A8B1294.5070805@etrafficsolutions.com>

Trevor Peirce wrote:
> Aug 14 15:24:58 localhost corosync[28324]:   [MAIN  ] parse error in 
> config: parse error in config: .
> Aug 14 15:24:58 localhost corosync[28324]:   [MAIN  ] Corosync Cluster 
> Engine exiting with status -9 at main.c:879.

Have I found a dead end?

Perhaps someone would be so kind as to send the most simple cluster.conf
file that is known to work so I can try it out?

Or any ideas what this error means?

Thanks,

-- 
Trevor Peirce
Systems Administrator
Etraffic Solutions Inc.

Suite 100 ? 710 Redbrick Street
Victoria, B.C. V8T 5J3
phone: 250.658.8238 x49
fax: 250.658.5936



From lhh at redhat.com  Tue Aug 18 21:29:36 2009
From: lhh at redhat.com (Lon Hohberger)
Date: Tue, 18 Aug 2009 17:29:36 -0400
Subject: [Linux-cluster] Min Quorum
In-Reply-To: <8b711df40908130842p7379016coaad5666d24bd49cc@mail.gmail.com>
References: <8b711df40908130830u6923f642td319b97a1e259861@mail.gmail.com>
	<4A843322.1050304@redhat.com>
	<8b711df40908130842p7379016coaad5666d24bd49cc@mail.gmail.com>
Message-ID: <1250630976.9711.229.camel@localhost.localdomain>

On Thu, 2009-08-13 at 10:42 -0500, Paras pradhan wrote:

> I have 3 nodes + 2 votes to quorum parition. So altogether the total
> votes should be 5 (?). I can see total votes to be only 3 .

<cman expected_votes="5" />

-- lon




From lhh at redhat.com  Tue Aug 18 21:33:16 2009
From: lhh at redhat.com (Lon Hohberger)
Date: Tue, 18 Aug 2009 17:33:16 -0400
Subject: [Linux-cluster] cluster-3.0.0 problems starting up
In-Reply-To: <4A8B1294.5070805@etrafficsolutions.com>
References: <4A85E1C7.1070502@etrafficsolutions.com>
	<d740f8db0908141523y7a0d4a7eta656459bf16bc145@mail.gmail.com>
	<4A85E4EF.4020501@etrafficsolutions.com>
	<4A8B1294.5070805@etrafficsolutions.com>
Message-ID: <1250631196.9711.232.camel@localhost.localdomain>

On Tue, 2009-08-18 at 13:44 -0700, Trevor Peirce wrote:

> Perhaps someone would be so kind as to send the most simple cluster.conf
> file that is known to work so I can try it out?
> 
> Or any ideas what this error means?

The XML looks valid; I'm not sure what the error would mean off the top
of my head.

-- Lon



From lhh at redhat.com  Tue Aug 18 21:35:47 2009
From: lhh at redhat.com (Lon Hohberger)
Date: Tue, 18 Aug 2009 17:35:47 -0400
Subject: [Linux-cluster] Rgmanager and Xen VMs
In-Reply-To: <7bc80d500908171651m48927838v21644fc03cd21fca@mail.gmail.com>
References: <7bc80d500908171651m48927838v21644fc03cd21fca@mail.gmail.com>
Message-ID: <1250631347.9711.238.camel@localhost.localdomain>

On Mon, 2009-08-17 at 16:51 -0700, Christopher Chen wrote:
> Hi there--
> 
> I'm wondering about the behavior of rgmanager with regards to Xen VMs
> and memory planning.

There isn't any "planning".

> I see there's a memory parameter defined in vm.sh. Does cluster
> manager take this into account when trying to start up guests? Right
> now it's undefined so my rgmanager will try to start up guests in the
> order defined for the failover domain--of course, at some point, VMs
> fail to start when the first Dom0's free memory gets too low, then
> they get started on the next machine, and so on...

We moved to using 'virsh' instead of 'xm', which will in theory allow us
to fix migration/start issues when memory is too low, but there's
currently no best-fit algorithm.

-- Lon



From lhh at redhat.com  Tue Aug 18 21:42:04 2009
From: lhh at redhat.com (Lon Hohberger)
Date: Tue, 18 Aug 2009 17:42:04 -0400
Subject: [Linux-cluster] Questions about rgmanager
In-Reply-To: <29ae894c0908180201l14ab18aexddefd0f04bf2ba29@mail.gmail.com>
References: <29ae894c0908180201l14ab18aexddefd0f04bf2ba29@mail.gmail.com>
Message-ID: <1250631724.9711.251.camel@localhost.localdomain>

On Tue, 2009-08-18 at 11:01 +0200, brem belguebli wrote:
> Hello,
>  
> I have a few questions about rgmanager:
>  
> 1) How to make it take into account resource metadata changes without
> restarting it. service rgmanager reload (or kill -HUP) doesn't. 
>     In case one needs to change some timers in a resource metadata,
> the only way to make it is to stop and start rgmanager which brings
> down the resources.

Need to change the configuration version in cluster.conf; rgmanager
doesn't record for example file mtimes and run stat() to see if metadata
changed.  Doing this wouldn't work anyway, as some resources have
external metadata.
 
> 2) Is there a way to log what "rules" rgmanager succeded to load at
> start time (or the ones it failed to), as having a "24 rules loaded"
> message isn't very explicit.

You can run: 

   rg_test rules | grep ^Agent

There's no way currently to have rgmanager log specifically which agent
it loaded at run-time.

> The last point concerns the potential replacement of rgmanager with
> pacemaker (as stated at
> http://sources.redhat.com/cluster/wiki/RGManager). Will this happen in
> a near future ?

Not immediately; we still have a bit to go:

* need configuration file format conversion utility
* need basic "work-alike" tools.  Ex: clustat.
* need to do a bit of work on rgmanager agents so they work
  on pacemaker - the standard by which rgmanager+pacemaker 
  operate was inadequate and the two programs have diverged
  in slightly incompatible ways

-- Lon



From lhh at redhat.com  Tue Aug 18 21:52:53 2009
From: lhh at redhat.com (Lon Hohberger)
Date: Tue, 18 Aug 2009 17:52:53 -0400
Subject: [Linux-cluster] RHCS on KVM
In-Reply-To: <5f61ab380908130707q5c936504k7351d0d6b3459090@mail.gmail.com>
References: <5f61ab380908130707q5c936504k7351d0d6b3459090@mail.gmail.com>
Message-ID: <1250632373.9711.282.camel@localhost.localdomain>

On Thu, 2009-08-13 at 10:07 -0400, Nehemias Jahcob wrote:
> Hi.
> 
>  How to create a cluster of 2 nodes in rhel5.4 (or Fedora 10) with
> KVM? 
> 
> With XEN follow this guide: 
> http://sources.redhat.com/cluster/wiki/VMClusterCookbook?highlight =
> (CategoryHowTo). 
> 
> Do you have a guide to implementation of RHCS in KVM? 

On Fedora 11, you need to add:

  /usr/sbin/fence_xvmd -LI virbr0

... to /etc/rc.d/rc.local on the "host" machine.

Other than that, you can just add a disk to two machines using
virt-manager (it complains, but does the right thing) to create a shared
"disk" for use by GFS2 or clustered LVM (or for use as a quorum disk).

If you're running a cluster of VMs on a single host using NATted
network, you don't even need to distribute fence_xvm.key (or any keys at
all).

I'll make a note to update that VMClusterCookbook page; it's out of date
and xen-specific.  This is bad because the cluster and associated tools
are all libvirt-based and *NOT* specific to any virtualization
technology.

-- Lon



From pradhanparas at gmail.com  Tue Aug 18 21:54:23 2009
From: pradhanparas at gmail.com (Paras pradhan)
Date: Tue, 18 Aug 2009 16:54:23 -0500
Subject: [Linux-cluster] Min Quorum
In-Reply-To: <1250630976.9711.229.camel@localhost.localdomain>
References: <8b711df40908130830u6923f642td319b97a1e259861@mail.gmail.com>
	<4A843322.1050304@redhat.com>
	<8b711df40908130842p7379016coaad5666d24bd49cc@mail.gmail.com>
	<1250630976.9711.229.camel@localhost.localdomain>
Message-ID: <8b711df40908181454t73c2f583r3d6e4b90d1609992@mail.gmail.com>

On Tue, Aug 18, 2009 at 4:29 PM, Lon Hohberger<lhh at redhat.com> wrote:
> On Thu, 2009-08-13 at 10:42 -0500, Paras pradhan wrote:
>
>> I have 3 nodes + 2 votes to quorum parition. So altogether the total
>> votes should be 5 (?). I can see total votes to be only 3 .
>
> <cman expected_votes="5" />

You mean total should be 5?

>
> -- lon
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>

Paras.



From lhh at redhat.com  Tue Aug 18 21:57:46 2009
From: lhh at redhat.com (Lon Hohberger)
Date: Tue, 18 Aug 2009 17:57:46 -0400
Subject: [Linux-cluster] quorum handling
In-Reply-To: <8b711df40908111230q45d1faafp839238d25d6c4e26@mail.gmail.com>
References: <8b711df40908111230q45d1faafp839238d25d6c4e26@mail.gmail.com>
Message-ID: <1250632666.9711.297.camel@localhost.localdomain>

On Tue, 2009-08-11 at 14:30 -0500, Paras pradhan wrote:
> I have a 3 nodes xen cluser under centos5.3 that host linux virtual
> machines. Node1 and Node2 got virtual machines but not on Node3. Node
> 3 is basically used when Node1 or Node2 goes down. This is working OK.
> Now when I turn off both  Node1 and Node2. Then quorum is dissolved
> and my cluser fails. I do not see any virtual machines in Node3 if the
> other 2 nodes die.
> 
> So, how to handle this situation? I need node 3 to be alive a working
> to host virtual machines when Node 1 and Node 2 is shut down.

1/3 is not enough to constitute a quorum; you will have to use a quorum
disk or some other method in order to allow a quorum of 1 (out of a
possible 3).

You could make that node a high weight (like 3 votes), but then when it
fails, the other 2 nodes will not be quorate.

The reason for this sort of operation is to prevent a split-brain:  If
the 1 node is alive and running virtual machines, and the other 2 nodes
come online but *cannot* talk to the already running node, they can
start up the virtual machines, causing data corruption.

-- Lon



From lhh at redhat.com  Tue Aug 18 21:58:42 2009
From: lhh at redhat.com (Lon Hohberger)
Date: Tue, 18 Aug 2009 17:58:42 -0400
Subject: [Linux-cluster] rgmanager ignoring monitor action for a custom
	resource script
In-Reply-To: <29ae894c0908101011i33f6bb9bl9dd0427fcb877701@mail.gmail.com>
References: <29ae894c0908101011i33f6bb9bl9dd0427fcb877701@mail.gmail.com>
Message-ID: <1250632722.9711.301.camel@localhost.localdomain>

On Mon, 2009-08-10 at 19:11 +0200, brem belguebli wrote:
> Hi,
>  
> I have written a custom resource script that starts,  stops and
> "meta-data" 's with the correct return  value (no error).
>  
> Everything seems to work fine except the fact that rgmanager
> never runs the custom monitor action defined in my resource metadata.
>  
> I have lowered the interval value for monitor action to 10 (  are the
> interval and timeout units in seconds by default ?) but nothing
> happens.
>  
> I have added a trace to the monitor case of my resource script that
> ocf_log debug "monitor started at `date +%H:%M:%S`" but it never gets
> there:

Change "monitor" to "status|monitor" and add "status" to the metadata.

-- Lon



From lhh at redhat.com  Tue Aug 18 21:59:54 2009
From: lhh at redhat.com (Lon Hohberger)
Date: Tue, 18 Aug 2009 17:59:54 -0400
Subject: [Linux-cluster] Re: rgmanager ignoring monitor action for a
	custom resource script
In-Reply-To: <29ae894c0908110323v6ee8033fh5117559262a34890@mail.gmail.com>
References: <29ae894c0908101011i33f6bb9bl9dd0427fcb877701@mail.gmail.com>
	<29ae894c0908110323v6ee8033fh5117559262a34890@mail.gmail.com>
Message-ID: <1250632794.9711.305.camel@localhost.localdomain>

On Tue, 2009-08-11 at 12:23 +0200, brem belguebli wrote:
> Problem resolved,
>  
> rgmanager seems to rely on the 'status' action rather than on the
> 'monitor' one.
>  
> Which doesn't seem to conform to the opencf API
> (http://www.opencf.org/cgi-bin/viewcvs.cgi/*checkout*/specs/ra/resource-agent-api.txt?rev=1.10) chapter 3.4.3 (is this doc outdated ?) that states that 'start' 'stop'  'monitor' and 'meta-data' are mandatory.

Rgmanager diverges from the standard.

There are a couple of ways it does this.  We will be resolving the
divergence issues shortly (i.e. in the next few months) as we move to
incorporate pacemaker into linux-cluster.

-- Lon




From lhh at redhat.com  Tue Aug 18 22:07:10 2009
From: lhh at redhat.com (Lon Hohberger)
Date: Tue, 18 Aug 2009 18:07:10 -0400
Subject: [Linux-cluster] vm.sh resource script
In-Reply-To: <20090814164550.16372xa21kmg4xz2@www.sjolshagen.net>
References: <20090814164550.16372xa21kmg4xz2@www.sjolshagen.net>
Message-ID: <1250633230.9711.326.camel@localhost.localdomain>

On Fri, 2009-08-14 at 16:45 -0400, Thomas Sjolshagen wrote:
> Hi,
> 
> I'm trying to use; "node1> clusvcadm -M <service> -m node2" in Fedora  
> 11 w/latest cluster updates installed and get the following error  
> message that I'm having a hard time deciphering/debugging:
> 
> Trying to migrate service:svc1 to node2...S/Lang Script Error
> 
> In /var/log/messages:
> 
> Aug 14 16:44:08 node2 rgmanager[3210]: [S/Lang] Variable Uninitialized Error
> Aug 14 16:44:08 node2 rgmanager[3210]: [S/Lang] Script Execution Failure
> 
> Not sure if there's something wrong with my cluster.conf or that error  
> is indicative of something else?

It's a known bug.

You can't mix central_processing + migrate operation right now.

The central_processing mode _requires_ synchronous operation, but with
the older (Xen) tools, the man page explicitly documents that most
operations were asynchronous - consequently, 'xm migrate' may or may not
have completed at the time the command returned.

With current virsh agent, we can fix this behavior because the virsh
operations (including migrate) are expected to be synchronous.

I talked with the libvirt developers about this recently - the fact that
virsh operations are synchronous (despite also being documented to be
asynchronous [GRRRRR] in the virsh man page) -greatly- simplifies the
steps we must take surrounding VM migration, so we can fix it now that
we use virsh instead of 'xm' :)

Can you file a bug so I don't forget to fix this for cluster-3.0.2 or
3.0.3 ?

-- Lon



From tpeirce at etrafficsolutions.com  Tue Aug 18 22:12:45 2009
From: tpeirce at etrafficsolutions.com (Trevor Peirce)
Date: Tue, 18 Aug 2009 15:12:45 -0700
Subject: [Linux-cluster] cluster-3.0.0 problems starting up
In-Reply-To: <1250631196.9711.232.camel@localhost.localdomain>
References: <4A85E1C7.1070502@etrafficsolutions.com>	<d740f8db0908141523y7a0d4a7eta656459bf16bc145@mail.gmail.com>	<4A85E4EF.4020501@etrafficsolutions.com>	<4A8B1294.5070805@etrafficsolutions.com>
	<1250631196.9711.232.camel@localhost.localdomain>
Message-ID: <4A8B275D.3060201@etrafficsolutions.com>

Lon Hohberger wrote:
> The XML looks valid; I'm not sure what the error would mean off the top
> of my head.

Hi Lon,

Thank you for the reply. Any suggestions on where to go from here...? 
Are there further debug settings I can enable, or any partial services I 
should try to start to see if they work individually?

Thanks,

-- 
Trevor Peirce
Systems Administrator
Etraffic Solutions Inc.

Suite 100 ? 710 Redbrick Street
Victoria, B.C. V8T 5J3
phone: 250.658.8238 x49
fax: 250.658.5936 




From brem.belguebli at gmail.com  Tue Aug 18 22:25:28 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Wed, 19 Aug 2009 00:25:28 +0200
Subject: [Linux-cluster] Questions about rgmanager
In-Reply-To: <1250631724.9711.251.camel@localhost.localdomain>
References: <29ae894c0908180201l14ab18aexddefd0f04bf2ba29@mail.gmail.com>
	<1250631724.9711.251.camel@localhost.localdomain>
Message-ID: <29ae894c0908181525n2ffad9b0m112f25fcd3b14ca5@mail.gmail.com>

thx Lon

2009/8/18 Lon Hohberger <lhh at redhat.com>

> On Tue, 2009-08-18 at 11:01 +0200, brem belguebli wrote:
> > Hello,
> >
> > I have a few questions about rgmanager:
> >
> > 1) How to make it take into account resource metadata changes without
> > restarting it. service rgmanager reload (or kill -HUP) doesn't.
> >     In case one needs to change some timers in a resource metadata,
> > the only way to make it is to stop and start rgmanager which brings
> > down the resources.
>
> Need to change the configuration version in cluster.conf; rgmanager
> doesn't record for example file mtimes and run stat() to see if metadata
> changed.  Doing this wouldn't work anyway, as some resources have
> external metadata.
>
> > 2) Is there a way to log what "rules" rgmanager succeded to load at
> > start time (or the ones it failed to), as having a "24 rules loaded"
> > message isn't very explicit.
>
> You can run:
>
>   rg_test rules | grep ^Agent
>
> There's no way currently to have rgmanager log specifically which agent
> it loaded at run-time.
>
> > The last point concerns the potential replacement of rgmanager with
> > pacemaker (as stated at
> > http://sources.redhat.com/cluster/wiki/RGManager). Will this happen in
> > a near future ?
>
> Not immediately; we still have a bit to go:
>
> * need configuration file format conversion utility
> * need basic "work-alike" tools.  Ex: clustat.
> * need to do a bit of work on rgmanager agents so they work
>  on pacemaker - the standard by which rgmanager+pacemaker
>  operate was inadequate and the two programs have diverged
>  in slightly incompatible ways
>
> -- Lon
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090819/5663585a/attachment.htm>

From sdake at redhat.com  Tue Aug 18 23:28:27 2009
From: sdake at redhat.com (Steven Dake)
Date: Tue, 18 Aug 2009 16:28:27 -0700
Subject: [Linux-cluster] cluster-3.0.0 problems starting up
In-Reply-To: <4A8B1294.5070805@etrafficsolutions.com>
References: <4A85E1C7.1070502@etrafficsolutions.com>
	<d740f8db0908141523y7a0d4a7eta656459bf16bc145@mail.gmail.com>
	<4A85E4EF.4020501@etrafficsolutions.com>
	<4A8B1294.5070805@etrafficsolutions.com>
Message-ID: <1250638107.19265.98.camel@localhost.localdomain>

On Tue, 2009-08-18 at 13:44 -0700, Trevor Peirce wrote:
> Trevor Peirce wrote:
> > Aug 14 15:24:58 localhost corosync[28324]:   [MAIN  ] parse error in 
> > config: parse error in config: .
> > Aug 14 15:24:58 localhost corosync[28324]:   [MAIN  ] Corosync Cluster 
> > Engine exiting with status -9 at main.c:879.
> 
> Have I found a dead end?
> 
> Perhaps someone would be so kind as to send the most simple cluster.conf
> file that is known to work so I can try it out?
> 
> Or any ideas what this error means?
> 
> Thanks,
> 

Trevor,

This means your XML format is invalid.  We are working on better error
reporting when the XML format is invalid rather then this cryptic error.
In the meantime, you can run strace -f cman, and libxml will print the
errors in write system calls.  If you need help analyzing this output,
send a link to the output from strace.

Regards
-steve



From thomas at sjolshagen.net  Wed Aug 19 00:48:09 2009
From: thomas at sjolshagen.net (Thomas Sjolshagen)
Date: Tue, 18 Aug 2009 20:48:09 -0400
Subject: [Linux-cluster] vm.sh resource script
In-Reply-To: <1250633230.9711.326.camel@localhost.localdomain>
References: <20090814164550.16372xa21kmg4xz2@www.sjolshagen.net>
	<1250633230.9711.326.camel@localhost.localdomain>
Message-ID: <20090818204809.19483k45ubb6z2yx@www.sjolshagen.net>


> Can you file a bug so I don't forget to fix this for cluster-3.0.2 or
> 3.0.3 ?

Will do!

Thanks,

// Thomas



----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.




From wen.gang.wang at oracle.com  Wed Aug 19 02:30:34 2009
From: wen.gang.wang at oracle.com (Wengang Wang)
Date: Wed, 19 Aug 2009 10:30:34 +0800
Subject: [Linux-cluster] gfs2 mount hangs
In-Reply-To: <20090818141442.GA6303@redhat.com>
References: <4A83B113.60609@oracle.com> <20090813141424.GA8148@redhat.com>
	<4A84CC39.3080809@oracle.com> <20090814145738.GA7094@redhat.com>
	<4A8A12D6.5050907@oracle.com> <20090818141442.GA6303@redhat.com>
Message-ID: <4A8B63CA.8090104@oracle.com>

Hi David,

> 
> That's the problem, there are a bunch of steps missing that come from
> mount.gfs2.  My guess is that you're missing /sbin/mount.gfs2 on this node.
> 
> It also seems that gfs2 no longer properly handles this error condition
> (missing/empty hostdata string).  In the past you would have gotten a nice
> message in /var/log/messages explaining that you were missing mount.gfs2.

Ohhhh, my fault!
I really didn't have mount.gfs2 on node "desk". while I didn't get logs 
saying that I am missing it, strange...

anyway, thank much for your help and patience!

regards,
wengang.



From jfriesse at redhat.com  Wed Aug 19 08:15:24 2009
From: jfriesse at redhat.com (Jan Friesse)
Date: Wed, 19 Aug 2009 10:15:24 +0200
Subject: [Linux-cluster] VirtualBox - fence agent
In-Reply-To: <20090811131030.0b13dac7@smicro.interstudio.homeunix.net>
References: <20090811031639.GA728@monsterjam.org>
	<20090811131030.0b13dac7@smicro.interstudio.homeunix.net>
Message-ID: <4A8BB49C.6060706@redhat.com>

Bob,
we don't have VirtualBox fence agent. If you have something working,
please feel free to send it to mailing list and I will take care to
include it in upstream. In case you don't have that agent, I can take
care to write it (one node only, simple ssh to DOM-0 and run
vbox-manager -> not for production use but good for simple testing cluster)

Regards,
  Honza

Bob Marcan wrote:
> Hi.
> Does anybody write a fence agent for the VirtualBox?
> Can we get this into the mainstream?
> 
> BR, Bob
> 
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster



From lhh at redhat.com  Wed Aug 19 13:25:08 2009
From: lhh at redhat.com (Lon Hohberger)
Date: Wed, 19 Aug 2009 09:25:08 -0400
Subject: [Linux-cluster] Cluster behavior
In-Reply-To: <8b711df40908171459q1afecda7k7b0dcc4773153b84@mail.gmail.com>
References: <8b711df40908171459q1afecda7k7b0dcc4773153b84@mail.gmail.com>
Message-ID: <1250688308.17779.5.camel@localhost.localdomain>

On Mon, 2009-08-17 at 16:59 -0500, Paras pradhan wrote:
> I have a 3 nodes cluster.
> 
> Node A - Vote 1
> 
> Node B - Vote 1
> 
> Node C - Votes 2
> 
> Qdisk - Votes 3
> 
> 
> Altogether the cluster has 7 votes. The required min quorum to run the
> cluster would be in this case 4. Now if I poweroff node 1, I can see
> Quorum Dissolved in the terminals of Node 2 and Node3 . This cluster
> has xen virtual machines.

I assume node 1 == node A, node 2 == node B, node 3 == node C.


> Whats wrong and how to do debug the problem?

Do they all "See" each other before you power off node 1 (where 1 ==
A) ?  Apart from having a node with 2 votes, the rest of the concept is
pretty similar to qdiskd's design for "all-but-1 die" case.

Expected behavior is that any of the following constitutes a quorate
cluster:

A+Qdisk      (4 votes)
B+Qdisk      (4 votes)
C+Qdisk      (4 votes)
A+B+Qdisk    (5 votes)
A+B+C        (4 votes)
A+B+C+Qdisk  (7 votes)

Effectively, you want to look at 'cman_tool status'.  Before losing a
node, you need to verify that each sees 7 votes.

-- Lon



From lhh at redhat.com  Wed Aug 19 13:39:16 2009
From: lhh at redhat.com (Lon Hohberger)
Date: Wed, 19 Aug 2009 09:39:16 -0400
Subject: [Linux-cluster] cluster-3.0.0 problems starting up
In-Reply-To: <1250638107.19265.98.camel@localhost.localdomain>
References: <4A85E1C7.1070502@etrafficsolutions.com>
	<d740f8db0908141523y7a0d4a7eta656459bf16bc145@mail.gmail.com>
	<4A85E4EF.4020501@etrafficsolutions.com>
	<4A8B1294.5070805@etrafficsolutions.com>
	<1250638107.19265.98.camel@localhost.localdomain>
Message-ID: <1250689156.17779.11.camel@localhost.localdomain>

On Tue, 2009-08-18 at 16:28 -0700, Steven Dake wrote:
> On Tue, 2009-08-18 at 13:44 -0700, Trevor Peirce wrote:
> > Trevor Peirce wrote:
> > > Aug 14 15:24:58 localhost corosync[28324]:   [MAIN  ] parse error in 
> > > config: parse error in config: .
> > > Aug 14 15:24:58 localhost corosync[28324]:   [MAIN  ] Corosync Cluster 
> > > Engine exiting with status -9 at main.c:879.
> > 
> > Have I found a dead end?
> > 
> > Perhaps someone would be so kind as to send the most simple cluster.conf
> > file that is known to work so I can try it out?
> > 
> > Or any ideas what this error means?
> > 
> > Thanks,
> > 
> 
> Trevor,
> 
> This means your XML format is invalid.  We are working on better error
> reporting when the XML format is invalid rather then this cryptic error.
> In the meantime, you can run strace -f cman, and libxml will print the
> errors in write system calls.  If you need help analyzing this output,
> send a link to the output from strace.

That's odd; I ran it through xmllint with no errors - that's why I'm
confused.

There is probably a bug in the schema (i.e. something should be
"required" that isn't) if the config is invalid:

[lhh at localhost xml]$ git status | grep branch
# On branch stable3
[lhh at localhost xml]$ xmllint --relaxng ./cluster.rng /tmp/trevor.xml 
<?xml version="1.0"?>
<cluster name="TestCluster" config_version="3">
  <clusternodes><clusternode name="guinneapig" votes="1"
nodeid="1"/></clusternodes>
  <rm>
    <failoverdomains/>
    <resources/>
  </rm>
</cluster>
/tmp/trevor.xml validates

-- Lon



From R.Schwierz at physik.tu-dresden.de  Wed Aug 19 15:36:15 2009
From: R.Schwierz at physik.tu-dresden.de (Rainer Schwierz)
Date: Wed, 19 Aug 2009 17:36:15 +0200
Subject: [Linux-cluster] RHCS TestCluster with ScientificLinux 5.2
Message-ID: <4A8C1BEF.3010200@physik.tu-dresden.de>

Hello experts,

In preparation of a new production system I have setup a testsystem
with RHCS under ScientificLinux 5.2.
It consists of two identical nodes FSC/RX200, a Brocade FibreChannel 
switch, a FSC/SX80 FibreChannel-raidarray, and a APC-powerswitch.
The configuration is attached at the end.
I want to have three (GFS) filesystems
- exported via nfs to a number of clients, each service has its own IP
- backup the filesystems via TSM to a TSM-server

I see some problems I need an explanation/solution:
1) if I connect the nfs-clients to the IP of the configured nfs-service
   started e.g. on tnode02, the filesystem is mounted, but I see a
   strange lock problem
     tnode02 kernel: portmap: server "client-IP" not\
         responding, timed out
     tnode02 kernel: lockd: server "client-IP" not responding,\
         timed out
    It goes away, if I bind the nfs-clients direct to the IP of the
    the node tnode02. If I start the services on tnode01, it is exactly
    the same problem, solved by binding the clients direct to tnode01. It
    does not depend on firewall configuration, it is the same if I switch
    off iptables on both tnode0[12] and clients.

2) tnode02 died with kernel-panic; no real helpfull logs found regarding
    the panic, I only see a lot of messages regarding problems nfs
    locking over gfs :

   kernel: lockd: grant for unknown block
   kernel: dlm: dlm_plock_callback: lock granted after lock request failed

   before the kernel paniced, but is this a real reason to panic?

   At this point tnod01 tried to take over the cluster and to fence
   tnode02, which gave an error, I do not understand, because fence_apc
   runnig by hand (On, Off, Status) is properly working

tnode01 fenced[3127]: fencing node "tnode02.phy.tu-dresden.de"
tnode01 fenced[3127]: agent "fence_apc" reports: Traceback (most recent 
call last):   File "/sbin/fence_apc", line 829, in ?     main()   File 
"/sbin/fence_apc", line 349, in main     do_power_off(sock)   File 
"/sbin/fence_apc", line 813, in do_power_off     x = 
do_power_switch(sock, "off")   File "/sbi
tnode01 fenced[3127]: agent "fence_apc" reports: n/fence_apc", line 611, 
in do_power_switch     result_code, response = power_off(txt + ndbuf) 
File "/sbin/fence_apc", line 817, in power_off     x = 
power_switch(buffer, False, "2", "3");   File "/sbin/fence_apc", line 
810, in power_switch     raise "un
tnode01 fenced[3127]: agent "fence_apc" reports: known screen 
encountered in \n" + str(lines) + "\n" unknown screen encountered in 
['', '> 2', '', '', '------- Configure Outlet 
------------------------------------------------------', '', '    # 
State  Ph  Name                     Pwr On Dly  Pwr Off D
tnode01 fenced[3127]: agent "fence_apc" reports: ly  Reboot Dur.', ' 
----------------------------------------------------------------------------', 
'    2  ON     1   Outlet 2                 0 sec       0 sec        5 
sec', '', '     1- Outlet Name         : Outlet 2', '     2- Power On 
Delay(sec) :
tnode01 fenced[3127]: agent "fence_apc" reports: 0', '     3- Power Off 
Delay(sec): 0', '     4- Reboot Duration(sec): 5', '     5- Accept 
Changes      : ', '', '     ?- Help, <ESC>- Back, <ENTER>- Refresh, 
<CTRL-L>- Event Log']

   So tnode01 did not stop fencing tnod02 and so it was not able to take
   over the cluster services. Via system-config-cluster one was also not
   able to stop any service. Stopping processes did not really help. The
   only solution at this point was to power down both nodes and restart
   the cluster.

so my questions:

Is there a solution for the locking problem if one bind the nfs clients 
to the configured nfs service IP ?

Is there an explanation/solution of the nfs (dlm) GFS locking problem ?

Is there a signifivant update to fence_apc I have missed ?

Why do I have to configure the GFS resources with the "force umount" option?
   I was under the impression that one can mount GFS filesystems
   simultanously on a number of nodes. If I define the GFS resources
   without "force umount", the filesystem is not mounted at all. But
   running the defined TSM service depends on all mounted filesystems.

Thanks for any help,  Rainer

The configuration is
Scientific Linux SL release 5.2 (Boron)
kernel 2.6.18-128.4.1.el5 #1 SMP Tue Aug 4 12:51:10 EDT 2009 x86_64 
x86_64 x86_64 GNU/Linux
device-mapper-multipath-0.4.7-23.el5_3.2.x86_64
rgmanager-2.0.38-2.el5_2.1.x86_64
system-config-cluster-1.0.52-1.1.noarch
cman-2.0.84-2.el5.x86_64
kmod-gfs-0.1.23-5.el5_2.4.x86_64
gfs2-utils-0.1.44-1.el5.x86_64
gfs-utils-0.1.17-1.el5.x86_64
lvm2-cluster-2.02.32-4.el5.x86_64
modcluster-0.12.0-7.el5.x86_64
ricci-0.12.0-7.el5.x86_64
openais-0.80.3-15.el5.x86_64

cluster.conf
<?xml version="1.0"?>
<cluster alias="tstw_HA2" config_version="115" name="tstw_HA2">
         <fence_daemon clean_start="0" post_fail_delay="0" 
post_join_delay="3"/>
         <clusternodes>
                 <clusternode name="tnode02.tst.tu-dresden.de" 
nodeid="1" votes="1">
                         <fence>
                                 <method name="1">
                                         <device name="HA_APC" port="2"/>
                                 </method>
                         </fence>
                 </clusternode>
                 <clusternode name="tnode01.tst.tu-dresden.de" 
nodeid="2" votes="1">
                         <fence>
                                 <method name="1">
                                         <device name="HA_APC" port="1"/>
                                 </method>
                         </fence>
                 </clusternode>
         </clusternodes>
         <cman expected_votes="1" two_node="1"/>
         <fencedevices>
                 <fencedevice agent="fence_apc" ipaddr="192.168.0.10" 
login="xxx" name="HA_APC" passwd="yy-xxxx"/>
         </fencedevices>
         <rm>
                 <failoverdomains>
                         <failoverdomain name="HA_new_failover" 
ordered="1" restricted="1">
                                 <failoverdomainnode 
name="tnode01.tst.tu-dresden.de" priority="1"/>
                                 <failoverdomainnode 
name="tnode02.tst.tu-dresden.de" priority="2"/>
                         </failoverdomain>
                 </failoverdomains>
                 <resources>
                         <clusterfs device="/dev/VG1/LV00" 
force_unmount="1" fsid="53422" fstype="gfs" mountpoint="/global_home" 
name="home_GFS" options=""/>
                         <nfsexport name="home_nfsexport"/>
                         <nfsclient name="tstw_home" 
options="rw,root_squash" path="/global_home" 
target="tstw*.tst.tu-dresden.de"/>
                         <ip address="111.22.33.32" monitor_link="1"/>
                         <ip address="192.168.20.30" monitor_link="1"/>
                         <nfsclient name="fast_nfs_home_clients" 
options="rw,root_squash" path="/global_home" target="192.168.20.0/24"/>
                         <nfsexport name="cluster_nfsexport"/>
                         <nfsclient name="tstw_cluster" 
options="no_root_squash,ro" path="/global_cluster" 
target="tstw*.tst.tu-dresden.de"/>
                         <nfsclient name="fast_nfs_cluster_clients" 
options="no_root_squash,ro" path="/global_cluster" 
target="192.168.20.0/24"/>
                         <script file="/etc/rc.d/init.d/tsm" 
name="TSM_backup"/>
                         <clusterfs device="/dev/VG1/LV10" 
force_unmount="1" fsid="192" fstype="gfs" mountpoint="/global_cluster" 
name="cluster_GFS" options=""/>
                         <clusterfs device="/dev/VG1/LV20" 
force_unmount="1" fsid="63016" fstype="gfs" mountpoint="/global_soft" 
name="software_GFS" options=""/>
                         <nfsexport name="soft_nfsexport"/>
                         <nfsclient name="tstw_soft" 
options="rw,root_squash" path="/global_soft" 
target="tstw*.tst.tu-dresden.de"/>
                         <nfsclient name="fast_nfs_soft_clients" 
options="rw,root_squash" path="/global_soft" target="192.168.20.0/24"/>
                         <nfsclient name="tsts_home" 
options="no_root_squash,rw" path="/global_home" 
target="tsts0*.tst.tu-dresden.de"/>
                         <nfsclient name="tsts_cluster" 
options="rw,root_squash" path="/global_cluster" 
target="tsts0*.tst.tu-dresden.de"/>
                         <nfsclient name="tsts_soft" 
options="rw,root_squash" path="/global_soft" 
target="tsts0*.tst.tu-dresden.de"/>
                         <nfsclient name="tstf_home" 
options="rw,root_squash" path="/global_home" 
target="tstf*.tst.tu-dresden.de"/>
                         <nfsclient name="tstf_cluster" 
options="rw,root_squash" path="/global_cluster" 
target="tstf*.tst.tu-dresden.de"/>
                         <nfsclient name="tstf_soft" 
options="rw,root_squash" path="/global_soft" 
target="tstf*.tst.tu-dresden.de"/>
                         <ip address="111.22.33.31" monitor_link="1"/>
                         <ip address="111.22.33.30" monitor_link="1"/>
                         <ip address="192.168.20.31" monitor_link="1"/>
                         <ip address="192.168.20.32" monitor_link="1"/>
                         <clusterfs device="/dev/VG1/LV20" 
force_unmount="0" fsid="11728" fstype="gfs" mountpoint="/global_soft" 
name="Software_GFS" options=""/>
                         <clusterfs device="/dev/VG1/LV10" 
force_unmount="0" fsid="36631" fstype="gfs" mountpoint="/global_cluster" 
name="Cluster_GFS" options=""/>
                         <clusterfs device="/dev/VG1/LV00" 
force_unmount="0" fsid="45816" fstype="gfs" mountpoint="/global_home" 
name="Home_GFS" options=""/>
                 </resources>
                 <service autostart="1" domain="HA_new_failover" 
name="service_nfs_home">
                         <nfsexport ref="home_nfsexport"/>
                         <nfsclient ref="tstw_home"/>
                         <ip ref="111.22.33.32"/>
                         <nfsclient ref="tsts_home"/>
                         <nfsclient ref="tstf_home"/>
                         <clusterfs ref="home_GFS"/>
                 </service>
                 <service autostart="1" domain="HA_new_failover" 
name="service_nfs_home_fast">
                         <nfsexport ref="home_nfsexport"/>
                         <nfsclient ref="fast_nfs_home_clients"/>
                         <ip ref="192.168.20.32"/>
                         <clusterfs ref="Home_GFS"/>
                 </service>
                 <service autostart="1" domain="HA_new_failover" 
name="service_nfs_cluster">
                         <nfsexport ref="cluster_nfsexport"/>
                         <nfsclient ref="tstw_cluster"/>
                         <nfsclient ref="tsts_cluster"/>
                         <nfsclient ref="tstf_cluster"/>
                         <ip ref="111.22.33.30"/>
                         <clusterfs ref="cluster_GFS"/>
                 </service>
                 <service autostart="1" name="service_nfs_cluster_fast">
                         <nfsexport ref="cluster_nfsexport"/>
                         <ip ref="192.168.20.30"/>
                         <nfsclient ref="fast_nfs_cluster_clients"/>
                         <clusterfs ref="Cluster_GFS"/>
                 </service>
                 <service autostart="1" domain="HA_new_failover" 
name="service_TSM">
                         <ip ref="111.22.33.31"/>
                         <script ref="TSM_backup"/>
                         <clusterfs ref="Software_GFS"/>
                         <clusterfs ref="Cluster_GFS"/>
                         <clusterfs ref="Home_GFS"/>
                 </service>
                 <service autostart="1" domain="HA_new_failover" 
name="service_nfs_soft">
                         <nfsexport ref="soft_nfsexport"/>
                         <nfsclient ref="tstw_soft"/>
                         <nfsclient ref="tsts_soft"/>
                         <nfsclient ref="tstf_soft"/>
                         <ip ref="111.22.33.31"/>
                         <clusterfs ref="software_GFS"/>
                 </service>
                 <service autostart="1" domain="HA_new_failover" 
name="service_nfs_soft_fast">
                         <nfsexport ref="soft_nfsexport"/>
                         <nfsclient ref="fast_nfs_soft_clients"/>
                         <ip ref="192.168.20.31"/>
                         <clusterfs ref="Software_GFS"/>
                 </service>
         </rm>
</cluster>

-- 
| R.Schwierz at physik.tu-dresden.de                     |
| Rainer  Schwierz, Inst. f. Kern- und Teilchenphysik |
| TU Dresden,       D-01062 Dresden                   |
| Tel. ++49 351 463 32957    FAX ++49 351 463 37292   |
| http://iktp.tu-dresden.de/~schwierz/                |



From george.saji00 at gmail.com  Wed Aug 19 15:54:42 2009
From: george.saji00 at gmail.com (saji george)
Date: Wed, 19 Aug 2009 21:24:42 +0530
Subject: [Linux-cluster] Apache Cluster
Message-ID: <2460eaad0908190854w6610a90ft9c6a9b10d576b7bf@mail.gmail.com>

Hi
  I have configured the GFS cluster successfully in RHEL 5.3
But I am not able to configure the cluster for apache.while enabling the
service I am getting the below error

# clusvcadm -e webcluster
Local machine trying to enable service:webcluster...Aborted; service failed

/log/messages

Aug 19 21:12:49 cmsstorage-web1 clurgmgrd[32766]: <err> #43: Service
service:webcluster has failed; can not start.
Aug 19 21:12:49 cmsstorage-web1 clurgmgrd[32766]: <crit> #13: Service
service:webcluster failed to stop cleanly

cluster.conf

<failoverdomain name="GFS-apache" ordered="1" restricted="1">
                                <failoverdomainnode name="cmsstorage-web3"
priority="3"/>
                                <failoverdomainnode name="cmsstorage-web2"
priority="2"/>
                                <failoverdomainnode name="cmsstorage-web1"
priority="1"/>
                        </failoverdomain>

<resources>
<apache config_file="conf/httpd.conf" httpd_options="" name="webcluster"
server_root="/usr/local/apache/" shutdown_wait="2"/>
</resources>

 <service autostart="1" domain="GFS-apache" name="webcluster"
recovery="restart">
                        <apache ref="webcluster"/>
                </service>

Is there anything missing...

Thanks in advance.

Regards,
Saji George
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090819/7c950961/attachment.htm>

From brem.belguebli at gmail.com  Wed Aug 19 17:05:35 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Wed, 19 Aug 2009 19:05:35 +0200
Subject: [Linux-cluster] Questions about rgmanager
In-Reply-To: <29ae894c0908181525n2ffad9b0m112f25fcd3b14ca5@mail.gmail.com>
References: <29ae894c0908180201l14ab18aexddefd0f04bf2ba29@mail.gmail.com>
	<1250631724.9711.251.camel@localhost.localdomain>
	<29ae894c0908181525n2ffad9b0m112f25fcd3b14ca5@mail.gmail.com>
Message-ID: <29ae894c0908191005y3dbe7152l5729e7f3b9bf1fa8@mail.gmail.com>

Hi,

Changing the cluster version reloads the rgmanager configuration, but
it restarts my custom resources despite the fact that they were "started"
(in clustat).

Just changing the cluster.conf version, not doing anything to my custom
resource makes this happen.

In the log I get the message  :
<info> Stopping changed resources
<info> Restarting changed resources
<info> Starting changed resources

which may be generic messages and then I see my resources being stopped and
restarted.

Sometimes, I also can notice that my service after being stopped, I have a
monitor action that is triggered that fails (normal service is stopped) and
that disables (status failed) completely the service.

I'm running Rhel 5.3 (I saw the post about VM's on RHel 5.2 which seems to
be similar).

I've tried to run rgmanager in debug (clurgmgrd -d) mode but it seems to get
stuck

Any idea on what I may be missing ?


2009/8/19, brem belguebli <brem.belguebli at gmail.com>:
>
> thx Lon
>
> 2009/8/18 Lon Hohberger <lhh at redhat.com>
>
>> On Tue, 2009-08-18 at 11:01 +0200, brem belguebli wrote:
>> > Hello,
>> >
>> > I have a few questions about rgmanager:
>> >
>> > 1) How to make it take into account resource metadata changes without
>> > restarting it. service rgmanager reload (or kill -HUP) doesn't.
>> >     In case one needs to change some timers in a resource metadata,
>> > the only way to make it is to stop and start rgmanager which brings
>> > down the resources.
>>
>>
>> Need to change the configuration version in cluster.conf; rgmanager
>> doesn't record for example file mtimes and run stat() to see if metadata
>> changed.  Doing this wouldn't work anyway, as some resources have
>> external metadata.
>>
>> > 2) Is there a way to log what "rules" rgmanager succeded to load at
>> > start time (or the ones it failed to), as having a "24 rules loaded"
>> > message isn't very explicit.
>>
>>
>> You can run:
>>
>>   rg_test rules | grep ^Agent
>>
>> There's no way currently to have rgmanager log specifically which agent
>> it loaded at run-time.
>>
>> > The last point concerns the potential replacement of rgmanager with
>> > pacemaker (as stated at
>> > http://sources.redhat.com/cluster/wiki/RGManager). Will this happen in
>> > a near future ?
>>
>>
>> Not immediately; we still have a bit to go:
>>
>> * need configuration file format conversion utility
>> * need basic "work-alike" tools.  Ex: clustat.
>> * need to do a bit of work on rgmanager agents so they work
>>  on pacemaker - the standard by which rgmanager+pacemaker
>>  operate was inadequate and the two programs have diverged
>>  in slightly incompatible ways
>>
>> -- Lon
>>
>> --
>> Linux-cluster mailing list
>> Linux-cluster at redhat.com
>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090819/0269b754/attachment.htm>

From thomas at sjolshagen.net  Wed Aug 19 17:41:39 2009
From: thomas at sjolshagen.net (Thomas Sjolshagen)
Date: Wed, 19 Aug 2009 13:41:39 -0400
Subject: [Linux-cluster] vm.sh resource script
In-Reply-To: <1250633230.9711.326.camel@localhost.localdomain>
References: <20090814164550.16372xa21kmg4xz2@www.sjolshagen.net>
	<1250633230.9711.326.camel@localhost.localdomain>
Message-ID: <20090819134139.1128410mcaene3pv@www.sjolshagen.net>


Quoting Lon Hohberger <lhh at redhat.com>:
> Can you file a bug so I don't forget to fix this for cluster-3.0.2 or
> 3.0.3 ?

I've filed Bugzilla 518277.

Hope it contains all the info you need.

// Thomas


----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.




From pradhanparas at gmail.com  Wed Aug 19 18:18:03 2009
From: pradhanparas at gmail.com (Paras pradhan)
Date: Wed, 19 Aug 2009 13:18:03 -0500
Subject: [Linux-cluster] Cluster behavior
In-Reply-To: <1250688308.17779.5.camel@localhost.localdomain>
References: <8b711df40908171459q1afecda7k7b0dcc4773153b84@mail.gmail.com>
	<1250688308.17779.5.camel@localhost.localdomain>
Message-ID: <8b711df40908191118l566d5333mc2a4673e5e6aa4ff@mail.gmail.com>

On Wed, Aug 19, 2009 at 8:25 AM, Lon Hohberger<lhh at redhat.com> wrote:
> On Mon, 2009-08-17 at 16:59 -0500, Paras pradhan wrote:
>> I have a 3 nodes cluster.
>>
>> Node A - Vote 1
>>
>> Node B - Vote 1
>>
>> Node C - Votes 2
>>
>> Qdisk - Votes 3
>>
>>
>> Altogether the cluster has 7 votes. The required min quorum to run the
>> cluster would be in this case 4. Now if I poweroff node 1, I can see
>> Quorum Dissolved in the terminals of Node 2 and Node3 . This cluster
>> has xen virtual machines.
>
> I assume node 1 == node A, node 2 == node B, node 3 == node C.
>
>
>> Whats wrong and how to do debug the problem?
>
> Do they all "See" each other before you power off node 1 (where 1 ==
> A) ? ?Apart from having a node with 2 votes, the rest of the concept is
> pretty similar to qdiskd's design for "all-but-1 die" case.
>
> Expected behavior is that any of the following constitutes a quorate
> cluster:
>
> A+Qdisk ? ? ?(4 votes)
> B+Qdisk ? ? ?(4 votes)
> C+Qdisk ? ? ?(4 votes)
> A+B+Qdisk ? ?(5 votes)
> A+B+C ? ? ? ?(4 votes)
> A+B+C+Qdisk ?(7 votes)
>
> Effectively, you want to look at 'cman_tool status'. ?Before losing a
> node, you need to verify that each sees 7 votes.

Yes before loosing a node.. its 7.. cman_tool status says its 7.

Here is the output when all the nodes are active.

On Node B or Node 2:

root at cvtst2 ~]# cman_tool status
Version: 6.1.0
Config Version: 54
Cluster Name: cvtst
Cluster Id: 3338
Cluster Member: Yes
Cluster Generation: 2068
Membership state: Cluster-Member
Nodes: 3
Expected votes: 6
Quorum device votes: 3
Total votes: 7
Quorum: 4
Active subsystems: 10
Flags: Dirty
Ports Bound: 0 11 177
Node name: cvtst2
Node ID: 1
Multicast addresses: x.x.x.x
Node addresses: x.x.x.172
[root at cvtst2 ~]#


Now Powered off node A or node 1.

Here is the output.

Version: 6.1.0
Config Version: 54
Cluster Name: cvtst
Cluster Id: 3338
Cluster Member: Yes
Cluster Generation: 2080
Membership state: Cluster-Member
Nodes: 2
Expected votes: 6
Total votes: 3
Quorum: 2
Active subsystems: 10
Flags: Dirty
Ports Bound: 0 11 177
Node name: cvtst1
Node ID: 2
Multicast addresses: x.x.x.x
Node addresses: x.x.x.172

I have noticed after poweroff the total votes turns to 3 ( for few
seconds), Then I see quoram dissolved . But afer few seconds the
cluster is back and I can see the total votes to 6. What  might be the
issue here.








>
> -- Lon
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>

Thanks!
Paras.



From brem.belguebli at gmail.com  Wed Aug 19 20:50:26 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Wed, 19 Aug 2009 22:50:26 +0200
Subject: [Linux-cluster] Questions about rgmanager
In-Reply-To: <29ae894c0908191005y3dbe7152l5729e7f3b9bf1fa8@mail.gmail.com>
References: <29ae894c0908180201l14ab18aexddefd0f04bf2ba29@mail.gmail.com>
	<1250631724.9711.251.camel@localhost.localdomain>
	<29ae894c0908181525n2ffad9b0m112f25fcd3b14ca5@mail.gmail.com>
	<29ae894c0908191005y3dbe7152l5729e7f3b9bf1fa8@mail.gmail.com>
Message-ID: <29ae894c0908191350g570936d3i22e3bba73086e424@mail.gmail.com>

Hi,

What is the behaviour of rgmanager supposed to be in case one adds a new LVM
 and FS resources to an already running service?
Let's say I have a running database service on one node, and I need to add
some storage to this service.

After creating all the LVM things, the filesystems,, mount them on the right
node etc...
I edit the cluster.conf, setup the lines concerning  what I've added, and
upgrade file version.

Running then ccs_tool update /etc/cluster/cluster.conf will it restart my
modified service ?
Brem


2009/8/19 brem belguebli <brem.belguebli at gmail.com>

> Hi,
>
> Changing the cluster version reloads the rgmanager configuration, but
> it restarts my custom resources despite the fact that they were "started"
> (in clustat).
>
> Just changing the cluster.conf version, not doing anything to my custom
> resource makes this happen.
>
> In the log I get the message  :
> <info> Stopping changed resources
> <info> Restarting changed resources
> <info> Starting changed resources
>
> which may be generic messages and then I see my resources being stopped and
> restarted.
>
> Sometimes, I also can notice that my service after being stopped, I have a
> monitor action that is triggered that fails (normal service is stopped) and
> that disables (status failed) completely the service.
>
> I'm running Rhel 5.3 (I saw the post about VM's on RHel 5.2 which seems to
> be similar).
>
> I've tried to run rgmanager in debug (clurgmgrd -d) mode but it seems to
> get stuck
>
> Any idea on what I may be missing ?
>
>
> 2009/8/19, brem belguebli <brem.belguebli at gmail.com>:
>
>> thx Lon
>>
>> 2009/8/18 Lon Hohberger <lhh at redhat.com>
>>
>>> On Tue, 2009-08-18 at 11:01 +0200, brem belguebli wrote:
>>> > Hello,
>>> >
>>> > I have a few questions about rgmanager:
>>> >
>>> > 1) How to make it take into account resource metadata changes without
>>> > restarting it. service rgmanager reload (or kill -HUP) doesn't.
>>> >     In case one needs to change some timers in a resource metadata,
>>> > the only way to make it is to stop and start rgmanager which brings
>>> > down the resources.
>>>
>>>
>>> Need to change the configuration version in cluster.conf; rgmanager
>>> doesn't record for example file mtimes and run stat() to see if metadata
>>> changed.  Doing this wouldn't work anyway, as some resources have
>>> external metadata.
>>>
>>> > 2) Is there a way to log what "rules" rgmanager succeded to load at
>>> > start time (or the ones it failed to), as having a "24 rules loaded"
>>> > message isn't very explicit.
>>>
>>>
>>> You can run:
>>>
>>>   rg_test rules | grep ^Agent
>>>
>>> There's no way currently to have rgmanager log specifically which agent
>>> it loaded at run-time.
>>>
>>> > The last point concerns the potential replacement of rgmanager with
>>> > pacemaker (as stated at
>>> > http://sources.redhat.com/cluster/wiki/RGManager). Will this happen in
>>> > a near future ?
>>>
>>>
>>> Not immediately; we still have a bit to go:
>>>
>>> * need configuration file format conversion utility
>>> * need basic "work-alike" tools.  Ex: clustat.
>>> * need to do a bit of work on rgmanager agents so they work
>>>  on pacemaker - the standard by which rgmanager+pacemaker
>>>  operate was inadequate and the two programs have diverged
>>>  in slightly incompatible ways
>>>
>>> -- Lon
>>>
>>> --
>>> Linux-cluster mailing list
>>> Linux-cluster at redhat.com
>>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>>
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090819/a54cb457/attachment.htm>

From fdinitto at redhat.com  Thu Aug 20 11:26:20 2009
From: fdinitto at redhat.com (Fabio M. Di Nitto)
Date: Thu, 20 Aug 2009 13:26:20 +0200
Subject: [Linux-cluster] Cluster 3.0.1 stable release
Message-ID: <1250767580.8148.12.camel@cerberus.int.fabbione.net>

The cluster team and its community are proud to announce the 3.0.1
stable release from the STABLE3 branch.

In order to build the 3.0.1 release you will need:

- corosync 1.0.0
- openais 1.0.0
- linux kernel 2.6.30

The new source tarball can be downloaded here:

ftp://sources.redhat.com/pub/cluster/releases/cluster-3.0.1.tar.gz
https://fedorahosted.org/releases/c/l/cluster/cluster-3.0.1.tar.gz

At the same location is now possible to find separated tarballs for
fence-agents and resource-agents as previously announced
(http://www.redhat.com/archives/cluster-devel/2009-February/msg00003.htm)

To report bugs or issues:

   https://bugzilla.redhat.com/

Would you like to meet the cluster team or members of its community?

   Join us on IRC (irc.freenode.net #linux-cluster) and share your
   experience  with other sysadministrators or power users.

Thanks/congratulations to all people that contributed to achieve this
great milestone.

Happy clustering,
Fabio

Under the hood (from 3.0.0):

Andrew Price (4):
      libgfs2: Fix 'dubious one-bit signed bitfield' sparse errors
      libgfs2: Fix "Value stored is never read" warnings
      gfs2_quota: Fix sparse error
      fsck.gfs2: Make block_mounters static

Bob Peterson (8):
      fsck.gfs2 segfaults while fixing 'EA leaf block type' problem.
      gfs2_edit produces unaligned access
      GFS: Commited data evaporates
      Fix more man page references to gfs2_fsck
      "fsck.gfs2: invalid option -- a" on boot when mounting gfs2 root
      GFS2: gfs2_fsck segfault in rindex repair code
      GFS2: fsck.gfs2 sometimes needs to be run twice
      GFS: gfs_fsck sometimes needs to be run twice

Christine Caulfield (11):
      cman: Improve cman_tool startup error reporting
      cman: make cman_tool leave remove work even when no services are
running.
      cman: Fix a situation where cman could kill the wrong nodes
      cman: Enable error reporting from config modules
      cman: Tidy the shutdown code.
      config: Fix up some ccs_tool anachronisms
      config: Remove all references to ccsd from ccs_tool
      cman: Make disallowed state optional.
      cman: Add a single compatibilty switch
      cman: fix error if corosync takes >1s to start up
      cman: More join wait fixes

David Teigland (13):
      dlm_controld/gfs_controld: GETLK should free unused resource
      doc: update usage.txt
      doc: remove old gfs docs
      cluster.rng: various fixes
      cluster.rng: remove trailing whitespace
      cluster.rng: cluster_id and nodeid
      cluster.rng: move blocks around
      cluster.rng: comments around fence devices
      cluster.rng: dlm and gfs_controld descriptions
      cman: wait for quorum
      man page: cluster.conf
      cluster.rng: add stuff
      man page: cman

Fabio M. Di Nitto (13):
      build: drop dtest target in rgmanager daemon Makefile
      rgmanager: drop obsoleted tools
      update copyright file
      build: drop dependency on readline
      dlm_controld: fix protocol initialization
      fence agents: drop unrequired and wrong copyright statement
      gfs-kernel: update to 2.6.30
      build: make gfs.ko target phony and delegate everything to kbuild
      dlm: fix .pc file for libdlm to link against pthread
      build: update dependency tree
      fence: make libfenced shared
      cman init: don't wait for quorum twice
      cman init: fix cman startup issues

Florian Haas (1):
      rgmanager: Add resource agent for DRBD

Lars Marowsky-Bree (1):
      dlm_controld: fix skipping lock spaces

Lon Hohberger (5):
      rgmanager: Don't allow migrate of frozen services
      qdisk: Do not do a clean logout when exiting w/ error
      rgmanager: Fix clusterfs.sh to use meta_refcnt correctly
      rgmanager: Preserve incarnations across config changes
      rgmanager: Fix incarnation handling

Marek 'marx' Grac (7):
      fencing: Create a function that prints documentation
      fencing: Metadata suitable for automatic man pages generation
      fence_lpar: option -x for ssh was required
      fencing: Fencing fails if telnet/ssh is not available
      fencing.py: Adds device option login_eol_lf
      fencing: fence agents fails with pexpect exception
      [fencing] Fence agent for HP iLO2 MP

Paul Kennedy (1):
      cluster.rng: add descriptions

Ryan O'Hara (1):
      BZ 512998 - Add comment about fence_scsi limitations.

Shane Bradley (1):
      rgmanager: Assume 'no state' is OK for VMs

 Makefile                                        |    2 +-
 cman/cman_tool/join.c                           |   33 +-
 cman/cman_tool/main.c                           |    2 +
 cman/daemon/ais.c                               |    2 +
 cman/daemon/cman-preconfig.c                    |  103 ++-
 cman/daemon/cman.h                              |    1 +
 cman/daemon/cnxman-socket.h                     |    1 +
 cman/daemon/commands.c                          |  100 ++-
 cman/init.d/cman.in                             |   51 +-
 cman/lib/libcman.h                              |    1 +
 cman/man/cman.5                                 |   11 -
 cman/qdisk/main.c                               |    6 +-
 config/man/cluster.conf.5                       |    2 +-
 config/tools/ccs_tool/editconf.c                |  183 ++--
 config/tools/xml/cluster.rng                    | 1203
+++++++++++++++--------
 configure                                       |   16 +-
 dlm/libdlm/libdlm.pc.in                         |    2 +-
 doc/COPYRIGHT                                   |    4 -
 doc/Makefile                                    |    5 +-
 doc/gfs2.txt                                    |   45 -
 doc/journaling.txt                              |  155 ---
 doc/min-gfs.txt                                 |  159 ---
 doc/usage.txt                                   |  201 ++---
 fence/agents/alom/fence_alom.py                 |    4 +-
 fence/agents/apc/fence_apc.py                   |    2 +
 fence/agents/apc_snmp/fence_apc_snmp.py         |    2 +
 fence/agents/bladecenter/fence_bladecenter.py   |    4 +-
 fence/agents/cisco_mds/fence_cisco_mds.py       |    2 +
 fence/agents/drac/fence_drac5.py                |    4 +-
 fence/agents/eps/fence_eps.py                   |    2 +
 fence/agents/ibmblade/fence_ibmblade.py         |    2 +
 fence/agents/ifmib/fence_ifmib.py               |    2 +
 fence/agents/ilo/fence_ilo.py                   |    2 +
 fence/agents/ilo_mp/Makefile                    |    5 +
 fence/agents/ilo_mp/fence_ilo_mp.py             |   61 ++
 fence/agents/intelmodular/fence_intelmodular.py |    2 +
 fence/agents/ldom/fence_ldom.py                 |    2 +
 fence/agents/lib/fencing.py.py                  |  149 +++-
 fence/agents/lpar/fence_lpar.py                 |    8 +-
 fence/agents/rsa/fence_rsa.py                   |    4 +-
 fence/agents/virsh/fence_virsh.py               |    4 +-
 fence/agents/vmware/fence_vmware.py             |    4 +-
 fence/agents/wti/fence_wti.py                   |   13 +-
 fence/libfenced/Makefile                        |    2 -
 fence/libfenced/libfenced.h                     |   13 +
 fence/libfenced/libfenced.pc.in                 |   11 +
 fence/man/fence_scsi.8                          |    7 +
 gfs-kernel/src/gfs/Makefile                     |    2 +
 gfs-kernel/src/gfs/acl.c                        |    1 +
 gfs-kernel/src/gfs/ops_file.c                   |    2 +
 gfs-kernel/src/gfs/proc.c                       |    1 -
 gfs/gfs_fsck/block_list.c                       |   24 +-
 gfs/gfs_fsck/block_list.h                       |    3 +-
 gfs/gfs_fsck/fs_dir.c                           |   23 +-
 gfs/gfs_fsck/fs_dir.h                           |    2 +
 gfs/gfs_fsck/metawalk.c                         |  457 ++++++---
 gfs/gfs_fsck/metawalk.h                         |   13 +-
 gfs/gfs_fsck/pass1.c                            |  630 +++++++------
 gfs/gfs_fsck/pass1b.c                           |  115 ++-
 gfs/gfs_fsck/pass1c.c                           |  175 ++--
 gfs/gfs_fsck/pass2.c                            |  272 +++---
 gfs/gfs_fsck/pass3.c                            |   21 +-
 gfs/gfs_fsck/pass4.c                            |   47 +-
 gfs/gfs_fsck/pass5.c                            |   32 +-
 gfs/gfs_fsck/rgrp.c                             |    3 +-
 gfs2/edit/hexedit.c                             |    5 +-
 gfs2/edit/savemeta.c                            |   10 +-
 gfs2/fsck/fs_recovery.c                         |  101 ++-
 gfs2/fsck/fs_recovery.h                         |    4 +-
 gfs2/fsck/fsck.h                                |    4 +-
 gfs2/fsck/initialize.c                          |   52 +-
 gfs2/fsck/main.c                                |   33 +-
 gfs2/fsck/metawalk.c                            |  289 +++++--
 gfs2/fsck/metawalk.h                            |   15 +-
 gfs2/fsck/pass1.c                               |  373 ++++---
 gfs2/fsck/pass1b.c                              |   31 +-
 gfs2/fsck/pass1c.c                              |  215 ++---
 gfs2/fsck/pass2.c                               |   63 +-
 gfs2/fsck/pass4.c                               |   38 +-
 gfs2/fsck/pass5.c                               |   39 +-
 gfs2/fsck/rgrepair.c                            |    8 +-
 gfs2/libgfs2/fs_geometry.c                      |    1 -
 gfs2/libgfs2/gfs1.c                             |    3 -
 gfs2/libgfs2/libgfs2.h                          |    6 +-
 gfs2/libgfs2/misc.c                             |    6 +-
 gfs2/man/fsck.gfs2.8                            |   31 +-
 gfs2/man/gfs2_edit.8                            |    4 +-
 gfs2/man/gfs2_grow.8                            |    2 +-
 gfs2/quota/gfs2_quota.h                         |    4 +-
 gfs2/quota/main.c                               |    4 +-
 group/dlm_controld/config.c                     |    4 +-
 group/dlm_controld/cpg.c                        |   12 +-
 group/dlm_controld/plock.c                      |    1 +
 group/gfs_controld/plock.c                      |    1 +
 make/defines.mk.input                           |    2 -
 rgmanager/include/reslist.h                     |    1 +
 rgmanager/man/cluarp.8                          |   15 -
 rgmanager/man/clunfsops.8                       |   14 -
 rgmanager/src/daemons/Makefile                  |   42 +-
 rgmanager/src/daemons/groups.c                  |   27 +
 rgmanager/src/daemons/reslist.c                 |    1 +
 rgmanager/src/daemons/restree.c                 |   20 +-
 rgmanager/src/daemons/rg_state.c                |    5 +
 rgmanager/src/resources/Makefile                |    4 +-
 rgmanager/src/resources/clusterfs.sh            |    3 +-
 rgmanager/src/resources/drbd.metadata           |   51 +
 rgmanager/src/resources/drbd.sh                 |  144 +++
 rgmanager/src/resources/nfsexport.sh            |    7 -
 rgmanager/src/resources/vm.sh                   |    2 +-
 rgmanager/src/utils/Makefile                    |    2 -
 rgmanager/src/utils/cluarp.c                    |  158 ---
 rgmanager/src/utils/clunfsops.c                 |  224 -----
 rgmanager/src/utils/syscall.h                   |  165 ----
 113 files changed, 3660 insertions(+), 3008 deletions(-)

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090820/2abfbd92/attachment.sig>

From brem.belguebli at gmail.com  Thu Aug 20 13:56:08 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Thu, 20 Aug 2009 15:56:08 +0200
Subject: [Linux-cluster] Questions about rgmanager
In-Reply-To: <29ae894c0908191005y3dbe7152l5729e7f3b9bf1fa8@mail.gmail.com>
References: <29ae894c0908180201l14ab18aexddefd0f04bf2ba29@mail.gmail.com>
	<1250631724.9711.251.camel@localhost.localdomain>
	<29ae894c0908181525n2ffad9b0m112f25fcd3b14ca5@mail.gmail.com>
	<29ae894c0908191005y3dbe7152l5729e7f3b9bf1fa8@mail.gmail.com>
Message-ID: <29ae894c0908200656g3f35ca77u552865e94244fff0@mail.gmail.com>

Solved,

I was missing a primary key in my metadata.

Brem


2009/8/19, brem belguebli <brem.belguebli at gmail.com>:
>
> Hi,
>
> Changing the cluster version reloads the rgmanager configuration, but
> it restarts my custom resources despite the fact that they were "started"
> (in clustat).
>
> Just changing the cluster.conf version, not doing anything to my custom
> resource makes this happen.
>
> In the log I get the message  :
> <info> Stopping changed resources
> <info> Restarting changed resources
> <info> Starting changed resources
>
> which may be generic messages and then I see my resources being stopped and
> restarted.
>
> Sometimes, I also can notice that my service after being stopped, I have a
> monitor action that is triggered that fails (normal service is stopped) and
> that disables (status failed) completely the service.
>
> I'm running Rhel 5.3 (I saw the post about VM's on RHel 5.2 which seems to
> be similar).
>
> I've tried to run rgmanager in debug (clurgmgrd -d) mode but it seems to
> get stuck
>
> Any idea on what I may be missing ?
>
>
> 2009/8/19, brem belguebli <brem.belguebli at gmail.com>:
>>
>> thx Lon
>>
>> 2009/8/18 Lon Hohberger <lhh at redhat.com>
>>
>>> On Tue, 2009-08-18 at 11:01 +0200, brem belguebli wrote:
>>> > Hello,
>>> >
>>> > I have a few questions about rgmanager:
>>> >
>>> > 1) How to make it take into account resource metadata changes without
>>> > restarting it. service rgmanager reload (or kill -HUP) doesn't.
>>> >     In case one needs to change some timers in a resource metadata,
>>> > the only way to make it is to stop and start rgmanager which brings
>>> > down the resources.
>>>
>>>
>>> Need to change the configuration version in cluster.conf; rgmanager
>>> doesn't record for example file mtimes and run stat() to see if metadata
>>> changed.  Doing this wouldn't work anyway, as some resources have
>>> external metadata.
>>>
>>> > 2) Is there a way to log what "rules" rgmanager succeded to load at
>>> > start time (or the ones it failed to), as having a "24 rules loaded"
>>> > message isn't very explicit.
>>>
>>>
>>> You can run:
>>>
>>>   rg_test rules | grep ^Agent
>>>
>>> There's no way currently to have rgmanager log specifically which agent
>>> it loaded at run-time.
>>>
>>> > The last point concerns the potential replacement of rgmanager with
>>> > pacemaker (as stated at
>>> > http://sources.redhat.com/cluster/wiki/RGManager). Will this happen in
>>> > a near future ?
>>>
>>>
>>> Not immediately; we still have a bit to go:
>>>
>>> * need configuration file format conversion utility
>>> * need basic "work-alike" tools.  Ex: clustat.
>>> * need to do a bit of work on rgmanager agents so they work
>>>  on pacemaker - the standard by which rgmanager+pacemaker
>>>  operate was inadequate and the two programs have diverged
>>>  in slightly incompatible ways
>>>
>>> -- Lon
>>>
>>> --
>>> Linux-cluster mailing list
>>> Linux-cluster at redhat.com
>>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>>
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090820/c10c27b7/attachment.htm>

From linux-cluster at lists.grepular.com  Thu Aug 20 14:02:45 2009
From: linux-cluster at lists.grepular.com (Mike Cardwell)
Date: Thu, 20 Aug 2009 15:02:45 +0100
Subject: [Linux-cluster] Debian cluster suite packages
Message-ID: <20090820140245.GA15122@127.0.0.1>

Hi,

root at haven:~# mkqdisk -c /dev/sdb1 -l qdisk
mkqdisk v2.03.09

Writing new quorum disk label 'qdisk' to /dev/sdb1.
WARNING: About to destroy all data on /dev/sdb1; proceed [N/y] ? y
Warning: Initializing previously initialized partition
Initializing status block for node 1...
Initializing status block for node 2...
Initializing status block for node 3...
Initializing status block for node 4...
Initializing status block for node 5...
Initializing status block for node 6...
Initializing status block for node 7...
Initializing status block for node 8...
Initializing status block for node 9...
Initializing status block for node 10...
Initializing status block for node 11...
Initializing status block for node 12...
Initializing status block for node 13...
Initializing status block for node 14...
Initializing status block for node 15...
Initializing status block for node 16...
root at haven:~# qdiskd -fd
[15191] debug: Loading configuration information
[15191] debug: Heuristic: 'ping iscsi-target -c1 -t2' score=1 interval=2 tko=3
[15191] debug: 1 heuristics loaded
[15191] debug: Quorum Daemon: 1 heuristics, 1 interval, 10 tko, 1 votes
[15191] debug: Run Flags: 00000034
[15191] crit: Specified device /dev/sdb1 does match kernel's reported sector size (0 != -1)
root at haven:~#

I *think* that this is an old bug which was fixed in January here:
https://bugzilla.redhat.com/show_bug.cgi?id=470553

However. I'm using the latest Debian Lenny packaged version of the
various cluster suite tools:

root at haven:~# dpkg -l|grep 'Red Hat cluster suite'
ii  cman                                 2.20081102-1               Red Hat cluster suite - cluster manager
ii  gfs2-tools                           2.20081102-1               Red Hat cluster suite - global file system 2
ii  libcman2                             2.20081102-1               Red Hat cluster suite - cluster manager libr
ii  libdlm2                              2.20081102-1               Red Hat cluster suite - distributed lock man
root at haven:~#

So. I have two questions.

1.) Am I seeing the bug which was fixed by that bugzilla report?
2.) Does anyone know where I can get some newer Debian packages? I
*could* roll my own if necessary but I'm hoping I don't have to...

-- 
Mike Cardwell - IT Consultant and LAMP developer
Cardwell IT Ltd. (UK Reg'd Company #06920226) http://cardwellit.com/



From pradhanparas at gmail.com  Thu Aug 20 22:23:50 2009
From: pradhanparas at gmail.com (Paras pradhan)
Date: Thu, 20 Aug 2009 17:23:50 -0500
Subject: [Linux-cluster] Cluster and Network outage
Message-ID: <8b711df40908201523t48d0f942md9930fc3a41e0aea@mail.gmail.com>

Yesterday for around 14 minutes we have a network outage. Today
morning I saw that my redhat Linux cluster was stopped and had to
start the cluster from Conga. It was fine afterwards. I have a qdiskd
in which I  am using heuristics as well. When there was an outage I
think it was not able to ping my router (which I added to heuristics).
Does this bring my whole cluster down?

In Log I don't see anything interesting..

It says heuristics ip --- DOWN

and this
Aug 19 22:52:47 cvtst1 ccsd[3310]: Unable to connect to cluster
infrastructure after 210 seconds.

We usually have network outages. (2 or 3 in a month) How to get rid of this?

And does the network outage brings the cluster to stopped stage?

Thanks!
Paras.



From tageorgiou at gmail.com  Fri Aug 21 03:11:44 2009
From: tageorgiou at gmail.com (Thomas Georgiou)
Date: Thu, 20 Aug 2009 23:11:44 -0400
Subject: [Linux-cluster] DLM Issues
Message-ID: <6e4c20e70908202011y261bfa97i17bbf6acc85b9c40@mail.gmail.com>

Hi,
I am having trouble trying to get clvm working with corosync.
When I try to start clvmd, I get
  [  451.167435] dlm: no local IP address has been set
  [  451.167438] dlm: cannot start dlm lowcomms -107
and clvmd reports
  CLVMD[b9ae86f0]: Aug 20 23:10:13 CLVMD started
  CLVMD[b9ae86f0]: Aug 20 23:10:13 Can't initialise cluster interface
    Can't initialise cluster interface

I then tried starting dlm_controld, but I get errors with that to.
altair ~ # dlm_controld -D
1250824114 logging mode 3 syslog f 160 p 6 logfile p 6
/var/log/cluster/dlm_controld.log
1250824114 dlm_controld 3.0.1 started
1250824116 cman_admin_init error 2
1250824116 /sys/kernel/config/dlm/cluster/comms: opendir failed: 2
1250824116 /sys/kernel/config/dlm/cluster/spaces: opendir failed: 2
dlm is loaded
altair ~ # lsmod
Module                  Size  Used by
dlm                   123280  0

Any ideas?
This is using Linux 2.6.29 running Gentoo with openais,corosync,lvm2,
and cluster 3.0.1 installed manually (through custom ebuilds).

Thomas Georgiou



From fdinitto at redhat.com  Fri Aug 21 06:00:47 2009
From: fdinitto at redhat.com (Fabio M. Di Nitto)
Date: Fri, 21 Aug 2009 08:00:47 +0200
Subject: [Linux-cluster] cluster-3.0.0 problems starting up
In-Reply-To: <4A85E1C7.1070502@etrafficsolutions.com>
References: <4A85E1C7.1070502@etrafficsolutions.com>
Message-ID: <1250834447.8148.25.camel@cerberus.int.fabbione.net>

On Fri, 2009-08-14 at 15:14 -0700, Trevor Peirce wrote:
> Hello,
> 
> I've attempted to compile and install the cluster-3.0.0 suite and have 
> these problems when trying to start cman:
> 
> Aug 14 12:10:01 localhost corosync[27739]: parse error in config: parse 
> error in config: .
> Aug 14 12:10:01 localhost corosync[27739]: [MAIN ] Corosync Cluster 
> Engine ('trunk'): started and ready to provide service.
> Aug 14 12:10:01 localhost corosync[27739]: [MAIN ] Successfully read 
> config from /etc/cluster/cluster.conf
> Aug 14 12:10:01 localhost corosync[27739]: [MAIN ] Successfully parsed 
> cman config
> Aug 14 12:10:01 localhost corosync[27739]: [MAIN ] Successfully 
> configured openais services to load
> Aug 14 12:10:01 localhost corosync[27739]: [MAIN ] parse error in 
> config: parse error in config: .
> Aug 14 12:10:01 localhost corosync[27739]: [MAIN ] Corosync Cluster 
> Engine exiting with status -9 at main.c:879.
> 
> Here is cluster.conf, as generated by ccs_tool:
> 
> <?xml version="1.0"?>
> <cluster name="TestCluster" config_version="3">
> <clusternodes><clusternode name="guinneapig" votes="1" 
> nodeid="1"><fence><method name="single"><device 
> name="manual"/></method></fence></clusternode></clusternodes>
> <fencedevices><fencedevice name="manual" agent="manual"/></fencedevices>
> <rm>
> <failoverdomains/>
> <resources/>
> </rm>
> </cluster>

Check your hostname setting.

The name you use for clusternode _MUST_ resolve (either via DNS
or /etc/hosts) to the ip address of an ethernet interface. That hostname
set to localhost is suspicious.

Fabio



From fdinitto at redhat.com  Fri Aug 21 06:04:22 2009
From: fdinitto at redhat.com (Fabio M. Di Nitto)
Date: Fri, 21 Aug 2009 08:04:22 +0200
Subject: [Linux-cluster] Debian cluster suite packages
In-Reply-To: <20090820140245.GA15122@127.0.0.1>
References: <20090820140245.GA15122@127.0.0.1>
Message-ID: <1250834662.8148.29.camel@cerberus.int.fabbione.net>

On Thu, 2009-08-20 at 15:02 +0100, Mike Cardwell wrote:
> Hi,

> 
> I *think* that this is an old bug which was fixed in January here:
> https://bugzilla.redhat.com/show_bug.cgi?id=470553
> 
> However. I'm using the latest Debian Lenny packaged version of the
> various cluster suite tools:
> 
> root at haven:~# dpkg -l|grep 'Red Hat cluster suite'
> ii  cman                                 2.20081102-1               Red Hat cluster suite - cluster manager
> ii  gfs2-tools                           2.20081102-1               Red Hat cluster suite - global file system 2
> ii  libcman2                             2.20081102-1               Red Hat cluster suite - cluster manager libr
> ii  libdlm2                              2.20081102-1               Red Hat cluster suite - distributed lock man
> root at haven:~#
> 
> So. I have two questions.
> 
> 1.) Am I seeing the bug which was fixed by that bugzilla report?

Seems likely the same. The bug was fixed the 7th of Nov 2008. The
snapshot in debian is 5 days older.

> 2.) Does anyone know where I can get some newer Debian packages? I
> *could* roll my own if necessary but I'm hoping I don't have to...

You should file a bug on bugs.debian.org and get the debian maintainers
sort it out.

The patch is in git: c90643c8824ff2f12e2debc05f32f981facfcbd3

With those info it should be trivial for them to address it.

Fabio



From fdinitto at redhat.com  Fri Aug 21 06:06:30 2009
From: fdinitto at redhat.com (Fabio M. Di Nitto)
Date: Fri, 21 Aug 2009 08:06:30 +0200
Subject: [Linux-cluster] DLM Issues
In-Reply-To: <6e4c20e70908202011y261bfa97i17bbf6acc85b9c40@mail.gmail.com>
References: <6e4c20e70908202011y261bfa97i17bbf6acc85b9c40@mail.gmail.com>
Message-ID: <1250834790.8148.32.camel@cerberus.int.fabbione.net>

Hi,

On Thu, 2009-08-20 at 23:11 -0400, Thomas Georgiou wrote:
> Hi,
> I am having trouble trying to get clvm working with corosync.
> When I try to start clvmd, I get
>   [  451.167435] dlm: no local IP address has been set
>   [  451.167438] dlm: cannot start dlm lowcomms -107
> and clvmd reports
>   CLVMD[b9ae86f0]: Aug 20 23:10:13 CLVMD started
>   CLVMD[b9ae86f0]: Aug 20 23:10:13 Can't initialise cluster interface
>     Can't initialise cluster interface
> 
> I then tried starting dlm_controld, but I get errors with that to.
> altair ~ # dlm_controld -D
> 1250824114 logging mode 3 syslog f 160 p 6 logfile p 6
> /var/log/cluster/dlm_controld.log
> 1250824114 dlm_controld 3.0.1 started
> 1250824116 cman_admin_init error 2
> 1250824116 /sys/kernel/config/dlm/cluster/comms: opendir failed: 2
> 1250824116 /sys/kernel/config/dlm/cluster/spaces: opendir failed: 2
> dlm is loaded
> altair ~ # lsmod
> Module                  Size  Used by
> dlm                   123280  0
> 
> Any ideas?

Do one step at a time.

First of all check if cman_tool nodes output is sane and nodes can see
eachother.

If that works, then you need to make sure that dlm.ko is loaded and
configfs is mounted in the right place.

Basically follow the init script we provide step by step and see where
it starts failing.

Fabio



From fdinitto at redhat.com  Fri Aug 21 06:09:23 2009
From: fdinitto at redhat.com (Fabio M. Di Nitto)
Date: Fri, 21 Aug 2009 08:09:23 +0200
Subject: [Linux-cluster] Re: Submitting a mdadm resource script
In-Reply-To: <29ae894c0908170707p4f42cd78j24592b02a6e2983@mail.gmail.com>
References: <29ae894c0908170322u447ff473s5b0512e6b37267af@mail.gmail.com>
	<29ae894c0908170707p4f42cd78j24592b02a6e2983@mail.gmail.com>
Message-ID: <1250834963.8148.34.camel@cerberus.int.fabbione.net>


Hi Brem,

thanks for the submission.

Lon can you please take a look at this resource agent?

Fabio

On Mon, 2009-08-17 at 16:07 +0200, brem belguebli wrote:
> Hi,
>  
> there was  a bug in my script that occurs when debugging was set to
> off.
>  
> Please find attached  the corrected version.
>  
> Regards
> 
>  
> 2009/8/17, brem belguebli <brem.belguebli at gmail.com>: 
>         Hi,
>          
>         I have written a resource script (mdmonitor.sh) to manage
>         mdadm mirrors across a cluster.
>          
>         This is for testing purposes and still needs some additionnal
>         features and some deeper debugging.
>          
>         If someone is interested in hacking it to improve it or to
>         correct it, I'll be glad.
>          
>         A readme file is provided that explains the goal of this
>         resource script.
>          
>         Thanks to Rafael Mico Miranda for his lvm-cluster resource
>         script used on top of mdmonitor.
>          
>          Brem
>         
> 
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster



From fdinitto at redhat.com  Fri Aug 21 06:11:10 2009
From: fdinitto at redhat.com (Fabio M. Di Nitto)
Date: Fri, 21 Aug 2009 08:11:10 +0200
Subject: [Linux-cluster] fence_node fails for fence_wti agent - cluster 3.0
In-Reply-To: <20090814221316.89106gn410mavk58@www.sjolshagen.net>
References: <20090814221316.89106gn410mavk58@www.sjolshagen.net>
Message-ID: <1250835070.8148.36.camel@cerberus.int.fabbione.net>

Marek any idea on this problem?

Fabio

On Fri, 2009-08-14 at 22:13 -0400, Thomas Sjolshagen wrote:
> Hi,
> 
> I've got a WTI NPS power switch that I've configured as a fence device  
> for my 2-node cluster. If I call 'fence_wti' directly with  
> action=status or action=reboot, it does what I expect it to do; fence  
> the node I'm telling it to fence.
> 
> However, if I issue "fence_node -vv <node>", the command fails with  
> the following error:
> 
> # fence_node -vv node1
> fence node1 dev 0.0 agent fence_wti result: error from agent
> agent args: plug=1 agent=fence_wti ipaddr=1.1.1.1  
> debug-file=/tmp/wti-fence.debug inet4-only=1 verbose=1 ipport=23  
> action=reboot
> fence node1 failed
> 
> Also, there is no /tmp/wti-fence* file and the only entry in the error  
> logs is in /var/log/messages:
> 
> fence_node[19172]: fence virt1-backup.sjolshagen.net failed
> 
> Suggestions on how I can further troubleshoot this or fix the problem?
> 
> Thanks in advance,
> 
> // Thomas
> 
> My cluster.conf file contains the following fencedevice entry:
> 
> <fencedevices>
>     <!-- Definition of fence device named "power", referenced previously -->
>     <fencedevice agent="fence_wti" name="power" ipaddr="1.1.1.1"  
> debug-file="/tmp/wti-fence.debug" inet4-only="1" verbose="1"  
> ipport="23" action="reboot"/>
> </fencedevices>
> `
> And for the clusternodes definitions:
> 
> <clusternodes>
>          <clusternode name="node2" nodeid="1" votes="1">
>                  <fence>
>                          <method name="1">
>                                  <device name="power" plug="1"/>
>                          </method>
>                  </fence>
>          </clusternode>
>          <clusternode name="node2" nodeid="3" votes="1">
>                  <fence>
>                          <method name="1">
>                                  <device name="power" plug="8"/>
>                          </method>
>                  </fence>
>          </clusternode>
> </clusternodes>
> 
> 
> 
> ----------------------------------------------------------------
> This message was sent using IMP, the Internet Messaging Program.
> 
> 
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster



From brem.belguebli at gmail.com  Fri Aug 21 08:41:39 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Fri, 21 Aug 2009 10:41:39 +0200
Subject: [Linux-cluster] Re: Submitting a mdadm resource script
In-Reply-To: <1250834963.8148.34.camel@cerberus.int.fabbione.net>
References: <29ae894c0908170322u447ff473s5b0512e6b37267af@mail.gmail.com>
	<29ae894c0908170707p4f42cd78j24592b02a6e2983@mail.gmail.com>
	<1250834963.8148.34.camel@cerberus.int.fabbione.net>
Message-ID: <29ae894c0908210141k7f227c87vdee5478867d536dc@mail.gmail.com>

Hi Fabio,

I have corrected a few bugs.

The new version (2008-08-20.20:30) is attached. It contains the script
itself (metadata is now inside the script) and the changed readme.


Thanks and regards.

Brem


2009/8/21, Fabio M. Di Nitto <fdinitto at redhat.com>:
>
>
> Hi Brem,
>
> thanks for the submission.
>
> Lon can you please take a look at this resource agent?
>
> Fabio
>
> On Mon, 2009-08-17 at 16:07 +0200, brem belguebli wrote:
> > Hi,
> >
> > there was  a bug in my script that occurs when debugging was set to
> > off.
> >
> > Please find attached  the corrected version.
> >
> > Regards
> >
> >
> > 2009/8/17, brem belguebli <brem.belguebli at gmail.com>:
> >         Hi,
> >
> >         I have written a resource script (mdmonitor.sh) to manage
> >         mdadm mirrors across a cluster.
> >
> >         This is for testing purposes and still needs some additionnal
> >         features and some deeper debugging.
> >
> >         If someone is interested in hacking it to improve it or to
> >         correct it, I'll be glad.
> >
> >         A readme file is provided that explains the goal of this
> >         resource script.
> >
> >         Thanks to Rafael Mico Miranda for his lvm-cluster resource
> >         script used on top of mdmonitor.
> >
> >          Brem
> >
> >
> > --
> > Linux-cluster mailing list
> > Linux-cluster at redhat.com
> > https://www.redhat.com/mailman/listinfo/linux-cluster
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090821/c90ffb12/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: mdmonitor.sh
Type: application/x-sh
Size: 37162 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090821/c90ffb12/attachment.sh>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: mdmonitor.README
Type: application/octet-stream
Size: 12027 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090821/c90ffb12/attachment.obj>

From brem.belguebli at gmail.com  Fri Aug 21 09:27:07 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Fri, 21 Aug 2009 11:27:07 +0200
Subject: [Linux-cluster] Fencing question in geo cluster (dual sites
	clustering)
Message-ID: <29ae894c0908210227r85df80fm173af6452d22a5b2@mail.gmail.com>

Hi,

I'm trying to find out what best fencing solution could fit a dual sites
cluster.

Cluster is equally sized on each site (2 nodes/site), each site hosting a
SAN array so that each node from any site can see the 2 arrays.

Quorum  disk (iscsi LUN) is hosted on a 3rd site.

SAN and LAN using the same telco infrastructure (2 redundant DWDM loops).

In case something happens at Telco level (both DWDM loops are broken) that
makes 1 of the 2 sites completely isolated from the rest of the world,
the nodes at the good site (the one still operationnal) won't be able to
fence any node from the wrong site (the one that is isolated) as there is no
way for them to reach their ILO's or do any SAN fencing as the switches at
the wrong site are no more reachable.

As qdiskd is not reachable from the wrong nodes, they end up being rebooted
by  qdisk, but there is a short time (a few seconds) during which the wrong
nodes are still seing their local SAN array storage and may potentially have
written data on it.

Any ideas or comments on how to ensure data integrity in such setup ?

Regards

Brem
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090821/f8dbed47/attachment.htm>

From cryptogrid at gmail.com  Fri Aug 21 19:26:51 2009
From: cryptogrid at gmail.com (crypto grid)
Date: Fri, 21 Aug 2009 16:26:51 -0300
Subject: [Linux-cluster] Apache Cluster
In-Reply-To: <2460eaad0908190854w6610a90ft9c6a9b10d576b7bf@mail.gmail.com>
References: <2460eaad0908190854w6610a90ft9c6a9b10d576b7bf@mail.gmail.com>
Message-ID: <a9f464b80908211226h39b5d60av7c2c07823b72c778@mail.gmail.com>

On Wed, Aug 19, 2009 at 12:54 PM, saji george <george.saji00 at gmail.com>wrote:

> Hi
>   I have configured the GFS cluster successfully in RHEL 5.3
> But I am not able to configure the cluster for apache.while enabling the
> service I am getting the below error
>
> # clusvcadm -e webcluster
> Local machine trying to enable service:webcluster...Aborted; service failed
>
> /log/messages
>
> Aug 19 21:12:49 cmsstorage-web1 clurgmgrd[32766]: <err> #43: Service
> service:webcluster has failed; can not start.
> Aug 19 21:12:49 cmsstorage-web1 clurgmgrd[32766]: <crit> #13: Service
> service:webcluster failed to stop cleanly
>
> cluster.conf
>
> <failoverdomain name="GFS-apache" ordered="1" restricted="1">
>                                 <failoverdomainnode name="cmsstorage-web3"
> priority="3"/>
>                                 <failoverdomainnode name="cmsstorage-web2"
> priority="2"/>
>                                 <failoverdomainnode name="cmsstorage-web1"
> priority="1"/>
>                         </failoverdomain>
>
> <resources>
> <apache config_file="conf/httpd.conf" httpd_options="" name="webcluster"
> server_root="/usr/local/apache/" shutdown_wait="2"/>
> </resources>
>
>  <service autostart="1" domain="GFS-apache" name="webcluster"
> recovery="restart">
>                         <apache ref="webcluster"/>
>                 </service>
>
> Is there anything missing...



You are missing the definition of an ip address in the resource section, and
after that, the use of the defined ip address into the service definition.

Regards,



>
>
> Thanks in advance.
>
> Regards,
> Saji George
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090821/6a284eed/attachment.htm>

From rmicmirregs at gmail.com  Sat Aug 22 08:13:45 2009
From: rmicmirregs at gmail.com (rmicmirregs)
Date: Sat, 22 Aug 2009 10:13:45 +0200
Subject: [Linux-cluster] CLVMD without GFS
In-Reply-To: <29ae894c0908180719tf00e87di485d691fd8d84679@mail.gmail.com>
References: <MJ2mY0lYjZGc@NBjvyXzB> <1249079337.6494.10.camel@mecatol>
	<29ae894c0907311725k4759e631m332ffd38857a6f99@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C46B9@rkamsem701.emea.roche.com>
	<29ae894c0908180209i46e23ed1o8fc2d372c8371e12@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C4700@rkamsem701.emea.roche.com>
	<29ae894c0908180235g756ce310nc678a922ce5f718e@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C4841@rkamsem701.emea.roche.com>
	<29ae894c0908180719tf00e87di485d691fd8d84679@mail.gmail.com>
Message-ID: <1250928825.6676.17.camel@quann>

Hi,

El mar, 18-08-2009 a las 16:19 +0200, brem belguebli escribi?:
> Hi,
>  
> I see what you mean.
>  
> On the other hand, it would may be lessen some flexibility. The
> typical case being the fact that you do not want to use FS on top of
> your LV's but raw devices.
>  
> Brem
> 
>  
> 2009/8/18, Moralejo, Alfredo <alfredo.moralejo at roche.com>: 
>         Hi,
>         
>          
>         
>         Yes, I understand that. 
>         
>          
>         
>         However, what I meant is that, from a configuration simplicity
>         point of view, it could be convenient to add the functionality
>         into a only resource that take care of the entire cycle,
>         logical volume activation in exclusive mode and mounting it. I
>         think this is the most common usage of the fs resource. In
>         this way, I don?t need to create two resources for each
>         filesystem (one for the logical volume and one other for the
>         fs) and I can be sure a filesystem can not be mounted by hand
>         in another node (as it has been activate exclusively in the
>         node running the service).
>         
>          
>         
>         Best regards,
>         
>          
>         
>         Alfredo
>         
>          
>         
>          

What Alfredo asks is exactly the point i wanted to reach at the end of
my lvm-cluster resource script: a filesystem resource with grants of
exclusive use, avoiding administrators make the mistake of, by hand,
mount concurrently the same FS on two different nodes.

As far as i know, and correct me please if i'm wrong, there is no
"exclusive usage" flag on non-clustered FS, even maybe nor in clustered
FS. So i moved the protection to a lower layer into the storage stack:
the LVM layer.

This approach is the same i was accustomed to use with the Linux-HA +
EVMS (Enterprise Volume Manager System) approach. I used to use
Clustered Segment Managers to build the /dev EVMS devices that were used
as mountpoints, so if an administrator wanted to mount the FS by hand he
was needed to assure the CSM was owned by the node he was working on. 

I agree with both of you that this would be much simpler is there was no
need of implicating LVM layer on the protection of the FS, but i have
not found any other way of doing this.

In the other hand, there is also the "bug" of the overlapping of the LVM
exclusive flag that Brem has opened in the bugzilla. If this gets fixed
in the way we think it should be, i think we would have reached a good
complexity/success ratio for this security solution.

Cheers,

Rafael


-- 
Rafael Mic? Miranda



From edsonmarquezani at gmail.com  Sun Aug 23 15:35:39 2009
From: edsonmarquezani at gmail.com (Edson Marquezani Filho)
Date: Sun, 23 Aug 2009 12:35:39 -0300
Subject: [Linux-cluster] Preventing LVM from concurrent access
Message-ID: <2fc5f090908230835p20acd5e0ya2bd2a3ba909350b@mail.gmail.com>

Hello,

I'm a begginer on that cluster stuff, so I ask you for help.

I'm setting up a environment with virtual machines running upon Xen,
two servers, and a SAN storage. I want to set up a very simple
high-availability scenario, where once my master server goes down,
slave creates again those virtual machines on it. (What would
correspond to a restart of all machines, off course.)

Ok, I could do that with Heartbeat, with a very simple configuration
based on its version 1. It worked as expected.

My problem now is related to SAN data access.

I want to prevent data to be simultaneously mounted by both servers,
in a situation when I don't have master server down, altough it could
look like this. I'm connecting servers through a crossover UTP cable.
If it get disconnected or any of those dedicated interfaces goes down,
slave server would think that master server is down, and would
recreate all my virtual machines, altough it isn't down, corrupting
LVM and filesystem data.

Alright, I have found out that I would have to use either HA-LVM or
CLVMD. Actually, I think that HA-LVM would be more appropriate to me,
accordingly to what I've seen around. But, I'm really not sure about
that.

What should I use, exactly? Does HA-LVM do what I want? Does it
prevent LVM volumes from being mounted by more than a node
simultaneously? I have thought about locking VGs on each server, with
CLMD, but I'm not sure if it is the correct way for that.

I can say for sure that I just need data access at one server at a time.

What would be the simplest and more correct way for doing this?

Thank you.



From davidhollister at comcast.net  Sun Aug 23 18:20:46 2009
From: davidhollister at comcast.net (David Hollister)
Date: Sun, 23 Aug 2009 18:20:46 +0000
Subject: [Linux-cluster] Re: Linux-cluster Digest,
	Preventing LVM from concurrent access
In-Reply-To: <20090823160008.CA2B36194C4@hormel.redhat.com>
References: <20090823160008.CA2B36194C4@hormel.redhat.com>
Message-ID: <1269910516-1251051757-cardhu_decombobulator_blackberry.rim.net-1598207711-@bxe1291.bisx.prod.on.blackberry>

We use HA-LVM in environments where we do not use GFS. It works well in preventing shared storage from being mounted on both nodes simultaneouly. However I am not sure how it would react in a split brain conditon, which appears to be what you are describing. . You should probably consider creating redundancy in your heartbeat via IP Bonding. Basically having two nics on each host bound to one ip.
Sent via BlackBerry by AT&T

-----Original Message-----
From: linux-cluster-request at redhat.com

Date: Sun, 23 Aug 2009 12:00:08 
To: <linux-cluster at redhat.com>
Subject: Linux-cluster Digest, Vol 64, Issue 37


Send Linux-cluster mailing list submissions to
	linux-cluster at redhat.com

To subscribe or unsubscribe via the World Wide Web, visit
	https://www.redhat.com/mailman/listinfo/linux-cluster
or, via email, send a message with subject or body 'help' to
	linux-cluster-request at redhat.com

You can reach the person managing the list at
	linux-cluster-owner at redhat.com

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Linux-cluster digest..."


Today's Topics:

   1. Preventing LVM from concurrent access (Edson Marquezani Filho)


----------------------------------------------------------------------

Message: 1
Date: Sun, 23 Aug 2009 12:35:39 -0300
From: Edson Marquezani Filho <edsonmarquezani at gmail.com>
Subject: [Linux-cluster] Preventing LVM from concurrent access
To: linux-cluster at redhat.com
Message-ID:
	<2fc5f090908230835p20acd5e0ya2bd2a3ba909350b at mail.gmail.com>
Content-Type: text/plain; charset=ISO-8859-1

Hello,

I'm a begginer on that cluster stuff, so I ask you for help.

I'm setting up a environment with virtual machines running upon Xen,
two servers, and a SAN storage. I want to set up a very simple
high-availability scenario, where once my master server goes down,
slave creates again those virtual machines on it. (What would
correspond to a restart of all machines, off course.)

Ok, I could do that with Heartbeat, with a very simple configuration
based on its version 1. It worked as expected.

My problem now is related to SAN data access.

I want to prevent data to be simultaneously mounted by both servers,
in a situation when I don't have master server down, altough it could
look like this. I'm connecting servers through a crossover UTP cable.
If it get disconnected or any of those dedicated interfaces goes down,
slave server would think that master server is down, and would
recreate all my virtual machines, altough it isn't down, corrupting
LVM and filesystem data.

Alright, I have found out that I would have to use either HA-LVM or
CLVMD. Actually, I think that HA-LVM would be more appropriate to me,
accordingly to what I've seen around. But, I'm really not sure about
that.

What should I use, exactly? Does HA-LVM do what I want? Does it
prevent LVM volumes from being mounted by more than a node
simultaneously? I have thought about locking VGs on each server, with
CLMD, but I'm not sure if it is the correct way for that.

I can say for sure that I just need data access at one server at a time.

What would be the simplest and more correct way for doing this?

Thank you.



------------------------------

--
Linux-cluster mailing list
Linux-cluster at redhat.com
https://www.redhat.com/mailman/listinfo/linux-cluster

End of Linux-cluster Digest, Vol 64, Issue 37
*********************************************



From edsonmarquezani at gmail.com  Sun Aug 23 19:01:33 2009
From: edsonmarquezani at gmail.com (Edson Marquezani Filho)
Date: Sun, 23 Aug 2009 16:01:33 -0300
Subject: [Linux-cluster] Re: Linux-cluster Digest, Preventing LVM from 
	concurrent access
In-Reply-To: <1269910516-1251051757-cardhu_decombobulator_blackberry.rim.net-1598207711-@bxe1291.bisx.prod.on.blackberry>
References: <20090823160008.CA2B36194C4@hormel.redhat.com>
	<1269910516-1251051757-cardhu_decombobulator_blackberry.rim.net-1598207711-@bxe1291.bisx.prod.on.blackberry>
Message-ID: <2fc5f090908231201j534884f0ke1648bf5426d0c3e@mail.gmail.com>

On Sun, Aug 23, 2009 at 15:20, David
Hollister<davidhollister at comcast.net> wrote:
> We use HA-LVM in environments where we do not use GFS. It works well in preventing shared storage from being mounted on both nodes simultaneouly. However I am not sure how it would react in a split brain conditon, which appears to be what you are describing. . You should probably consider creating redundancy in your heartbeat via IP Bonding. Basically having two nics on each host bound to one ip.

Indeed, I won't use GFS. Once I don't need any volume to be mounted at
diferent points simultaneously, there is no reason to use it, and I
want to keep my system as simple as possible for my needs.

When you say that HA-LVM prevents shares storage from being mounted on
more than one node, do you mean that I could not do that, even
manually? Or you mean that this won't be done by the cluster
infrastrucuture? AFAIK, when using HA-LVM, the LVM locking keeps
local, right? So, that's the problem to me, because some unware people
could break things.

I have already thought about bonding two interfaces, but the simple
idea of having cables disconnected still sounds very dangerous to me.
Or even if I get master server restarted, when it is back, it would
try to launch virtual machines, but them would be mount on slave. Do
you understand?

I think that's not the correct way for doing what I want, and I'm not
getting how to do this.

How would you do that? (Forget about Heartbeat, it was just the
easiest way I have found for doing this, but what I need is things
working well, it don't matter how. But, I want to keep things not more
complex than necessary.)

Thank you.



From esanchezvela.redhatcluster at gmail.com  Mon Aug 24 03:28:17 2009
From: esanchezvela.redhatcluster at gmail.com (Enrique Sanchez)
Date: Sun, 23 Aug 2009 23:28:17 -0400
Subject: [Linux-cluster] problems with gfs and openais cluster.
In-Reply-To: <1250550383.19265.86.camel@localhost.localdomain>
References: <1247a04a0908142311o5ff51b2am3f245a736eda45b8@mail.gmail.com>
	<1250317407.2649.16.camel@localhost.localdomain>
	<1247a04a0908150512g2966f1cnefffb0c8dc5d033e@mail.gmail.com>
	<1250550383.19265.86.camel@localhost.localdomain>
Message-ID: <1247a04a0908232028s4a426490nae07ff9e38e191f0@mail.gmail.com>

On Mon, Aug 17, 2009 at 7:06 PM, Steven Dake<sdake at redhat.com> wrote:
> On Sat, 2009-08-15 at 08:12 -0400, Enrique Sanchez wrote:
>> On Sat, Aug 15, 2009 at 2:23 AM, Steven Dake<sdake at redhat.com> wrote:
>> >
>> > The openais.conf file is not used when starting with the cman service.
>> > Instead cman uses its own configuration file
>> > in /etc/cluster/cluster.conf.  openais should not be chkconfig'ed on
>> > unless running without cman.
>> >
>> > One reason you could have the error is that you have multiple clusters
>> > on the same multicast address as specified in /etc/cluster/cluster.conf.
>> >
>> > Regards
>> > -steve
>> >
>>
>> steve,
>>
>> thanks for the quick reply: the mcast address is the same and it
>> looks like it is using the same VLAN here are the details....
>>
>> this is one cluster (cluster one):
>>
>>         <cman expected_votes="1" two_node="1">
>>                 <multicast addr="224.0.0.1"/>
>>         </cman>
>>
>> and this is in the 2nd cluster (cluster two)......
>>
>>         <cman expected_votes="1" two_node="1">
>>                 <multicast addr="224.0.0.1" interface="eth1"/>
>>         </cman>
>>
>>
>> unfortunately eth1 is on the same VLAN for both clusters...is cluster
>> 1 using the mcast address on both adapters?
>>
>> thnx...
>> enrique.
>
> clusters are uniquely identified by their multicast address:port pair.
> I suggest changing one of the cluster multicast addresses to 224.0.0.2.
>
> Regards
> -steve
>>
>>
>
>

I think I need to use the <command> ccsd -m 224.0.0.2 </command> to
make the change, just I am not sure whether I need to do a cold start
on the cluster, will this bring the service down or can this be done
concurrently without affecting the service.

thanks again,
esv.

-- 
Enrique Sanchez Vela
------------------------------------------
"What you have been obliged to discover
by yourself leaves a path in your mind
which you can use again when the need
arises."    --G. C. Lichtenberg
http://themathcircle.org/



From fdinitto at redhat.com  Mon Aug 24 04:37:45 2009
From: fdinitto at redhat.com (Fabio M. Di Nitto)
Date: Mon, 24 Aug 2009 06:37:45 +0200
Subject: [Linux-cluster] Cluster 3.0.2 stable release
Message-ID: <1251088665.8148.60.camel@cerberus.int.fabbione.net>

The cluster team and its community are proud to announce the 3.0.2
stable release from the STABLE3 branch.

This release contains a few major bug fixes. We strongly recommend
people to update your clusters.

In order to build the 3.0.2 release you will need:

- corosync 1.0.0
- openais 1.0.0
- linux kernel 2.6.30

The new source tarball can be downloaded here:

ftp://sources.redhat.com/pub/cluster/releases/cluster-3.0.2.tar.gz
https://fedorahosted.org/releases/c/l/cluster/cluster-3.0.2.tar.gz

At the same location is now possible to find separated tarballs for
fence-agents and resource-agents as previously announced
(http://www.redhat.com/archives/cluster-devel/2009-February/msg00003.htm)

To report bugs or issues:

   https://bugzilla.redhat.com/

Would you like to meet the cluster team or members of its community?

   Join us on IRC (irc.freenode.net #linux-cluster) and share your
   experience  with other sysadministrators or power users.

Thanks/congratulations to all people that contributed to achieve this
great milestone.

Happy clustering,
Fabio

Under the hood (from 3.0.1):

Bob Peterson (1):
      The mount -o remount option failed due to new relatime option

Christine Caulfield (1):
      cman: Assign stderr to /dev/null rather than closing it

Lon Hohberger (1):
      rgmanager: Fix referencing bug

Marek 'marx' Grac (1):
      fencing: Fencing fails when using implicit connection method other
then telnet

 cman/daemon/cman-preconfig.c      |   17 +++++++++++++----
 fence/agents/alom/fence_alom.py   |    8 ++++----
 fence/agents/ilo/fence_ilo.py     |    7 ++++---
 fence/agents/ldom/fence_ldom.py   |    6 +++---
 fence/agents/lpar/fence_lpar.py   |    5 +++--
 fence/agents/virsh/fence_virsh.py |    6 +++---
 gfs2/mount/util.c                 |    1 +
 rgmanager/src/daemons/reslist.c   |   13 ++++++++-----
 8 files changed, 39 insertions(+), 24 deletions(-)

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090824/82289deb/attachment.sig>

From alfredo.moralejo at roche.com  Mon Aug 24 07:09:12 2009
From: alfredo.moralejo at roche.com (Moralejo, Alfredo)
Date: Mon, 24 Aug 2009 09:09:12 +0200
Subject: [Linux-cluster] CLVMD without GFS
In-Reply-To: <1250928825.6676.17.camel@quann>
References: <MJ2mY0lYjZGc@NBjvyXzB> <1249079337.6494.10.camel@mecatol>
	<29ae894c0907311725k4759e631m332ffd38857a6f99@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C46B9@rkamsem701.emea.roche.com>
	<29ae894c0908180209i46e23ed1o8fc2d372c8371e12@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C4700@rkamsem701.emea.roche.com>
	<29ae894c0908180235g756ce310nc678a922ce5f718e@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C4841@rkamsem701.emea.roche.com>
	<29ae894c0908180719tf00e87di485d691fd8d84679@mail.gmail.com>
	<1250928825.6676.17.camel@quann>
Message-ID: <C64734E4E1C80E49955AD539DB2FBC3A0236CC26@rkamsem703.emea.roche.com>

Hi,

By default, when clvmd service is restarted, all logical volumes in clustered filesystems are activated. I understand this is a problem to implement the exclusive activation, as it should work in the opposite way, all logical volumes should be deactivated until the resource taking care of it is started in a node where it'd be exclusively activated.

Has anyone found this problem while implemented the lvm-cluster.sh resource or a similar solution? Is there a way to fix it via lvm configuration or something like that?

Best regards,

Alfredo


-----Original Message-----
From: linux-cluster-bounces at redhat.com [mailto:linux-cluster-bounces at redhat.com] On Behalf Of rmicmirregs
Sent: Saturday, August 22, 2009 10:14 AM
To: linux clustering
Subject: Re: [Linux-cluster] CLVMD without GFS

Hi,

El mar, 18-08-2009 a las 16:19 +0200, brem belguebli escribi?:
> Hi,
>  
> I see what you mean.
>  
> On the other hand, it would may be lessen some flexibility. The
> typical case being the fact that you do not want to use FS on top of
> your LV's but raw devices.
>  
> Brem
> 
>  
> 2009/8/18, Moralejo, Alfredo <alfredo.moralejo at roche.com>: 
>         Hi,
>         
>          
>         
>         Yes, I understand that. 
>         
>          
>         
>         However, what I meant is that, from a configuration simplicity
>         point of view, it could be convenient to add the functionality
>         into a only resource that take care of the entire cycle,
>         logical volume activation in exclusive mode and mounting it. I
>         think this is the most common usage of the fs resource. In
>         this way, I don't need to create two resources for each
>         filesystem (one for the logical volume and one other for the
>         fs) and I can be sure a filesystem can not be mounted by hand
>         in another node (as it has been activate exclusively in the
>         node running the service).
>         
>          
>         
>         Best regards,
>         
>          
>         
>         Alfredo
>         
>          
>         
>          

What Alfredo asks is exactly the point i wanted to reach at the end of
my lvm-cluster resource script: a filesystem resource with grants of
exclusive use, avoiding administrators make the mistake of, by hand,
mount concurrently the same FS on two different nodes.

As far as i know, and correct me please if i'm wrong, there is no
"exclusive usage" flag on non-clustered FS, even maybe nor in clustered
FS. So i moved the protection to a lower layer into the storage stack:
the LVM layer.

This approach is the same i was accustomed to use with the Linux-HA +
EVMS (Enterprise Volume Manager System) approach. I used to use
Clustered Segment Managers to build the /dev EVMS devices that were used
as mountpoints, so if an administrator wanted to mount the FS by hand he
was needed to assure the CSM was owned by the node he was working on. 

I agree with both of you that this would be much simpler is there was no
need of implicating LVM layer on the protection of the FS, but i have
not found any other way of doing this.

In the other hand, there is also the "bug" of the overlapping of the LVM
exclusive flag that Brem has opened in the bugzilla. If this gets fixed
in the way we think it should be, i think we would have reached a good
complexity/success ratio for this security solution.

Cheers,

Rafael


-- 
Rafael Mic? Miranda

--
Linux-cluster mailing list
Linux-cluster at redhat.com
https://www.redhat.com/mailman/listinfo/linux-cluster



From rmicmirregs at gmail.com  Mon Aug 24 11:39:28 2009
From: rmicmirregs at gmail.com (rmicmirregs)
Date: Mon, 24 Aug 2009 13:39:28 +0200
Subject: [Linux-cluster] CLVMD without GFS
In-Reply-To: <C64734E4E1C80E49955AD539DB2FBC3A0236CC26@rkamsem703.emea.roche.com>
References: <MJ2mY0lYjZGc@NBjvyXzB> <1249079337.6494.10.camel@mecatol>
	<29ae894c0907311725k4759e631m332ffd38857a6f99@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C46B9@rkamsem701.emea.roche.com>
	<29ae894c0908180209i46e23ed1o8fc2d372c8371e12@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C4700@rkamsem701.emea.roche.com>
	<29ae894c0908180235g756ce310nc678a922ce5f718e@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C4841@rkamsem701.emea.roche.com>
	<29ae894c0908180719tf00e87di485d691fd8d84679@mail.gmail.com>
	<1250928825.6676.17.camel@quann>
	<C64734E4E1C80E49955AD539DB2FBC3A0236CC26@rkamsem703.emea.roche.com>
Message-ID: <1251113968.5844.7.camel@quann>

Hi Alfredo

El lun, 24-08-2009 a las 09:09 +0200, Moralejo, Alfredo escribi?:
> Hi,
> 
> By default, when clvmd service is restarted, all logical volumes in clustered filesystems are activated. I understand this is a problem to implement the exclusive activation, as it should work in the opposite way, all logical volumes should be deactivated until the resource taking care of it is started in a node where it'd be exclusively activated.
> 
> Has anyone found this problem while implemented the lvm-cluster.sh resource or a similar solution? Is there a way to fix it via lvm configuration or something like that?
> 
> Best regards,
> 
> Alfredo

To use the lvm-cluster.sh resource script, you need to:

1.- Configure your cluster.conf file to use this new resource script.

2.- Start cluster services CMAN and CLVMD, but do not start RGMANAGER

3.- By hand, deactivate all the Logical Volumes you have configured to
be used with the "exclusive" flag. This can be done with "lvchange -an
volgrp01/logvol01"

4.- After that, you can start the RGMANAGER service. Now it should find
all the volumes deactivated and they can be activated with the exclusive
flag.

This instructions are not provided with the script because i was waiting
some kind of "information requests" by the CMAN  project managers, in
wich i would detail this kind of information.

I hope this helps, please let me know if you find any problems.

Cheers,

Rafael

-- 
Rafael Mic? Miranda



From rmicmirregs at gmail.com  Mon Aug 24 12:06:38 2009
From: rmicmirregs at gmail.com (rmicmirregs)
Date: Mon, 24 Aug 2009 14:06:38 +0200
Subject: [Linux-cluster] Re: Linux-cluster Digest, Preventing LVM from 
	concurrent access
In-Reply-To: <2fc5f090908231201j534884f0ke1648bf5426d0c3e@mail.gmail.com>
References: <20090823160008.CA2B36194C4@hormel.redhat.com>
	<1269910516-1251051757-cardhu_decombobulator_blackberry.rim.net-1598207711-@bxe1291.bisx.prod.on.blackberry>
	<2fc5f090908231201j534884f0ke1648bf5426d0c3e@mail.gmail.com>
Message-ID: <1251115598.5844.31.camel@quann>

Hi Edson

El dom, 23-08-2009 a las 16:01 -0300, Edson Marquezani Filho escribi?:
> On Sun, Aug 23, 2009 at 15:20, David
> Hollister<davidhollister at comcast.net> wrote:
> > We use HA-LVM in environments where we do not use GFS. It works well in preventing shared storage from being mounted on both nodes simultaneouly. However I am not sure how it would react in a split brain conditon, which appears to be what you are describing. . You should probably consider creating redundancy in your heartbeat via IP Bonding. Basically having two nics on each host bound to one ip.
> 
> Indeed, I won't use GFS. Once I don't need any volume to be mounted at
> diferent points simultaneously, there is no reason to use it, and I
> want to keep my system as simple as possible for my needs.
> 
> When you say that HA-LVM prevents shares storage from being mounted on
> more than one node, do you mean that I could not do that, even
> manually? Or you mean that this won't be done by the cluster
> infrastrucuture? AFAIK, when using HA-LVM, the LVM locking keeps
> local, right? So, that's the problem to me, because some unware people
> could break things.
> 
> I have already thought about bonding two interfaces, but the simple
> idea of having cables disconnected still sounds very dangerous to me.
> Or even if I get master server restarted, when it is back, it would
> try to launch virtual machines, but them would be mount on slave. Do
> you understand?
> 
> I think that's not the correct way for doing what I want, and I'm not
> getting how to do this.
> 
> How would you do that? (Forget about Heartbeat, it was just the
> easiest way I have found for doing this, but what I need is things
> working well, it don't matter how. But, I want to keep things not more
> complex than necessary.)
> 
> Thank you.
> 
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster

You have some interesting and different ideas here:

1.- Shared storage with non-clustered filesystem: i have no experience
with HA-LVM, i have just checked its knowledge base on RHEL to refresh
my ideas.

If i'm not wrong, what you want is to have a shared storage with
protection from concurrent mounts for your non-clustered filesystems,
but EVEN for admins doing, by hand, erratic mounts, am i right? 

You can use a different approach to HA-LVM with CLVMD and a resource
script i have developed called "lvm-cluster". You can find it here:

https://www.redhat.com/archives/cluster-devel/2009-June/msg00065.html

It's not included in the main project yet, but some people in the list
are testing it successfully.

Actually it would not avoid an admin doing, by hand, an erratic mount
but a bugzilla ticket has been opened by Brem Belguebli to fix an LVM
issue that is causing that behaviour. Apart from this admin issue, the
script would protect the concurrent mount of your filesystem by the
usage of the LVM layer. 

I do like more this approach because you don't need to generate new
initrd images for your nodes every time you configure new volumes to be
into the HA-LVM.

2.- Communications interruption and split-brain: you should use some
fencing system to make the 2nd node "kill" the main one and take the
resources. 

Having the main node booted back again AND the comms not reestablished
will cause you a split-brain again, with a new fencing action. I usually
do not set cluster services (CMAN, RGMANAGER, etc.) to be automatically
started on the nodes. I mean they must be re-run and activated by an
admin.

In the other hand you can use a Quorum Disk, provided my CMAN too, as
the 3rd vote on your cluster and keep the quorum. I use both approaches
at the same time (no auto-start of cluster services and qdisk).

3.- Communications channels: there's no way to set on CMAN more than one
comm channel, but you can set more than one network interface to be part
of that channel, via bonding. I think this is one of the lacks of CMAN
actually.

I hope this helps. Just ask whatever you want.

Cheers,

Rafael 

-- 
Rafael Mic? Miranda



From brem.belguebli at gmail.com  Mon Aug 24 12:23:33 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Mon, 24 Aug 2009 14:23:33 +0200
Subject: [Linux-cluster] CLVMD without GFS
In-Reply-To: <1251113968.5844.7.camel@quann>
References: <MJ2mY0lYjZGc@NBjvyXzB>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C46B9@rkamsem701.emea.roche.com>
	<29ae894c0908180209i46e23ed1o8fc2d372c8371e12@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C4700@rkamsem701.emea.roche.com>
	<29ae894c0908180235g756ce310nc678a922ce5f718e@mail.gmail.com>
	<FD716ADFEE797543AA8B2AB0B3F9148E01A09C4841@rkamsem701.emea.roche.com>
	<29ae894c0908180719tf00e87di485d691fd8d84679@mail.gmail.com>
	<1250928825.6676.17.camel@quann>
	<C64734E4E1C80E49955AD539DB2FBC3A0236CC26@rkamsem703.emea.roche.com>
	<1251113968.5844.7.camel@quann>
Message-ID: <29ae894c0908240523h61e15a23g11dbf97647742300@mail.gmail.com>

Hi,

This is mainly due to clvmd init script that activates all the VG's it
finds.

I kind off have a problem with all those init scripts (including rc.sysinit)
that want to automatically do things
that an admin (well myself ;-) ) wouldn't like them to be active.

Ideally, there should be a file (/etc/sysconfig/clvm.conf)  that contains
only what we want to automatically activate at startup. If empty everythings
gets activated.

I have the same problem with automatic cluster startup, it is something
that, IMHO, could lead to undesirable situations (imagine the node that
keeps on rebooting after automatic join the cluster).

Brem


2009/8/24, rmicmirregs <rmicmirregs at gmail.com>:
>
> Hi Alfredo
>
> El lun, 24-08-2009 a las 09:09 +0200, Moralejo, Alfredo escribi?:
> > Hi,
> >
> > By default, when clvmd service is restarted, all logical volumes in
> clustered filesystems are activated. I understand this is a problem to
> implement the exclusive activation, as it should work in the opposite way,
> all logical volumes should be deactivated until the resource taking care of
> it is started in a node where it'd be exclusively activated.
> >
> > Has anyone found this problem while implemented the lvm-cluster.sh
> resource or a similar solution? Is there a way to fix it via lvm
> configuration or something like that?
> >
> > Best regards,
> >
> > Alfredo
>
> To use the lvm-cluster.sh resource script, you need to:
>
> 1.- Configure your cluster.conf file to use this new resource script.
>
> 2.- Start cluster services CMAN and CLVMD, but do not start RGMANAGER
>
> 3.- By hand, deactivate all the Logical Volumes you have configured to
> be used with the "exclusive" flag. This can be done with "lvchange -an
> volgrp01/logvol01"
>
> 4.- After that, you can start the RGMANAGER service. Now it should find
> all the volumes deactivated and they can be activated with the exclusive
> flag.
>
> This instructions are not provided with the script because i was waiting
> some kind of "information requests" by the CMAN  project managers, in
> wich i would detail this kind of information.
>
> I hope this helps, please let me know if you find any problems.
>
> Cheers,
>
> Rafael
>
> --
> Rafael Mic? Miranda
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090824/d22bc3d4/attachment.htm>

From sdake at redhat.com  Mon Aug 24 13:41:37 2009
From: sdake at redhat.com (Steven Dake)
Date: Mon, 24 Aug 2009 06:41:37 -0700
Subject: [Linux-cluster] problems with gfs and openais cluster.
In-Reply-To: <1247a04a0908232028s4a426490nae07ff9e38e191f0@mail.gmail.com>
References: <1247a04a0908142311o5ff51b2am3f245a736eda45b8@mail.gmail.com>
	<1250317407.2649.16.camel@localhost.localdomain>
	<1247a04a0908150512g2966f1cnefffb0c8dc5d033e@mail.gmail.com>
	<1250550383.19265.86.camel@localhost.localdomain>
	<1247a04a0908232028s4a426490nae07ff9e38e191f0@mail.gmail.com>
Message-ID: <1251121297.2706.22.camel@localhost.localdomain>

On Sun, 2009-08-23 at 23:28 -0400, Enrique Sanchez wrote:
> On Mon, Aug 17, 2009 at 7:06 PM, Steven Dake<sdake at redhat.com> wrote:
> > On Sat, 2009-08-15 at 08:12 -0400, Enrique Sanchez wrote:
> >> On Sat, Aug 15, 2009 at 2:23 AM, Steven Dake<sdake at redhat.com> wrote:
> >> >
> >> > The openais.conf file is not used when starting with the cman service.
> >> > Instead cman uses its own configuration file
> >> > in /etc/cluster/cluster.conf.  openais should not be chkconfig'ed on
> >> > unless running without cman.
> >> >
> >> > One reason you could have the error is that you have multiple clusters
> >> > on the same multicast address as specified in /etc/cluster/cluster.conf.
> >> >
> >> > Regards
> >> > -steve
> >> >
> >>
> >> steve,
> >>
> >> thanks for the quick reply: the mcast address is the same and it
> >> looks like it is using the same VLAN here are the details....
> >>
> >> this is one cluster (cluster one):
> >>
> >>         <cman expected_votes="1" two_node="1">
> >>                 <multicast addr="224.0.0.1"/>
> >>         </cman>
> >>
> >> and this is in the 2nd cluster (cluster two)......
> >>
> >>         <cman expected_votes="1" two_node="1">
> >>                 <multicast addr="224.0.0.1" interface="eth1"/>
> >>         </cman>
> >>
> >>
> >> unfortunately eth1 is on the same VLAN for both clusters...is cluster
> >> 1 using the mcast address on both adapters?
> >>
> >> thnx...
> >> enrique.
> >
> > clusters are uniquely identified by their multicast address:port pair.
> > I suggest changing one of the cluster multicast addresses to 224.0.0.2.
> >
> > Regards
> > -steve
> >>
> >>
> >
> >
> 
> I think I need to use the <command> ccsd -m 224.0.0.2 </command> to
> make the change, just I am not sure whether I need to do a cold start
> on the cluster, will this bring the service down or can this be done
> concurrently without affecting the service.
> 
> thanks again,
> esv.
> 
Enrique,

I am not sure how you change that in a running cluster.  Best to ask
Chrissie Caulfield.

Regards
-steve



From ccaulfie at redhat.com  Mon Aug 24 13:59:26 2009
From: ccaulfie at redhat.com (Christine Caulfield)
Date: Mon, 24 Aug 2009 14:59:26 +0100
Subject: [Linux-cluster] problems with gfs and openais cluster.
In-Reply-To: <1251121297.2706.22.camel@localhost.localdomain>
References: <1247a04a0908142311o5ff51b2am3f245a736eda45b8@mail.gmail.com>	<1250317407.2649.16.camel@localhost.localdomain>	<1247a04a0908150512g2966f1cnefffb0c8dc5d033e@mail.gmail.com>	<1250550383.19265.86.camel@localhost.localdomain>	<1247a04a0908232028s4a426490nae07ff9e38e191f0@mail.gmail.com>
	<1251121297.2706.22.camel@localhost.localdomain>
Message-ID: <4A929CBE.5030708@redhat.com>

On 24/08/09 14:41, Steven Dake wrote:
> On Sun, 2009-08-23 at 23:28 -0400, Enrique Sanchez wrote:
>> On Mon, Aug 17, 2009 at 7:06 PM, Steven Dake<sdake at redhat.com>  wrote:
>>> On Sat, 2009-08-15 at 08:12 -0400, Enrique Sanchez wrote:
>>>> On Sat, Aug 15, 2009 at 2:23 AM, Steven Dake<sdake at redhat.com>  wrote:
>>>>>
>>>>> The openais.conf file is not used when starting with the cman service.
>>>>> Instead cman uses its own configuration file
>>>>> in /etc/cluster/cluster.conf.  openais should not be chkconfig'ed on
>>>>> unless running without cman.
>>>>>
>>>>> One reason you could have the error is that you have multiple clusters
>>>>> on the same multicast address as specified in /etc/cluster/cluster.conf.
>>>>>
>>>>> Regards
>>>>> -steve
>>>>>
>>>>
>>>> steve,
>>>>
>>>> thanks for the quick reply: the mcast address is the same and it
>>>> looks like it is using the same VLAN here are the details....
>>>>
>>>> this is one cluster (cluster one):
>>>>
>>>>          <cman expected_votes="1" two_node="1">
>>>>                  <multicast addr="224.0.0.1"/>
>>>>          </cman>
>>>>
>>>> and this is in the 2nd cluster (cluster two)......
>>>>
>>>>          <cman expected_votes="1" two_node="1">
>>>>                  <multicast addr="224.0.0.1" interface="eth1"/>
>>>>          </cman>
>>>>
>>>>
>>>> unfortunately eth1 is on the same VLAN for both clusters...is cluster
>>>> 1 using the mcast address on both adapters?
>>>>
>>>> thnx...
>>>> enrique.
>>>
>>> clusters are uniquely identified by their multicast address:port pair.
>>> I suggest changing one of the cluster multicast addresses to 224.0.0.2.
>>>
>>> Regards
>>> -steve
>>>>
>>>>
>>>
>>>
>>
>> I think I need to use the<command>  ccsd -m 224.0.0.2</command>  to
>> make the change, just I am not sure whether I need to do a cold start
>> on the cluster, will this bring the service down or can this be done
>> concurrently without affecting the service.

That command just changes the multicast address that ccsd uses to 
communicate. You might need that in addition to the entry in 
cluster.conf if you have to change the address for ccsd itself. 
Otherwise just adding the <multicast addr > tag in cluster.conf will work.

If you change the multicast address of a cluster you need to take down 
all nodes and restart them. Multicast is how they communicate...it's not 
possible to change without a cluster restart, sorry.


Chrissie



From tomas.lavicky at racom.eu  Mon Aug 24 14:31:33 2009
From: tomas.lavicky at racom.eu (Tomas Lavicky)
Date: Mon, 24 Aug 2009 16:31:33 +0200
Subject: [Linux-cluster] OpenVPN - event_wait : Interrupted system call
	(code=4)
Message-ID: <200908241631.33562.tomas.lavicky@racom.eu>

I'm trying to add OpenVPN server as service to RHEL 5.3 Cluster. I installed 
OpenVPN via Dag Wieers repository. I created script resource OpenVPN with 
script file /etc/init.d/openvpn and add it to VPN service. Tunnel looks like 
working well but "event_wait : Interrupted system call (code=4)" message 
appears in /var/log/openvpn.log twice in a minute:

Mon Aug 24 16:24:17 2009 event_wait : Interrupted system call (code=4)
Mon Aug 24 16:24:17 2009 OpenVPN CLIENT LIST
Mon Aug 24 16:24:17 2009 Updated,Mon Aug 24 16:24:17 2009
Mon Aug 24 16:24:17 2009 Common Name,Real Address,Bytes Received,Bytes 
Sent,Connected Since
Mon Aug 24 16:24:17 2009 Racom_to_RWE,192.168.0.59:43778,13508,13192,Mon Aug 
24 16:06:18 2009
Mon Aug 24 16:24:17 2009 ROUTING TABLE
Mon Aug 24 16:24:17 2009 Virtual Address,Common Name,Real Address,Last Ref
Mon Aug 24 16:24:17 2009 172.19.7.6,Racom_to_RWE,192.168.0.59:43778,Mon Aug 24 
16:09:47 2009
Mon Aug 24 16:24:17 2009 GLOBAL STATS
Mon Aug 24 16:24:17 2009 Max bcast/mcast queue length,0
Mon Aug 24 16:24:17 2009 END
Mon Aug 24 16:24:47 2009 event_wait : Interrupted system call (code=4)
Mon Aug 24 16:24:47 2009 OpenVPN CLIENT LIST
Mon Aug 24 16:24:47 2009 Updated,Mon Aug 24 16:24:47 2009
Mon Aug 24 16:24:47 2009 Common Name,Real Address,Bytes Received,Bytes 
Sent,Connected Since
Mon Aug 24 16:24:47 2009 Racom_to_RWE,192.168.0.59:43778,13667,13351,Mon Aug 
24 16:06:18 2009
Mon Aug 24 16:24:47 2009 ROUTING TABLE
Mon Aug 24 16:24:47 2009 Virtual Address,Common Name,Real Address,Last Ref
Mon Aug 24 16:24:47 2009 172.19.7.6,Racom_to_RWE,192.168.0.59:43778,Mon Aug 24 
16:09:47 2009
Mon Aug 24 16:24:47 2009 GLOBAL STATS
Mon Aug 24 16:24:47 2009 Max bcast/mcast queue length,0
Mon Aug 24 16:24:47 2009 END

No similar messages if I remove VPN service and run OpenVPN out of hand.
Any idea, please?



From edsonmarquezani at gmail.com  Mon Aug 24 15:13:20 2009
From: edsonmarquezani at gmail.com (Edson Marquezani Filho)
Date: Mon, 24 Aug 2009 12:13:20 -0300
Subject: [Linux-cluster] Re: Linux-cluster Digest, Preventing LVM from 
	concurrent access
In-Reply-To: <1251115598.5844.31.camel@quann>
References: <20090823160008.CA2B36194C4@hormel.redhat.com>
	<1269910516-1251051757-cardhu_decombobulator_blackberry.rim.net-1598207711-@bxe1291.bisx.prod.on.blackberry>
	<2fc5f090908231201j534884f0ke1648bf5426d0c3e@mail.gmail.com>
	<1251115598.5844.31.camel@quann>
Message-ID: <2fc5f090908240813n45137862s876ec7ffa8c4a625@mail.gmail.com>

>
> You have some interesting and different ideas here:
>
> 1.- Shared storage with non-clustered filesystem: i have no experience
> with HA-LVM, i have just checked its knowledge base on RHEL to refresh
> my ideas.
>
> If i'm not wrong, what you want is to have a shared storage with
> protection from concurrent mounts for your non-clustered filesystems,
> but EVEN for admins doing, by hand, erratic mounts, am i right?

Yes, that's it.

> You can use a different approach to HA-LVM with CLVMD and a resource
> script i have developed called "lvm-cluster". You can find it here:
>
> https://www.redhat.com/archives/cluster-devel/2009-June/msg00065.html
>
> It's not included in the main project yet, but some people in the list
> are testing it successfully.
>
> Actually it would not avoid an admin doing, by hand, an erratic mount
> but a bugzilla ticket has been opened by Brem Belguebli to fix an LVM
> issue that is causing that behaviour. Apart from this admin issue, the
> script would protect the concurrent mount of your filesystem by the
> usage of the LVM layer.
>
> I do like more this approach because you don't need to generate new
> initrd images for your nodes every time you configure new volumes to be
> into the HA-LVM.

Hum... that's good too. By the way, I was concerned about that
requisite, which is absolutely not much pleaseant to do.

Anyway, I'm trying to set up your script here, but I realized that
actualy I still don't know how to do that. =/
(I'm using that tool called Luci, which looks good, specialy to
someone that knows almost nothing about cluster. =)

Ok, I have defined a script resource with the path of your script, and
I guess that now I have to create a service, that's it?
But how, if I don't have a single service to be watched. (Actualy, my
service is my whole server, we could say.)

Could you give me a configuration example? I'm really confused here yet.

Wait, should I write a script for this task and use it on service
configuration? I have tought about a script that always returns OK to
verify-all, and status, and lauch VMs for start. Does it make any
sense?

> 2.- Communications interruption and split-brain: you should use some
> fencing system to make the 2nd node "kill" the main one and take the
> resources.

Ok, I have understood that there's no way to avoid that problem unless
turning off a server. Anyway, this is acceptable, once I don't expect
such a situation happening oftenly.

> Having the main node booted back again AND the comms not reestablished
> will cause you a split-brain again, with a new fencing action. I usually
> do not set cluster services (CMAN, RGMANAGER, etc.) to be automatically
> started on the nodes. I mean they must be re-run and activated by an
> admin.
>
> In the other hand you can use a Quorum Disk, provided my CMAN too, as
> the 3rd vote on your cluster and keep the quorum. I use both approaches
> at the same time (no auto-start of cluster services and qdisk).
>
> 3.- Communications channels: there's no way to set on CMAN more than one
> comm channel, but you can set more than one network interface to be part
> of that channel, via bonding. I think this is one of the lacks of CMAN
> actually.
>
> I hope this helps. Just ask whatever you want.
>

Thank you, very much.



From Alain.Moulle at bull.net  Mon Aug 24 15:37:58 2009
From: Alain.Moulle at bull.net (Alain.Moulle)
Date: Mon, 24 Aug 2009 17:37:58 +0200
Subject: [Linux-cluster] ordered/ restricted and clusvcadm -F
Message-ID: <4A92B3D6.6060903@bull.net>

Hi,
given a failover domain like this one :
<failoverdomain name="FailoverDomain-1" ordered="*???*" restricted="*???*">
         <failoverdomainnode name="node1" priority="1"/>
         <failoverdomainnode name="node2" priority="2"/>
         <failoverdomainnode name="node3" priority="3"/>
</failoverdomain>
and a service /serv1/ configured on this failoverdomain.

I would like that the command:
 /clusvcadm -e serv1
/executed on whatever node of the three nodes starts the /serv1/ on the
prefered node  node1 (priority1)

And  in the case /serv1/ is started on node2 (or node3) (after a failover
for example) and that node1 transitions online , I don't want an
automatic migration of /serv1/ onto node1.

My understanding was to set ordered=1 and restricted=0  but
it seems that clusvcadm -e serv1 starts it on the node where clusvcadm is
executed, even if it is node the "priority 1" node (node1).

Option -F of clusvcadm seemed to give a response to my question, but
it is announced as "deprecated" ...

Thanks for your response.
Regards
Alain
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090824/2c5c9a9d/attachment.htm>

From rmicmirregs at gmail.com  Mon Aug 24 16:04:06 2009
From: rmicmirregs at gmail.com (Rafael =?ISO-8859-1?Q?Mic=F3?= Miranda)
Date: Mon, 24 Aug 2009 18:04:06 +0200
Subject: [Linux-cluster] Re: Linux-cluster Digest, Preventing LVM from 
	concurrent access
In-Reply-To: <2fc5f090908240813n45137862s876ec7ffa8c4a625@mail.gmail.com>
References: <20090823160008.CA2B36194C4@hormel.redhat.com>
	<1269910516-1251051757-cardhu_decombobulator_blackberry.rim.net-1598207711-@bxe1291.bisx.prod.on.blackberry>
	<2fc5f090908231201j534884f0ke1648bf5426d0c3e@mail.gmail.com>
	<1251115598.5844.31.camel@quann>
	<2fc5f090908240813n45137862s876ec7ffa8c4a625@mail.gmail.com>
Message-ID: <1251129846.5898.15.camel@quann>

Hi Edson

El lun, 24-08-2009 a las 12:13 -0300, Edson Marquezani Filho escribi?:

[...]
> 
> Hum... that's good too. By the way, I was concerned about that
> requisite, which is absolutely not much pleaseant to do.
> 
> Anyway, I'm trying to set up your script here, but I realized that
> actualy I still don't know how to do that. =/
> (I'm using that tool called Luci, which looks good, specialy to
> someone that knows almost nothing about cluster. =)


I know about Luci and Ricci. I'm pretty sure you would not be able to
use my script with then, either with the system-config-cluster tool
which i have tested, because the "don't know" about this resource
script. You will need to use it, meanwhile this script gets into the
project or not, configuring the cluster.conf file by hand.

> Ok, I have defined a script resource with the path of your script, and
> I guess that now I have to create a service, that's it?
> But how, if I don't have a single service to be watched. (Actualy, my
> service is my whole server, we could say.)


You dont need to create a new script that uses mine. Take a look to the
cluster.conf file and the RedHat manuals about Cluster Administration
and configure my resource script as part of a service, as your are
suspecting.

You need to define a service the cluster has to handle. You plan to run
Xen Virtual Machines, don't you? You'll need a service, at least, formed
by:

1.- An IP Address where the service is offered
2.- A shared storage where the virtual machines lay
3.- A "binary" or "system call out" to start the Xen service when the
two preceding items are taken. 

> Could you give me a configuration example? I'm really confused here yet.
> 
> Wait, should I write a script for this task and use it on service
> configuration? I have tought about a script that always returns OK to
> verify-all, and status, and lauch VMs for start. Does it make any
> sense?
> 

I provide you as an attached file an example cluster.conf file I have
just prepared to show you as an example of lvm-cluster usage. It is for
a NFS export service to a client. Fencing is not configured, but qdisk
is.

> > 2.- Communications interruption and split-brain: you should use some
> > fencing system to make the 2nd node "kill" the main one and take the
> > resources.
> 
> Ok, I have understood that there's no way to avoid that problem unless
> turning off a server. Anyway, this is acceptable, once I don't expect
> such a situation happening oftenly.
> 

Yes, you should always use fencing when there is an attached shared
storage.

> > Having the main node booted back again AND the comms not reestablished
> > will cause you a split-brain again, with a new fencing action. I usually
> > do not set cluster services (CMAN, RGMANAGER, etc.) to be automatically
> > started on the nodes. I mean they must be re-run and activated by an
> > admin.
> >
> > In the other hand you can use a Quorum Disk, provided my CMAN too, as
> > the 3rd vote on your cluster and keep the quorum. I use both approaches
> > at the same time (no auto-start of cluster services and qdisk).
> >
> > 3.- Communications channels: there's no way to set on CMAN more than one
> > comm channel, but you can set more than one network interface to be part
> > of that channel, via bonding. I think this is one of the lacks of CMAN
> > actually.
> >
> > I hope this helps. Just ask whatever you want.
> >
> 
> Thank you, very much.
> 
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster

Thanks for your interest. Cheers,

Rafael

-- 
Rafael Mic? Miranda
-------------- next part --------------
A non-text attachment was scrubbed...
Name: cluster.conf.example
Type: application/xml
Size: 1661 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090824/41b33968/attachment.wsdl>

From alfredo.moralejo at roche.com  Mon Aug 24 16:11:30 2009
From: alfredo.moralejo at roche.com (Moralejo, Alfredo)
Date: Mon, 24 Aug 2009 18:11:30 +0200
Subject: [Linux-cluster] LVM Mirror in clustered environments
Message-ID: <C64734E4E1C80E49955AD539DB2FBC3A0242AEB0@rkamsem703.emea.roche.com>

Hi,

I'm testing clustered mirror with RHEL 5.3 and I'm finding weird behavior. I've installed and enabled cmirror and created a couple of mirrored logical volumes (after some fight with clvmd errors...):

[root at rmamseslab07 usr]# lvs -a -o +devices
  lvol8                vg00      -wi-ao 4.00G                                                  /dev/cciss/c0d0p2(736)
  lv_mirror            vgcluster mwi-ao 1.00G                    lv_mirror_mlog 100.00         lv_mirror_mimage_0(0),lv_mirror_mimage_1(0)
  [lv_mirror_mimage_0] vgcluster iwi-ao 1.00G                                                  /dev/mapper/mpath0p2(0)
  [lv_mirror_mimage_1] vgcluster iwi-ao 1.00G                                                  /dev/mapper/mpqdiskp5(512)
  [lv_mirror_mlog]     vgcluster lwi-ao 4.00M                                                  /dev/mapper/mpath0p1(0)
  lv_test1             vgcluster -wi-ao 1.00G                                                  /dev/mapper/mpqdiskp5(256)
  lv_test4             vgcluster mwi-ao 1.00G                    lv_test4_mlog  100.00         lv_test4_mimage_0(0),lv_test4_mimage_1(0)
  [lv_test4_mimage_0]  vgcluster iwi-ao 1.00G                                                  /dev/mapper/mpqdiskp5(0)
  [lv_test4_mimage_1]  vgcluster iwi-ao 1.00G                                                  /dev/mapper/mpath0p2(256)
  [lv_test4_mlog]      vgcluster lwi-ao 4.00M                                                  /dev/mapper/mpath0p1(1)
[root at rmamseslab07 usr]#

I'm simulating the failure of a disk by disabling one of the disks having a mirror leg from the storage array or from dm-multipath and the result I get is not the desired one, simply, any lvm command hangs and filesystem appears as hang as well. After enabling the disk again, everything is recovered.

Is anyone using successfully lvm mirror in a RH cluster with RHEL 5.3 with dm-multipath devices?

Is there any alternative for software mirroring to lvm mirror for RHCS. i.e. VxVM or something like that?

Best regards,



Alfredo Moralejo
Business Platforms Engineering - OS Servers - UNIX Senior Specialist
F. Hoffmann-La Roche Ltd.
Global Informatics Group Infrastructure
Josefa Valc?rcel, 40
28027 Madrid SPAIN

Phone: +34 91 305 97 87

alfredo.moralejo at roche.com<mailto:alfredo.moralejo at roche.com>

Confidentiality Note: This message is intended only for the use of the named recipient(s) and may contain confidential and/or proprietary information. If you are not the intended recipient, please contact the sender and delete this message. Any unauthorized use of the information contained in this message is prohibited.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090824/10522fb6/attachment.htm>

From alfredo.moralejo at roche.com  Mon Aug 24 16:23:09 2009
From: alfredo.moralejo at roche.com (Moralejo, Alfredo)
Date: Mon, 24 Aug 2009 18:23:09 +0200
Subject: [Linux-cluster] RE: LVM Mirror in clustered environments
In-Reply-To: <C64734E4E1C80E49955AD539DB2FBC3A0242AEB0@rkamsem703.emea.roche.com>
References: <C64734E4E1C80E49955AD539DB2FBC3A0242AEB0@rkamsem703.emea.roche.com>
Message-ID: <C64734E4E1C80E49955AD539DB2FBC3A0242AEB4@rkamsem703.emea.roche.com>

Hi,

After some additional testing I've checked that the problem is related with dm-multipath, using /dev/sd devices instead of the multipath devices everything went fine.

Is cmirror supposed to work fine with dm-multipath or is it not supported?, is this a bug?

Best regards,

Alfredo



________________________________
From: linux-cluster-bounces at redhat.com [mailto:linux-cluster-bounces at redhat.com] On Behalf Of Moralejo, Alfredo
Sent: Monday, August 24, 2009 6:12 PM
To: linux-cluster at redhat.com
Subject: [Linux-cluster] LVM Mirror in clustered environments

Hi,

I'm testing clustered mirror with RHEL 5.3 and I'm finding weird behavior. I've installed and enabled cmirror and created a couple of mirrored logical volumes (after some fight with clvmd errors...):

[root at rmamseslab07 usr]# lvs -a -o +devices
  lvol8                vg00      -wi-ao 4.00G                                                  /dev/cciss/c0d0p2(736)
  lv_mirror            vgcluster mwi-ao 1.00G                    lv_mirror_mlog 100.00         lv_mirror_mimage_0(0),lv_mirror_mimage_1(0)
  [lv_mirror_mimage_0] vgcluster iwi-ao 1.00G                                                  /dev/mapper/mpath0p2(0)
  [lv_mirror_mimage_1] vgcluster iwi-ao 1.00G                                                  /dev/mapper/mpqdiskp5(512)
  [lv_mirror_mlog]     vgcluster lwi-ao 4.00M                                                  /dev/mapper/mpath0p1(0)
  lv_test1             vgcluster -wi-ao 1.00G                                                  /dev/mapper/mpqdiskp5(256)
  lv_test4             vgcluster mwi-ao 1.00G                    lv_test4_mlog  100.00         lv_test4_mimage_0(0),lv_test4_mimage_1(0)
  [lv_test4_mimage_0]  vgcluster iwi-ao 1.00G                                                  /dev/mapper/mpqdiskp5(0)
  [lv_test4_mimage_1]  vgcluster iwi-ao 1.00G                                                  /dev/mapper/mpath0p2(256)
  [lv_test4_mlog]      vgcluster lwi-ao 4.00M                                                  /dev/mapper/mpath0p1(1)
[root at rmamseslab07 usr]#

I'm simulating the failure of a disk by disabling one of the disks having a mirror leg from the storage array or from dm-multipath and the result I get is not the desired one, simply, any lvm command hangs and filesystem appears as hang as well. After enabling the disk again, everything is recovered.

Is anyone using successfully lvm mirror in a RH cluster with RHEL 5.3 with dm-multipath devices?

Is there any alternative for software mirroring to lvm mirror for RHCS. i.e. VxVM or something like that?

Best regards,



Alfredo Moralejo
Business Platforms Engineering - OS Servers - UNIX Senior Specialist
F. Hoffmann-La Roche Ltd.
Global Informatics Group Infrastructure
Josefa Valc?rcel, 40
28027 Madrid SPAIN

Phone: +34 91 305 97 87

alfredo.moralejo at roche.com<mailto:alfredo.moralejo at roche.com>

Confidentiality Note: This message is intended only for the use of the named recipient(s) and may contain confidential and/or proprietary information. If you are not the intended recipient, please contact the sender and delete this message. Any unauthorized use of the information contained in this message is prohibited.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090824/50a80a5e/attachment.htm>

From rhurst at bidmc.harvard.edu  Mon Aug 24 16:32:39 2009
From: rhurst at bidmc.harvard.edu (Robert Hurst)
Date: Mon, 24 Aug 2009 12:32:39 -0400
Subject: [Linux-cluster] RHEL 4.7 fenced fails -- stuck join state: S-2,2,1
In-Reply-To: <1250002549.2782.36.camel@WSBID06223.bidmc.harvard.edu>
References: <1250002549.2782.36.camel@WSBID06223.bidmc.harvard.edu>
Message-ID: <1251131559.2795.47.camel@WSBID06223.bidmc.harvard.edu>

RHEL support pointed me to a document suggesting this may be an
implementation issue:

http://kbase.redhat.com/faq/docs/DOC-5935


        "DNS is not a reliable way to get name resolution for the
        cluster. All cluster nodes must be defined in /etc/hosts with
        the name that matches cluster.conf and uname -n."




But we use a local DNS service on all of our hosts (for years) and
leave /etc/hosts alone with only the localhost entry in it.  Our servers
have multiple bonded NICs, so I put the "hosts" in their own private
domain / IP address, and each clusternode entry in cluster.conf is
simply added as: acropolis, cerberus, rycon, and solaria.  I let DNS
(and reverse DNS) resolve those names, i.e.,

/etc/resolve 
search blade  ccc.cluster  bidmc.harvard.edu  bidn.caregroup.org 
nameserver 127.0.0.1

$ host acropolis 
acropolis.blade has address 192.168.2.1 
$ host cerberus 
cerberus.blade has address 192.168.2.4 
$ host rycon 
rycon.blade has address 192.168.2.11 
$ host solaria 
solaria.blade has address 192.168.2.12 
 
root at acropolis [~]$ netstat -a | grep 6809 
udp    0    0 acropolis.blade:6809    *:* 
udp    0    0 192.168.255.255:6809    *:* 
 
[root at cerberus ~]# netstat -a | grep 6809 
udp    0    0 cerberus.blade:6809     *:* 
udp    0    0 192.168.255.255:6809    *:* 
 
[root at solaria ~]# netstat -a | grep 6809 
udp    0    0 solaria.blade:6809      *:* 
udp    0    0 192.168.255.255:6809    *:*


... even though each of those server's $( uname -n )
has .bidmc.harvard.edu (for corporate LAN-facing NICs) appended to them.
Is this REALLY a cause for concern?  If so, could this introduce a
failure (if not at join) during some later event?  Any feedback is
welcome!


On Tue, 2009-08-11 at 10:55 -0400, Robert Hurst wrote:

> Simple 4-node cluster, 2-nodes have a GFS shared home directory
> mounted for over a month.  Today, I wanted to mount /home on a 3rd
> node, so:
> 
> # service fenced start                [failed]
> 
> Weird.  Checking /var/log/messages show:
> 
> Aug 11 10:19:06 cerberus kernel: Lock_Harness 2.6.9-80.9.el4_7.10
> (built Jan 22 2009 18:39:16) installed
> Aug 11 10:19:06 cerberus kernel: GFS 2.6.9-80.9.el4_7.10 (built Jan 22
> 2009 18:39:32) installed
> Aug 11 10:19:06 cerberus kernel: GFS: Trying to join cluster
> "lock_dlm", "ccc_cluster47:home"
> Aug 11 10:19:06 cerberus kernel: Lock_DLM (built Jan 22 2009 18:39:18)
> installed
> Aug 11 10:19:06 cerberus kernel: lock_dlm: fence domain not found;
> check fenced
> Aug 11 10:19:06 cerberus kernel: GFS: can't mount proto = lock_dlm,
> table = ccc_cluster47:home, hostdata = 
> 
> # cman_tool services
> Service          Name                              GID LID State
> Code
> Fence Domain:    "default"                           0   2 join
> S-2,2,1
> []
> 
> So, a fenced process is now hung:
> 
> root     28302  0.0  0.0  3668  192 ?        Ss   10:19   0:00 fenced
> -t 120 -w
> 
> Q: Any idea how to "recover" from this state, without rebooting?
> 
> The other two servers are unaffected by this (thankfully) and show
> normal operations:
> 
> $ cman_tool services
> 
> Service          Name                              GID LID State
> Code
> Fence Domain:    "default"                           2   2 run       -
> [1 12]
> 
> DLM Lock Space:  "home"                              5   5 run       -
> [1 12]
> 
> GFS Mount Group: "home"                              6   6 run       -
> [1 12]
> 
> 
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090824/f0702401/attachment.htm>

From tpeirce at etrafficsolutions.com  Mon Aug 24 19:24:14 2009
From: tpeirce at etrafficsolutions.com (Trevor Peirce)
Date: Mon, 24 Aug 2009 12:24:14 -0700
Subject: [Linux-cluster] cluster-3.0.0 problems starting up
In-Reply-To: <1250638107.19265.98.camel@localhost.localdomain>
References: <4A85E1C7.1070502@etrafficsolutions.com>	<d740f8db0908141523y7a0d4a7eta656459bf16bc145@mail.gmail.com>	<4A85E4EF.4020501@etrafficsolutions.com>	<4A8B1294.5070805@etrafficsolutions.com>
	<1250638107.19265.98.camel@localhost.localdomain>
Message-ID: <4A92E8DE.30809@etrafficsolutions.com>

Steven Dake wrote:
> This means your XML format is invalid.  We are working on better error
> reporting when the XML format is invalid rather then this cryptic error.
> In the meantime, you can run strace -f cman, and libxml will print the
> errors in write system calls.  If you need help analyzing this output,
> send a link to the output from strace.
>   

Ha! The problem was /var/log/cluster/ did not exist. Once I created this
directory cman_tool was able to create corosync.log inside there and
away we went.

Thanks for the guidance, strace pointed this out when I saw an error
about a file not opening.

-- 
Trevor Peirce
Systems Administrator
Etraffic Solutions Inc.

Suite 100 ? 710 Redbrick Street
Victoria, B.C. V8T 5J3
phone: 250.658.8238 x49
fax: 250.658.5936



From edsonmarquezani at gmail.com  Mon Aug 24 19:27:37 2009
From: edsonmarquezani at gmail.com (Edson Marquezani Filho)
Date: Mon, 24 Aug 2009 16:27:37 -0300
Subject: [Linux-cluster] Re: Linux-cluster Digest, Preventing LVM from 
	concurrent access
In-Reply-To: <1251129846.5898.15.camel@quann>
References: <20090823160008.CA2B36194C4@hormel.redhat.com>
	<1269910516-1251051757-cardhu_decombobulator_blackberry.rim.net-1598207711-@bxe1291.bisx.prod.on.blackberry>
	<2fc5f090908231201j534884f0ke1648bf5426d0c3e@mail.gmail.com>
	<1251115598.5844.31.camel@quann>
	<2fc5f090908240813n45137862s876ec7ffa8c4a625@mail.gmail.com>
	<1251129846.5898.15.camel@quann>
Message-ID: <2fc5f090908241227t1c709415t2475c3cf91126667@mail.gmail.com>

>
> I know about Luci and Ricci. I'm pretty sure you would not be able to
> use my script with then, either with the system-config-cluster tool
> which i have tested, because the "don't know" about this resource
> script. You will need to use it, meanwhile this script gets into the
> project or not, configuring the cluster.conf file by hand.
>

Unfortunately, all RedHat documentation I've have seen just tells how
to do administration tasks with those graphical tools. That's why I'm
facing some difficulties to understand the file, although its
structure is simple.

> You dont need to create a new script that uses mine. Take a look to the
> cluster.conf file and the RedHat manuals about Cluster Administration
> and configure my resource script as part of a service, as your are
> suspecting.
>
> You need to define a service the cluster has to handle. You plan to run
> Xen Virtual Machines, don't you? You'll need a service, at least, formed
> by:
>
> 1.- An IP Address where the service is offered
> 2.- A shared storage where the virtual machines lay
> 3.- A "binary" or "system call out" to start the Xen service when the
> two preceding items are taken.
>

Maybe I have not been clear enough. When I said that I was thinking
about writing a script, I was refering to a service script, like those
found in /etc/init.d.

>
> I provide you as an attached file an example cluster.conf file I have
> just prepared to show you as an example of lvm-cluster usage. It is for
> a NFS export service to a client. Fencing is not configured, but qdisk
> is.
>

Ok, I have tried to make my own cluster.conf, and I would apreciate if
you could validate it to me. I took the conf generated here by Luci,
your example, and another example from here[1], and I have made
mine[2].
It has a fence action, but not a quorum disk yet, the two lvm-cluster
resources I need, and one service definition, which would launch the
VMs with a script made by me. I didn't specified a logical volume,
because I want the whole VG locked. I can do that, can't I ?

Obs: nobody has told me, but I guess that I should put your
lvm-cluster.sh in /usr/share/cluster/, am I right?

Am I thinking correctly now?

Thank you, again.

Obs: I don't need an IP section because all my services (that are
IP-dependant, off course) run inside the VMs, and they have their own
IPs, it don't matter which phisical server they are running on. Ok?

[1] http://sources.redhat.com/cluster/wiki/QuickStart-Portuguese
[2] http://pastebin.com/f1601e50e



From rmicmirregs at gmail.com  Mon Aug 24 20:05:03 2009
From: rmicmirregs at gmail.com (Rafael =?ISO-8859-1?Q?Mic=F3?= Miranda)
Date: Mon, 24 Aug 2009 22:05:03 +0200
Subject: [Linux-cluster] Re: Linux-cluster Digest, Preventing LVM from 
	concurrent access
In-Reply-To: <2fc5f090908241227t1c709415t2475c3cf91126667@mail.gmail.com>
References: <20090823160008.CA2B36194C4@hormel.redhat.com>
	<1269910516-1251051757-cardhu_decombobulator_blackberry.rim.net-1598207711-@bxe1291.bisx.prod.on.blackberry>
	<2fc5f090908231201j534884f0ke1648bf5426d0c3e@mail.gmail.com>
	<1251115598.5844.31.camel@quann>
	<2fc5f090908240813n45137862s876ec7ffa8c4a625@mail.gmail.com>
	<1251129846.5898.15.camel@quann>
	<2fc5f090908241227t1c709415t2475c3cf91126667@mail.gmail.com>
Message-ID: <1251144303.5840.28.camel@quann>

Hi Edson,

El lun, 24-08-2009 a las 16:27 -0300, Edson Marquezani Filho escribi?:

> >
> > You need to define a service the cluster has to handle. You plan to run
> > Xen Virtual Machines, don't you? You'll need a service, at least, formed
> > by:
> >
> > 1.- An IP Address where the service is offered
> > 2.- A shared storage where the virtual machines lay
> > 3.- A "binary" or "system call out" to start the Xen service when the
> > two preceding items are taken.
> >
> 
> Maybe I have not been clear enough. When I said that I was thinking
> about writing a script, I was refering to a service script, like those
> found in /etc/init.d.
> 

I'm sorry I did not understand you. Yes, you could use your own script
to start the Xen services as the last resource you need to run to start
your service.

You will need to include that resource into your service as a "script"
resource. Take a look at it in this link:

http://www.redhat.com/docs/en-US/Red_Hat_Enterprise_Linux/5/html/Cluster_Administration/ap-ha-resource-params-CA.html#tb-script-resource-CA

Remember to build your script LSB compliant as the page says. I can't
find any documentation relative to RedHat about this LSB compliance,
maybe it has its "own standard".


> 
> Ok, I have tried to make my own cluster.conf, and I would apreciate if
> you could validate it to me. I took the conf generated here by Luci,
> your example, and another example from here[1], and I have made
> mine[2].

Sorry I cant validate it, my lab is not available now. I took a look at
it and there are some points you must change: 

1.- lvm-cluster resource definition: the arguments vg_name and lv_name
are mandatory. You must activate a logical volume into the volume group.
As far as i know, you cant build the filesystem on top of the volume
group directly, and I built the script with some checks to avoid that. 

2.- Filesystem resource: you need to mount the filesystem stored into
the logical volume you activate. You need a filesystem to store the VMs,
don't you? Or do you use a "raw device" in your Xen service?

I updated your pastebin file. I'm not sure if it will be fully
funcional. 

Maybe you should even nest the resources building your service to be
sure they are started in order. It will depend on your service. Check my
example.



> It has a fence action, but not a quorum disk yet, the two lvm-cluster
> resources I need, and one service definition, which would launch the
> VMs with a script made by me. I didn't specified a logical volume,
> because I want the whole VG locked. I can do that, can't I ?
> 
> Obs: nobody has told me, but I guess that I should put your
> lvm-cluster.sh in /usr/share/cluster/, am I right?

Yes, you are right. Keep an eye on the rights of the file just to be the
same as the others.

> 
> Am I thinking correctly now?
> 
> Thank you, again.
> 
> Obs: I don't need an IP section because all my services (that are
> IP-dependant, off course) run inside the VMs, and they have their own
> IPs, it don't matter which phisical server they are running on. Ok?

Yes, but if you have a management console for your VM service you will
need a Virtual IP which "floats" with the service, so you can always
connect to the same IP to check the status of your virtualization
service. You can make that attaching a IP resource to your service.

> 
> [1] http://sources.redhat.com/cluster/wiki/QuickStart-Portuguese
> [2] http://pastebin.com/f1601e50e
> 
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster

Cheers,

Rafael

-- 
Rafael Mic? Miranda



From edsonmarquezani at gmail.com  Mon Aug 24 20:48:47 2009
From: edsonmarquezani at gmail.com (Edson Marquezani Filho)
Date: Mon, 24 Aug 2009 17:48:47 -0300
Subject: [Linux-cluster] Re: Linux-cluster Digest, Preventing LVM from 
	concurrent access
In-Reply-To: <1251144303.5840.28.camel@quann>
References: <20090823160008.CA2B36194C4@hormel.redhat.com>
	<1269910516-1251051757-cardhu_decombobulator_blackberry.rim.net-1598207711-@bxe1291.bisx.prod.on.blackberry>
	<2fc5f090908231201j534884f0ke1648bf5426d0c3e@mail.gmail.com>
	<1251115598.5844.31.camel@quann>
	<2fc5f090908240813n45137862s876ec7ffa8c4a625@mail.gmail.com>
	<1251129846.5898.15.camel@quann>
	<2fc5f090908241227t1c709415t2475c3cf91126667@mail.gmail.com>
	<1251144303.5840.28.camel@quann>
Message-ID: <2fc5f090908241348p73461b62u76835829be10a91d@mail.gmail.com>

>
> Sorry I cant validate it, my lab is not available now. I took a look at
> it and there are some points you must change:
>
> 1.- lvm-cluster resource definition: the arguments vg_name and lv_name
> are mandatory. You must activate a logical volume into the volume group.
> As far as i know, you cant build the filesystem on top of the volume
> group directly, and I built the script with some checks to avoid that.

When I said validate, I meant just check it out. I didn't expect you
to run it. Probabily it was not the appropiat word to say what I
wanted. (I'm sorry for this poor english.)

> 2.- Filesystem resource: you need to mount the filesystem stored into
> the logical volume you activate. You need a filesystem to store the VMs,
> don't you? Or do you use a "raw device" in your Xen service?

No, I don't and can't do that. LVs are passed directly to VMs'
configuration files, and Xen takes care of it. Actualy, each LV is
seen like a phisical disk by the VM. It's Xen that access and "mounts"
it.

> I updated your pastebin file. I'm not sure if it will be fully
> funcional.
>
> Maybe you should even nest the resources building your service to be
> sure they are started in order. It will depend on your service. Check my
> example.

Hum... so nesting indicates order. It's good to know. =)

>
> Yes, but if you have a management console for your VM service you will
> need a Virtual IP which "floats" with the service, so you can always
> connect to the same IP to check the status of your virtualization
> service. You can make that attaching a IP resource to your service.
>

What are the changes that needs to be done to lvm.conf, as advised in
the script code?

I have seen that there is already a resource called vm. Maybe,
lvm-cluster and vm resouces can do everything I need. I hope that it
can. =)



From rmicmirregs at gmail.com  Tue Aug 25 10:36:26 2009
From: rmicmirregs at gmail.com (Rafael =?ISO-8859-1?Q?Mic=F3?= Miranda)
Date: Tue, 25 Aug 2009 12:36:26 +0200
Subject: [Linux-cluster] Re: Linux-cluster Digest, Preventing LVM from 
	concurrent access
In-Reply-To: <2fc5f090908241348p73461b62u76835829be10a91d@mail.gmail.com>
References: <20090823160008.CA2B36194C4@hormel.redhat.com>
	<1269910516-1251051757-cardhu_decombobulator_blackberry.rim.net-1598207711-@bxe1291.bisx.prod.on.blackberry>
	<2fc5f090908231201j534884f0ke1648bf5426d0c3e@mail.gmail.com>
	<1251115598.5844.31.camel@quann>
	<2fc5f090908240813n45137862s876ec7ffa8c4a625@mail.gmail.com>
	<1251129846.5898.15.camel@quann>
	<2fc5f090908241227t1c709415t2475c3cf91126667@mail.gmail.com>
	<1251144303.5840.28.camel@quann>
	<2fc5f090908241348p73461b62u76835829be10a91d@mail.gmail.com>
Message-ID: <1251196586.5835.6.camel@quann>

Hi Edson,

El lun, 24-08-2009 a las 17:48 -0300, Edson Marquezani Filho escribi?:

> 
> When I said validate, I meant just check it out. I didn't expect you
> to run it. Probabily it was not the appropiat word to say what I
> wanted. (I'm sorry for this poor english.)
> 

Ok, don't worry about that. I can't clearly explain myself too.

> > 2.- Filesystem resource: you need to mount the filesystem stored into
> > the logical volume you activate. You need a filesystem to store the VMs,
> > don't you? Or do you use a "raw device" in your Xen service?
> 
> No, I don't and can't do that. LVs are passed directly to VMs'
> configuration files, and Xen takes care of it. Actualy, each LV is
> seen like a phisical disk by the VM. It's Xen that access and "mounts"
> it.
> 

Ok, so you need to use "raw device" for Xen. In this case, your raw
device are the Logical Volumes on top of the Volume Groups. You wont
need any filesystem on them, but anyway you need the LVs. Use the
lvm-cluster resource to take exclusively that LVs and after that use
them as raw devices for Xen. 

> > I updated your pastebin file. I'm not sure if it will be fully
> > funcional.
> >
> > Maybe you should even nest the resources building your service to be
> > sure they are started in order. It will depend on your service. Check my
> > example.
> 
> Hum... so nesting indicates order. It's good to know. =)
> 
> >
> > Yes, but if you have a management console for your VM service you will
> > need a Virtual IP which "floats" with the service, so you can always
> > connect to the same IP to check the status of your virtualization
> > service. You can make that attaching a IP resource to your service.
> >
> 
> What are the changes that needs to be done to lvm.conf, as advised in
> the script code?
> 

You need, at least to set "locking_type = 3" and possibly change your
blacklist of devices. Take a look at the RH documentation:

http://www.redhat.com/docs/en-US/Red_Hat_Enterprise_Linux/5/html/Cluster_Logical_Volume_Manager/lvmconf_file.html



> I have seen that there is already a resource called vm. Maybe,
> lvm-cluster and vm resouces can do everything I need. I hope that it
> can. =)
> 

I don't have any knowledge about that resource script, I'm sorry I can't
help.

> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster

Cheers,

Rafael
-- 
Rafael Mic? Miranda



From linux-cluster at lists.grepular.com  Tue Aug 25 11:19:42 2009
From: linux-cluster at lists.grepular.com (Mike Cardwell)
Date: Tue, 25 Aug 2009 12:19:42 +0100
Subject: [Linux-cluster] NFS client failover
Message-ID: <4A93C8CE.7010202@lists.grepular.com>

Hi,

I have a small cluster which shares a filesystem via GFS on a SAN, with 
iSCSI. There are a couple of hosts external to this cluster which I 
would like to have access to the GFS filesystem via NFS. I have exported 
the same mountpoint on each of the hosts that have the GFS mount.

I can connect via NFS to one of these cluster hosts, but when it goes 
down, I have to unmount it, and then remount against another host. I 
suspect I could do something with a floating IP? but I was wondering if 
it would be possible to do this with client side logic so that the nfs 
mount automatically moves to a different working host if one stops 
responding?

I tried creating multiple A records in the DNS for a single host, and 
connecting to that hostname, but that doesn't seem to work. It seems to 
pick one of the IPs and stick to it. Any ideas?

-- 
Mike Cardwell - IT Consultant and LAMP developer
Cardwell IT Ltd. (UK Reg'd Company #06920226) http://cardwellit.com/



From RJM002 at shsu.edu  Tue Aug 25 14:51:58 2009
From: RJM002 at shsu.edu (Marti, Robert)
Date: Tue, 25 Aug 2009 09:51:58 -0500
Subject: [Linux-cluster] NFS client failover
In-Reply-To: <4A93C8CE.7010202@lists.grepular.com>
References: <4A93C8CE.7010202@lists.grepular.com>
Message-ID: <8FAC1E47484E43469AA28DBF35C955E4A498CE2089@EXMBX.SHSU.EDU>

Assign a virtual IP to that service, and it move with the "hosting" node in the cluster.

Rob Marti


-----Original Message-----
From: linux-cluster-bounces at redhat.com [mailto:linux-cluster-bounces at redhat.com] On Behalf Of Mike Cardwell
Sent: Tuesday, August 25, 2009 6:20 AM
To: linux-cluster at redhat.com
Subject: [Linux-cluster] NFS client failover

Hi,

I have a small cluster which shares a filesystem via GFS on a SAN, with iSCSI. There are a couple of hosts external to this cluster which I would like to have access to the GFS filesystem via NFS. I have exported the same mountpoint on each of the hosts that have the GFS mount.

I can connect via NFS to one of these cluster hosts, but when it goes down, I have to unmount it, and then remount against another host. I suspect I could do something with a floating IP? but I was wondering if it would be possible to do this with client side logic so that the nfs mount automatically moves to a different working host if one stops responding?

I tried creating multiple A records in the DNS for a single host, and connecting to that hostname, but that doesn't seem to work. It seems to pick one of the IPs and stick to it. Any ideas?

--
Mike Cardwell - IT Consultant and LAMP developer Cardwell IT Ltd. (UK Reg'd Company #06920226) http://cardwellit.com/

--
Linux-cluster mailing list
Linux-cluster at redhat.com
https://www.redhat.com/mailman/listinfo/linux-cluster



From cryptogrid at gmail.com  Tue Aug 25 19:14:30 2009
From: cryptogrid at gmail.com (crypto grid)
Date: Tue, 25 Aug 2009 16:14:30 -0300
Subject: [Linux-cluster] NFS client failover
In-Reply-To: <4A93C8CE.7010202@lists.grepular.com>
References: <4A93C8CE.7010202@lists.grepular.com>
Message-ID: <a9f464b80908251214q33650614p52e449d62ee30d7f@mail.gmail.com>

On Tue, Aug 25, 2009 at 8:19 AM, Mike Cardwell <
linux-cluster at lists.grepular.com> wrote:

> Hi,
>
> I have a small cluster which shares a filesystem via GFS on a SAN, with
> iSCSI. There are a couple of hosts external to this cluster which I would
> like to have access to the GFS filesystem via NFS. I have exported the same
> mountpoint on each of the hosts that have the GFS mount.
>
> I can connect via NFS to one of these cluster hosts, but when it goes down,
> I have to unmount it, and then remount against another host. I suspect I
> could do something with a floating IP? but I was wondering if it would be
> possible to do this with client side logic so that the nfs mount
> automatically moves to a different working host if one stops responding?


I don't see why you want to resolve this from the client side. You must
configure an ip address resource.
Take a look at this document:
http://www.redhat.com/docs/en-US/Red_Hat_Enterprise_Linux/5/pdf/Configuration_Example_-_NFS_Over_GFS.pdf
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090825/33fe8333/attachment.htm>

From linux-cluster at lists.grepular.com  Tue Aug 25 19:40:33 2009
From: linux-cluster at lists.grepular.com (Mike Cardwell)
Date: Tue, 25 Aug 2009 20:40:33 +0100
Subject: [Linux-cluster] NFS client failover
In-Reply-To: <a9f464b80908251214q33650614p52e449d62ee30d7f@mail.gmail.com>
References: <4A93C8CE.7010202@lists.grepular.com>
	<a9f464b80908251214q33650614p52e449d62ee30d7f@mail.gmail.com>
Message-ID: <4A943E31.3080209@lists.grepular.com>

crypto grid wrote:

>     I have a small cluster which shares a filesystem via GFS on a SAN,
>     with iSCSI. There are a couple of hosts external to this cluster
>     which I would like to have access to the GFS filesystem via NFS. I
>     have exported the same mountpoint on each of the hosts that have the
>     GFS mount.
> 
>     I can connect via NFS to one of these cluster hosts, but when it
>     goes down, I have to unmount it, and then remount against another
>     host. I suspect I could do something with a floating IP? but I was
>     wondering if it would be possible to do this with client side logic
>     so that the nfs mount automatically moves to a different working
>     host if one stops responding?
> 
> I don't see why you want to resolve this from the client side.

I figured that failover would happen more smoothly if the client was 
aware of and in control of what was going on. If the IP suddenly moves 
to another NFS server I don't know how the NFS client will cope with that.

> You must configure an ip address resource.

Yeah, that's what I was going to try if there wasn't a client side solution.

> Take a look at this document:
> http://www.redhat.com/docs/en-US/Red_Hat_Enterprise_Linux/5/pdf/Configuration_Example_-_NFS_Over_GFS.pdf

That document looks very useful. Thank you for pointing it out to me.

-- 
Mike Cardwell - IT Consultant and LAMP developer
Cardwell IT Ltd. (UK Reg'd Company #06920226) http://cardwellit.com/



From carlopmart at gmail.com  Tue Aug 25 19:47:13 2009
From: carlopmart at gmail.com (carlopmart)
Date: Tue, 25 Aug 2009 21:47:13 +0200
Subject: [Linux-cluster] Using iscsi devices with scsi reservation and
	fence_scsi agent
Message-ID: <4A943FC1.3060104@gmail.com>

Hi all,

  Somebody knows if can I use an iscsi target (iscsitarget.sf.net) with 
scsi fence reservation?? Does it owrks on production environments??

Thanks
-- 
CL Martinez
carlopmart {at} gmail {d0t} com



From gordan at bobich.net  Tue Aug 25 22:43:17 2009
From: gordan at bobich.net (Gordan Bobic)
Date: Tue, 25 Aug 2009 23:43:17 +0100
Subject: [Linux-cluster] Fencing Logs
Message-ID: <4A946905.4040002@bobich.net>

I have a really strange problem on one of my clusters. It exhibits all 
signs of fencing being broken, but the fencing agents work when tested 
manually, and I cannot find anything in syslog to even suggest that 
fencing is being attempted by the surviving node (which just locks up on 
GFS access until the other node returns).

Has anybody got any suggestions on how to troubleshoot this?

The relevant extract from my cluster.conf is:

<clusternodes>
         <clusternode name="hades-cls" nodeid="1" votes="1">
...
                 <fence>
                         <method name = "1">
                                 <device name = "hades-oob"/>
                         </method>
                 </fence>
...
         </clusternode>
         <clusternode name="persephone-cls" nodeid="2" votes="1">
...
                 <fence>
                         <method name = "1">
                                 <device name ="persephone-oob"/>
                         </method>
                 </fence>
         </clusternode>
</clusternodes>
<fencedevices>
         <fencedevice agent="fence_eric" ipaddr="10.1.254.251" 
login="fence" passwd="some_password" name="hades-oob"/>
         <fencedevice agent="fence_eric" ipaddr="10.1.254.252" 
login="fence" passwd="some_password" name="persephone-oob"/>
</fencedevices>
...

I have a near identical setup on all my other clusters, so this is 
somewhat baffling. What else could be relevant to this, specifically in 
the context of no fencing attempts even showing up in the logs? I have 
set up scores of RHCS clusters and never seen anything like this before. 
The only unusual thing about this cluster is that I had to write a 
bespoke fencing agent for the machines, but these test true when I use 
them to down/reboot the machines.

TIA.

Gordan



From george.saji00 at gmail.com  Wed Aug 26 05:17:13 2009
From: george.saji00 at gmail.com (saji george)
Date: Wed, 26 Aug 2009 10:47:13 +0530
Subject: [Linux-cluster] Apache Cluster
In-Reply-To: <a9f464b80908211226h39b5d60av7c2c07823b72c778@mail.gmail.com>
References: <2460eaad0908190854w6610a90ft9c6a9b10d576b7bf@mail.gmail.com>
	<a9f464b80908211226h39b5d60av7c2c07823b72c778@mail.gmail.com>
Message-ID: <2460eaad0908252217w1c959990i6aca6b67c71ba1a1@mail.gmail.com>

Hi,
    I have added the ip, still its not working with the Resource
Apache server, But the same is working with the Resource script. And
in script sections only the /etc/init.d/httpd script is working.

Regards
Saji George

On 8/22/09, crypto grid <cryptogrid at gmail.com> wrote:
> On Wed, Aug 19, 2009 at 12:54 PM, saji george
> <george.saji00 at gmail.com>wrote:
>
>> Hi
>>   I have configured the GFS cluster successfully in RHEL 5.3
>> But I am not able to configure the cluster for apache.while enabling the
>> service I am getting the below error
>>
>> # clusvcadm -e webcluster
>> Local machine trying to enable service:webcluster...Aborted; service
>> failed
>>
>> /log/messages
>>
>> Aug 19 21:12:49 cmsstorage-web1 clurgmgrd[32766]: <err> #43: Service
>> service:webcluster has failed; can not start.
>> Aug 19 21:12:49 cmsstorage-web1 clurgmgrd[32766]: <crit> #13: Service
>> service:webcluster failed to stop cleanly
>>
>> cluster.conf
>>
>> <failoverdomain name="GFS-apache" ordered="1" restricted="1">
>>                                 <failoverdomainnode name="cmsstorage-web3"
>> priority="3"/>
>>                                 <failoverdomainnode name="cmsstorage-web2"
>> priority="2"/>
>>                                 <failoverdomainnode name="cmsstorage-web1"
>> priority="1"/>
>>                         </failoverdomain>
>>
>> <resources>
>> <apache config_file="conf/httpd.conf" httpd_options="" name="webcluster"
>> server_root="/usr/local/apache/" shutdown_wait="2"/>
>> </resources>
>>
>>  <service autostart="1" domain="GFS-apache" name="webcluster"
>> recovery="restart">
>>                         <apache ref="webcluster"/>
>>                 </service>
>>
>> Is there anything missing...
>
>
>
> You are missing the definition of an ip address in the resource section, and
> after that, the use of the defined ip address into the service definition.
>
> Regards,
>
>
>
>>
>>
>> Thanks in advance.
>>
>> Regards,
>> Saji George
>>
>>
>> --
>> Linux-cluster mailing list
>> Linux-cluster at redhat.com
>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>
>



From linux-cluster at lists.grepular.com  Wed Aug 26 08:55:24 2009
From: linux-cluster at lists.grepular.com (Mike Cardwell)
Date: Wed, 26 Aug 2009 09:55:24 +0100
Subject: [Linux-cluster] NFS client failover
In-Reply-To: <4A943E31.3080209@lists.grepular.com>
References: <4A93C8CE.7010202@lists.grepular.com>	<a9f464b80908251214q33650614p52e449d62ee30d7f@mail.gmail.com>
	<4A943E31.3080209@lists.grepular.com>
Message-ID: <4A94F87C.7030708@lists.grepular.com>

On 25/08/2009 20:40, Mike Cardwell wrote:

> I figured that failover would happen more smoothly if the client was
> aware of and in control of what was going on. If the IP suddenly moves
> to another NFS server I don't know how the NFS client will cope with that.

Well, it seems to cope quite well. The nfs mount "hangs" for a few 
seconds whilst the IP moves from one server to another (unavoidable 
obviously), but it then picks up from where it was. I suspect there will 
be file corruption issues with files that are partially written when the 
failover happens, but I guess that can't be avoided without a client 
side solution.

-- 
Mike Cardwell - IT Consultant and LAMP developer
Cardwell IT Ltd. (UK Reg'd Company #06920226) http://cardwellit.com/



From jakov.sosic at srce.hr  Wed Aug 26 13:19:03 2009
From: jakov.sosic at srce.hr (Jakov Sosic)
Date: Wed, 26 Aug 2009 15:19:03 +0200
Subject: [Linux-cluster] CLVM not activating LVs
Message-ID: <20090826151903.6e4e7759@pc-jsosic.srce.hr>

Hi! CLVM not activating logical volumes.



I have a major issues with CLVM. It is not activating volumes in my
VG's. I have 2 iSCSI volumes, and one SAS volume with 3 VG's. On
node01, all logical volumes in 1 iSCSI and on SAS are activated. Other
iSCSI - zero. On node02 same situation. On node03 only lv's from SAS
are activated. lvm.conf is same on the machines....

This is very strange, because when I boot the machines, all the
services are stopped, so logical volumes shouldn't be activated.

Here is the situation:

[root at node01 lvm]# vgs
  VG          #PV #LV #SN Attr   VSize  VFree
  VolGroupC0    1   7   0 wz--nc  3.41T 1.48T
  VolGroupC1    1   0   0 wz--nc  3.41T 3.41T
  VolGroupSAS   1   2   0 wz--nc 20.63G 4.63G

[root at node01 lvm]# lvs
  LV            VG          Attr   LSize
  nered1        VolGroupC0  -wi-a- 200.00G
  nered2        VolGroupC0  -wi-a- 200.00G
  nered3        VolGroupC0  -wi-a-   1.46T
  nered4        VolGroupC0  -wi-a-  20.00G
  nered5        VolGroupC0  -wi-a-  20.00G
  nered6        VolGroupC0  -wi-a-  20.00G
  nered7        VolGroupC0  -wi-a-  20.00G
  sasnered0     VolGroupSAS -wi-a-   8.00G
  sasnered1     VolGroupSAS -wi-a-   8.00G                     

[root at node03 cache]# vgs
  VG          #PV #LV #SN Attr   VSize  VFree
  VolGroupC0    1   0   0 wz--nc  3.41T 3.41T
  VolGroupC1    1   0   0 wz--nc  3.41T 3.41T
  VolGroupSAS   1   2   0 wz--nc 20.63G 4.63G

[root at node03 lvm]# lvs
  LV          VG          Attr   LSize Origin
  sasnered0   VolGroupSAS -wi-a-   8.00G
  sasnered1   VolGroupSAS -wi-a-   8.00G


here is my lvm.conf:

[root at node01 lvm]# lvm dumpconfig
  devices {
  	dir="/dev"
  	scan="/dev"
  	preferred_names=[]
  	filter=["a|^/dev/mapper/controller0$|",
"a|^/dev/mapper/controller1$|", "a|^/dev/mapper/sas-xen$|", "r|.*|"]
  	cache_dir="/etc/lvm/cache"
  	cache_file_prefix=""
  	write_cache_state=0
  	sysfs_scan=1
  	md_component_detection=1
  	md_chunk_alignment=1
  	ignore_suspended_devices=0
  }
  dmeventd {
  	mirror_library="libdevmapper-event-lvm2mirror.so"
  	snapshot_library="libdevmapper-event-lvm2snapshot.so"
  }
  activation {
  	missing_stripe_filler="error"
  	reserved_stack=256
  	reserved_memory=8192
  	process_priority=-18
  	mirror_region_size=512
  	readahead="auto"
  	mirror_log_fault_policy="allocate"
  	mirror_device_fault_policy="remove"
  }
  global {
  	library_dir="/usr/lib64"
  	umask=63
  	test=0
  	units="h"
  	activation=1
  	proc="/proc"
  	locking_type=3
  	fallback_to_clustered_locking=1
  	fallback_to_local_locking=1
  	locking_dir="/var/lock/lvm"
  }
  shell {
  	history_size=100
  }
  backup {
  	backup=1
  	backup_dir="/etc/lvm/backup"
  	archive=1
  	archive_dir="/etc/lvm/archive"
  	retain_min=10
  	retain_days=30
  }
  log {
  	verbose=0
  	syslog=1
  	overwrite=0
  	level=0
  	indent=1
  	command_names=0
  	prefix="  "
  }


Note that logical volumes form C1 were present on node01 and node02,
but after the node03 joined cluster they dissapeared. I'm running
CentOS 5.3.

This is really dissapointing. Enterprise Linux? Linux maybe but not
Enterprise... After much trouble with linux dm-multipath issues with my
storage - which are unresolved and are waiting for RHEL 5.4, now clvmd.

Note that locking (DLM), cman, rgmanager, qdisk and all the other
cluster services are working without problems. I just don't get it why
is CLVM behaving this way? 

I'm thinking about switching to non-clustered LVM - but are there
issues with possible corruption of metadata? I won't create any new
volumes nor snapshots or anything similar. Setup is done and it should
work like this for the extended period of time.... But are there issues
with activation or something else changing metadata?



-- 
|    Jakov Sosic    |    ICQ: 28410271    |   PGP: 0x965CAE2D   |
=================================================================
| start fighting cancer -> http://www.worldcommunitygrid.org/   |



From ccaulfie at redhat.com  Wed Aug 26 14:09:11 2009
From: ccaulfie at redhat.com (Christine Caulfield)
Date: Wed, 26 Aug 2009 15:09:11 +0100
Subject: [Linux-cluster] CLVM not activating LVs
In-Reply-To: <20090826151903.6e4e7759@pc-jsosic.srce.hr>
References: <20090826151903.6e4e7759@pc-jsosic.srce.hr>
Message-ID: <4A954207.9000609@redhat.com>

On 26/08/09 14:19, Jakov Sosic wrote:
> Hi! CLVM not activating logical volumes.
>
>
>
> I have a major issues with CLVM. It is not activating volumes in my
> VG's. I have 2 iSCSI volumes, and one SAS volume with 3 VG's. On
> node01, all logical volumes in 1 iSCSI and on SAS are activated. Other
> iSCSI - zero. On node02 same situation. On node03 only lv's from SAS
> are activated. lvm.conf is same on the machines....
>
> This is very strange, because when I boot the machines, all the
> services are stopped, so logical volumes shouldn't be activated.
>
> Here is the situation:
>
> [root at node01 lvm]# vgs
>    VG          #PV #LV #SN Attr   VSize  VFree
>    VolGroupC0    1   7   0 wz--nc  3.41T 1.48T
>    VolGroupC1    1   0   0 wz--nc  3.41T 3.41T
>    VolGroupSAS   1   2   0 wz--nc 20.63G 4.63G
>
> [root at node01 lvm]# lvs
>    LV            VG          Attr   LSize
>    nered1        VolGroupC0  -wi-a- 200.00G
>    nered2        VolGroupC0  -wi-a- 200.00G
>    nered3        VolGroupC0  -wi-a-   1.46T
>    nered4        VolGroupC0  -wi-a-  20.00G
>    nered5        VolGroupC0  -wi-a-  20.00G
>    nered6        VolGroupC0  -wi-a-  20.00G
>    nered7        VolGroupC0  -wi-a-  20.00G
>    sasnered0     VolGroupSAS -wi-a-   8.00G
>    sasnered1     VolGroupSAS -wi-a-   8.00G
>
> [root at node03 cache]# vgs
>    VG          #PV #LV #SN Attr   VSize  VFree
>    VolGroupC0    1   0   0 wz--nc  3.41T 3.41T
>    VolGroupC1    1   0   0 wz--nc  3.41T 3.41T
>    VolGroupSAS   1   2   0 wz--nc 20.63G 4.63G
>
> [root at node03 lvm]# lvs
>    LV          VG          Attr   LSize Origin
>    sasnered0   VolGroupSAS -wi-a-   8.00G
>    sasnered1   VolGroupSAS -wi-a-   8.00G
>
>
> here is my lvm.conf:
>
> [root at node01 lvm]# lvm dumpconfig
>    devices {
>    	dir="/dev"
>    	scan="/dev"
>    	preferred_names=[]
>    	filter=["a|^/dev/mapper/controller0$|",
> "a|^/dev/mapper/controller1$|", "a|^/dev/mapper/sas-xen$|", "r|.*|"]
>    	cache_dir="/etc/lvm/cache"
>    	cache_file_prefix=""
>    	write_cache_state=0
>    	sysfs_scan=1
>    	md_component_detection=1
>    	md_chunk_alignment=1
>    	ignore_suspended_devices=0
>    }
>    dmeventd {
>    	mirror_library="libdevmapper-event-lvm2mirror.so"
>    	snapshot_library="libdevmapper-event-lvm2snapshot.so"
>    }
>    activation {
>    	missing_stripe_filler="error"
>    	reserved_stack=256
>    	reserved_memory=8192
>    	process_priority=-18
>    	mirror_region_size=512
>    	readahead="auto"
>    	mirror_log_fault_policy="allocate"
>    	mirror_device_fault_policy="remove"
>    }
>    global {
>    	library_dir="/usr/lib64"
>    	umask=63
>    	test=0
>    	units="h"
>    	activation=1
>    	proc="/proc"
>    	locking_type=3
>    	fallback_to_clustered_locking=1
>    	fallback_to_local_locking=1
>    	locking_dir="/var/lock/lvm"
>    }
>    shell {
>    	history_size=100
>    }
>    backup {
>    	backup=1
>    	backup_dir="/etc/lvm/backup"
>    	archive=1
>    	archive_dir="/etc/lvm/archive"
>    	retain_min=10
>    	retain_days=30
>    }
>    log {
>    	verbose=0
>    	syslog=1
>    	overwrite=0
>    	level=0
>    	indent=1
>    	command_names=0
>    	prefix="  "
>    }
>
>
> Note that logical volumes form C1 were present on node01 and node02,
> but after the node03 joined cluster they dissapeared. I'm running
> CentOS 5.3.
>
> This is really dissapointing. Enterprise Linux? Linux maybe but not
> Enterprise... After much trouble with linux dm-multipath issues with my
> storage - which are unresolved and are waiting for RHEL 5.4, now clvmd.
>
> Note that locking (DLM), cman, rgmanager, qdisk and all the other
> cluster services are working without problems. I just don't get it why
> is CLVM behaving this way?
>
> I'm thinking about switching to non-clustered LVM - but are there
> issues with possible corruption of metadata? I won't create any new
> volumes nor snapshots or anything similar. Setup is done and it should
> work like this for the extended period of time.... But are there issues
> with activation or something else changing metadata?
>


You need to mark the shared VGs clustered using the command

# vgchange -cy <VGname>


If you created them while clvmd was active then this is the default. If 
not then you will have to add it yourself as above.


Chrissie




From jakov.sosic at srce.hr  Wed Aug 26 14:11:28 2009
From: jakov.sosic at srce.hr (Jakov Sosic)
Date: Wed, 26 Aug 2009 16:11:28 +0200
Subject: [Linux-cluster] 3 node cluster and quorum disk?
Message-ID: <20090826161128.1e32721c@pc-jsosic.srce.hr>

Hi.

I have a situation - when two nodes are up in 3 node cluster, and one
node goes down, cluster looses quorate - although I'm using qdiskd...

I think that problem is in switching qdisk master from one node to
another. In that case, rgmanager disables all running services, which is
not acceptable situation. Services are currently set to
autostart="0" because cluster is in evaluation phase.

Here is my config:

<?xml version="1.0"?>
<cluster alias="cluster-c00" config_version="56" name="cluster-c00">
	<fence_daemon post_fail_delay="0" post_join_delay="120"/>
	<!-- DEFINE CLUSTER NODES, AND FENCE DEVICES  -->
	<clusternodes>
		<clusternode name="node01" nodeid="1" votes="1">
			<fence>
				<method name="1">
					<device name="node01-ipmi"/>
				</method>
			</fence>
		</clusternode>
		<clusternode name="node02" nodeid="2" votes="1">
			<fence>
				<method name="1">
					<device name="node02-ipmi"/>
				</method>
			</fence>
		</clusternode>
		<clusternode name="node03" nodeid="3" votes="1">
			<fence>
				<method name="1">
					<device name="node03-ipmi"/>
				</method>
			</fence>
		</clusternode>
	</clusternodes>

	<!-- DEFINE CLUSTER MANAGER BEHAVIOUR -->
	<cman expected_votes="3" deadnode_timeout="80"/>
<!--		<multicast addr="224.0.0.1"/>	</cman> -->

	<!-- Token -->
	<totem token="55000"/>

	<!-- Quorum Disk -->
	<quorumd interval="5" tko="5" votes="2"
	label="SAS-qdisk" status_file="/tmp/qdisk"/>

	<!-- DEFINE FENCE DEVICES -->
	<fencedevices>
		<fencedevice agent="fence_ipmilan" auth="password"
	ipaddr="" login="" passwd="" name="node01-ipmi"/>
		<fencedevice agent="fence_ipmilan" auth="password"
	ipaddr="" login="" passwd="" name="node02-ipmi"/>
		<fencedevice agent="fence_ipmilan" auth="password"
	ipaddr="" login="" passwd="" name="node03-ipmi"/>
	</fencedevices>

</cluster>

Should I change any of the timeouts?






-- 
|    Jakov Sosic    |    ICQ: 28410271    |   PGP: 0x965CAE2D   |
=================================================================
| start fighting cancer -> http://www.worldcommunitygrid.org/   |



From jakov.sosic at srce.hr  Wed Aug 26 14:13:23 2009
From: jakov.sosic at srce.hr (Jakov Sosic)
Date: Wed, 26 Aug 2009 16:13:23 +0200
Subject: [Linux-cluster] CLVM not activating LVs
In-Reply-To: <4A954207.9000609@redhat.com>
References: <20090826151903.6e4e7759@pc-jsosic.srce.hr>
	<4A954207.9000609@redhat.com>
Message-ID: <20090826161323.41bef066@pc-jsosic.srce.hr>

On Wed, 26 Aug 2009 15:09:11 +0100
Christine Caulfield <ccaulfie at redhat.com> wrote:

> You need to mark the shared VGs clustered using the command
> 
> # vgchange -cy <VGname>
> 
> 
> If you created them while clvmd was active then this is the default.
> If not then you will have to add it yourself as above.

They were created as clustered.

[root at node03 ~]# vgchange -cy VolGroupC0
  Volume group "VolGroupC0" is already clustered
[root at node03 ~]# vgchange -cy VolGroupC1
  Volume group "VolGroupC1" is already clustered
[root at node03 ~]# vgchange -cy VolGroupSAS
  Volume group "VolGroupSAS" is already clustered



-- 
|    Jakov Sosic    |    ICQ: 28410271    |   PGP: 0x965CAE2D   |
=================================================================
| start fighting cancer -> http://www.worldcommunitygrid.org/   |



From waugh at metservice.com  Wed Aug 26 16:27:21 2009
From: waugh at metservice.com (waugh)
Date: Thu, 27 Aug 2009 04:27:21 +1200
Subject: [Linux-cluster] Re: Linux-cluster Digest, Vol 64, Issue 42
In-Reply-To: <20090826160016.EBC4A61A134@hormel.redhat.com>
References: <20090826160016.EBC4A61A134@hormel.redhat.com>
Message-ID: <1251304041.15503.4.camel@amd64.newlands.nz>

On Thu, 2009-08-27 at 04:00 +1200, linux-cluster-request at redhat.com
wrote:
> Send Linux-cluster mailing list submissions to
>         linux-cluster at redhat.com
> 
> To subscribe or unsubscribe via the World Wide Web, visit
>         https://www.redhat.com/mailman/listinfo/linux-cluster
> or, via email, send a message with subject or body 'help' to
>         linux-cluster-request at redhat.com
> 
> You can reach the person managing the list at
>         linux-cluster-owner at redhat.com
> 
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Linux-cluster digest..."
> 
> 
> Today's Topics:
> 
>    1. Re: NFS client failover (crypto grid)
>    2. Re: NFS client failover (Mike Cardwell)
>    3. Using iscsi devices with scsi reservation and     fence_scsi
>       agent (carlopmart)
>    4. Fencing Logs (Gordan Bobic)
>    5. Re: Apache Cluster (saji george)
>    6. Re: NFS client failover (Mike Cardwell)
>    7. CLVM not activating LVs (Jakov Sosic)
>    8. Re: CLVM not activating LVs (Christine Caulfield)
>    9. 3 node cluster and quorum disk? (Jakov Sosic)
>   10. Re: CLVM not activating LVs (Jakov Sosic)
> 
> 
> ----------------------------------------------------------------------
> 
> Message: 1
> Date: Tue, 25 Aug 2009 16:14:30 -0300
> From: crypto grid <cryptogrid at gmail.com>
> Subject: Re: [Linux-cluster] NFS client failover
> To: linux clustering <linux-cluster at redhat.com>
> Message-ID:
>         <a9f464b80908251214q33650614p52e449d62ee30d7f at mail.gmail.com>
> Content-Type: text/plain; charset="iso-8859-1"
> 
> On Tue, Aug 25, 2009 at 8:19 AM, Mike Cardwell <
> linux-cluster at lists.grepular.com> wrote:
> 
> > Hi,
> >
> > I have a small cluster which shares a filesystem via GFS on a SAN,
> with
> > iSCSI. There are a couple of hosts external to this cluster which I
> would
> > like to have access to the GFS filesystem via NFS. I have exported
> the same
> > mountpoint on each of the hosts that have the GFS mount.
> >
> > I can connect via NFS to one of these cluster hosts, but when it
> goes down,
> > I have to unmount it, and then remount against another host. I
> suspect I
> > could do something with a floating IP? but I was wondering if it
> would be
> > possible to do this with client side logic so that the nfs mount
> > automatically moves to a different working host if one stops
> responding?
> 
> 
> I don't see why you want to resolve this from the client side. You
> must
> configure an ip address resource.
> Take a look at this document:
> http://www.redhat.com/docs/en-US/Red_Hat_Enterprise_Linux/5/pdf/Configuration_Example_-_NFS_Over_GFS.pdf
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
> https://www.redhat.com/archives/linux-cluster/attachments/20090825/33fe8333/attachment.html
> 
> ------------------------------
> 
> Message: 2
> Date: Tue, 25 Aug 2009 20:40:33 +0100
> From: Mike Cardwell <linux-cluster at lists.grepular.com>
> Subject: Re: [Linux-cluster] NFS client failover
> To: linux-cluster at redhat.com
> Message-ID: <4A943E31.3080209 at lists.grepular.com>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
> 
> crypto grid wrote:
> 
> >     I have a small cluster which shares a filesystem via GFS on a
> SAN,
> >     with iSCSI. There are a couple of hosts external to this cluster
> >     which I would like to have access to the GFS filesystem via NFS.
> I
> >     have exported the same mountpoint on each of the hosts that have
> the
> >     GFS mount.
> >
> >     I can connect via NFS to one of these cluster hosts, but when it
> >     goes down, I have to unmount it, and then remount against
> another
> >     host. I suspect I could do something with a floating IP? but I
> was
> >     wondering if it would be possible to do this with client side
> logic
> >     so that the nfs mount automatically moves to a different working
> >     host if one stops responding?
> >
> > I don't see why you want to resolve this from the client side.
> 
> I figured that failover would happen more smoothly if the client was
> aware of and in control of what was going on. If the IP suddenly moves
> to another NFS server I don't know how the NFS client will cope with
> that.
> 
> > You must configure an ip address resource.
> 
> Yeah, that's what I was going to try if there wasn't a client side
> solution.
> 
> > Take a look at this document:
> >
> http://www.redhat.com/docs/en-US/Red_Hat_Enterprise_Linux/5/pdf/Configuration_Example_-_NFS_Over_GFS.pdf
> 
> That document looks very useful. Thank you for pointing it out to me.
> 
> --
> Mike Cardwell - IT Consultant and LAMP developer
> Cardwell IT Ltd. (UK Reg'd Company #06920226) http://cardwellit.com/
> 
> 
> 
> ------------------------------
> 
> Message: 3
> Date: Tue, 25 Aug 2009 21:47:13 +0200
> From: carlopmart <carlopmart at gmail.com>
> Subject: [Linux-cluster] Using iscsi devices with scsi reservation and
>         fence_scsi agent
> To: linux clustering <linux-cluster at redhat.com>
> Message-ID: <4A943FC1.3060104 at gmail.com>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
> 
> Hi all,
> 
>   Somebody knows if can I use an iscsi target (iscsitarget.sf.net)
> with
> scsi fence reservation?? Does it owrks on production environments??
> 
> Thanks
> --
> CL Martinez
> carlopmart {at} gmail {d0t} com
> 
> 
> 
> ------------------------------
> 
> Message: 4
> Date: Tue, 25 Aug 2009 23:43:17 +0100
> From: Gordan Bobic <gordan at bobich.net>
> Subject: [Linux-cluster] Fencing Logs
> To: linux clustering <linux-cluster at redhat.com>
> Message-ID: <4A946905.4040002 at bobich.net>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
> 
> I have a really strange problem on one of my clusters. It exhibits all
> signs of fencing being broken, but the fencing agents work when tested
> manually, and I cannot find anything in syslog to even suggest that
> fencing is being attempted by the surviving node (which just locks up
> on
> GFS access until the other node returns).
> 
> Has anybody got any suggestions on how to troubleshoot this?
> 
> The relevant extract from my cluster.conf is:
> 
> <clusternodes>
>          <clusternode name="hades-cls" nodeid="1" votes="1">
> ...
>                  <fence>
>                          <method name = "1">
>                                  <device name = "hades-oob"/>
>                          </method>
>                  </fence>
> ...
>          </clusternode>
>          <clusternode name="persephone-cls" nodeid="2" votes="1">
> ...
>                  <fence>
>                          <method name = "1">
>                                  <device name ="persephone-oob"/>
>                          </method>
>                  </fence>
>          </clusternode>
> </clusternodes>
> <fencedevices>
>          <fencedevice agent="fence_eric" ipaddr="10.1.254.251"
> login="fence" passwd="some_password" name="hades-oob"/>
>          <fencedevice agent="fence_eric" ipaddr="10.1.254.252"
> login="fence" passwd="some_password" name="persephone-oob"/>
> </fencedevices>
> ...
> 
> I have a near identical setup on all my other clusters, so this is
> somewhat baffling. What else could be relevant to this, specifically
> in
> the context of no fencing attempts even showing up in the logs? I have
> set up scores of RHCS clusters and never seen anything like this
> before.
> The only unusual thing about this cluster is that I had to write a
> bespoke fencing agent for the machines, but these test true when I use
> them to down/reboot the machines.
> 
> TIA.
> 
> Gordan
> 
> 
> 
> ------------------------------
> 
> Message: 5
> Date: Wed, 26 Aug 2009 10:47:13 +0530
> From: saji george <george.saji00 at gmail.com>
> Subject: Re: [Linux-cluster] Apache Cluster
> To: linux clustering <linux-cluster at redhat.com>
> Message-ID:
>         <2460eaad0908252217w1c959990i6aca6b67c71ba1a1 at mail.gmail.com>
> Content-Type: text/plain; charset=ISO-8859-1
> 
> Hi,
>     I have added the ip, still its not working with the Resource
> Apache server, But the same is working with the Resource script. And
> in script sections only the /etc/init.d/httpd script is working.
> 
> Regards
> Saji George
> 
> On 8/22/09, crypto grid <cryptogrid at gmail.com> wrote:
> > On Wed, Aug 19, 2009 at 12:54 PM, saji george
> > <george.saji00 at gmail.com>wrote:
> >
> >> Hi
> >>   I have configured the GFS cluster successfully in RHEL 5.3
> >> But I am not able to configure the cluster for apache.while
> enabling the
> >> service I am getting the below error
> >>
> >> # clusvcadm -e webcluster
> >> Local machine trying to enable service:webcluster...Aborted;
> service
> >> failed
> >>
> >> /log/messages
> >>
> >> Aug 19 21:12:49 cmsstorage-web1 clurgmgrd[32766]: <err> #43:
> Service
> >> service:webcluster has failed; can not start.
> >> Aug 19 21:12:49 cmsstorage-web1 clurgmgrd[32766]: <crit> #13:
> Service
> >> service:webcluster failed to stop cleanly
> >>
> >> cluster.conf
> >>
> >> <failoverdomain name="GFS-apache" ordered="1" restricted="1">
> >>                                 <failoverdomainnode
> name="cmsstorage-web3"
> >> priority="3"/>
> >>                                 <failoverdomainnode
> name="cmsstorage-web2"
> >> priority="2"/>
> >>                                 <failoverdomainnode
> name="cmsstorage-web1"
> >> priority="1"/>
> >>                         </failoverdomain>
> >>
> >> <resources>
> >> <apache config_file="conf/httpd.conf" httpd_options=""
> name="webcluster"
> >> server_root="/usr/local/apache/" shutdown_wait="2"/>
> >> </resources>
> >>
> >>  <service autostart="1" domain="GFS-apache" name="webcluster"
> >> recovery="restart">
> >>                         <apache ref="webcluster"/>
> >>                 </service>
> >>
> >> Is there anything missing...
> >
> >
> >
> > You are missing the definition of an ip address in the resource
> section, and
> > after that, the use of the defined ip address into the service
> definition.
> >
> > Regards,
> >
> >
> >
> >>
> >>
> >> Thanks in advance.
> >>
> >> Regards,
> >> Saji George
> >>
> >>
> >> --
> >> Linux-cluster mailing list
> >> Linux-cluster at redhat.com
> >> https://www.redhat.com/mailman/listinfo/linux-cluster
> >>
> >
> 
> 
> 
> ------------------------------
> 
> Message: 6
> Date: Wed, 26 Aug 2009 09:55:24 +0100
> From: Mike Cardwell <linux-cluster at lists.grepular.com>
> Subject: Re: [Linux-cluster] NFS client failover
> To: linux clustering <linux-cluster at redhat.com>
> Message-ID: <4A94F87C.7030708 at lists.grepular.com>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
> 
> On 25/08/2009 20:40, Mike Cardwell wrote:
> 
> > I figured that failover would happen more smoothly if the client was
> > aware of and in control of what was going on. If the IP suddenly
> moves
> > to another NFS server I don't know how the NFS client will cope with
> that.
> 
> Well, it seems to cope quite well. The nfs mount "hangs" for a few
> seconds whilst the IP moves from one server to another (unavoidable
> obviously), but it then picks up from where it was. I suspect there
> will
> be file corruption issues with files that are partially written when
> the
> failover happens, but I guess that can't be avoided without a client
> side solution.
> 

NFS seems to handle all this quite well. It has been around a while.
> --
> Mike Cardwell - IT Consultant and LAMP developer
> Cardwell IT Ltd. (UK Reg'd Company #06920226) http://cardwellit.com/
> 
> 
> 
> ------------------------------
> 
> Message: 7
> Date: Wed, 26 Aug 2009 15:19:03 +0200
> From: Jakov Sosic <jakov.sosic at srce.hr>
> Subject: [Linux-cluster] CLVM not activating LVs
> To: linux-cluster at redhat.com
> Message-ID: <20090826151903.6e4e7759 at pc-jsosic.srce.hr>
> Content-Type: text/plain; charset=US-ASCII
> 
> Hi! CLVM not activating logical volumes.
> 
> 
> 
> I have a major issues with CLVM. It is not activating volumes in my
> VG's. I have 2 iSCSI volumes, and one SAS volume with 3 VG's. On
> node01, all logical volumes in 1 iSCSI and on SAS are activated. Other
> iSCSI - zero. On node02 same situation. On node03 only lv's from SAS
> are activated. lvm.conf is same on the machines....
> 
> This is very strange, because when I boot the machines, all the
> services are stopped, so logical volumes shouldn't be activated.
> 
> Here is the situation:
> 
> [root at node01 lvm]# vgs
>   VG          #PV #LV #SN Attr   VSize  VFree
>   VolGroupC0    1   7   0 wz--nc  3.41T 1.48T
>   VolGroupC1    1   0   0 wz--nc  3.41T 3.41T
>   VolGroupSAS   1   2   0 wz--nc 20.63G 4.63G
> 
> [root at node01 lvm]# lvs
>   LV            VG          Attr   LSize
>   nered1        VolGroupC0  -wi-a- 200.00G
>   nered2        VolGroupC0  -wi-a- 200.00G
>   nered3        VolGroupC0  -wi-a-   1.46T
>   nered4        VolGroupC0  -wi-a-  20.00G
>   nered5        VolGroupC0  -wi-a-  20.00G
>   nered6        VolGroupC0  -wi-a-  20.00G
>   nered7        VolGroupC0  -wi-a-  20.00G
>   sasnered0     VolGroupSAS -wi-a-   8.00G
>   sasnered1     VolGroupSAS -wi-a-   8.00G                    
> 
> [root at node03 cache]# vgs
>   VG          #PV #LV #SN Attr   VSize  VFree
>   VolGroupC0    1   0   0 wz--nc  3.41T 3.41T
>   VolGroupC1    1   0   0 wz--nc  3.41T 3.41T
>   VolGroupSAS   1   2   0 wz--nc 20.63G 4.63G
> 
> [root at node03 lvm]# lvs
>   LV          VG          Attr   LSize Origin
>   sasnered0   VolGroupSAS -wi-a-   8.00G
>   sasnered1   VolGroupSAS -wi-a-   8.00G
> 
> 
> here is my lvm.conf:
> 
> [root at node01 lvm]# lvm dumpconfig
>   devices {
>         dir="/dev"
>         scan="/dev"
>         preferred_names=[]
>         filter=["a|^/dev/mapper/controller0$|",
> "a|^/dev/mapper/controller1$|", "a|^/dev/mapper/sas-xen$|", "r|.*|"]
>         cache_dir="/etc/lvm/cache"
>         cache_file_prefix=""
>         write_cache_state=0
>         sysfs_scan=1
>         md_component_detection=1
>         md_chunk_alignment=1
>         ignore_suspended_devices=0
>   }
>   dmeventd {
>         mirror_library="libdevmapper-event-lvm2mirror.so"
>         snapshot_library="libdevmapper-event-lvm2snapshot.so"
>   }
>   activation {
>         missing_stripe_filler="error"
>         reserved_stack=256
>         reserved_memory=8192
>         process_priority=-18
>         mirror_region_size=512
>         readahead="auto"
>         mirror_log_fault_policy="allocate"
>         mirror_device_fault_policy="remove"
>   }
>   global {
>         library_dir="/usr/lib64"
>         umask=63
>         test=0
>         units="h"
>         activation=1
>         proc="/proc"
>         locking_type=3
>         fallback_to_clustered_locking=1
>         fallback_to_local_locking=1
>         locking_dir="/var/lock/lvm"
>   }
>   shell {
>         history_size=100
>   }
>   backup {
>         backup=1
>         backup_dir="/etc/lvm/backup"
>         archive=1
>         archive_dir="/etc/lvm/archive"
>         retain_min=10
>         retain_days=30
>   }
>   log {
>         verbose=0
>         syslog=1
>         overwrite=0
>         level=0
>         indent=1
>         command_names=0
>         prefix="  "
>   }
> 
> 
> Note that logical volumes form C1 were present on node01 and node02,
> but after the node03 joined cluster they dissapeared. I'm running
> CentOS 5.3.
> 
> This is really dissapointing. Enterprise Linux? Linux maybe but not
> Enterprise... After much trouble with linux dm-multipath issues with
> my
> storage - which are unresolved and are waiting for RHEL 5.4, now
> clvmd.
> 
> Note that locking (DLM), cman, rgmanager, qdisk and all the other
> cluster services are working without problems. I just don't get it why
> is CLVM behaving this way?
> 
> I'm thinking about switching to non-clustered LVM - but are there
> issues with possible corruption of metadata? I won't create any new
> volumes nor snapshots or anything similar. Setup is done and it should
> work like this for the extended period of time.... But are there
> issues
> with activation or something else changing metadata?
> 
> 
> 
> --
> |    Jakov Sosic    |    ICQ: 28410271    |   PGP: 0x965CAE2D   |
> =================================================================
> | start fighting cancer -> http://www.worldcommunitygrid.org/   |
> 
> 
> 
> ------------------------------
> 
> Message: 8
> Date: Wed, 26 Aug 2009 15:09:11 +0100
> From: Christine Caulfield <ccaulfie at redhat.com>
> Subject: Re: [Linux-cluster] CLVM not activating LVs
> To: linux clustering <linux-cluster at redhat.com>
> Message-ID: <4A954207.9000609 at redhat.com>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
> 
> On 26/08/09 14:19, Jakov Sosic wrote:
> > Hi! CLVM not activating logical volumes.
> >
> >
> >
> > I have a major issues with CLVM. It is not activating volumes in my
> > VG's. I have 2 iSCSI volumes, and one SAS volume with 3 VG's. On
> > node01, all logical volumes in 1 iSCSI and on SAS are activated.
> Other
> > iSCSI - zero. On node02 same situation. On node03 only lv's from SAS
> > are activated. lvm.conf is same on the machines....
> >
> > This is very strange, because when I boot the machines, all the
> > services are stopped, so logical volumes shouldn't be activated.
> >
> > Here is the situation:
> >
> > [root at node01 lvm]# vgs
> >    VG          #PV #LV #SN Attr   VSize  VFree
> >    VolGroupC0    1   7   0 wz--nc  3.41T 1.48T
> >    VolGroupC1    1   0   0 wz--nc  3.41T 3.41T
> >    VolGroupSAS   1   2   0 wz--nc 20.63G 4.63G
> >
> > [root at node01 lvm]# lvs
> >    LV            VG          Attr   LSize
> >    nered1        VolGroupC0  -wi-a- 200.00G
> >    nered2        VolGroupC0  -wi-a- 200.00G
> >    nered3        VolGroupC0  -wi-a-   1.46T
> >    nered4        VolGroupC0  -wi-a-  20.00G
> >    nered5        VolGroupC0  -wi-a-  20.00G
> >    nered6        VolGroupC0  -wi-a-  20.00G
> >    nered7        VolGroupC0  -wi-a-  20.00G
> >    sasnered0     VolGroupSAS -wi-a-   8.00G
> >    sasnered1     VolGroupSAS -wi-a-   8.00G
> >
> > [root at node03 cache]# vgs
> >    VG          #PV #LV #SN Attr   VSize  VFree
> >    VolGroupC0    1   0   0 wz--nc  3.41T 3.41T
> >    VolGroupC1    1   0   0 wz--nc  3.41T 3.41T
> >    VolGroupSAS   1   2   0 wz--nc 20.63G 4.63G
> >
> > [root at node03 lvm]# lvs
> >    LV          VG          Attr   LSize Origin
> >    sasnered0   VolGroupSAS -wi-a-   8.00G
> >    sasnered1   VolGroupSAS -wi-a-   8.00G
> >
> >
> > here is my lvm.conf:
> >
> > [root at node01 lvm]# lvm dumpconfig
> >    devices {
> >       dir="/dev"
> >       scan="/dev"
> >       preferred_names=[]
> >       filter=["a|^/dev/mapper/controller0$|",
> > "a|^/dev/mapper/controller1$|", "a|^/dev/mapper/sas-xen$|", "r|.*|"]
> >       cache_dir="/etc/lvm/cache"
> >       cache_file_prefix=""
> >       write_cache_state=0
> >       sysfs_scan=1
> >       md_component_detection=1
> >       md_chunk_alignment=1
> >       ignore_suspended_devices=0
> >    }
> >    dmeventd {
> >       mirror_library="libdevmapper-event-lvm2mirror.so"
> >       snapshot_library="libdevmapper-event-lvm2snapshot.so"
> >    }
> >    activation {
> >       missing_stripe_filler="error"
> >       reserved_stack=256
> >       reserved_memory=8192
> >       process_priority=-18
> >       mirror_region_size=512
> >       readahead="auto"
> >       mirror_log_fault_policy="allocate"
> >       mirror_device_fault_policy="remove"
> >    }
> >    global {
> >       library_dir="/usr/lib64"
> >       umask=63
> >       test=0
> >       units="h"
> >       activation=1
> >       proc="/proc"
> >       locking_type=3
> >       fallback_to_clustered_locking=1
> >       fallback_to_local_locking=1
> >       locking_dir="/var/lock/lvm"
> >    }
> >    shell {
> >       history_size=100
> >    }
> >    backup {
> >       backup=1
> >       backup_dir="/etc/lvm/backup"
> >       archive=1
> >       archive_dir="/etc/lvm/archive"
> >       retain_min=10
> >       retain_days=30
> >    }
> >    log {
> >       verbose=0
> >       syslog=1
> >       overwrite=0
> >       level=0
> >       indent=1
> >       command_names=0
> >       prefix="  "
> >    }
> >
> >
> > Note that logical volumes form C1 were present on node01 and node02,
> > but after the node03 joined cluster they dissapeared. I'm running
> > CentOS 5.3.
> >
> > This is really dissapointing. Enterprise Linux? Linux maybe but not
> > Enterprise... After much trouble with linux dm-multipath issues with
> my
> > storage - which are unresolved and are waiting for RHEL 5.4, now
> clvmd.
> >
> > Note that locking (DLM), cman, rgmanager, qdisk and all the other
> > cluster services are working without problems. I just don't get it
> why
> > is CLVM behaving this way?
> >
> > I'm thinking about switching to non-clustered LVM - but are there
> > issues with possible corruption of metadata? I won't create any new
> > volumes nor snapshots or anything similar. Setup is done and it
> should
> > work like this for the extended period of time.... But are there
> issues
> > with activation or something else changing metadata?
> >
> 
> 
> You need to mark the shared VGs clustered using the command
> 
> # vgchange -cy <VGname>
> 
> 
> If you created them while clvmd was active then this is the default.
> If
> not then you will have to add it yourself as above.
> 
> 
> Chrissie
> 
> 
> 
> 
> ------------------------------
> 
> Message: 9
> Date: Wed, 26 Aug 2009 16:11:28 +0200
> From: Jakov Sosic <jakov.sosic at srce.hr>
> Subject: [Linux-cluster] 3 node cluster and quorum disk?
> To: linux-cluster at redhat.com
> Message-ID: <20090826161128.1e32721c at pc-jsosic.srce.hr>
> Content-Type: text/plain; charset=US-ASCII
> 
> Hi.
> 
> I have a situation - when two nodes are up in 3 node cluster, and one
> node goes down, cluster looses quorate - although I'm using qdiskd...
> 
> I think that problem is in switching qdisk master from one node to
> another. In that case, rgmanager disables all running services, which
> is
> not acceptable situation. Services are currently set to
> autostart="0" because cluster is in evaluation phase.
> 
> Here is my config:
> 
> <?xml version="1.0"?>
> <cluster alias="cluster-c00" config_version="56" name="cluster-c00">
>         <fence_daemon post_fail_delay="0" post_join_delay="120"/>
>         <!-- DEFINE CLUSTER NODES, AND FENCE DEVICES  -->
>         <clusternodes>
>                 <clusternode name="node01" nodeid="1" votes="1">
>                         <fence>
>                                 <method name="1">
>                                         <device name="node01-ipmi"/>
>                                 </method>
>                         </fence>
>                 </clusternode>
>                 <clusternode name="node02" nodeid="2" votes="1">
>                         <fence>
>                                 <method name="1">
>                                         <device name="node02-ipmi"/>
>                                 </method>
>                         </fence>
>                 </clusternode>
>                 <clusternode name="node03" nodeid="3" votes="1">
>                         <fence>
>                                 <method name="1">
>                                         <device name="node03-ipmi"/>
>                                 </method>
>                         </fence>
>                 </clusternode>
>         </clusternodes>
> 
>         <!-- DEFINE CLUSTER MANAGER BEHAVIOUR -->
>         <cman expected_votes="3" deadnode_timeout="80"/>
> <!--            <multicast addr="224.0.0.1"/>   </cman> -->
> 
>         <!-- Token -->
>         <totem token="55000"/>
> 
>         <!-- Quorum Disk -->
>         <quorumd interval="5" tko="5" votes="2"
>         label="SAS-qdisk" status_file="/tmp/qdisk"/>
> 
>         <!-- DEFINE FENCE DEVICES -->
>         <fencedevices>
>                 <fencedevice agent="fence_ipmilan" auth="password"
>         ipaddr="" login="" passwd="" name="node01-ipmi"/>
>                 <fencedevice agent="fence_ipmilan" auth="password"
>         ipaddr="" login="" passwd="" name="node02-ipmi"/>
>                 <fencedevice agent="fence_ipmilan" auth="password"
>         ipaddr="" login="" passwd="" name="node03-ipmi"/>
>         </fencedevices>
> 
> </cluster>
> 
> Should I change any of the timeouts?
> 
> 
> 
> 
> 
> 
> --
> |    Jakov Sosic    |    ICQ: 28410271    |   PGP: 0x965CAE2D   |
> =================================================================
> | start fighting cancer -> http://www.worldcommunitygrid.org/   |
> 
> 
> 
> ------------------------------
> 
> Message: 10
> Date: Wed, 26 Aug 2009 16:13:23 +0200
> From: Jakov Sosic <jakov.sosic at srce.hr>
> Subject: Re: [Linux-cluster] CLVM not activating LVs
> To: linux-cluster at redhat.com
> Message-ID: <20090826161323.41bef066 at pc-jsosic.srce.hr>
> Content-Type: text/plain; charset=US-ASCII
> 
> On Wed, 26 Aug 2009 15:09:11 +0100
> Christine Caulfield <ccaulfie at redhat.com> wrote:
> 
> > You need to mark the shared VGs clustered using the command
> >
> > # vgchange -cy <VGname>
> >
> >
> > If you created them while clvmd was active then this is the default.
> > If not then you will have to add it yourself as above.
> 
> They were created as clustered.
> 
> [root at node03 ~]# vgchange -cy VolGroupC0
>   Volume group "VolGroupC0" is already clustered
> [root at node03 ~]# vgchange -cy VolGroupC1
>   Volume group "VolGroupC1" is already clustered
> [root at node03 ~]# vgchange -cy VolGroupSAS
>   Volume group "VolGroupSAS" is already clustered
> 
> 
> 
> --
> |    Jakov Sosic    |    ICQ: 28410271    |   PGP: 0x965CAE2D   |
> =================================================================
> | start fighting cancer -> http://www.worldcommunitygrid.org/   |
> 
> 
> 
> ------------------------------
> 
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
> 
> End of Linux-cluster Digest, Vol 64, Issue 42
> *********************************************
> 
> 



From marko.jung at oucs.ox.ac.uk  Wed Aug 26 17:00:32 2009
From: marko.jung at oucs.ox.ac.uk (Marko Jung)
Date: Wed, 26 Aug 2009 18:00:32 +0100
Subject: [Linux-cluster] GFS2 hangs on one node of a two node cluster
Message-ID: <4A956A30.7040701@oucs.ox.ac.uk>

Hello,

this is the second time this week all my gfs2 filesystems are unaccessible 
on one of my two node cluster. I am using CentOS 5.3 (the evil AS ;-) ) and 
  the cluster otherwise works well. What is used: clvmd to manage 40TB SAN 
storage in combination with GFS2. No other repositories except Fedora EPEL5 
used.  Kernel is latest (2.6.18-128.4.1.el5) and all provided software 
updates are in place.

The issue can be described as all gfs2 filesystems being not accessible 
without any error messages by user space applications, except bash 
complaining at logins: "-bash: cd: /san/home/USERNAME: Input/output error
". Trying to unmount the filesystems also did not succeed. Just a reboot 
seems to fix the issue.

The first time the problem occurred the system got a load of >800 (!!) but 
just because all processes were waiting for IO.

Find an excerpt of the messages that seem to be related to that issue 
attached. Any hints, ideas and comments highly appreciated. Thank you for 
your good work and help!

Marko
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: gfs2-hang.log
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090826/06d190c0/attachment.log>

From swhiteho at redhat.com  Wed Aug 26 17:05:15 2009
From: swhiteho at redhat.com (Steven Whitehouse)
Date: Wed, 26 Aug 2009 18:05:15 +0100
Subject: [Linux-cluster] GFS2 hangs on one node of a two node cluster
In-Reply-To: <4A956A30.7040701@oucs.ox.ac.uk>
References: <4A956A30.7040701@oucs.ox.ac.uk>
Message-ID: <1251306315.6052.5.camel@localhost.localdomain>

Hi,

On Wed, 2009-08-26 at 18:00 +0100, Marko Jung wrote:
> Hello,
> 
> this is the second time this week all my gfs2 filesystems are unaccessible 
> on one of my two node cluster. I am using CentOS 5.3 (the evil AS ;-) ) and 
>   the cluster otherwise works well. What is used: clvmd to manage 40TB SAN 
> storage in combination with GFS2. No other repositories except Fedora EPEL5 
> used.  Kernel is latest (2.6.18-128.4.1.el5) and all provided software 
> updates are in place.
> 
> The issue can be described as all gfs2 filesystems being not accessible 
> without any error messages by user space applications, except bash 
> complaining at logins: "-bash: cd: /san/home/USERNAME: Input/output error
> ". Trying to unmount the filesystems also did not succeed. Just a reboot 
> seems to fix the issue.
> 
> The first time the problem occurred the system got a load of >800 (!!) but 
> just because all processes were waiting for IO.
> 
> Find an excerpt of the messages that seem to be related to that issue 
> attached. Any hints, ideas and comments highly appreciated. Thank you for 
> your good work and help!
> 
> Marko
> plain text document attachment (gfs2-hang.log)
> server kernel: GFS2: fsid=: Trying to join cluster "lock_dlm", "dps:sb01vg01lv02"
> server kernel: GFS2: fsid=dps:sb01vg01lv02.1: Joined cluster. Now mounting FS...
> server kernel: GFS2: fsid=dps:sb01vg01lv02.1: jid=1, already locked for use
> server kernel: GFS2: fsid=dps:sb01vg01lv02.1: jid=1: Looking at journal...
> server kernel: GFS2: fsid=dps:sb01vg01lv02.1: jid=1: Done
> [...other non cluster/gfs related stuff removed...]]
> server kernel: GFS2: fsid=dps:sb01vg01lv01.0: fatal: filesystem consistency error
Any time you see this:                               ^^^^^^^^^^^^^^^^^

It means that you'll have to use fsck.gfs2 to fix the problem. Its
always tricky to track down what caused the issue since that is probable
some time in the past, but it might be a fencing issue or something
along those lines. It could be a real bug too, so if you keep hitting
it, even after fsck, then any information that you can give as to the
cause would be very useful.

Steve.

> server kernel: GFS2: fsid=dps:sb01vg01lv01.0:   RG = 9635245
> server kernel: GFS2: fsid=dps:sb01vg01lv01.0:   function = gfs2_free_uninit_di, file = fs/gfs2/rgrp.c, line = 1653
> server kernel: GFS2: fsid=dps:sb01vg01lv01.0: about to withdraw this file system
> server kernel: GFS2: fsid=dps:sb01vg01lv01.0: telling LM to withdraw
> server qdiskd[10309]: <warning> qdisk cycle took more than 1 second to complete (13.760000) 
> server kernel: dlm: sb01vg01lv01: group leave failed -512 0
> server dlm_controld[10331]: open "/sys/kernel/dlm/sb01vg01lv01/event_done" error -1 2
> server kernel: GFS2: fsid=dps:sb01vg01lv01.0: withdrawn
> server kernel: 
> server kernel: Call Trace:
> server kernel:  [<ffffffff885e252a>] :gfs2:gfs2_lm_withdraw+0xc1/0xd0
> server kernel:  [<ffffffff8002ce63>] wake_up_bit+0x11/0x22
> server kernel:  [<ffffffff800646f6>] __down_read+0x12/0x92
> server kernel:  [<ffffffff885de02c>] :gfs2:do_promote+0x108/0x137
> server kernel:  [<ffffffff80063bb7>] mutex_lock+0xd/0x1d
> server kernel:  [<ffffffff885e51a3>] :gfs2:buf_lo_add+0x71/0x106
> server kernel:  [<ffffffff885f56bf>] :gfs2:gfs2_consist_rgrpd_i+0x34/0x39
> server kernel:  [<ffffffff885f2368>] :gfs2:gfs2_free_di+0x7c/0xef
> server kernel:  [<ffffffff885e1549>] :gfs2:gfs2_dinode_dealloc+0x141/0x1a7
> server kernel:  [<ffffffff885edd7b>] :gfs2:gfs2_delete_inode+0xeb/0x191
> server kernel:  [<ffffffff885edcd6>] :gfs2:gfs2_delete_inode+0x46/0x191
> server kernel:  [<ffffffff885edc90>] :gfs2:gfs2_delete_inode+0x0/0x191
> server kernel:  [<ffffffff8002f191>] generic_delete_inode+0xc6/0x143
> server kernel:  [<ffffffff885f2c88>] :gfs2:gfs2_inplace_reserve_i+0x63b/0x691
> server kernel:  [<ffffffff885de019>] :gfs2:do_promote+0xf5/0x137
> server kernel:  [<ffffffff885e740d>] :gfs2:gfs2_write_begin+0x116/0x33f
> server kernel:  [<ffffffff885e8e8f>] :gfs2:gfs2_file_buffered_write+0x14b/0x2e5
> server kernel:  [<ffffffff8000cdf5>] file_read_actor+0x0/0x154
> server kernel:  [<ffffffff885e92c5>] :gfs2:__gfs2_file_aio_write_nolock+0x29c/0x2d4
> server kernel:  [<ffffffff885e9468>] :gfs2:gfs2_file_write_nolock+0xaa/0x10f
> server kernel:  [<ffffffff800c3206>] generic_file_read+0xac/0xc5
> server kernel:  [<ffffffff8009dbae>] autoremove_wake_function+0x0/0x2e
> server kernel:  [<ffffffff8009dbae>] autoremove_wake_function+0x0/0x2e
> server kernel:  [<ffffffff885e95b8>] :gfs2:gfs2_file_write+0x49/0xa7
> server kernel:  [<ffffffff80016591>] vfs_write+0xce/0x174
> server kernel:  [<ffffffff80016e5e>] sys_write+0x45/0x6e
> server kernel:  [<ffffffff8005d28d>] tracesys+0xd5/0xe0
> server kernel: 
> server kernel: GFS2: fsid=dps:sb01vg01lv01.0: gfs2_delete_inode: -5
> [...then no logs for more than two hours until i started diagnosing the issue...]
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster



From jakov.sosic at srce.hr  Wed Aug 26 18:51:25 2009
From: jakov.sosic at srce.hr (Jakov Sosic)
Date: Wed, 26 Aug 2009 20:51:25 +0200
Subject: [Linux-cluster] CLVM not activating LVs
In-Reply-To: <20090826161323.41bef066@pc-jsosic.srce.hr>
References: <20090826151903.6e4e7759@pc-jsosic.srce.hr>
	<4A954207.9000609@redhat.com>
	<20090826161323.41bef066@pc-jsosic.srce.hr>
Message-ID: <20090826205125.72a95f72@pc-jsosic.srce.hr>

On Wed, 26 Aug 2009 16:13:23 +0200
Jakov Sosic <jakov.sosic at srce.hr> wrote:

> They were created as clustered.
> 
> [root at node03 ~]# vgchange -cy VolGroupC0
>   Volume group "VolGroupC0" is already clustered
> [root at node03 ~]# vgchange -cy VolGroupC1
>   Volume group "VolGroupC1" is already clustered
> [root at node03 ~]# vgchange -cy VolGroupSAS
>   Volume group "VolGroupSAS" is already clustered

OK, I think I solved this one... Issue was on the storage side - I used
drbd primary/primary with two tgtd's, and multipath failover on the
client side... Problem was somewhere in the tgt layer... one side was
not reporting correct data.

I switched storages to drbd primary/secondary in RHEL cluster, and
changed tgtd to IET (iscsi enterprise target) from livna repositories.

If I encounter any more trouble I'll post to the list.


-- 
|    Jakov Sosic    |    ICQ: 28410271    |   PGP: 0x965CAE2D   |
=================================================================
| start fighting cancer -> http://www.worldcommunitygrid.org/   |



From teigland at redhat.com  Wed Aug 26 19:25:50 2009
From: teigland at redhat.com (David Teigland)
Date: Wed, 26 Aug 2009 14:25:50 -0500
Subject: [Linux-cluster] Fencing Logs
In-Reply-To: <4A946905.4040002@bobich.net>
References: <4A946905.4040002@bobich.net>
Message-ID: <20090826192550.GA29464@redhat.com>

On Tue, Aug 25, 2009 at 11:43:17PM +0100, Gordan Bobic wrote:
>                         <method name = "1">
>                                 <device name = "hades-oob"/>

I've never seen anyone use spaces before/after " = ".

Otherwise, the fenced debug log might help,
group_tool dump fence (cluster2/rhel5)
fence_tool dump (cluster3)

Dave



From alan.zg at gmail.com  Wed Aug 26 23:36:26 2009
From: alan.zg at gmail.com (Alan A)
Date: Wed, 26 Aug 2009 18:36:26 -0500
Subject: [Linux-cluster] Can not establish new cluster 5.3 with luci -
	quorum error
Message-ID: <fac531740908261636j29165920odcdc169f28923f22@mail.gmail.com>

I have tried almost everything at this point to try and troubleshoot this
further. I can't create new cluster with luci.

I broke and tried to reconfigure 3 node cluster at least 6 times.

I have noticed nodes taking expectational long on initializing fencing upon
cman start. I tried with defined and undefined fencing, the amount of time
needed is still the same. Even after the fencing is overcome in
/var/log/messages nodes refuse to join cluster due to the state of 'not in
quorum' during joining process. I uped the post_join_delay as much as 150
but the result is the same.

Fencing - I use APC PW Switches - I can login into apc PWS from the node, I
can even fence the other node, but when cman is started it looks like it is
almost timign out on staring fencing.

If I issue cman_tool nodes it gives me the local node name as the member of
the cluster and the other two with state 'X'. If I try cman_tool join
clustername - it tells me the nodes are already in that cluster but cluster
as the whole does not register. Each node thinks it's the only working
member of the cluster.


Any pointers?



-- 
Alan A.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090826/e8cb0a0a/attachment.htm>

From Alain.Moulle at bull.net  Thu Aug 27 06:31:59 2009
From: Alain.Moulle at bull.net (Alain.Moulle)
Date: Thu, 27 Aug 2009 08:31:59 +0200
Subject: [Linux-cluster] cluster.conf in another place ?
Message-ID: <4A96285F.6010302@bull.net>

Hi,
I'd need to put the cluster.conf in another place than /etc/cluster/.
(without any symbolic link)
Is-it possible ?
Thanks
Alain
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090827/28938ac6/attachment.htm>

From xavier.montagutelli at unilim.fr  Thu Aug 27 07:08:21 2009
From: xavier.montagutelli at unilim.fr (Xavier Montagutelli)
Date: Thu, 27 Aug 2009 09:08:21 +0200
Subject: [Linux-cluster] Can not establish new cluster 5.3 with luci -
	quorum error
In-Reply-To: <fac531740908261636j29165920odcdc169f28923f22@mail.gmail.com>
References: <fac531740908261636j29165920odcdc169f28923f22@mail.gmail.com>
Message-ID: <200908270908.21850.xavier.montagutelli@unilim.fr>

On Thursday 27 August 2009 01:36:26 Alan A wrote:
> I have tried almost everything at this point to try and troubleshoot this
> further. I can't create new cluster with luci.
>
> I broke and tried to reconfigure 3 node cluster at least 6 times.
>
> I have noticed nodes taking expectational long on initializing fencing upon
> cman start. I tried with defined and undefined fencing, the amount of time
> needed is still the same. Even after the fencing is overcome in
> /var/log/messages nodes refuse to join cluster due to the state of 'not in
> quorum' during joining process. I uped the post_join_delay as much as 150
> but the result is the same.
>
> Fencing - I use APC PW Switches - I can login into apc PWS from the node, I
> can even fence the other node, but when cman is started it looks like it is
> almost timign out on staring fencing.
>
> If I issue cman_tool nodes it gives me the local node name as the member of
> the cluster and the other two with state 'X'. If I try cman_tool join
> clustername - it tells me the nodes are already in that cluster but cluster
> as the whole does not register. Each node thinks it's the only working
> member of the cluster.
>
>
> Any pointers?

Yesterday, I have created a new cluster with a fresh install of CentOS 5.3, on 
three nodes. I used luci + ricci.

I disabled the firewall and SELinux (system-config-securitylevel-tui --disabled 
--selinux=disabled -q). I know it is supposed to work with SELinux now, but 
last time I tried it, it was still problematic.

You can also try the "clean_start="1" attribute in the "<fence_daemon >" tag 
(the fenced man page explains this parameter in the "Domain startup" section, 
perhaps that's why you have a long timeout during the intialization of the 
fencing)

Did you also try your fencing methods with the "fence_manual -n " command 
(useful to validate the fence device) ?

HTH

-- 
Xavier Montagutelli                      Tel : +33 (0)5 55 45 77 20
Service Commun Informatique              Fax : +33 (0)5 55 45 75 95
Universite de Limoges
123, avenue Albert Thomas
87060 Limoges cedex



From jakov.sosic at srce.hr  Thu Aug 27 08:27:52 2009
From: jakov.sosic at srce.hr (Jakov Sosic)
Date: Thu, 27 Aug 2009 10:27:52 +0200
Subject: [Linux-cluster] Can not establish new cluster 5.3 with luci -
	quorum error
In-Reply-To: <fac531740908261636j29165920odcdc169f28923f22@mail.gmail.com>
References: <fac531740908261636j29165920odcdc169f28923f22@mail.gmail.com>
Message-ID: <20090827102752.391915bc@nb-jsosic>

On Wed, 26 Aug 2009 18:36:26 -0500
Alan A <alan.zg at gmail.com> wrote:

> I have tried almost everything at this point to try and troubleshoot
> this further. I can't create new cluster with luci.
> 
> I broke and tried to reconfigure 3 node cluster at least 6 times.
> 
> I have noticed nodes taking expectational long on initializing
> fencing upon cman start. I tried with defined and undefined fencing,
> the amount of time needed is still the same. Even after the fencing
> is overcome in /var/log/messages nodes refuse to join cluster due to
> the state of 'not in quorum' during joining process. I uped the
> post_join_delay as much as 150 but the result is the same.
> 
> Fencing - I use APC PW Switches - I can login into apc PWS from the
> node, I can even fence the other node, but when cman is started it
> looks like it is almost timign out on staring fencing.
> 
> If I issue cman_tool nodes it gives me the local node name as the
> member of the cluster and the other two with state 'X'. If I try
> cman_tool join clustername - it tells me the nodes are already in
> that cluster but cluster as the whole does not register. Each node
> thinks it's the only working member of the cluster.
> 
> 
> Any pointers?

Looks like network issue to me.

Are you sure your network is operational in a sense of a multicast /
igmp? Try forcing igmp v1 in sysctl.conf - and if you have Cisco
equipment take a look at openais FAQ (mode sparse-dense).


-- 
|    Jakov Sosic    |    ICQ: 28410271    |   PGP: 0x965CAE2D   |
=================================================================
| start fighting cancer -> http://www.worldcommunitygrid.org/   |



From fdinitto at redhat.com  Thu Aug 27 10:28:24 2009
From: fdinitto at redhat.com (Fabio M. Di Nitto)
Date: Thu, 27 Aug 2009 12:28:24 +0200
Subject: [Linux-cluster] cluster.conf in another place ?
In-Reply-To: <4A96285F.6010302@bull.net>
References: <4A96285F.6010302@bull.net>
Message-ID: <1251368904.8148.276.camel@cerberus.int.fabbione.net>

On Thu, 2009-08-27 at 08:31 +0200, Alain.Moulle wrote:
> Hi,
> I'd need to put the cluster.conf in another place than /etc/cluster/.
> (without any symbolic link)
> Is-it possible ?

What version of the stack are you using?

Fabio

PS pretty please.. don't send HTML messages with colors.



From linux-cluster at lists.grepular.com  Thu Aug 27 12:33:53 2009
From: linux-cluster at lists.grepular.com (Mike Cardwell)
Date: Thu, 27 Aug 2009 13:33:53 +0100
Subject: [Linux-cluster] FS-Cache
Message-ID: <4A967D31.5050409@lists.grepular.com>

Hi,

Are there any plans to make GFS work with FS-Cache?

http://people.redhat.com/~dhowells/fscache/FS-Cache.pdf

-- 
Mike Cardwell - IT Consultant and LAMP developer
Cardwell IT Ltd. (UK Reg'd Company #06920226) http://cardwellit.com/



From swhiteho at redhat.com  Thu Aug 27 13:03:07 2009
From: swhiteho at redhat.com (Steven Whitehouse)
Date: Thu, 27 Aug 2009 14:03:07 +0100
Subject: [Linux-cluster] FS-Cache
In-Reply-To: <4A967D31.5050409@lists.grepular.com>
References: <4A967D31.5050409@lists.grepular.com>
Message-ID: <1251378187.6052.6.camel@localhost.localdomain>

Hi,

On Thu, 2009-08-27 at 13:33 +0100, Mike Cardwell wrote:
> Hi,
> 
> Are there any plans to make GFS work with FS-Cache?
> 
> http://people.redhat.com/~dhowells/fscache/FS-Cache.pdf
> 

Not that I'm aware of. I'm not sure it would make any sense to do so,

Steve.




From linux-cluster at lists.grepular.com  Thu Aug 27 13:22:26 2009
From: linux-cluster at lists.grepular.com (Mike Cardwell)
Date: Thu, 27 Aug 2009 14:22:26 +0100
Subject: [Linux-cluster] FS-Cache
In-Reply-To: <1251378187.6052.6.camel@localhost.localdomain>
References: <4A967D31.5050409@lists.grepular.com>
	<1251378187.6052.6.camel@localhost.localdomain>
Message-ID: <4A968892.4050506@lists.grepular.com>

On 27/08/2009 14:03, Steven Whitehouse wrote:

>> Are there any plans to make GFS work with FS-Cache?
>>
>> http://people.redhat.com/~dhowells/fscache/FS-Cache.pdf
>
> Not that I'm aware of. I'm not sure it would make any sense to do so,

Hi. Can you explain why it wouldn't make any sense please? I have a GFS 
filesystem mounted over 100Mb iSCSI (I'm not in a position to upgrade 
the network speed at this time, but will in future). It's read heavy. 
Particular files will be hit very regularly. I just thought it would be 
nice if those files could be cached locally rather than repeatedly 
pulling them over the network...?

-- 
Mike Cardwell - IT Consultant and LAMP developer
Cardwell IT Ltd. (UK Reg'd Company #06920226) http://cardwellit.com/



From swhiteho at redhat.com  Thu Aug 27 13:25:11 2009
From: swhiteho at redhat.com (Steven Whitehouse)
Date: Thu, 27 Aug 2009 14:25:11 +0100
Subject: [Linux-cluster] FS-Cache
In-Reply-To: <4A968892.4050506@lists.grepular.com>
References: <4A967D31.5050409@lists.grepular.com>
	<1251378187.6052.6.camel@localhost.localdomain>
	<4A968892.4050506@lists.grepular.com>
Message-ID: <1251379511.6052.8.camel@localhost.localdomain>

Hi,

On Thu, 2009-08-27 at 14:22 +0100, Mike Cardwell wrote:
> On 27/08/2009 14:03, Steven Whitehouse wrote:
> 
> >> Are there any plans to make GFS work with FS-Cache?
> >>
> >> http://people.redhat.com/~dhowells/fscache/FS-Cache.pdf
> >
> > Not that I'm aware of. I'm not sure it would make any sense to do so,
> 
> Hi. Can you explain why it wouldn't make any sense please? I have a GFS 
> filesystem mounted over 100Mb iSCSI (I'm not in a position to upgrade 
> the network speed at this time, but will in future). It's read heavy. 
> Particular files will be hit very regularly. I just thought it would be 
> nice if those files could be cached locally rather than repeatedly 
> pulling them over the network...?
> 

Unless you are doing something to push those files out of the cache,
then they will be cached locally anyway. GFS2 is a bit better at doing
that then GFS1 though,

Steve.




From jfaridzadegan at hotmail.com  Thu Aug 27 13:31:34 2009
From: jfaridzadegan at hotmail.com (Jalil Faridzadegan)
Date: Thu, 27 Aug 2009 18:01:34 +0430
Subject: [Linux-cluster] fencing devices in redhat cluster
Message-ID: <COL109-W52E68352DA3AC1BD5364D2AFF60@phx.gbl>


Hi,

Are fencing devices always required in a redhat cluster configuration or optional?

Thanks a lot,
J. Farid

_________________________________________________________________
With Windows Live, you can organize, edit, and share your photos.
http://www.microsoft.com/middleeast/windows/windowslive/products/photo-gallery-edit.aspx
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090827/d1ddbc98/attachment.htm>

From mike at ubxess.com  Thu Aug 27 13:35:04 2009
From: mike at ubxess.com (Mike Kemelmakher)
Date: Thu, 27 Aug 2009 16:35:04 +0300
Subject: [Linux-cluster] fencing devices in redhat cluster
In-Reply-To: <COL109-W52E68352DA3AC1BD5364D2AFF60@phx.gbl>
References: <COL109-W52E68352DA3AC1BD5364D2AFF60@phx.gbl>
Message-ID: <9c387a370908270635h49a669aerd047b6b7c7f94a22@mail.gmail.com>

Hi,
As far as i know - fencing is required when you have to create two-node
cluster.

Just finished such configuration on two IBM blades. We had to integrate
redhat cluster with IBM bladecenter management in order to get working
fencing.

regards,
-Mike

2009/8/27 Jalil Faridzadegan <jfaridzadegan at hotmail.com>

>  Hi,
>
> Are fencing devices always required in a redhat cluster configuration or
> optional?
>
> Thanks a lot,
> J. Farid
>
> ------------------------------
> With Windows Live, you can organize, edit, and share your photos.<http://www.microsoft.com/middleeast/windows/windowslive/products/photo-gallery-edit.aspx>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090827/8be3d7b7/attachment.htm>

From linux-cluster at lists.grepular.com  Thu Aug 27 13:39:38 2009
From: linux-cluster at lists.grepular.com (Mike Cardwell)
Date: Thu, 27 Aug 2009 14:39:38 +0100
Subject: [Linux-cluster] FS-Cache
In-Reply-To: <1251379511.6052.8.camel@localhost.localdomain>
References: <4A967D31.5050409@lists.grepular.com>	<1251378187.6052.6.camel@localhost.localdomain>	<4A968892.4050506@lists.grepular.com>
	<1251379511.6052.8.camel@localhost.localdomain>
Message-ID: <4A968C9A.9060901@lists.grepular.com>

On 27/08/2009 14:25, Steven Whitehouse wrote:

>>>> Are there any plans to make GFS work with FS-Cache?
>>>>
>>>> http://people.redhat.com/~dhowells/fscache/FS-Cache.pdf
>>>
>>> Not that I'm aware of. I'm not sure it would make any sense to do so,
>>
>> Hi. Can you explain why it wouldn't make any sense please? I have a GFS
>> filesystem mounted over 100Mb iSCSI (I'm not in a position to upgrade
>> the network speed at this time, but will in future). It's read heavy.
>> Particular files will be hit very regularly. I just thought it would be
>> nice if those files could be cached locally rather than repeatedly
>> pulling them over the network...?
>
> Unless you are doing something to push those files out of the cache,
> then they will be cached locally anyway. GFS2 is a bit better at doing
> that then GFS1 though,

Are you refering to some sort of in-memory cache? I've not been able to 
find any information about how GFS2 caches file data and how to tune it 
(if possible)? Each of my servers has over 100GB of unused local 
storage. It would be nice if I could use that as a cache...

-- 
Mike Cardwell - IT Consultant and LAMP developer
Cardwell IT Ltd. (UK Reg'd Company #06920226) http://cardwellit.com/



From swhiteho at redhat.com  Thu Aug 27 13:48:02 2009
From: swhiteho at redhat.com (Steven Whitehouse)
Date: Thu, 27 Aug 2009 14:48:02 +0100
Subject: [Linux-cluster] FS-Cache
In-Reply-To: <4A968C9A.9060901@lists.grepular.com>
References: <4A967D31.5050409@lists.grepular.com>
	<1251378187.6052.6.camel@localhost.localdomain>
	<4A968892.4050506@lists.grepular.com>
	<1251379511.6052.8.camel@localhost.localdomain>
	<4A968C9A.9060901@lists.grepular.com>
Message-ID: <1251380882.6052.15.camel@localhost.localdomain>

Hi,

On Thu, 2009-08-27 at 14:39 +0100, Mike Cardwell wrote:
> On 27/08/2009 14:25, Steven Whitehouse wrote:
> 
> >>>> Are there any plans to make GFS work with FS-Cache?
> >>>>
> >>>> http://people.redhat.com/~dhowells/fscache/FS-Cache.pdf
> >>>
> >>> Not that I'm aware of. I'm not sure it would make any sense to do so,
> >>
> >> Hi. Can you explain why it wouldn't make any sense please? I have a GFS
> >> filesystem mounted over 100Mb iSCSI (I'm not in a position to upgrade
> >> the network speed at this time, but will in future). It's read heavy.
> >> Particular files will be hit very regularly. I just thought it would be
> >> nice if those files could be cached locally rather than repeatedly
> >> pulling them over the network...?
> >
> > Unless you are doing something to push those files out of the cache,
> > then they will be cached locally anyway. GFS2 is a bit better at doing
> > that then GFS1 though,
> 
> Are you refering to some sort of in-memory cache? I've not been able to 
> find any information about how GFS2 caches file data and how to tune it 
> (if possible)? Each of my servers has over 100GB of unused local 
> storage. It would be nice if I could use that as a cache...
> 
Yes, the page cache, just like any other file system. If you want to
tune it, madvise, fadvise, etc are available. There is currently no
support for caching on other local storage as in most cases (I realise
your case may be an exception) the fastest storage is the shared disk.

If we were do to caching on local storage, then we'd need to make
significant changes to the glock subsystem (even if cachefs were used)
in order to ensure coherency. It would probably result in accesses from
remote nodes becoming so slow (due to the extended invalidation) that it
would only make sense on read-only mounts.

If your application is able to use read-only copies of the data most of
the time, then it might make more sense just to copy the data locally,
process it, and then copy back/delete the local copy as required,

Steve.




From alan.zg at gmail.com  Thu Aug 27 13:59:52 2009
From: alan.zg at gmail.com (Alan A)
Date: Thu, 27 Aug 2009 08:59:52 -0500
Subject: [Linux-cluster] Can not establish new cluster 5.3 with luci - 
	quorum error
In-Reply-To: <fac531740908270658x1c1bc9dctfb8a91fabbd08166@mail.gmail.com>
References: <fac531740908261636j29165920odcdc169f28923f22@mail.gmail.com>
	<20090827102752.391915bc@nb-jsosic>
	<fac531740908270658x1c1bc9dctfb8a91fabbd08166@mail.gmail.com>
Message-ID: <fac531740908270659v339fb159v84820a0832f78292@mail.gmail.com>

Srry about partial post - web browser did that for me.

On Thu, Aug 27, 2009 at 8:58 AM, Alan A <alan.zg at gmail.com> wrote:

> I decided this morning to start checking packages/versions first. Here are
> some details about the system thus far:
>
> CONF:
> <?xml version="1.0" ?>
> <cluster alias="mrcluster" config_version="2" name="mrcluster">
>     <fence_daemon post_fail_delay="0" post_join_delay="30"/>
>     <clusternodes>
>         <clusternode name="clxmrcati12.xxxxxx.com" nodeid="1" votes="1">
>             <fence>
>                 <method name="1">
>                     <device name="apcps05" option="off" port="3"
> switch="3"/>
>                     <device name="apcps06" option="off" port="3"
> switch="3"/>
>                     <device name="apcps05" option="on" port="3"
> switch="3"/>
>                     <device name="apcps06" option="on" port="3"
> switch="3"/>
>                 </method>
>             </fence>
>         </clusternode>
>         <clusternode name="clxmrcati11.xxxxxx.com" nodeid="2" votes="1">
>             <fence>
>                 <method name="1">
>                     <device name="apcps05" option="off" port="4"
> switch="4"/>
>                     <device name="apcps06" option="off" port="4"
> switch="4"/>
>                     <device name="apcps05" option="on" port="4"
> switch="4"/>
>                     <device name="apcps06" option="on" port="4"
> switch="4"/>
>                 </method>
>             </fence>
>         </clusternode>
>         <clusternode name="clxmrweb20.xxxxxx.com" nodeid="3" votes="1">
>             <fence>
>                 <method name="1">
>                     <device name="apcps05" option="off" port="2"
> switch="2"/>
>                     <device name="apcps06" option="off" port="2"
> switch="2"/>
>                     <device name="apcps05" option="on" port="2"
> switch="2"/>
>                     <device name="apcps06" option="on" port="2"
> switch="2"/>
>                 </method>
>             </fence>
>         </clusternode>
>     </clusternodes>
>     <cman/>
>     <fencedevices>
>         <fencedevice agent="fence_apc" ipaddr="172.XX.XX.27" login="apc"
> name="apcps05" passwd="xxx"/>
>         <fencedevice agent="fence_apc" ipaddr="172.XX.XX..28" login="apc"
> name="apcps06" passwd="xxx"/>
>     </fencedevices>
>     <rm>
>         <failoverdomains/>
>         <resources/>
>     </rm>
> </cluster>
>
> -------------------------------------------------------------------------------------------
> Host Files:
> From Luci Node clxmrcati11:
> 127.0.0.1    localhost.localdomain    localhost
> 172.XX.XX.18    clxmrcati11.xxxxxx.com       clxmrcati11
> 172.XX.XX.19    clxmrcati12.xxxxxx.com       clxmrcati12
> 172.XX.XX.20    clxmrrpt10.xxxxxx.com         clxmrrpt10
> 172.XX.XX.21    clxmrweb20.xxxxxx.com      clxmrweb20
>
> From ricci node clxmrcati12:
> 127.0.0.1    localhost.localdomain    localhost
> 172.XX.XX.19    clxmrcati12.maritz.com               fenclxmrcati12
> 172.XX.XX.21    clxmrweb20.maritz.com       I decided this morning to
> start checking packages/versions first. Here are some details about the
> system thus far:
>
> CONF:
> <?xml version="1.0" ?>
> <cluster alias="mrcluster" config_version="2" name="mrcluster">
>     <fence_daemon post_fail_delay="0" post_join_delay="30"/>
>     <clusternodes>
>         <clusternode name="clxmrcati12.xxxxxx.com" nodeid="1" votes="1">
>             <fence>
>                 <method name="1">
>                     <device name="apcps05" option="off" port="3"
> switch="3"/>
>                     <device name="apcps06" option="off" port="3"
> switch="3"/>
>                     <device name="apcps05" option="on" port="3"
> switch="3"/>
>                     <device name="apcps06" option="on" port="3"
> switch="3"/>
>                 </method>
>             </fence>
>         </clusternode>
>         <clusternode name="clxmrcati11.xxxxxx.com" nodeid="2" votes="1">
>             <fence>
>                 <method name="1">
>                     <device name="apcps05" option="off" port="4"
> switch="4"/>
>                     <device name="apcps06" option="off" port="4"
> switch="4"/>
>                     <device name="apcps05" option="on" port="4"
> switch="4"/>
>                     <device name="apcps06" option="on" port="4"
> switch="4"/>
>                 </method>
>             </fence>
>         </clusternode>
>         <clusternode name="clxmrweb20.xxxxxx.com" nodeid="3" votes="1">
>             <fence>
>                 <method name="1">
>                     <device name="apcps05" option="off" port="2"
> switch="2"/>
>                     <device name="apcps06" option="off" port="2"
> switch="2"/>
>                     <device name="apcps05" option="on" port="2"
> switch="2"/>
>                     <device name="apcps06" option="on" port="2"
> switch="2"/>
>                 </method>
>             </fence>
>         </clusternode>
>     </clusternodes>
>     <cman/>
>     <fencedevices>
>         <fencedevice agent="fence_apc" ipaddr="172.XX.XX.27" login="apc"
> name="apcps05" passwd="xxx"/>
>         <fencedevice agent="fence_apc" ipaddr="172.XX.XX..28" login="apc"
> name="apcps06" passwd="xxx"/>
>     </fencedevices>
>     <rm>
>         <failoverdomains/>
>         <resources/>
>     </rm>
> </cluster>
>
> -------------------------------------------------------------------------------------------
> Host Files:
> From Luci Node clxmrcati11:
> 127.0.0.1    localhost.localdomain    localhost
> 172.XX.XX.18    clxmrcati11.xxxxxx.com       clxmrcati11
> 172.XX.XX.19    clxmrcati12.xxxxxx.com       clxmrcati12
> 172.XX.XX.20    clxmrrpt10.xxxxxx.com         clxmrrpt10
> 172.XX.XX.21    clxmrweb20.xxxxxx.com      clxmrweb20
>
> From ricci node clxmrcati12:
> 127.0.0.1    localhost.localdomain    localhost
> 172.XX.XX.19    clxmrcati12.xxxxxx.com               clxmrcati12
> 172.XX.XX.21    clxmrweb20.xxxxxx.com              clxmrweb20
> 172.XX.XX.20    clxmrrpt10.xxxxxx.com                 clxmrrpt10
> 172.XX.XX.18    clxmrcati11.xxxxxx.com               clxmrcati11
>
> From ricci node clxmrweb20:
> 127.0.0.1    localhost.localdomain    localhost
> 172.XX.XX.21    clxmrweb20.xxxxxx.com             clxmrweb20
> 172.XX.XX.20    clxmrrpt10.xxxxxx.com                clxmrrpt10
> 172.XX.XX.18    clxmrcati11.xxxxxx.com              clxmrcati11
> 172.XX.XX.19    clxmrcati12.xxxxxx.com              clxmrcati12
>
> Mostly this in /var/log/messages:
> Aug 25 09:36:12 fenclxmrcati11 dlm_controld[2267]: connect to ccs error
> -111, check ccsd or cluster status
> Aug 25 09:36:12 fenclxmrcati11 ccsd[3758]: Cluster is not quorate.
> Refusing connection.
> Aug 25 09:36:12 fenclxmrcati11 ccsd[3758]: Error while processing connect:
> Connection refused
> Aug 25 09:36:12 fenclxmrcati11 gfs_controld[2273]: connect to ccs error
> -111, check ccsd or cluster status
> Aug 25 09:36:12 fenclxmrcati11 ccsd[3758]: Cluster is not quorate.
> Refusing connection.
> Aug 25 09:36:12 fenclxmrcati11 ccsd[3758]: Error while processing connect:
> Connection refused
> Aug 25 09:36:13 fenclxmrcati11 ccsd[3758]: Cluster is not quorate.
> Refusing connection.
> Aug 25 09:36:13 fenclxmrcati11 ccsd[3758]: Error while processing connect:
> Connection refused
> Aug 25 09:36:13 fenclxmrcati11 ccsd[3758]: Cluster is not quorate.
> Refusing connection.
> Aug 25 09:36:13 fenclxmrcati11 ccsd[3758]: Error while processing connect:
> Connection refused
> Aug 25 09:36:13 fenclxmrcati11 ccsd[3758]: Cluster is not quorate.
> Refusing connection.
> Aug 25 09:36:13 fenclxmrcati11 ccsd[3758]: Error while processing connect:
> Connection refused
> Aug 25 09:36:14 fenclxmrcati11 ccsd[3758]: Cluster is not quorate.
> Refusing connection.
> Aug 25 09:36:14 fenclxmrcati11 ccsd[3758]: Error while processing connect:
> Connection refused
> Aug 25 09:36:14 fenclxmrcati11 ccsd[3758]: Cluster is not quorate.
> Refusing connection.
> Aug 25 09:36:14 fenclxmrcati11 ccsd[3758]: Error while processing connect:
> Connection re
>
>
>
> On Thu, Aug 27, 2009 at 3:27 AM, Jakov Sosic <jakov.sosic at srce.hr> wrote:
>
>> On Wed, 26 Aug 2009 18:36:26 -0500
>> Alan A <alan.zg at gmail.com> wrote:
>>
>> > I have tried almost everything at this point to try and troubleshoot
>> > this further. I can't create new cluster with luci.
>> >
>>
>          fenclxmrweb20
> 172.XX.XX.20    clxmrrpt10.maritz.com                fenclxmrrpt10
> 172.XX.XX.18    clxmrcati11..com               clxmrcati11
>
>
>
>
>
>
> On Thu, Aug 27, 2009 at 3:27 AM, Jakov Sosic <jakov.sosic at srce.hr> wrote:
>
>> On Wed, 26 Aug 2009 18:36:26 -0500
>> Alan A <alan.zg at gmail.com> wrote:
>>
>> > I have tried almost everything at this point to try and troubleshoot
>> > this further. I can't create new cluster with luci.
>> >
>> > I broke and tried to reconfigure 3 node cluster at least 6 times.
>> >
>> > I have noticed nodes taking expectational long on initializing
>> > fencing upon cman start. I tried with defined and undefined fencing,
>> > the amount of time needed is still the same. Even after the fencing
>> > is overcome in /var/log/messages nodes refuse to join cluster due to
>> > the state of 'not in quorum' during joining process. I uped the
>> > post_join_delay as much as 150 but the result is the same.
>> >
>> > Fencing - I use APC PW Switches - I can login into apc PWS from the
>> > node, I can even fence the other node, but when cman is started it
>> > looks like it is almost timign out on staring fencing.
>> >
>> > If I issue cman_tool nodes it gives me the local node name as the
>> > member of the cluster and the other two with state 'X'. If I try
>> > cman_tool join clustername - it tells me the nodes are already in
>> > that cluster but cluster as the whole does not register. Each node
>> > thinks it's the only working member of the cluster.
>> >
>> >
>> > Any pointers?
>>
>> Looks like network issue to me.
>>
>> Are you sure your network is operational in a sense of a multicast /
>> igmp? Try forcing igmp v1 in sysctl.conf - and if you have Cisco
>> equipment take a look at openais FAQ (mode sparse-dense).
>>
>>
>> --
>> |    Jakov Sosic    |    ICQ: 28410271    |   PGP: 0x965CAE2D   |
>> =================================================================
>> | start fighting cancer -> http://www.worldcommunitygrid.org/   |
>>
>> --
>> Linux-cluster mailing list
>> Linux-cluster at redhat.com
>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>
>
>
>
> --
> Alan A.
>



-- 
Alan A.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090827/7db97f5f/attachment.htm>

From alan.zg at gmail.com  Thu Aug 27 14:05:41 2009
From: alan.zg at gmail.com (Alan A)
Date: Thu, 27 Aug 2009 09:05:41 -0500
Subject: [Linux-cluster] Can not establish new cluster 5.3 with luci - 
	quorum error
In-Reply-To: <20090827102752.391915bc@nb-jsosic>
References: <fac531740908261636j29165920odcdc169f28923f22@mail.gmail.com>
	<20090827102752.391915bc@nb-jsosic>
Message-ID: <fac531740908270705rea433f4m94988f31d25e975e@mail.gmail.com>

Host file looks OK, pardon my post for my partial hostname editing didn't
quite work as expected since my mouse played a game on me.
We do run Cisco equipment - any pointers on what exactly should I check. I
am planning on opening ticket with RH as well.

Xavier - thanks for your suggestion too - will try it.

On Thu, Aug 27, 2009 at 3:27 AM, Jakov Sosic <jakov.sosic at srce.hr> wrote:

> On Wed, 26 Aug 2009 18:36:26 -0500
> Alan A <alan.zg at gmail.com> wrote:
>
> > I have tried almost everything at this point to try and troubleshoot
> > this further. I can't create new cluster with luci.
> >
> > I broke and tried to reconfigure 3 node cluster at least 6 times.
> >
> > I have noticed nodes taking expectational long on initializing
> > fencing upon cman start. I tried with defined and undefined fencing,
> > the amount of time needed is still the same. Even after the fencing
> > is overcome in /var/log/messages nodes refuse to join cluster due to
> > the state of 'not in quorum' during joining process. I uped the
> > post_join_delay as much as 150 but the result is the same.
> >
> > Fencing - I use APC PW Switches - I can login into apc PWS from the
> > node, I can even fence the other node, but when cman is started it
> > looks like it is almost timign out on staring fencing.
> >
> > If I issue cman_tool nodes it gives me the local node name as the
> > member of the cluster and the other two with state 'X'. If I try
> > cman_tool join clustername - it tells me the nodes are already in
> > that cluster but cluster as the whole does not register. Each node
> > thinks it's the only working member of the cluster.
> >
> >
> > Any pointers?
>
> Looks like network issue to me.
>
> Are you sure your network is operational in a sense of a multicast /
> igmp? Try forcing igmp v1 in sysctl.conf - and if you have Cisco
> equipment take a look at openais FAQ (mode sparse-dense).
>
>
> --
> |    Jakov Sosic    |    ICQ: 28410271    |   PGP: 0x965CAE2D   |
> =================================================================
> | start fighting cancer -> http://www.worldcommunitygrid.org/   |
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>



-- 
Alan A.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090827/fcc23b59/attachment.htm>

From Alain.Moulle at bull.net  Thu Aug 27 14:06:29 2009
From: Alain.Moulle at bull.net (Alain.Moulle)
Date: Thu, 27 Aug 2009 16:06:29 +0200
Subject: [Linux-cluster] cluster.conf in another place ?
Message-ID: <4A9692E5.9030607@bull.net>

Hi Fabio,
cman-3.0.0-15.rc1.fc11.x86_64
rgmanager-3.0.0-15.rc1.fc11.x86_64
Alain
> On Thu, 2009-08-27 at 08:31 +0200, Alain.Moulle wrote:
>> > Hi,
>> > I'd need to put the cluster.conf in another place than /etc/cluster/.
>> > (without any symbolic link)
>> > Is-it possible ?
>>     
>
> What version of the stack are you using?
> Fabio
>
>   
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090827/098ee38a/attachment.htm>

From Eric.Johnson at mtsallstream.com  Thu Aug 27 14:25:00 2009
From: Eric.Johnson at mtsallstream.com (Johnson, Eric)
Date: Thu, 27 Aug 2009 09:25:00 -0500
Subject: [Linux-cluster] GFS2 subdirectory hang
Message-ID: <CD9C931A046A4A41876F7F5E134298B40206F392@PTEEXB02.mtsallstream.com>

I have a 32-bit RHEL 5.3 Cluster Suite setup of two nodes with GFS2 file
systems on FC attached SAN. I have run into this issue twice now, where
attempts to access a certain directory within one of the GFS2 file
systems never return. Other directories and paths within that file
system work just fine.

The first time it happened, I had to crash the node to get it to release
the FS, then unmount it on both nodes, fsck it, remount it, and it was
fine. It has happened again (different path, different file system). A
simple "ls" in the directory (which has maybe 20 files in it) leaves the
process in an uninterruptible sleep state. I left it all night and it
never returned.

I'm not sure what other info would be useful on this, but this is what I
see from a gfs2_tool lockdump output for ls PID on that node:

G:  s:UN n:2/bf1df f:l t:SH d:EX/0 l:0 a:0 r:4
 H: s:SH f:aW e:0 p:9938 [ls] gfs2_lookup+0x44/0x90 [gfs2]
G:  s:SH n:5/bf9ce f: t:SH d:EX/0 l:0 a:0 r:3
 H: s:SH f:EH e:0 p:9938 [ls] gfs2_inode_lookup+0x12d/0x1f0 [gfs2]
G:  s:SH n:5/bf1df f: t:SH d:EX/0 l:0 a:0 r:3
 H: s:SH f:EH e:0 p:9938 [ls] gfs2_inode_lookup+0x12d/0x1f0 [gfs2]
G:  s:SH n:5/20c9b f: t:SH d:EX/0 l:0 a:0 r:3
 H: s:SH f:EH e:0 p:9938 [ls] gfs2_inode_lookup+0x12d/0x1f0 [gfs2]

I know there have been a few updates that have been released that
address a couple of GFS2 issues, but I don't recall seeing one for this
type of problem. I've got kernel 2.6.18-128.1.14el5PAE and
gfs2-utils-0.1.53-1.el5_3.3.

Thanks,
Eric



From alan.zg at gmail.com  Thu Aug 27 14:44:27 2009
From: alan.zg at gmail.com (Alan A)
Date: Thu, 27 Aug 2009 09:44:27 -0500
Subject: [Linux-cluster] Can not establish new cluster 5.3 with luci - 
	quorum error
In-Reply-To: <fac531740908270705rea433f4m94988f31d25e975e@mail.gmail.com>
References: <fac531740908261636j29165920odcdc169f28923f22@mail.gmail.com>
	<20090827102752.391915bc@nb-jsosic>
	<fac531740908270705rea433f4m94988f31d25e975e@mail.gmail.com>
Message-ID: <fac531740908270744o62f4d06ct87e0a993c149c9fb@mail.gmail.com>

I checked packages and they are all up to date on all nodes:

[root at clxmrweb20 ~]# yum groupinstall Clustering
Loaded plugins: rhnplugin, security
Setting up Group Process
Package ricci-0.12.1-7.3.el5_3.i386 already installed and latest version
Package system-config-cluster-1.0.55-1.0.noarch already installed and latest
version
Package ipvsadm-1.24-8.1.i386 already installed and latest version
Package piranha-0.8.4-11.el5.i386 already installed and latest version
Package cluster-snmp-0.12.1-2.el5.i386 already installed and latest version
Package modcluster-0.12.1-2.el5.i386 already installed and latest version
Package rgmanager-2.0.46-1.el5_3.4.i386 already installed and latest version
Package Cluster_Administration-en-US-5.2-1.noarch already installed and
latest version
Package cluster-cim-0.12.1-2.el5.i386 already installed and latest version
Package luci-0.12.1-7.3.el5_3.i386 already installed and latest version
No packages in any requested group available to install or update

[root at clxmrweb20 ~]# yum groupinstall "Cluster Storage"
Loaded plugins: rhnplugin, security
Setting up Group Process
Package kmod-gnbd-0.1.5-2.el5.i686 already installed and latest version
Package lvm2-cluster-2.02.40-7.el5.i386 already installed and latest version
Package gfs-utils-0.1.18-1.el5.i386 already installed and latest version
Package kmod-gfs-0.1.31-3.el5_3.1.i686 already installed and latest version
Package gnbd-1.1.7-1.el5.i386 already installed and latest version
Package Global_File_System-en-US-5.2-1.noarch already installed and latest
version
No packages in any requested group available to install or update


On Thu, Aug 27, 2009 at 9:05 AM, Alan A <alan.zg at gmail.com> wrote:

> Host file looks OK, pardon my post for my partial hostname editing didn't
> quite work as expected since my mouse played a game on me.
> We do run Cisco equipment - any pointers on what exactly should I check. I
> am planning on opening ticket with RH as well.
>
> Xavier - thanks for your suggestion too - will try it.
>
> On Thu, Aug 27, 2009 at 3:27 AM, Jakov Sosic <jakov.sosic at srce.hr> wrote:
>
>> On Wed, 26 Aug 2009 18:36:26 -0500
>> Alan A <alan.zg at gmail.com> wrote:
>>
>> > I have tried almost everything at this point to try and troubleshoot
>> > this further. I can't create new cluster with luci.
>> >
>> > I broke and tried to reconfigure 3 node cluster at least 6 times.
>> >
>> > I have noticed nodes taking expectational long on initializing
>> > fencing upon cman start. I tried with defined and undefined fencing,
>> > the amount of time needed is still the same. Even after the fencing
>> > is overcome in /var/log/messages nodes refuse to join cluster due to
>> > the state of 'not in quorum' during joining process. I uped the
>> > post_join_delay as much as 150 but the result is the same.
>> >
>> > Fencing - I use APC PW Switches - I can login into apc PWS from the
>> > node, I can even fence the other node, but when cman is started it
>> > looks like it is almost timign out on staring fencing.
>> >
>> > If I issue cman_tool nodes it gives me the local node name as the
>> > member of the cluster and the other two with state 'X'. If I try
>> > cman_tool join clustername - it tells me the nodes are already in
>> > that cluster but cluster as the whole does not register. Each node
>> > thinks it's the only working member of the cluster.
>> >
>> >
>> > Any pointers?
>>
>> Looks like network issue to me.
>>
>> Are you sure your network is operational in a sense of a multicast /
>> igmp? Try forcing igmp v1 in sysctl.conf - and if you have Cisco
>> equipment take a look at openais FAQ (mode sparse-dense).
>>
>>
>> --
>> |    Jakov Sosic    |    ICQ: 28410271    |   PGP: 0x965CAE2D   |
>> =================================================================
>> | start fighting cancer -> http://www.worldcommunitygrid.org/   |
>>
>> --
>> Linux-cluster mailing list
>> Linux-cluster at redhat.com
>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>
>
>
>
> --
> Alan A.
>



-- 
Alan A.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090827/b8c4f36c/attachment.htm>

From swhiteho at redhat.com  Thu Aug 27 14:46:23 2009
From: swhiteho at redhat.com (Steven Whitehouse)
Date: Thu, 27 Aug 2009 15:46:23 +0100
Subject: [Linux-cluster] GFS2 subdirectory hang
In-Reply-To: <CD9C931A046A4A41876F7F5E134298B40206F392@PTEEXB02.mtsallstream.com>
References: <CD9C931A046A4A41876F7F5E134298B40206F392@PTEEXB02.mtsallstream.com>
Message-ID: <1251384383.6052.24.camel@localhost.localdomain>

Hi,

On Thu, 2009-08-27 at 09:25 -0500, Johnson, Eric wrote:
> I have a 32-bit RHEL 5.3 Cluster Suite setup of two nodes with GFS2 file
> systems on FC attached SAN. I have run into this issue twice now, where
> attempts to access a certain directory within one of the GFS2 file
> systems never return. Other directories and paths within that file
> system work just fine.
> 
> The first time it happened, I had to crash the node to get it to release
> the FS, then unmount it on both nodes, fsck it, remount it, and it was
> fine. It has happened again (different path, different file system). A
> simple "ls" in the directory (which has maybe 20 files in it) leaves the
> process in an uninterruptible sleep state. I left it all night and it
> never returned.
> 
> I'm not sure what other info would be useful on this, but this is what I
> see from a gfs2_tool lockdump output for ls PID on that node:
> 
> G:  s:UN n:2/bf1df f:l t:SH d:EX/0 l:0 a:0 r:4
>  H: s:SH f:aW e:0 p:9938 [ls] gfs2_lookup+0x44/0x90 [gfs2]
              ^ The W flag indicates that this is waiting for a glock

Currently the glock is in the UN (unlocked) state, and its trying to get
a SH (shared) lock. The next step in the investigation is to look for
the same glock number 2/bf1df on the other nodes, and see what is
holding that lock. This particular node will hang until the lock is
released on whichever other node is holding it.

If there is nothing on any other node apparently holding that lock in
the glock dumps, then looking at dlm lock dumps would be the next step,

Steve.


> G:  s:SH n:5/bf9ce f: t:SH d:EX/0 l:0 a:0 r:3
>  H: s:SH f:EH e:0 p:9938 [ls] gfs2_inode_lookup+0x12d/0x1f0 [gfs2]
> G:  s:SH n:5/bf1df f: t:SH d:EX/0 l:0 a:0 r:3
>  H: s:SH f:EH e:0 p:9938 [ls] gfs2_inode_lookup+0x12d/0x1f0 [gfs2]
> G:  s:SH n:5/20c9b f: t:SH d:EX/0 l:0 a:0 r:3
>  H: s:SH f:EH e:0 p:9938 [ls] gfs2_inode_lookup+0x12d/0x1f0 [gfs2]
> 
> I know there have been a few updates that have been released that
> address a couple of GFS2 issues, but I don't recall seeing one for this
> type of problem. I've got kernel 2.6.18-128.1.14el5PAE and
> gfs2-utils-0.1.53-1.el5_3.3.
> 
> Thanks,
> Eric
> 
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster



From alan.zg at gmail.com  Thu Aug 27 13:58:02 2009
From: alan.zg at gmail.com (Alan A)
Date: Thu, 27 Aug 2009 08:58:02 -0500
Subject: [Linux-cluster] Can not establish new cluster 5.3 with luci - 
	quorum error
In-Reply-To: <20090827102752.391915bc@nb-jsosic>
References: <fac531740908261636j29165920odcdc169f28923f22@mail.gmail.com>
	<20090827102752.391915bc@nb-jsosic>
Message-ID: <fac531740908270658x1c1bc9dctfb8a91fabbd08166@mail.gmail.com>

I decided this morning to start checking packages/versions first. Here are
some details about the system thus far:

CONF:
<?xml version="1.0" ?>
<cluster alias="mrcluster" config_version="2" name="mrcluster">
    <fence_daemon post_fail_delay="0" post_join_delay="30"/>
    <clusternodes>
        <clusternode name="clxmrcati12.xxxxxx.com" nodeid="1" votes="1">
            <fence>
                <method name="1">
                    <device name="apcps05" option="off" port="3"
switch="3"/>
                    <device name="apcps06" option="off" port="3"
switch="3"/>
                    <device name="apcps05" option="on" port="3" switch="3"/>
                    <device name="apcps06" option="on" port="3" switch="3"/>
                </method>
            </fence>
        </clusternode>
        <clusternode name="clxmrcati11.xxxxxx.com" nodeid="2" votes="1">
            <fence>
                <method name="1">
                    <device name="apcps05" option="off" port="4"
switch="4"/>
                    <device name="apcps06" option="off" port="4"
switch="4"/>
                    <device name="apcps05" option="on" port="4" switch="4"/>
                    <device name="apcps06" option="on" port="4" switch="4"/>
                </method>
            </fence>
        </clusternode>
        <clusternode name="clxmrweb20.xxxxxx.com" nodeid="3" votes="1">
            <fence>
                <method name="1">
                    <device name="apcps05" option="off" port="2"
switch="2"/>
                    <device name="apcps06" option="off" port="2"
switch="2"/>
                    <device name="apcps05" option="on" port="2" switch="2"/>
                    <device name="apcps06" option="on" port="2" switch="2"/>
                </method>
            </fence>
        </clusternode>
    </clusternodes>
    <cman/>
    <fencedevices>
        <fencedevice agent="fence_apc" ipaddr="172.XX.XX.27" login="apc"
name="apcps05" passwd="xxx"/>
        <fencedevice agent="fence_apc" ipaddr="172.XX.XX..28" login="apc"
name="apcps06" passwd="xxx"/>
    </fencedevices>
    <rm>
        <failoverdomains/>
        <resources/>
    </rm>
</cluster>
-------------------------------------------------------------------------------------------
Host Files:
>From Luci Node clxmrcati11:
127.0.0.1    localhost.localdomain    localhost
172.XX.XX.18    clxmrcati11.xxxxxx.com       clxmrcati11
172.XX.XX.19    clxmrcati12.xxxxxx.com       clxmrcati12
172.XX.XX.20    clxmrrpt10.xxxxxx.com         clxmrrpt10
172.XX.XX.21    clxmrweb20.xxxxxx.com      clxmrweb20

>From ricci node clxmrcati12:
127.0.0.1    localhost.localdomain    localhost
172.XX.XX.19    clxmrcati12.maritz.com               fenclxmrcati12
172.XX.XX.21    clxmrweb20.maritz.com       I decided this morning to start
checking packages/versions first. Here are some details about the system
thus far:

CONF:
<?xml version="1.0" ?>
<cluster alias="mrcluster" config_version="2" name="mrcluster">
    <fence_daemon post_fail_delay="0" post_join_delay="30"/>
    <clusternodes>
        <clusternode name="clxmrcati12.xxxxxx.com" nodeid="1" votes="1">
            <fence>
                <method name="1">
                    <device name="apcps05" option="off" port="3"
switch="3"/>
                    <device name="apcps06" option="off" port="3"
switch="3"/>
                    <device name="apcps05" option="on" port="3" switch="3"/>
                    <device name="apcps06" option="on" port="3" switch="3"/>
                </method>
            </fence>
        </clusternode>
        <clusternode name="clxmrcati11.xxxxxx.com" nodeid="2" votes="1">
            <fence>
                <method name="1">
                    <device name="apcps05" option="off" port="4"
switch="4"/>
                    <device name="apcps06" option="off" port="4"
switch="4"/>
                    <device name="apcps05" option="on" port="4" switch="4"/>
                    <device name="apcps06" option="on" port="4" switch="4"/>
                </method>
            </fence>
        </clusternode>
        <clusternode name="clxmrweb20.xxxxxx.com" nodeid="3" votes="1">
            <fence>
                <method name="1">
                    <device name="apcps05" option="off" port="2"
switch="2"/>
                    <device name="apcps06" option="off" port="2"
switch="2"/>
                    <device name="apcps05" option="on" port="2" switch="2"/>
                    <device name="apcps06" option="on" port="2" switch="2"/>
                </method>
            </fence>
        </clusternode>
    </clusternodes>
    <cman/>
    <fencedevices>
        <fencedevice agent="fence_apc" ipaddr="172.XX.XX.27" login="apc"
name="apcps05" passwd="xxx"/>
        <fencedevice agent="fence_apc" ipaddr="172.XX.XX..28" login="apc"
name="apcps06" passwd="xxx"/>
    </fencedevices>
    <rm>
        <failoverdomains/>
        <resources/>
    </rm>
</cluster>
-------------------------------------------------------------------------------------------
Host Files:
>From Luci Node clxmrcati11:
127.0.0.1    localhost.localdomain    localhost
172.XX.XX.18    clxmrcati11.xxxxxx.com       clxmrcati11
172.XX.XX.19    clxmrcati12.xxxxxx.com       clxmrcati12
172.XX.XX.20    clxmrrpt10.xxxxxx.com         clxmrrpt10
172.XX.XX.21    clxmrweb20.xxxxxx.com      clxmrweb20

>From ricci node clxmrcati12:
127.0.0.1    localhost.localdomain    localhost
172.XX.XX.19    clxmrcati12.xxxxxx.com               clxmrcati12
172.XX.XX.21    clxmrweb20.xxxxxx.com              clxmrweb20
172.XX.XX.20    clxmrrpt10.xxxxxx.com                 clxmrrpt10
172.XX.XX.18    clxmrcati11.xxxxxx.com               clxmrcati11

>From ricci node clxmrweb20:
127.0.0.1    localhost.localdomain    localhost
172.XX.XX.21    clxmrweb20.xxxxxx.com             clxmrweb20
172.XX.XX.20    clxmrrpt10.xxxxxx.com                clxmrrpt10
172.XX.XX.18    clxmrcati11.xxxxxx.com              clxmrcati11
172.XX.XX.19    clxmrcati12.xxxxxx.com              clxmrcati12

Mostly this in /var/log/messages:
Aug 25 09:36:12 fenclxmrcati11 dlm_controld[2267]: connect to ccs error
-111, check ccsd or cluster status
Aug 25 09:36:12 fenclxmrcati11 ccsd[3758]: Cluster is not quorate.  Refusing
connection.
Aug 25 09:36:12 fenclxmrcati11 ccsd[3758]: Error while processing connect:
Connection refused
Aug 25 09:36:12 fenclxmrcati11 gfs_controld[2273]: connect to ccs error
-111, check ccsd or cluster status
Aug 25 09:36:12 fenclxmrcati11 ccsd[3758]: Cluster is not quorate.  Refusing
connection.
Aug 25 09:36:12 fenclxmrcati11 ccsd[3758]: Error while processing connect:
Connection refused
Aug 25 09:36:13 fenclxmrcati11 ccsd[3758]: Cluster is not quorate.  Refusing
connection.
Aug 25 09:36:13 fenclxmrcati11 ccsd[3758]: Error while processing connect:
Connection refused
Aug 25 09:36:13 fenclxmrcati11 ccsd[3758]: Cluster is not quorate.  Refusing
connection.
Aug 25 09:36:13 fenclxmrcati11 ccsd[3758]: Error while processing connect:
Connection refused
Aug 25 09:36:13 fenclxmrcati11 ccsd[3758]: Cluster is not quorate.  Refusing
connection.
Aug 25 09:36:13 fenclxmrcati11 ccsd[3758]: Error while processing connect:
Connection refused
Aug 25 09:36:14 fenclxmrcati11 ccsd[3758]: Cluster is not quorate.  Refusing
connection.
Aug 25 09:36:14 fenclxmrcati11 ccsd[3758]: Error while processing connect:
Connection refused
Aug 25 09:36:14 fenclxmrcati11 ccsd[3758]: Cluster is not quorate.  Refusing
connection.
Aug 25 09:36:14 fenclxmrcati11 ccsd[3758]: Error while processing connect:
Connection re



On Thu, Aug 27, 2009 at 3:27 AM, Jakov Sosic <jakov.sosic at srce.hr> wrote:

> On Wed, 26 Aug 2009 18:36:26 -0500
> Alan A <alan.zg at gmail.com> wrote:
>
> > I have tried almost everything at this point to try and troubleshoot
> > this further. I can't create new cluster with luci.
> >
>
         fenclxmrweb20
172.XX.XX.20    clxmrrpt10.maritz.com                fenclxmrrpt10
172.XX.XX.18    clxmrcati11..com               clxmrcati11





On Thu, Aug 27, 2009 at 3:27 AM, Jakov Sosic <jakov.sosic at srce.hr> wrote:

> On Wed, 26 Aug 2009 18:36:26 -0500
> Alan A <alan.zg at gmail.com> wrote:
>
> > I have tried almost everything at this point to try and troubleshoot
> > this further. I can't create new cluster with luci.
> >
> > I broke and tried to reconfigure 3 node cluster at least 6 times.
> >
> > I have noticed nodes taking expectational long on initializing
> > fencing upon cman start. I tried with defined and undefined fencing,
> > the amount of time needed is still the same. Even after the fencing
> > is overcome in /var/log/messages nodes refuse to join cluster due to
> > the state of 'not in quorum' during joining process. I uped the
> > post_join_delay as much as 150 but the result is the same.
> >
> > Fencing - I use APC PW Switches - I can login into apc PWS from the
> > node, I can even fence the other node, but when cman is started it
> > looks like it is almost timign out on staring fencing.
> >
> > If I issue cman_tool nodes it gives me the local node name as the
> > member of the cluster and the other two with state 'X'. If I try
> > cman_tool join clustername - it tells me the nodes are already in
> > that cluster but cluster as the whole does not register. Each node
> > thinks it's the only working member of the cluster.
> >
> >
> > Any pointers?
>
> Looks like network issue to me.
>
> Are you sure your network is operational in a sense of a multicast /
> igmp? Try forcing igmp v1 in sysctl.conf - and if you have Cisco
> equipment take a look at openais FAQ (mode sparse-dense).
>
>
> --
> |    Jakov Sosic    |    ICQ: 28410271    |   PGP: 0x965CAE2D   |
> =================================================================
> | start fighting cancer -> http://www.worldcommunitygrid.org/   |
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>



-- 
Alan A.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090827/03ba1ca7/attachment.htm>

From edsonmarquezani at gmail.com  Thu Aug 27 15:14:09 2009
From: edsonmarquezani at gmail.com (Edson Marquezani Filho)
Date: Thu, 27 Aug 2009 12:14:09 -0300
Subject: [Linux-cluster] Re: Linux-cluster Digest, Preventing LVM from 
	concurrent access
In-Reply-To: <1251196586.5835.6.camel@quann>
References: <20090823160008.CA2B36194C4@hormel.redhat.com>
	<1269910516-1251051757-cardhu_decombobulator_blackberry.rim.net-1598207711-@bxe1291.bisx.prod.on.blackberry>
	<2fc5f090908231201j534884f0ke1648bf5426d0c3e@mail.gmail.com>
	<1251115598.5844.31.camel@quann>
	<2fc5f090908240813n45137862s876ec7ffa8c4a625@mail.gmail.com>
	<1251129846.5898.15.camel@quann>
	<2fc5f090908241227t1c709415t2475c3cf91126667@mail.gmail.com>
	<1251144303.5840.28.camel@quann>
	<2fc5f090908241348p73461b62u76835829be10a91d@mail.gmail.com>
	<1251196586.5835.6.camel@quann>
Message-ID: <2fc5f090908270814q1e7a63a3g602a67fa1dd99655@mail.gmail.com>

Ok, after a while, things is starting to work here. I have defined a
service, with a lvm-cluster resource, and my (still very simple)
script that launchs the VM and destroys them. Service is moved from
one server to another, when some goes down, and VM are recreated.

But, I could launch the same VM (with the same LV as its disk)
manually on the slave node, even though it was mounted on master. I
guess that it was expected, right?, once you have warned about that
before.

"Actually it would not avoid an admin doing, by hand, an erratic mount
but a bugzilla ticket has been opened by Brem Belguebli to fix an LVM
issue that is causing that behaviour."

But, rgmanager was able to mount it on both servers simultaneosly too.
I runned a test, disconecting the heartbeat link, making one server to
be fenced, and the VM launched on the "winner" as expected. But when
the "loser" server came back, still without heartbeat link, it
launched the same VM again, and service appeared as running locally on
boths nodes. I guess lvm-cluster should avoid this, shouldn't ?

I have not understood, because this behauvior is the same as if I had
not defined any lvm or lvm-cluster resource to the service. Both nodes
see the LV, anyway, and could mount it whenever them want.

What could be wrong?

Besides that, I can't relocate the service to another node by hand,
with clusvcadm, what causes the service to fail and become inactive,
forcing me to disable and enable it again.

Thanks.



From rmicmirregs at gmail.com  Thu Aug 27 16:34:25 2009
From: rmicmirregs at gmail.com (Rafael =?ISO-8859-1?Q?Mic=F3?= Miranda)
Date: Thu, 27 Aug 2009 18:34:25 +0200
Subject: [Linux-cluster] Re: Linux-cluster Digest, Preventing LVM from 
	concurrent access
In-Reply-To: <2fc5f090908270814q1e7a63a3g602a67fa1dd99655@mail.gmail.com>
References: <20090823160008.CA2B36194C4@hormel.redhat.com>
	<1269910516-1251051757-cardhu_decombobulator_blackberry.rim.net-1598207711-@bxe1291.bisx.prod.on.blackberry>
	<2fc5f090908231201j534884f0ke1648bf5426d0c3e@mail.gmail.com>
	<1251115598.5844.31.camel@quann>
	<2fc5f090908240813n45137862s876ec7ffa8c4a625@mail.gmail.com>
	<1251129846.5898.15.camel@quann>
	<2fc5f090908241227t1c709415t2475c3cf91126667@mail.gmail.com>
	<1251144303.5840.28.camel@quann>
	<2fc5f090908241348p73461b62u76835829be10a91d@mail.gmail.com>
	<1251196586.5835.6.camel@quann>
	<2fc5f090908270814q1e7a63a3g602a67fa1dd99655@mail.gmail.com>
Message-ID: <1251390865.5849.3.camel@quann>

Hi Edson,

El jue, 27-08-2009 a las 12:14 -0300, Edson Marquezani Filho escribi?:
> Ok, after a while, things is starting to work here. I have defined a
> service, with a lvm-cluster resource, and my (still very simple)
> script that launchs the VM and destroys them. Service is moved from
> one server to another, when some goes down, and VM are recreated.
> 
> But, I could launch the same VM (with the same LV as its disk)
> manually on the slave node, even though it was mounted on master. I
> guess that it was expected, right?, once you have warned about that
> before.
> 
> "Actually it would not avoid an admin doing, by hand, an erratic mount
> but a bugzilla ticket has been opened by Brem Belguebli to fix an LVM
> issue that is causing that behaviour."
> 

Yes, you can bypass the exclusive flag by hand if you dont take it into
account when activating the exclusive LV on a second node.

While the exclusive LV is mounted in other node, the command "lvchange
-aey XXX/YYY" should give you an error message. "lvchange -ay XXX/YYY"
will bypass it.

> But, rgmanager was able to mount it on both servers simultaneosly too.
> I runned a test, disconecting the heartbeat link, making one server to
> be fenced, and the VM launched on the "winner" as expected. But when
> the "loser" server came back, still without heartbeat link, it
> launched the same VM again, and service appeared as running locally on
> boths nodes. I guess lvm-cluster should avoid this, shouldn't ?
> 

This should not happen. Have you set "exclusive=yes" into the resource
definitions in cluster.conf? Can we have a copy of your current
cluster.conf?

> I have not understood, because this behauvior is the same as if I had
> not defined any lvm or lvm-cluster resource to the service. Both nodes
> see the LV, anyway, and could mount it whenever them want.
> 
> What could be wrong?
> 
> Besides that, I can't relocate the service to another node by hand,
> with clusvcadm, what causes the service to fail and become inactive,
> forcing me to disable and enable it again.
> 

I tested it and it should work. Same answer as before: give us a copy of
your current cluster.conf.


> Thanks.
> 
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster

Cheers,

Rafael

-- 
Rafael Mic? Miranda



From jakov.sosic at srce.hr  Thu Aug 27 17:59:04 2009
From: jakov.sosic at srce.hr (Jakov Sosic)
Date: Thu, 27 Aug 2009 19:59:04 +0200
Subject: [Linux-cluster] Can not establish new cluster 5.3 with luci - 
	quorum error
In-Reply-To: <fac531740908270705rea433f4m94988f31d25e975e@mail.gmail.com>
References: <fac531740908261636j29165920odcdc169f28923f22@mail.gmail.com>
	<20090827102752.391915bc@nb-jsosic>
	<fac531740908270705rea433f4m94988f31d25e975e@mail.gmail.com>
Message-ID: <20090827195904.0bf14900@pc-jsosic.srce.hr>

On Thu, 27 Aug 2009 09:05:41 -0500
Alan A <alan.zg at gmail.com> wrote:

> Host file looks OK, pardon my post for my partial hostname editing
> didn't quite work as expected since my mouse played a game on me.
> We do run Cisco equipment - any pointers on what exactly should I
> check. I am planning on opening ticket with RH as well.

Try this one...

http://www.openais.org/doku.php?id=faq:cisco_switches


We had same situation before we configured Cisco to behave this way.




-- 
|    Jakov Sosic    |    ICQ: 28410271    |   PGP: 0x965CAE2D   |
=================================================================
| start fighting cancer -> http://www.worldcommunitygrid.org/   |



From Eric.Johnson at mtsallstream.com  Thu Aug 27 18:16:03 2009
From: Eric.Johnson at mtsallstream.com (Johnson, Eric)
Date: Thu, 27 Aug 2009 13:16:03 -0500
Subject: [Linux-cluster] RE: GFS2 subdirectory hang
In-Reply-To: <20090827145704.5A56D61A5AB@hormel.redhat.com>
References: <20090827145704.5A56D61A5AB@hormel.redhat.com>
Message-ID: <CD9C931A046A4A41876F7F5E134298B40206F54B@PTEEXB02.mtsallstream.com>


On Thu, 2009-08-27 at 09:25 -0500, Johnson, Eric wrote:
>> I have a 32-bit RHEL 5.3 Cluster Suite setup of two nodes with GFS2
file
>> systems on FC attached SAN. I have run into this issue twice now,
where
>> attempts to access a certain directory within one of the GFS2 file
>> systems never return. Other directories and paths within that file
>> system work just fine.
>> 
>> The first time it happened, I had to crash the node to get it to
release
>> the FS, then unmount it on both nodes, fsck it, remount it, and it
was
>> fine. It has happened again (different path, different file system).
A
>> simple "ls" in the directory (which has maybe 20 files in it) leaves
the
>> process in an uninterruptible sleep state. I left it all night and it
>> never returned.
>> 
>> I'm not sure what other info would be useful on this, but this is
what I
>> see from a gfs2_tool lockdump output for ls PID on that node:
>> 
>> G:  s:UN n:2/bf1df f:l t:SH d:EX/0 l:0 a:0 r:4
>>  H: s:SH f:aW e:0 p:9938 [ls] gfs2_lookup+0x44/0x90 [gfs2]
>              ^ The W flag indicates that this is waiting for a glock
>
>Currently the glock is in the UN (unlocked) state, and its trying to
get
>a SH (shared) lock. The next step in the investigation is to look for
>the same glock number 2/bf1df on the other nodes, and see what is
>holding that lock. This particular node will hang until the lock is
>released on whichever other node is holding it.
>
>If there is nothing on any other node apparently holding that lock in
>the glock dumps, then looking at dlm lock dumps would be the next step,
>
>Steve.

Thanks for the response, Steve. I found this reference to that lock on
the other node:

G:  s:EX n:2/bf1df f:dy t:EX d:SH/0 l:0 a:0 r:4
 I: n:1155192/782815 t:8 f:0x00000010

I'm having trouble finding documentation that describes what each of
these fields are. There's no obvious process ID here, and all I can
determine is that it's an exclusive lock.

Eric



From edsonmarquezani at gmail.com  Thu Aug 27 20:47:15 2009
From: edsonmarquezani at gmail.com (Edson Marquezani Filho)
Date: Thu, 27 Aug 2009 17:47:15 -0300
Subject: [Linux-cluster] Re: Linux-cluster Digest, Preventing LVM from 
	concurrent access
In-Reply-To: <1251390865.5849.3.camel@quann>
References: <20090823160008.CA2B36194C4@hormel.redhat.com>
	<1251115598.5844.31.camel@quann>
	<2fc5f090908240813n45137862s876ec7ffa8c4a625@mail.gmail.com>
	<1251129846.5898.15.camel@quann>
	<2fc5f090908241227t1c709415t2475c3cf91126667@mail.gmail.com>
	<1251144303.5840.28.camel@quann>
	<2fc5f090908241348p73461b62u76835829be10a91d@mail.gmail.com>
	<1251196586.5835.6.camel@quann>
	<2fc5f090908270814q1e7a63a3g602a67fa1dd99655@mail.gmail.com>
	<1251390865.5849.3.camel@quann>
Message-ID: <2fc5f090908271347m35c1f500s6373a9868ab3408c@mail.gmail.com>

>
> Yes, you can bypass the exclusive flag by hand if you dont take it into
> account when activating the exclusive LV on a second node.
>
> While the exclusive LV is mounted in other node, the command "lvchange
> -aey XXX/YYY" should give you an error message. "lvchange -ay XXX/YYY"
> will bypass it.
>

Ok, but I was able to access that LV without changing any flag, while
both nodes were joined to the cluster.

But now, I don't know why, it's working as it should be, and I can't
use the LV to launch the VM while it is running on another node, under
rgmanager control. OK. That's what I wanted.

I could move the service through servers, but sometimes it has failed.
I have done this many times in sequence, from master to slave, and
vice-versa, just waiting a few seconds between each attempt, and it
seems that sometimes the LV is not released. Anyway, such a test would
not match a real situation, so I think this can be ignored.

( I have been changing my cluster.conf a lot, so, it's difficult to
say what was wrong, if it was. Forget about. =/ )

>> I runned a test, disconecting the heartbeat link, making one server to
>> be fenced, and the VM launched on the "winner" as expected. But when
>> the "loser" server came back, still without heartbeat link, it
>> launched the same VM again, and service appeared as running locally on
>> boths nodes. I guess lvm-cluster should avoid this, shouldn't ?
>>
>
> This should not happen. Have you set "exclusive=yes" into the resource
> definitions in cluster.conf? Can we have a copy of your current
> cluster.conf?

Yes, and I can say for sure that this "problem" (if I can call it like
this) keeps happening to me. Look, remember that this happens when I
bring back the server that was fenced, still without any connection
with the other, which VM was "moved" to.
 I can see that cman, clvmd, rgmanger and etc are running normaly, and
each server thinks that he is alone.

Let me try to make it easier to undestand, listing the steps I did:

1) Server A and B is up, VM service is started on cluster, and get
running on server A;
2) Cable of heartbeat interface is disconnected;
3) Server A is fenced by B, and server B starts the VM service;
4( Clabe is kept disconnected;
5) Server A is turned on again and after it has completed the
initialization, I can see that VM service is running, as well as
server B, both getting access to the same LV.

That's it, I hope you understand what I meant.

My cluster.conf with which I have runned those test is here [1].

[1] http://pastebin.com/m6a23734a



From alan.zg at gmail.com  Thu Aug 27 20:50:11 2009
From: alan.zg at gmail.com (Alan A)
Date: Thu, 27 Aug 2009 15:50:11 -0500
Subject: [Linux-cluster] Can not establish new cluster 5.3 with luci - 
	quorum error
In-Reply-To: <20090827195904.0bf14900@pc-jsosic.srce.hr>
References: <fac531740908261636j29165920odcdc169f28923f22@mail.gmail.com>
	<20090827102752.391915bc@nb-jsosic>
	<fac531740908270705rea433f4m94988f31d25e975e@mail.gmail.com>
	<20090827195904.0bf14900@pc-jsosic.srce.hr>
Message-ID: <fac531740908271350t3a1174c6y5a727bed02bd60fd@mail.gmail.com>

I spoke to our LAN guys and they are saying IGMPv2 should work. I forced the
cluster into version two (you can do the same by
http://kbase.redhat.com/faq/docs/DOC-3111).
So far no luck, trying version 1.

On Thu, Aug 27, 2009 at 12:59 PM, Jakov Sosic <jakov.sosic at srce.hr> wrote:

> On Thu, 27 Aug 2009 09:05:41 -0500
> Alan A <alan.zg at gmail.com> wrote:
>
> > Host file looks OK, pardon my post for my partial hostname editing
> > didn't quite work as expected since my mouse played a game on me.
> > We do run Cisco equipment - any pointers on what exactly should I
> > check. I am planning on opening ticket with RH as well.
>
> Try this one...
>
> http://www.openais.org/doku.php?id=faq:cisco_switches
>
>
> We had same situation before we configured Cisco to behave this way.
>
>
>
>
> --
> |    Jakov Sosic    |    ICQ: 28410271    |   PGP: 0x965CAE2D   |
> =================================================================
> | start fighting cancer -> http://www.worldcommunitygrid.org/   |
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>



-- 
Alan A.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090827/0e8c974d/attachment.htm>

From jakov.sosic at srce.hr  Thu Aug 27 22:02:08 2009
From: jakov.sosic at srce.hr (Jakov Sosic)
Date: Fri, 28 Aug 2009 00:02:08 +0200
Subject: [Linux-cluster] Can not establish new cluster 5.3 with luci - 
	quorum error
In-Reply-To: <fac531740908271350t3a1174c6y5a727bed02bd60fd@mail.gmail.com>
References: <fac531740908261636j29165920odcdc169f28923f22@mail.gmail.com>
	<20090827102752.391915bc@nb-jsosic>
	<fac531740908270705rea433f4m94988f31d25e975e@mail.gmail.com>
	<20090827195904.0bf14900@pc-jsosic.srce.hr>
	<fac531740908271350t3a1174c6y5a727bed02bd60fd@mail.gmail.com>
Message-ID: <20090828000208.63278e94@nb-jsosic>

On Thu, 27 Aug 2009 15:50:11 -0500
Alan A <alan.zg at gmail.com> wrote:

> I spoke to our LAN guys and they are saying IGMPv2 should work. I
> forced the cluster into version two (you can do the same by
> http://kbase.redhat.com/faq/docs/DOC-3111).
> So far no luck, trying version 1.

I spoke to my LAN guys too, but cluster simply cannot work unless
multicast is set to sparse-dense mode on the heartbeat vlan. Force your
guys to set it up like the openais page says - believe me... IGMP is
working, but not in a mode with multiple senders and multiple receivers
- which is one of the RHCS requests.

I went and FORCED my LAN guys to type commands from openais page in
front of me... After that everything worked out as a charm :D


-- 
|    Jakov Sosic    |    ICQ: 28410271    |   PGP: 0x965CAE2D   |
=================================================================
| start fighting cancer -> http://www.worldcommunitygrid.org/   |



From plebeuz at ig.com.br  Fri Aug 28 01:50:16 2009
From: plebeuz at ig.com.br (Daniel Viana Auler(Plebeuz))
Date: Thu, 27 Aug 2009 22:50:16 -0300
Subject: [Linux-cluster] GFS - cluster
Message-ID: <4A9737D8.3090009@ig.com.br>

Hello,
        I need to fence tha machines to use the same gfs?
Att,

Plebeuz



From jakov.sosic at srce.hr  Fri Aug 28 05:46:29 2009
From: jakov.sosic at srce.hr (Jakov Sosic)
Date: Fri, 28 Aug 2009 07:46:29 +0200
Subject: [Linux-cluster] GFS - cluster
In-Reply-To: <4A9737D8.3090009@ig.com.br>
References: <4A9737D8.3090009@ig.com.br>
Message-ID: <20090828074629.5c371237@nb-jsosic>

On Thu, 27 Aug 2009 22:50:16 -0300
"Daniel Viana Auler(Plebeuz)" <plebeuz at ig.com.br> wrote:

> Hello,
>         I need to fence tha machines to use the same gfs?

GFS is working on top of the cluster suite, so to use GFS you have to
have working cluster - which means fencing also. You can try to use it
without fencing although it is not recommended. 



-- 
|    Jakov Sosic    |    ICQ: 28410271    |   PGP: 0x965CAE2D   |
=================================================================
| start fighting cancer -> http://www.worldcommunitygrid.org/   |



From swhiteho at redhat.com  Fri Aug 28 07:46:30 2009
From: swhiteho at redhat.com (Steven Whitehouse)
Date: Fri, 28 Aug 2009 08:46:30 +0100
Subject: [Linux-cluster] RE: GFS2 subdirectory hang
In-Reply-To: <CD9C931A046A4A41876F7F5E134298B40206F54B@PTEEXB02.mtsallstream.com>
References: <20090827145704.5A56D61A5AB@hormel.redhat.com>
	<CD9C931A046A4A41876F7F5E134298B40206F54B@PTEEXB02.mtsallstream.com>
Message-ID: <1251445590.6052.55.camel@localhost.localdomain>

Hi,

On Thu, 2009-08-27 at 13:16 -0500, Johnson, Eric wrote:
> On Thu, 2009-08-27 at 09:25 -0500, Johnson, Eric wrote:
> >> I have a 32-bit RHEL 5.3 Cluster Suite setup of two nodes with GFS2
> file
> >> systems on FC attached SAN. I have run into this issue twice now,
> where
> >> attempts to access a certain directory within one of the GFS2 file
> >> systems never return. Other directories and paths within that file
> >> system work just fine.
> >> 
> >> The first time it happened, I had to crash the node to get it to
> release
> >> the FS, then unmount it on both nodes, fsck it, remount it, and it
> was
> >> fine. It has happened again (different path, different file system).
> A
> >> simple "ls" in the directory (which has maybe 20 files in it) leaves
> the
> >> process in an uninterruptible sleep state. I left it all night and it
> >> never returned.
> >> 
> >> I'm not sure what other info would be useful on this, but this is
> what I
> >> see from a gfs2_tool lockdump output for ls PID on that node:
> >> 
> >> G:  s:UN n:2/bf1df f:l t:SH d:EX/0 l:0 a:0 r:4
> >>  H: s:SH f:aW e:0 p:9938 [ls] gfs2_lookup+0x44/0x90 [gfs2]
> >              ^ The W flag indicates that this is waiting for a glock
> >
> >Currently the glock is in the UN (unlocked) state, and its trying to
> get
> >a SH (shared) lock. The next step in the investigation is to look for
> >the same glock number 2/bf1df on the other nodes, and see what is
> >holding that lock. This particular node will hang until the lock is
> >released on whichever other node is holding it.
> >
> >If there is nothing on any other node apparently holding that lock in
> >the glock dumps, then looking at dlm lock dumps would be the next step,
> >
> >Steve.
> 
> Thanks for the response, Steve. I found this reference to that lock on
> the other node:
> 
> G:  s:EX n:2/bf1df f:dy t:EX d:SH/0 l:0 a:0 r:4
>  I: n:1155192/782815 t:8 f:0x00000010
> 
> I'm having trouble finding documentation that describes what each of
> these fields are. There's no obvious process ID here, and all I can
> determine is that it's an exclusive lock.
> 
> Eric
> 
There is now some documentation in the upstream kernels under
linux/Documentation/filesystems/gfs2-glocks.txt and I hope to gradually
expand the docs available. There was more info in my Linux Symposium
paper too, but I see that they haven't appeared online yet.

The issue with the glock dump file is that it tells you how things are
now, and not how they got like that. In the case of the above example,
there is a glock in EX mode which has received a demote request and it
has been deferred (thats probably because the node had only just
received the lock when the demote request arrived) and for some reason
it looks like this has not been acted upon.

What should have happened is that upon expiry of the time interval, the
glock workqueue should have been scheduled to perform the demotion of
the glock. As it is, it looks like the glock is sitting there happily
caching the data and ignoring the pending demote.

You might be able to unstick this by unmounting & remounting or by
remounting read-only (and then you can remount back to rw right
afterwards) as that should be enough to release that lock and allow the
other node to make progress.

The real issue though, is why the workqueue didn't get run when it
should have done. Its a tricky issue to track down as there are a number
of possible code paths, and we need a record of which ones were taken.

This can be done in the upstream code with the new tracing code, but it
is not easily possible in RHEL5 kernels. If you are in a position where
you can run test kernels, then we can help track down the source of the
issue,

Steve.

> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster



From brem.belguebli at gmail.com  Fri Aug 28 14:10:14 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Fri, 28 Aug 2009 16:10:14 +0200
Subject: [Linux-cluster] Use alternate network interfaces for heartbeat in
	RHCS
Message-ID: <29ae894c0908280710w3f999b0as1438451bf5869a8e@mail.gmail.com>

Hi,

I'm trying to send cluster communications thru dedicated LAN interfaces (out
of band network) separate from the production interfaces but it doesn't seem
to be functional.

I have a 2 nodes cluster (RHEL 5.3) with 2 bonding interfaces each, one
bonding for production on a given Vlan and the other on a dedicated and
isolated LAN (out of band LAN).

According to some cman doc (that may be outdated), the way to do so is
to declare explicitely at each clusternode level  the multicast addr of the
cluster and the interface on which the coms are sent.

 <clusternode name="node1.mydomain" nodeid="1" votes="1">
                        <multicast addr="239.128.2.1" interface="bond1"/>
</clusternode>
<clusternode name="node2.mydomain" nodeid="2" votes="1">
                        <multicast addr="239.128.2.1" interface="bond1"/>
</clusternode>

and the cman multicast addr:

<cman expected_votes="3" two_node="0">
                <multicast addr="239.128.2.1"/>
        </cman>
(I have a qdiskd).

After restarting the cluster, traffic is still sent through bond0 which is
the production interface.(tcpdump -i bond0 host 239.128.2.1)

Any idea ?

Brem
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090828/3bef3154/attachment.htm>

From ccaulfie at redhat.com  Fri Aug 28 14:15:22 2009
From: ccaulfie at redhat.com (Christine Caulfield)
Date: Fri, 28 Aug 2009 15:15:22 +0100
Subject: [Linux-cluster] Use alternate network interfaces for heartbeat
	in	RHCS
In-Reply-To: <29ae894c0908280710w3f999b0as1438451bf5869a8e@mail.gmail.com>
References: <29ae894c0908280710w3f999b0as1438451bf5869a8e@mail.gmail.com>
Message-ID: <4A97E67A.1030506@redhat.com>

On 28/08/09 15:10, brem belguebli wrote:
> Hi,
> I'm trying to send cluster communications thru dedicated LAN interfaces
> (out of band network) separate from the production interfaces but it
> doesn't seem to be functional.
> I have a 2 nodes cluster (RHEL 5.3) with 2 bonding interfaces each, one
> bonding for production on a given Vlan and the other on a dedicated and
> isolated LAN (out of band LAN).
> According to some cman doc (that may be outdated), the way to do so is
> to declare explicitely at each clusternode level  the multicast addr of
> the cluster and the interface on which the coms are sent.
> <clusternode name="node1.mydomain" nodeid="1" votes="1">
> <multicast addr="239.128.2.1" interface="bond1"/>
> </clusternode>
> <clusternode name="node2.mydomain" nodeid="2" votes="1">
> <multicast addr="239.128.2.1" interface="bond1"/>
> </clusternode>
> and the cman multicast addr:
> <cman expected_votes="3" two_node="0">
> <multicast addr="239.128.2.1"/>
> </cman>
> (I have a qdiskd).
> After restarting the cluster, traffic is still sent through bond0 which
> is the production interface.(tcpdump -i bond0 host 239.128.2.1)


I don't know where that document is but it's rather out of date, as you 
suspected ;-)

The wiki has a up-to-date item on how to assigned different ethernet 
interfaces for cman:

http://sources.redhat.com/cluster/wiki/FAQ/CMAN#cman_heartbeat_nic

Chrissie



From brem.belguebli at gmail.com  Fri Aug 28 14:24:28 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Fri, 28 Aug 2009 16:24:28 +0200
Subject: [Linux-cluster] Use alternate network interfaces for heartbeat in
	RHCS
In-Reply-To: <4A97E67A.1030506@redhat.com>
References: <29ae894c0908280710w3f999b0as1438451bf5869a8e@mail.gmail.com>
	<4A97E67A.1030506@redhat.com>
Message-ID: <29ae894c0908280724p4fc8cbe4g6943a3138f278c1b@mail.gmail.com>

Hi Chrissie,

Are you pointing me to the paragraph "what's the right way to ....eth0 ?"

I've tried this at first adding adding a suffix to the interfaces but
nothing happened. Suffix -p may be hardcoded (I've used -hb)

Here's an outputof my /etc/hosts (identical on both nodes):

10.146.15.184   node1 node1.mydomain
10.146.15.175   node2 node2.mydomain
192.168.84.50   node1-hb
192.168.84.51   node2-hb

Still using bond0 ....

Brem



2009/8/28, Christine Caulfield <ccaulfie at redhat.com>:
>
> On 28/08/09 15:10, brem belguebli wrote:
>
>> Hi,
>> I'm trying to send cluster communications thru dedicated LAN interfaces
>> (out of band network) separate from the production interfaces but it
>> doesn't seem to be functional.
>> I have a 2 nodes cluster (RHEL 5.3) with 2 bonding interfaces each, one
>> bonding for production on a given Vlan and the other on a dedicated and
>> isolated LAN (out of band LAN).
>> According to some cman doc (that may be outdated), the way to do so is
>> to declare explicitely at each clusternode level  the multicast addr of
>> the cluster and the interface on which the coms are sent.
>> <clusternode name="node1.mydomain" nodeid="1" votes="1">
>> <multicast addr="239.128.2.1" interface="bond1"/>
>> </clusternode>
>> <clusternode name="node2.mydomain" nodeid="2" votes="1">
>> <multicast addr="239.128.2.1" interface="bond1"/>
>> </clusternode>
>> and the cman multicast addr:
>> <cman expected_votes="3" two_node="0">
>> <multicast addr="239.128.2.1"/>
>> </cman>
>> (I have a qdiskd).
>> After restarting the cluster, traffic is still sent through bond0 which
>> is the production interface.(tcpdump -i bond0 host 239.128.2.1)
>>
>
>
> I don't know where that document is but it's rather out of date, as you
> suspected ;-)
>
> The wiki has a up-to-date item on how to assigned different ethernet
> interfaces for cman:
>
> http://sources.redhat.com/cluster/wiki/FAQ/CMAN#cman_heartbeat_nic
>
> Chrissie
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090828/a0fb0e63/attachment.htm>

From ccaulfie at redhat.com  Fri Aug 28 14:53:40 2009
From: ccaulfie at redhat.com (Christine Caulfield)
Date: Fri, 28 Aug 2009 15:53:40 +0100
Subject: [Linux-cluster] Use alternate network interfaces for heartbeat
	in	RHCS
In-Reply-To: <29ae894c0908280724p4fc8cbe4g6943a3138f278c1b@mail.gmail.com>
References: <29ae894c0908280710w3f999b0as1438451bf5869a8e@mail.gmail.com>	<4A97E67A.1030506@redhat.com>
	<29ae894c0908280724p4fc8cbe4g6943a3138f278c1b@mail.gmail.com>
Message-ID: <4A97EF74.6090904@redhat.com>

On 28/08/09 15:24, brem belguebli wrote:
> Hi Chrissie,
> Are you pointing me to the paragraph "what's the right way to ....eth0 ?"
> I've tried this at first adding adding a suffix to the interfaces but
> nothing happened. Suffix -p may be hardcoded (I've used -hb)
> Here's an outputof my /etc/hosts (identical on both nodes):
> 10.146.15.184   node1 node1.mydomain
> 10.146.15.175   node2 node2.mydomain
> 192.168.84.50   node1-hb
> 192.168.84.51   node2-hb
> Still using bond0 ....
> Brem
>


The suffix isn't hard-coded or anything to do with cman really, it's 
just a way of distinguishing interfaces.

You need to edit cluster.conf to tell it to use the different name.

If you get despertate then you can always put the IP address in 
cluster.conf, but the output from cman_tool nodes doesn't look as nice!

Chrissie



From ccaulfie at redhat.com  Fri Aug 28 14:58:24 2009
From: ccaulfie at redhat.com (Christine Caulfield)
Date: Fri, 28 Aug 2009 15:58:24 +0100
Subject: [Linux-cluster] Use alternate network interfaces for heartbeat
	in	RHCS
In-Reply-To: <4A97EF74.6090904@redhat.com>
References: <29ae894c0908280710w3f999b0as1438451bf5869a8e@mail.gmail.com>	<4A97E67A.1030506@redhat.com>	<29ae894c0908280724p4fc8cbe4g6943a3138f278c1b@mail.gmail.com>
	<4A97EF74.6090904@redhat.com>
Message-ID: <4A97F090.9080508@redhat.com>

On 28/08/09 15:53, Christine Caulfield wrote:
> On 28/08/09 15:24, brem belguebli wrote:
>> Hi Chrissie,
>> Are you pointing me to the paragraph "what's the right way to ....eth0 ?"
>> I've tried this at first adding adding a suffix to the interfaces but
>> nothing happened. Suffix -p may be hardcoded (I've used -hb)
>> Here's an outputof my /etc/hosts (identical on both nodes):
>> 10.146.15.184 node1 node1.mydomain
>> 10.146.15.175 node2 node2.mydomain
>> 192.168.84.50 node1-hb
>> 192.168.84.51 node2-hb
>> Still using bond0 ....
>> Brem
>>
>
>
> The suffix isn't hard-coded or anything to do with cman really, it's
> just a way of distinguishing interfaces.
>
> You need to edit cluster.conf to tell it to use the different name.
>
> If you get despertate then you can always put the IP address in
> cluster.conf, but the output from cman_tool nodes doesn't look as nice!

I haven't read that article in detail before but you're right, it makes 
no mention of changing cluster.conf!

I've fixed that now, thank you.

Chrissie



From brem.belguebli at gmail.com  Fri Aug 28 17:38:53 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Fri, 28 Aug 2009 19:38:53 +0200
Subject: [Linux-cluster] Use alternate network interfaces for heartbeat in
	RHCS
In-Reply-To: <4A97EF74.6090904@redhat.com>
References: <29ae894c0908280710w3f999b0as1438451bf5869a8e@mail.gmail.com>
	<4A97E67A.1030506@redhat.com>
	<29ae894c0908280724p4fc8cbe4g6943a3138f278c1b@mail.gmail.com>
	<4A97EF74.6090904@redhat.com>
Message-ID: <29ae894c0908281038i20f408f7gf14483ad6f73ca5e@mail.gmail.com>

Hi
the clusternodes defined in cluster.conf are :

node1.mydomain
node2.mydomain

which correpond to the bond0 interfaces on both nodes.

I expect to use node1-hb and node2-hb as heartbeat interfaces. (bond1)

I may have misunderstood something, but are you telling me that I have to
use the nodeX-hb as clusternodes in cluster.conf ?

Brem

2009/8/28 Christine Caulfield <ccaulfie at redhat.com>

> On 28/08/09 15:24, brem belguebli wrote:
>
>> Hi Chrissie,
>> Are you pointing me to the paragraph "what's the right way to ....eth0 ?"
>> I've tried this at first adding adding a suffix to the interfaces but
>> nothing happened. Suffix -p may be hardcoded (I've used -hb)
>> Here's an outputof my /etc/hosts (identical on both nodes):
>> 10.146.15.184   node1 node1.mydomain
>> 10.146.15.175   node2 node2.mydomain
>> 192.168.84.50   node1-hb
>> 192.168.84.51   node2-hb
>> Still using bond0 ....
>> Brem
>>
>>
>
> The suffix isn't hard-coded or anything to do with cman really, it's just a
> way of distinguishing interfaces.
>
> You need to edit cluster.conf to tell it to use the different name.
>
> If you get despertate then you can always put the IP address in
> cluster.conf, but the output from cman_tool nodes doesn't look as nice!
>
>
> Chrissie
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090828/8428e6d0/attachment.htm>

From brem.belguebli at gmail.com  Fri Aug 28 17:45:05 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Fri, 28 Aug 2009 19:45:05 +0200
Subject: [Linux-cluster] Use alternate network interfaces for heartbeat in
	RHCS
In-Reply-To: <4A97F090.9080508@redhat.com>
References: <29ae894c0908280710w3f999b0as1438451bf5869a8e@mail.gmail.com>
	<4A97E67A.1030506@redhat.com>
	<29ae894c0908280724p4fc8cbe4g6943a3138f278c1b@mail.gmail.com>
	<4A97EF74.6090904@redhat.com> <4A97F090.9080508@redhat.com>
Message-ID: <29ae894c0908281045l6c93c7dbo2d0e4f27c5bab14e@mail.gmail.com>

Ok,

It answers my last question.

I have been confused by some mention somewhere in a post or doc saying that
cman has a built-in algorithm to determine, just by adding entries in
/etc/hosts, the right interfaces to use .

Brem



2009/8/28 Christine Caulfield <ccaulfie at redhat.com>

> On 28/08/09 15:53, Christine Caulfield wrote:
>
>> On 28/08/09 15:24, brem belguebli wrote:
>>
>>> Hi Chrissie,
>>> Are you pointing me to the paragraph "what's the right way to ....eth0 ?"
>>> I've tried this at first adding adding a suffix to the interfaces but
>>> nothing happened. Suffix -p may be hardcoded (I've used -hb)
>>> Here's an outputof my /etc/hosts (identical on both nodes):
>>> 10.146.15.184 node1 node1.mydomain
>>> 10.146.15.175 node2 node2.mydomain
>>> 192.168.84.50 node1-hb
>>> 192.168.84.51 node2-hb
>>> Still using bond0 ....
>>> Brem
>>>
>>>
>>
>> The suffix isn't hard-coded or anything to do with cman really, it's
>> just a way of distinguishing interfaces.
>>
>> You need to edit cluster.conf to tell it to use the different name.
>>
>> If you get despertate then you can always put the IP address in
>> cluster.conf, but the output from cman_tool nodes doesn't look as nice!
>>
>
> I haven't read that article in detail before but you're right, it makes no
> mention of changing cluster.conf!
>
> I've fixed that now, thank you.
>
>
> Chrissie
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090828/e6699607/attachment.htm>

From fdinitto at redhat.com  Sat Aug 29 06:57:34 2009
From: fdinitto at redhat.com (Fabio M. Di Nitto)
Date: Sat, 29 Aug 2009 08:57:34 +0200
Subject: [Linux-cluster] cluster.conf in another place ?
In-Reply-To: <4A9692E5.9030607@bull.net>
References: <4A9692E5.9030607@bull.net>
Message-ID: <1251529054.339.2.camel@cerberus.int.fabbione.net>

On Thu, 2009-08-27 at 16:06 +0200, Alain.Moulle wrote:
> Hi Fabio,
> cman-3.0.0-15.rc1.fc11.x86_64
> rgmanager-3.0.0-15.rc1.fc11.x86_64

Please update to the new packages available in fedora 11 updates as
first thing. The ones you are using are release candidates. Updates have
the final versions.

You can use alternative config files setting the correct envvar:

export COROSYNC_CLUSTER_CONFIG_FILE=/path/to/your/config/file in either
/etc/sysconfig/cluster or /etc/sysconfig/cman.

Cheers
Fabio



From asanabria at linuxdynasty.org  Sat Aug 29 23:36:07 2009
From: asanabria at linuxdynasty.org (Allen Sanabria)
Date: Sat, 29 Aug 2009 19:36:07 -0400
Subject: [Linux-cluster] Cluster Monitoring
Message-ID: <4A99BB67.4000608@linuxdynasty.org>

Not sure if this went through the 1st time...

For those of you who need to monitor your RedHat Clusters. I created a 
new Check command that works under Zenoss or Nagios, that will monitor 
your cluster through snmp.

You will need pysnmp and pyasn1 python modules for this command to work, 
but I felt it was better then using a check by ssh command..

Here is the link if you want to try it out...
http://www.linuxdynasty.org/howto-monitor-redhat-cluster-using-snmp-and-python.html 


I also created the equivalent of a clustat command using python and 
snmp, so you do not have to be logged into a node in the cluster to run 
it...


Code:
[root at gfs1 python]# python rh_cluster_check.py -d gfs3 -c public -t cluster
WARNING, MyCluster is Quorate and Some services not running

[root at gfs1 python]# python rh_cluster_check.py -d gfs1 -c public -t 
service -s CIM
OK, CIM is running on gfs1

[root at gfs1 python]# python rh_cluster_check.py -d gfs1 -c public -t node 
-n "gfs3"
OK, gfs3 is Participating in cluster



clustat_snmp.py -d gfs1 -c public
Cluster Status for MyCluster @ Fri Aug 28 11:59:40 2009
Member Status: Quorate
Total Nodes:  3
Total Votes:  5
Votes Needed For Quorum:  3

Member Name                               Status
------ ----                               ------
gfs1                                      Participating in cluster
gfs2                                      Participating in cluster
gfs3                                      Participating in cluster

Service Name                              Owner                         
State                   Start Up Mode
------- ----                              -----                         
-----                   -------------
Pirahna                                                                 
stopped                 automatic
CIM                                       gfs1                          
running                 automatic


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090829/84496ea7/attachment.htm>

From edsonmarquezani at gmail.com  Sun Aug 30 03:23:11 2009
From: edsonmarquezani at gmail.com (Edson Marquezani Filho)
Date: Sun, 30 Aug 2009 00:23:11 -0300
Subject: [Linux-cluster] ordered/ restricted and clusvcadm -F
In-Reply-To: <4A92B3D6.6060903@bull.net>
References: <4A92B3D6.6060903@bull.net>
Message-ID: <2fc5f090908292023k4e26fa5p58123275679a83a9@mail.gmail.com>

On Mon, Aug 24, 2009 at 12:37, Alain.Moulle<Alain.Moulle at bull.net> wrote:
> Hi,
> given a failover domain like this one :
> <failoverdomain name="FailoverDomain-1" ordered="???" restricted="???">
> ???????? <failoverdomainnode name="node1" priority="1"/>
> ???????? <failoverdomainnode name="node2" priority="2"/>
> ???????? <failoverdomainnode name="node3" priority="3"/>
> </failoverdomain>
> and a service serv1 configured on this failoverdomain.
>
> I would like that the command:
> ?clusvcadm -e serv1
> executed on whatever node of the three nodes starts the serv1 on the
> prefered node? node1 (priority1)

I wanted that two, but it seems like there's no way do that exactly
like this. See below.

> And? in the case serv1 is started on node2 (or node3) (after a failover
> for example) and that node1 transitions online , I don't want an
> automatic migration of serv1 onto node1.

This can be done with nofailback="1" on <failoverdomain> tag. I want
things working like this too, I use that option, and it works.

> My understanding was to set ordered=1 and restricted=0? but
> it seems that clusvcadm -e serv1 starts it on the node where clusvcadm is
> executed, even if it is node the "priority 1" node (node1).

Yes, acording to this[1], clusvcadm will always start the service on
the node where the command was issued. The only way to make it to run
on a specific node is setting it with -m option. ( FAQ says -n, but as
far as I have seen, it's -m that specifies the cluster member, and
actualy, there is no -n option at all.)

> Option -F of clusvcadm seemed to give a response to my question, but
> it is announced as "deprecated" ...

That option seemed not work to me too.

> Thanks for your response.

I'm a very begginer in RHCS, but I have been playing around with
rgmanager and that's what I could figure out during these last days.

I hope I could help you.

[1] http://sources.redhat.com/cluster/wiki/FAQ/RGManager#rgm_svcstart



From rmicmirregs at gmail.com  Sun Aug 30 14:21:14 2009
From: rmicmirregs at gmail.com (Rafael =?ISO-8859-1?Q?Mic=F3?= Miranda)
Date: Sun, 30 Aug 2009 16:21:14 +0200
Subject: [Linux-cluster] Re: Linux-cluster Digest, Preventing LVM from 
	concurrent access
In-Reply-To: <2fc5f090908271347m35c1f500s6373a9868ab3408c@mail.gmail.com>
References: <20090823160008.CA2B36194C4@hormel.redhat.com>
	<1251115598.5844.31.camel@quann>
	<2fc5f090908240813n45137862s876ec7ffa8c4a625@mail.gmail.com>
	<1251129846.5898.15.camel@quann>
	<2fc5f090908241227t1c709415t2475c3cf91126667@mail.gmail.com>
	<1251144303.5840.28.camel@quann>
	<2fc5f090908241348p73461b62u76835829be10a91d@mail.gmail.com>
	<1251196586.5835.6.camel@quann>
	<2fc5f090908270814q1e7a63a3g602a67fa1dd99655@mail.gmail.com>
	<1251390865.5849.3.camel@quann>
	<2fc5f090908271347m35c1f500s6373a9868ab3408c@mail.gmail.com>
Message-ID: <1251642074.6488.16.camel@mecatol>

Hi Edson,

El jue, 27-08-2009 a las 17:47 -0300, Edson Marquezani Filho escribi?:
> >
> > Yes, you can bypass the exclusive flag by hand if you dont take it into
> > account when activating the exclusive LV on a second node.
> >
> > While the exclusive LV is mounted in other node, the command "lvchange
> > -aey XXX/YYY" should give you an error message. "lvchange -ay XXX/YYY"
> > will bypass it.
> >
> 
> Ok, but I was able to access that LV without changing any flag, while
> both nodes were joined to the cluster.
> 

This should not happen. You should only have the Logical Volume on one
node if you use the exclusive flag correctly.

> But now, I don't know why, it's working as it should be, and I can't
> use the LV to launch the VM while it is running on another node, under
> rgmanager control. OK. That's what I wanted.

Ok

> 
> I could move the service through servers, but sometimes it has failed.
> I have done this many times in sequence, from master to slave, and
> vice-versa, just waiting a few seconds between each attempt, and it
> seems that sometimes the LV is not released. Anyway, such a test would
> not match a real situation, so I think this can be ignored.
> 

Maybe this is caused because your VM script. Maybe the lvm-cluster
resource tries to deport a device that is in-use, so it will fail. Did
you check your logs for any error messages? I built the resource very
"verbose".

> ( I have been changing my cluster.conf a lot, so, it's difficult to
> say what was wrong, if it was. Forget about. =/ )
> 
> >> I runned a test, disconecting the heartbeat link, making one server to
> >> be fenced, and the VM launched on the "winner" as expected. But when
> >> the "loser" server came back, still without heartbeat link, it
> >> launched the same VM again, and service appeared as running locally on
> >> boths nodes. I guess lvm-cluster should avoid this, shouldn't ?
> >>
> >
> > This should not happen. Have you set "exclusive=yes" into the resource
> > definitions in cluster.conf? Can we have a copy of your current
> > cluster.conf?
> 
> Yes, and I can say for sure that this "problem" (if I can call it like
> this) keeps happening to me. Look, remember that this happens when I
> bring back the server that was fenced, still without any connection
> with the other, which VM was "moved" to.
>  I can see that cman, clvmd, rgmanger and etc are running normaly, and
> each server thinks that he is alone.

> Let me try to make it easier to undestand, listing the steps I did:
> 
> 1) Server A and B is up, VM service is started on cluster, and get
> running on server A;
> 2) Cable of heartbeat interface is disconnected;
> 3) Server A is fenced by B, and server B starts the VM service;
> 4( Clabe is kept disconnected;
> 5) Server A is turned on again and after it has completed the
> initialization, I can see that VM service is running, as well as
> server B, both getting access to the same LV.
> 
> That's it, I hope you understand what I meant.

I'm pretty sure this is caused by an ineffective fencing configuration
and wrong configuration of LVM. What i think is happening to you,
without any info of your logs, is:

1.- Primary node gets fenced out. VM is moved to alive secondary node.
2.- Fenced node boots again. 
3.- While comms are interrupted, cluster services are started again on
restarted node. This services should be, at least, CMAN, CLVMD and
RGMANAGER.
4.- Restarted node thinks it is alone because comms are interrupted. It
applies fencing on secondary node.
5a.- If fencing is OK, secondary node gets shut down, and primary node
gets the VM again. 
5b.- If fencing is not working properly and primary node believes the
secondary node has been fenced, it will try to get control of the LVM
and the VM. The exclusive flag protection should forbade this.

Check your fencing devices. Maybe they always return "success" code
error and its not true. 

On the other hand, about the LVM volume, please check this in order to
establish your service:

https://www.redhat.com/archives/linux-cluster/2009-July/msg00259.html


> 
> My cluster.conf with which I have runned those test is here [1].
> 
> [1] http://pastebin.com/m6a23734a
> 
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster

Cheers,

Rafael
-- 
Rafael Mic? Miranda



From dwight.hubbard at efausol.com  Mon Aug 31 03:35:01 2009
From: dwight.hubbard at efausol.com (Dwight Hubbard)
Date: Sun, 30 Aug 2009 20:35:01 -0700
Subject: [Linux-cluster] NFS client failover
In-Reply-To: <4A94F87C.7030708@lists.grepular.com>
References: <4A93C8CE.7010202@lists.grepular.com>
	<4A943E31.3080209@lists.grepular.com>
	<4A94F87C.7030708@lists.grepular.com>
Message-ID: <200908302035.01848.dwight.hubbard@efausol.com>

With clustered NFS the writes should be synchronous so they get committed to 
stable storage before the client gets confirmation the write completed.  As a 
result the write is on disk so there is no corruption after a reboot or 
failover.  

On Wednesday 26 August 2009 01:55:24 am Mike Cardwell wrote:
> On 25/08/2009 20:40, Mike Cardwell wrote:
> > I figured that failover would happen more smoothly if the client was
> > aware of and in control of what was going on. If the IP suddenly moves
> > to another NFS server I don't know how the NFS client will cope with
> > that.
>
> Well, it seems to cope quite well. The nfs mount "hangs" for a few
> seconds whilst the IP moves from one server to another (unavoidable
> obviously), but it then picks up from where it was. I suspect there will
> be file corruption issues with files that are partially written when the
> failover happens, but I guess that can't be avoided without a client
> side solution.

-- 
Dwight Hubbard
Owner Effective Automation Solutions
Website: http://effectiveautomationsolutions.com
Email: dwight at dwighthubbard.com
Phone: (503) 941-0327
Redhat Certified Engineer - RHCE #804007137224095
VMware Certified Professional - VCP #18529



From brem.belguebli at gmail.com  Mon Aug 31 07:56:29 2009
From: brem.belguebli at gmail.com (brem belguebli)
Date: Mon, 31 Aug 2009 09:56:29 +0200
Subject: [Linux-cluster] Use alternate network interfaces for heartbeat in
	RHCS
In-Reply-To: <29ae894c0908281045l6c93c7dbo2d0e4f27c5bab14e@mail.gmail.com>
References: <29ae894c0908280710w3f999b0as1438451bf5869a8e@mail.gmail.com>
	<4A97E67A.1030506@redhat.com>
	<29ae894c0908280724p4fc8cbe4g6943a3138f278c1b@mail.gmail.com>
	<4A97EF74.6090904@redhat.com> <4A97F090.9080508@redhat.com>
	<29ae894c0908281045l6c93c7dbo2d0e4f27c5bab14e@mail.gmail.com>
Message-ID: <29ae894c0908310056u68c1272dud8babe5ac3542f9d@mail.gmail.com>

Hi,

I was wondering if there is a way with cman to configure 2 heartbeat
channels (let's say my prod bond0 and my outband bond1) as it seems to be
possible with openais and their redundant ring interfaces configuration.

Brem


2009/8/28, brem belguebli <brem.belguebli at gmail.com>:
>
> Ok,
>
> It answers my last question.
>
> I have been confused by some mention somewhere in a post or doc saying that
> cman has a built-in algorithm to determine, just by adding entries in
> /etc/hosts, the right interfaces to use .
>
> Brem
>
>
>
> 2009/8/28 Christine Caulfield <ccaulfie at redhat.com>
>
>> On 28/08/09 15:53, Christine Caulfield wrote:
>>
>>> On 28/08/09 15:24, brem belguebli wrote:
>>>
>>>> Hi Chrissie,
>>>> Are you pointing me to the paragraph "what's the right way to ....eth0
>>>> ?"
>>>> I've tried this at first adding adding a suffix to the interfaces but
>>>> nothing happened. Suffix -p may be hardcoded (I've used -hb)
>>>> Here's an outputof my /etc/hosts (identical on both nodes):
>>>> 10.146.15.184 node1 node1.mydomain
>>>> 10.146.15.175 node2 node2.mydomain
>>>> 192.168.84.50 node1-hb
>>>> 192.168.84.51 node2-hb
>>>> Still using bond0 ....
>>>> Brem
>>>>
>>>>
>>>
>>> The suffix isn't hard-coded or anything to do with cman really, it's
>>> just a way of distinguishing interfaces.
>>>
>>> You need to edit cluster.conf to tell it to use the different name.
>>>
>>> If you get despertate then you can always put the IP address in
>>> cluster.conf, but the output from cman_tool nodes doesn't look as nice!
>>>
>>
>> I haven't read that article in detail before but you're right, it makes no
>> mention of changing cluster.conf!
>>
>> I've fixed that now, thank you.
>>
>>
>> Chrissie
>>
>> --
>> Linux-cluster mailing list
>> Linux-cluster at redhat.com
>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090831/d4cf5337/attachment.htm>

From ntadmin at fi.upm.es  Mon Aug 31 12:48:27 2009
From: ntadmin at fi.upm.es (Administrador de NT)
Date: Mon, 31 Aug 2009 14:48:27 +0200
Subject: [Linux-cluster] Can not establish new cluster 5.3 with luci -
	quorum error
In-Reply-To: <20090828000208.63278e94@nb-jsosic>
References: <fac531740908261636j29165920odcdc169f28923f22@mail.gmail.com>	<20090827102752.391915bc@nb-jsosic>	<fac531740908270705rea433f4m94988f31d25e975e@mail.gmail.com>	<20090827195904.0bf14900@pc-jsosic.srce.hr>	<fac531740908271350t3a1174c6y5a727bed02bd60fd@mail.gmail.com>
	<20090828000208.63278e94@nb-jsosic>
Message-ID: <4A9BC69B.2000603@fi.upm.es>

Jakov Sosic escribi?:
> On Thu, 27 Aug 2009 15:50:11 -0500
> Alan A <alan.zg at gmail.com> wrote:
>
>   
>> I spoke to our LAN guys and they are saying IGMPv2 should work. I
>> forced the cluster into version two (you can do the same by
>> http://kbase.redhat.com/faq/docs/DOC-3111).
>> So far no luck, trying version 1.
>>     
>
> I spoke to my LAN guys too, but cluster simply cannot work unless
> multicast is set to sparse-dense mode on the heartbeat vlan. Force your
> guys to set it up like the openais page says - believe me... IGMP is
> working, but not in a mode with multiple senders and multiple receivers
> - which is one of the RHCS requests.
>
> I went and FORCED my LAN guys to type commands from openais page in
> front of me... After that everything worked out as a charm :D
>
>
>   
In our case, we have 3Com switches which only understand IGMPv1. Due to 
version 1 requirements for other applications, I had to use a 
administrative multicast address (224.0.0.X) and I didn't change any 
kernel parameters.

Bye



From tageorgiou at gmail.com  Mon Aug 31 13:12:12 2009
From: tageorgiou at gmail.com (Thomas Georgiou)
Date: Mon, 31 Aug 2009 09:12:12 -0400
Subject: [Linux-cluster] Problem with Pacemaker and Corosync
Message-ID: <6e4c20e70908310612o120933cema2609513f13be78c@mail.gmail.com>

Hi,

I have installed Pacemaker 1.0.5, Corosync 1.0.0, and Openais 1.0.1
from source according to the Clusterlabs docs.  However, when I go to
start corosync/pacemaker, I get error messages pertaining to lrm and
cibadmin -Q hangs and complains that the remote node is not available.
 Attached is the corosync log.

Any ideas?

Thomas Georgiou
-------------- next part --------------
A non-text attachment was scrubbed...
Name: corosync.log
Type: application/octet-stream
Size: 54947 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090831/88426927/attachment.obj>

From alan.zg at gmail.com  Mon Aug 31 14:47:12 2009
From: alan.zg at gmail.com (Alan A)
Date: Mon, 31 Aug 2009 09:47:12 -0500
Subject: [Linux-cluster] Can not establish new cluster 5.3 with luci - 
	quorum error
In-Reply-To: <4A9BC69B.2000603@fi.upm.es>
References: <fac531740908261636j29165920odcdc169f28923f22@mail.gmail.com>
	<20090827102752.391915bc@nb-jsosic>
	<fac531740908270705rea433f4m94988f31d25e975e@mail.gmail.com>
	<20090827195904.0bf14900@pc-jsosic.srce.hr>
	<fac531740908271350t3a1174c6y5a727bed02bd60fd@mail.gmail.com>
	<20090828000208.63278e94@nb-jsosic> <4A9BC69B.2000603@fi.upm.es>
Message-ID: <fac531740908310747p4556c0e3hed06dfbb661eeb38@mail.gmail.com>

 We are working on setting up VLAN with IGMP. It might take day or two to
setup.

On Mon, Aug 31, 2009 at 7:48 AM, Administrador de NT <ntadmin at fi.upm.es>wrote:

> Jakov Sosic escribi?:
>
>> On Thu, 27 Aug 2009 15:50:11 -0500
>> Alan A <alan.zg at gmail.com> wrote:
>>
>>
>>
>>> I spoke to our LAN guys and they are saying IGMPv2 should work. I
>>> forced the cluster into version two (you can do the same by
>>> http://kbase.redhat.com/faq/docs/DOC-3111).
>>> So far no luck, trying version 1.
>>>
>>>
>>
>> I spoke to my LAN guys too, but cluster simply cannot work unless
>> multicast is set to sparse-dense mode on the heartbeat vlan. Force your
>> guys to set it up like the openais page says - believe me... IGMP is
>> working, but not in a mode with multiple senders and multiple receivers
>> - which is one of the RHCS requests.
>>
>> I went and FORCED my LAN guys to type commands from openais page in
>> front of me... After that everything worked out as a charm :D
>>
>>
>>
>>
> In our case, we have 3Com switches which only understand IGMPv1. Due to
> version 1 requirements for other applications, I had to use a administrative
> multicast address (224.0.0.X) and I didn't change any kernel parameters.
>
> Bye
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>



-- 
Alan A.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090831/c612a6f3/attachment.htm>

From jakov.sosic at srce.hr  Mon Aug 31 20:30:53 2009
From: jakov.sosic at srce.hr (Jakov Sosic)
Date: Mon, 31 Aug 2009 22:30:53 +0200
Subject: [Linux-cluster] How to disable node?
Message-ID: <20090831223053.5461ad55@nb-jsosic>

Hi.

I have a problem that one of my 3 nodes is sent to RMA to hardware
dealer. I want to know is it possible to disable that one node, because
clurgmgrd is not starting any services because cman is unable to fence
the missing node...

Aug 31 22:02:14 node03 fenced[5082]: fence "node02" failed
Aug 31 22:02:19 node03 fenced[5082]: fencing node "node02"
Aug 31 22:04:38 node03 fenced[5082]: agent "fence_ipmilan" reports:
Rebooting machine @ IPMI:x.y.z.w...ipmilan: Failed to connect
after 30 seconds Failed 





-- 
|    Jakov Sosic    |    ICQ: 28410271    |   PGP: 0x965CAE2D   |
=================================================================
| start fighting cancer -> http://www.worldcommunitygrid.org/   |



From alan.zg at gmail.com  Mon Aug 31 20:30:13 2009
From: alan.zg at gmail.com (Alan A)
Date: Mon, 31 Aug 2009 15:30:13 -0500
Subject: [Linux-cluster] How to disable node?
In-Reply-To: <20090831223053.5461ad55@nb-jsosic>
References: <20090831223053.5461ad55@nb-jsosic>
Message-ID: <fac531740908311330q6827548dlf1966605f3ff4e10@mail.gmail.com>

You have to give your cluster 2 node setting on remaining two nodes.

How can I define a two-node cluster if a majority is needed to reach quorum?

We had to allow two-node clusters, so we made a special exception to the
quorum rules. There is a special setting "two_node" in the /etc/cluster.conf
file that looks like this:

<cman expected_votes="1" two_node="1"/>

This will allow one node to be considered enough to establish a quorum. Note
that if you configure a quorum disk/partition, you don't want two_node="1".

http://sources.redhat.com/cluster/wiki/FAQ/CMAN#two_node

On Mon, Aug 31, 2009 at 3:30 PM, Jakov Sosic <jakov.sosic at srce.hr> wrote:

> Hi.
>
> I have a problem that one of my 3 nodes is sent to RMA to hardware
> dealer. I want to know is it possible to disable that one node, because
> clurgmgrd is not starting any services because cman is unable to fence
> the missing node...
>
> Aug 31 22:02:14 node03 fenced[5082]: fence "node02" failed
> Aug 31 22:02:19 node03 fenced[5082]: fencing node "node02"
> Aug 31 22:04:38 node03 fenced[5082]: agent "fence_ipmilan" reports:
> Rebooting machine @ IPMI:x.y.z.w...ipmilan: Failed to connect
> after 30 seconds Failed
>
>
>
>
>
> --
> |    Jakov Sosic    |    ICQ: 28410271    |   PGP: 0x965CAE2D   |
> =================================================================
> | start fighting cancer -> http://www.worldcommunitygrid.org/   |
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>



-- 
Alan A.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090831/56113ce4/attachment.htm>

From jruemker at redhat.com  Mon Aug 31 20:49:06 2009
From: jruemker at redhat.com (John Ruemker)
Date: Mon, 31 Aug 2009 16:49:06 -0400
Subject: [Linux-cluster] How to disable node?
In-Reply-To: <20090831223053.5461ad55@nb-jsosic>
References: <20090831223053.5461ad55@nb-jsosic>
Message-ID: <4A9C3742.3000905@redhat.com>

On 08/31/2009 04:30 PM, Jakov Sosic wrote:
> Hi.
>
> I have a problem that one of my 3 nodes is sent to RMA to hardware
> dealer. I want to know is it possible to disable that one node, because
> clurgmgrd is not starting any services because cman is unable to fence
> the missing node...
>
> Aug 31 22:02:14 node03 fenced[5082]: fence "node02" failed
> Aug 31 22:02:19 node03 fenced[5082]: fencing node "node02"
> Aug 31 22:04:38 node03 fenced[5082]: agent "fence_ipmilan" reports:
> Rebooting machine @ IPMI:x.y.z.w...ipmilan: Failed to connect
> after 30 seconds Failed
>

The best course of action to take would be to remove that missing node 
from your cluster configuration using conga, system-config-cluster, or 
by hand editing /etc/cluster/cluster.conf.  As long as it exists in the 
configuration then the other nodes will expect it to join the cluster, 
and they will attempt to fence it when they try to join the cluster and 
see it is not present.  When you remove it from the configuration by 
hand, you will also want to set two_node="1 and expected_votes="1" in 
the cman tag like so

    <cman two_node="1" expected_votes="1"/>

If you are using one of the graphical tools these options should be 
added for you.  They ensure that if one of your remaining two nodes 
dies, the one that is left can maintain quorum by itself.

After making these changes and propagating the new configuration to both 
nodes, you will want to restart the cluster service on both nodes or 
just reboot both nodes.

-John



From jakov.sosic at srce.hr  Mon Aug 31 20:59:18 2009
From: jakov.sosic at srce.hr (Jakov Sosic)
Date: Mon, 31 Aug 2009 22:59:18 +0200
Subject: [Linux-cluster] How to disable node?
In-Reply-To: <fac531740908311330q6827548dlf1966605f3ff4e10@mail.gmail.com>
References: <20090831223053.5461ad55@nb-jsosic>
	<fac531740908311330q6827548dlf1966605f3ff4e10@mail.gmail.com>
Message-ID: <20090831225918.33c892eb@nb-jsosic>

On Mon, 31 Aug 2009 15:30:13 -0500
Alan A <alan.zg at gmail.com> wrote:

> You have to give your cluster 2 node setting on remaining two nodes.
> 
> How can I define a two-node cluster if a majority is needed to reach
> quorum?
> 
> We had to allow two-node clusters, so we made a special exception to
> the quorum rules. There is a special setting "two_node" in
> the /etc/cluster.conf file that looks like this:
> 
> <cman expected_votes="1" two_node="1"/>
> 
> This will allow one node to be considered enough to establish a
> quorum. Note that if you configure a quorum disk/partition, you don't
> want two_node="1".
> 
> http://sources.redhat.com/cluster/wiki/FAQ/CMAN#two_node

Yeah I know how to set up two_node cluster, but didn't know that
reconfiguration is the only way out when one node is missing :( I'm
disappointed with this limitation, I hope that developers will sort it
out for RHEL6 :(



-- 
|    Jakov Sosic    |    ICQ: 28410271    |   PGP: 0x965CAE2D   |
=================================================================
| start fighting cancer -> http://www.worldcommunitygrid.org/   |



From ricks at nerd.com  Mon Aug 31 21:22:07 2009
From: ricks at nerd.com (Rick Stevens)
Date: Mon, 31 Aug 2009 14:22:07 -0700
Subject: [Linux-cluster] How to disable node?
In-Reply-To: <20090831225918.33c892eb@nb-jsosic>
References: <20090831223053.5461ad55@nb-jsosic>	<fac531740908311330q6827548dlf1966605f3ff4e10@mail.gmail.com>
	<20090831225918.33c892eb@nb-jsosic>
Message-ID: <4A9C3EFF.1000006@nerd.com>

Jakov Sosic wrote:
> On Mon, 31 Aug 2009 15:30:13 -0500
> Alan A <alan.zg at gmail.com> wrote:
> 
>> You have to give your cluster 2 node setting on remaining two nodes.
>>
>> How can I define a two-node cluster if a majority is needed to reach
>> quorum?
>>
>> We had to allow two-node clusters, so we made a special exception to
>> the quorum rules. There is a special setting "two_node" in
>> the /etc/cluster.conf file that looks like this:
>>
>> <cman expected_votes="1" two_node="1"/>
>>
>> This will allow one node to be considered enough to establish a
>> quorum. Note that if you configure a quorum disk/partition, you don't
>> want two_node="1".
>>
>> http://sources.redhat.com/cluster/wiki/FAQ/CMAN#two_node
> 
> Yeah I know how to set up two_node cluster, but didn't know that
> reconfiguration is the only way out when one node is missing :( I'm
> disappointed with this limitation, I hope that developers will sort it
> out for RHEL6 :(

I don't see that there's anything to fix.  You had a three-node cluster
so you needed a majority of nodes up to maintain a quorum.  One node
died, killing quorum and thus stopping the cluster (the expected and
correct behavior).  As a three-node cluster, it's dead.  It can't be run 
as a three-node cluster until the third node is fixed.  Those are the
rules.

A two node cluster requires special handling of things to prevent the
dread split-brain situation, which is what two_node does.  Running the
surviving nodes as a two-node cluster is, by definition, a
reconfiguration.  I'd say simply requiring you to set two_node is pretty
damned innocuous to let you run a dead (ok, mortally wounded) cluster.

If you pulled a drive out of a RAID6--thus degrading it to a RAID5--
would you complain because it didn't remain a RAID6?
----------------------------------------------------------------------
- Rick Stevens, Systems Engineer                      ricks at nerd.com -
- AIM/Skype: therps2        ICQ: 22643734            Yahoo: origrps2 -
-                                                                    -
-      Do you know how to save five drowning lawyers?  No?  GOOD!    -
----------------------------------------------------------------------



From mad at wol.de  Mon Aug 31 21:26:06 2009
From: mad at wol.de (Marc - A. Dahlhaus)
Date: Mon, 31 Aug 2009 23:26:06 +0200
Subject: [Linux-cluster] How to disable node?
In-Reply-To: <20090831225918.33c892eb@nb-jsosic>
References: <20090831223053.5461ad55@nb-jsosic>	<fac531740908311330q6827548dlf1966605f3ff4e10@mail.gmail.com>
	<20090831225918.33c892eb@nb-jsosic>
Message-ID: <4A9C3FEE.3000309@wol.de>

Jakov Sosic schrieb:
> On Mon, 31 Aug 2009 15:30:13 -0500
> Alan A <alan.zg at gmail.com> wrote:
>
>   
>> You have to give your cluster 2 node setting on remaining two nodes.
>>
>> How can I define a two-node cluster if a majority is needed to reach
>> quorum?
>>
>> We had to allow two-node clusters, so we made a special exception to
>> the quorum rules. There is a special setting "two_node" in
>> the /etc/cluster.conf file that looks like this:
>>
>> <cman expected_votes="1" two_node="1"/>
>>
>> This will allow one node to be considered enough to establish a
>> quorum. Note that if you configure a quorum disk/partition, you don't
>> want two_node="1".
>>
>> http://sources.redhat.com/cluster/wiki/FAQ/CMAN#two_node
>>     
>
> Yeah I know how to set up two_node cluster, but didn't know that
> reconfiguration is the only way out when one node is missing :( I'm
> disappointed with this limitation, I hope that developers will sort it
> out for RHEL6 :(
>   
I think your so called 'limitation' is more related to mistakes that was 
made during the planing phase of your cluster setup than to missing 
functionality.
Please take a look at the qdisk manpage and aditionaly to the cman faq 
sections about tiebraker, qdisks and especially the last man standing 
setup...


Marc



From ricks at nerd.com  Mon Aug 31 21:35:22 2009
From: ricks at nerd.com (Rick Stevens)
Date: Mon, 31 Aug 2009 14:35:22 -0700
Subject: [Linux-cluster] How to disable node?
In-Reply-To: <4A9C3EFF.1000006@nerd.com>
References: <20090831223053.5461ad55@nb-jsosic>	<fac531740908311330q6827548dlf1966605f3ff4e10@mail.gmail.com>
	<20090831225918.33c892eb@nb-jsosic> <4A9C3EFF.1000006@nerd.com>
Message-ID: <4A9C421A.8020003@nerd.com>

Rick Stevens wrote:
> Jakov Sosic wrote:
>> On Mon, 31 Aug 2009 15:30:13 -0500
>> Alan A <alan.zg at gmail.com> wrote:
>>
>>> You have to give your cluster 2 node setting on remaining two nodes.
>>>
>>> How can I define a two-node cluster if a majority is needed to reach
>>> quorum?
>>>
>>> We had to allow two-node clusters, so we made a special exception to
>>> the quorum rules. There is a special setting "two_node" in
>>> the /etc/cluster.conf file that looks like this:
>>>
>>> <cman expected_votes="1" two_node="1"/>
>>>
>>> This will allow one node to be considered enough to establish a
>>> quorum. Note that if you configure a quorum disk/partition, you don't
>>> want two_node="1".
>>>
>>> http://sources.redhat.com/cluster/wiki/FAQ/CMAN#two_node
>>
>> Yeah I know how to set up two_node cluster, but didn't know that
>> reconfiguration is the only way out when one node is missing :( I'm
>> disappointed with this limitation, I hope that developers will sort it
>> out for RHEL6 :(
> 
> I don't see that there's anything to fix.  You had a three-node cluster
> so you needed a majority of nodes up to maintain a quorum.  One node
> died, killing quorum and thus stopping the cluster (the expected and
> correct behavior).  As a three-node cluster, it's dead.  It can't be run 
> as a three-node cluster until the third node is fixed.  Those are the
> rules.
> 
> A two node cluster requires special handling of things to prevent the
> dread split-brain situation, which is what two_node does.  Running the
> surviving nodes as a two-node cluster is, by definition, a
> reconfiguration.  I'd say simply requiring you to set two_node is pretty
> damned innocuous to let you run a dead (ok, mortally wounded) cluster.
> 
> If you pulled a drive out of a RAID6--thus degrading it to a RAID5--
> would you complain because it didn't remain a RAID6?

On re-reading my response, it seemed unintentionally harsh.  I didn't
mean any disrespect, sir.  I was simply questioning the concept that a
reconfiguration of a cluster shouldn't be required when, indeed the
cluster was being reconfigured.  The other response I saw to this
thread regarding planning, and things such as last-man-standing was much
better worded.

My apologies if it seemed I was jumping down your throat.  I wasn't.
----------------------------------------------------------------------
- Rick Stevens, Systems Engineer                      ricks at nerd.com -
- AIM/Skype: therps2        ICQ: 22643734            Yahoo: origrps2 -
-                                                                    -
-     Is that a buffer overflow or are you just happy to see me?     -
----------------------------------------------------------------------



From mad at wol.de  Mon Aug 31 21:43:54 2009
From: mad at wol.de (Marc - A. Dahlhaus)
Date: Mon, 31 Aug 2009 23:43:54 +0200
Subject: [Linux-cluster] How to disable node?
In-Reply-To: <4A9C3FEE.3000309@wol.de>
References: <20090831223053.5461ad55@nb-jsosic>	<fac531740908311330q6827548dlf1966605f3ff4e10@mail.gmail.com>	<20090831225918.33c892eb@nb-jsosic>
	<4A9C3FEE.3000309@wol.de>
Message-ID: <4A9C441A.2080909@wol.de>

Marc - A. Dahlhaus schrieb:
> Jakov Sosic schrieb:
>> On Mon, 31 Aug 2009 15:30:13 -0500
>> Alan A <alan.zg at gmail.com> wrote:
>>
>>  
>>> You have to give your cluster 2 node setting on remaining two nodes.
>>>
>>> How can I define a two-node cluster if a majority is needed to reach
>>> quorum?
>>>
>>> We had to allow two-node clusters, so we made a special exception to
>>> the quorum rules. There is a special setting "two_node" in
>>> the /etc/cluster.conf file that looks like this:
>>>
>>> <cman expected_votes="1" two_node="1"/>
>>>
>>> This will allow one node to be considered enough to establish a
>>> quorum. Note that if you configure a quorum disk/partition, you don't
>>> want two_node="1".
>>>
>>> http://sources.redhat.com/cluster/wiki/FAQ/CMAN#two_node
>>>     
>>
>> Yeah I know how to set up two_node cluster, but didn't know that
>> reconfiguration is the only way out when one node is missing :( I'm
>> disappointed with this limitation, I hope that developers will sort it
>> out for RHEL6 :(
>>   
> I think your so called 'limitation' is more related to mistakes that was 
> made during the planing phase of your cluster setup than to missing 
> functionality.
> Please take a look at the qdisk manpage and aditionaly to the cman faq 
> sections about tiebraker, qdisks and especially the last man standing 
> setup...

I forgot to mention that if you can't add a qdisk to your setup, you 
could still try to bring your cluster into running state by starting 
fenced in clean start without fencing mode (read fenced manpage for the 
details). Your quorum is reached at a score of two so you can start it 
that way.

Marc