experimenting with dmraid - disk status

Sandra Escandor sescandor at evertz.com
Tue May 3 17:46:18 UTC 2011


Ok, I've set it up so that I transfer a large file onto the RAID, and
then during the transfer, I pull out a member disk to simulate a disk
failure. I would like to confirm if I should be seeing what I'm seeing:

-The output of "dmraid -n" shows disk[0].status: 0x53a for all of the
disks, even the one pulled out. I expected dmeventd to change one of
them, since I'm doing I/O as you had mentioned. Is what I was expecting
incorrect?
-The output of "dmraid -n" shows isw_dev[0].vol.map[0].failed_disk_num:
255. I expected this to be the number of the disk that I have removed.
Is what I am expecting for this field incorrect?

I do see that dmsetup and dm_dso_reg_tool give the proper output (as
mentioned in the document "How to Guide: DMRAID Eventing - What To
Expect"), so that's good. So, does this mean that dmeventd doesn't
modify the metadata at all as what I had expected above?

Thanks,
Sandra

-----Original Message-----
From: ataraid-list-bounces at redhat.com
[mailto:ataraid-list-bounces at redhat.com] On Behalf Of Heinz Mauelshagen
Sent: Tuesday, May 03, 2011 8:46 AM
To: ATARAID (eg, Promise Fasttrak, Highpoint 370) related discussions
Subject: RE: experimenting with dmraid - disk status

On Tue, 2011-05-03 at 07:58 -0400, Sandra Escandor wrote:
> So, if I understood what you have said correctly, then the following
> would be true:
> 
> - When there is no I/O being performed on the RAID, and then one of
the
> member disks becomes disconnected for some reason, then a dmeventd
event
> would not be generated.

Correct.

> - After a while, I/O does get performed on the RAID (with a member
disk
> disconnected), then that will be the time that a dmeventd event gets
> generated.

Correct.

Any generated events are being based on io, not on _any_ other events
such as unplugging a drive.

Heinz

> 
> Is that right?
> 
> -----Original Message-----
> From: ataraid-list-bounces at redhat.com
> [mailto:ataraid-list-bounces at redhat.com] On Behalf Of Heinz
Mauelshagen
> Sent: Monday, May 02, 2011 4:53 PM
> To: ATARAID (eg, Promise Fasttrak, Highpoint 370) related discussions
> Subject: RE: experimenting with dmraid - disk status
> 
> On Mon, 2011-05-02 at 15:19 -0400, Sandra Escandor wrote:
> > After I installed the dmeventd package, I notice the following:
> > (general setup: ICH10R controller, raid0+1, 4 member disks)
> > - When I pull out a drive, then run dm_dso_reg_tool -m, the section
> > "Error Events Recorded by Kernel" for the superset and two subsets
are
> > still 0. Is this expected?
> 
> Unless you performed io on the set.
> 
> > - (Drive still physically disconnected) when I run dmraid -n, the
> > metadata shows that the status of the disk that I pulled out still
has
> > the same status as the other disks which are plugged in (i.e
> > disk[0].status: 0x53a is shown for all of the disks, even the one
that
> > I pulled out).
> 
> Same as above.
> 
> Heinz
> 
> > 
> > The steps that I took to set up the device mapper raid event
> monitoring
> > are basically the same as the one found in: "How to Setup Device
> Mapper
> > Raid event Monitoring_0.74.pdf"
> > (http://sources.redhat.com/lvm2/wiki/DMRAID_Eventing)
> > EXCEPT: 
> > * I obtained the dmeventd package from the Ubuntu repository
> (Maverick)
> > * I didn't install the patch (since I'm using kernel 2.6.35)
> > * I changed the makefile of libdmraid-events so that it would link
> with
> > the pthread library (i.e added -lpthread to the makefile).
> > 
> > Is there another package that I need to install, or have I not set
> this
> > up properly?
> > 
> > Thanks for the clarifications,
> > Sandra
> > 
> > -----Original Message-----
> > From: Heinz Mauelshagen [mailto:heinzm at redhat.com] 
> > Sent: Friday, April 29, 2011 10:14 AM
> > To: Sandra Escandor
> > Subject: RE: experimenting with dmraid - disk status
> > 
> > On Fri, 2011-04-29 at 08:00 -0400, Sandra Escandor wrote:
> > > I forgot to mention my metadata type. Yes, I'm using Intel
software
> > raid
> > > (so it's in isw format). Where would I be able to get this plugin?
> > 
> > You should get them with any distros dmraid-events package.
> > 
> > Heinz
> > 
> > > 
> > > Thanks,
> > > Sandra
> > > 
> > > -----Original Message-----
> > > From: ataraid-list-bounces at redhat.com
> > > [mailto:ataraid-list-bounces at redhat.com] On Behalf Of Heinz
> > Mauelshagen
> > > Sent: Friday, April 29, 2011 3:58 AM
> > > To: ataraid-list at redhat.com
> > > Subject: Re: experimenting with dmraid - disk status
> > > 
> > > On Thu, 2011-04-28 at 09:30 -0400, Sandra Escandor wrote:
> > > > Hello,
> > > > 
> > > >  
> > > > 
> > > > I was doing a bit of experimenting with dmraid. What I have done
> is
> > to
> > > > create a raid0+1 array, and then physically pull one of the
disks
> > out.
> > > > I then use the command "dmraid -n" to look at the metadata on
the
> > > > member disks. I notice that the member disk listed by the output
> > that
> > > > has the specific serial number of the disk that I pulled, whose
> > status
> > > > I expect to be different from the other disks listed, still has
> the
> > > > same status as the other still physically plugged in. (i.e
> > > > disk[0].status: 0x53a is shown for all of the disks, even the
one
> > that
> > > > I pulled out). Is this normal?
> > > 
> > > Yes, unless it's isw format, where a plugin exists to update the
> > > metadata.
> > > 
> > > Heinz
> > > 
> > > > 
> > > >  
> > > > 
> > > > The following is the output of "dmraid -V":
> > > > 
> > > > dmraid version: 1.0.0.rc16 (2009.09.16) shared
> > > > 
> > > > dmraid library version:  1.0.0.rc16 (2009.09.16) 
> > > > 
> > > > device-mapper version: 4.13.0
> > > > 
> > > >  
> > > > 
> > > > Thanks for clarifying,
> > > > 
> > > > Sandra 
> > > > 
> > > > 
> > > > _______________________________________________
> > > > Ataraid-list mailing list
> > > > Ataraid-list at redhat.com
> > > > https://www.redhat.com/mailman/listinfo/ataraid-list
> > > 
> > > 
> > > _______________________________________________
> > > Ataraid-list mailing list
> > > Ataraid-list at redhat.com
> > > https://www.redhat.com/mailman/listinfo/ataraid-list
> > 
> > 
> > 
> > _______________________________________________
> > Ataraid-list mailing list
> > Ataraid-list at redhat.com
> > https://www.redhat.com/mailman/listinfo/ataraid-list
> 
> 
> _______________________________________________
> Ataraid-list mailing list
> Ataraid-list at redhat.com
> https://www.redhat.com/mailman/listinfo/ataraid-list
> 
> _______________________________________________
> Ataraid-list mailing list
> Ataraid-list at redhat.com
> https://www.redhat.com/mailman/listinfo/ataraid-list


_______________________________________________
Ataraid-list mailing list
Ataraid-list at redhat.com
https://www.redhat.com/mailman/listinfo/ataraid-list




More information about the Ataraid-list mailing list