[Linux-cluster] sudden unfencing problem

[laurence.schuler]

I will try the tcpdump, I've been able to do stupid stuff with the
fence_sanbox2 python script (repeat sequences of issue/expect; increase
the python delaybeforesend value) to get it to work most of the time,
but I don't believe the script code is wrong, there's something else
going on, perhaps the switch is in a funny state. Now, how would I
detect *that*!

Thanks,
--Larry

On 03/24/2013 12:00 PM, James Washer wrote:
> You might want to grab a tcpdump of the connection. Perhaps you'll be
> able to see a bit more of the conversation.
>
> On Sat, Mar 23, 2013 at 5:55 AM, Laurence Schuler
> <laurence.schuler at nasa.gov <mailto:laurence.schuler at nasa.gov>> wrote:
>
>     I have a two node cluster that has been running fine for a couple of
>     months (little to 0 reboots though). We recently updated the software
>     with the latest Centos 6 software but now the cluster will not
>     start. It
>     keeps throwing errors during startup when attempting to unfence the
>     disks. I have hard reset the fiber switch, and reset both hosts, but
>     when I run fence_sanbox2, I am unable to either enable, disable or
>     even
>     get status of the switch ports. This is the error I get.
>
>     > [root at web1 lschule3]# /usr/sbin/fence_sanbox2 -a 192.168.1.190 -l
>     > admin -S FCpass.sh -o enable -n 5 -v
>     > telnet> set binary
>     > Negotiating binary mode with remote host.
>     > telnet> open 192.168.1.190 -23
>     > Trying 192.168.1.190...
>     > Connected to 192.168.1.190.
>     > Escape character is '^]'.
>     >
>     > Firmware V8.0.13.8.0
>     >
>     > r3fc1 login:
>     >
>     >
>     >   Establishing connection...   Please wait.
>     >
>     >        *****************************************************
>     >        *                                                   *
>     >        *       Command Line Interface SHell  (CLISH)       *
>     >        *                                                   *
>     >        *****************************************************
>     >
>     >        SystemDescription   SANbox 5800 FC Switch
>     >        HostName            r3fc1
>     >        EthIPv4NetworkAddr  192.168.1.190
>     >        EthIPv6NetworkAddr  fe80::2c0:00:00:90b
>     >        MACAddress          00:c0:dd:77:10:0b
>     >        WorldWideName       10:00:00:c0:dd:24:09:0b
>     >        SerialNumber        1236H00833
>     >        SymbolicName        r3fc1
>     >        ActiveSWVersion     V8.0.13.8.0
>     >        ActiveTimestamp     Mon Apr  2 18:32:33 2012
>     >        POSTStatus          Passed
>     >        LicensedPorts       12
>     >        SwitchMode          Full Fabric
>     >
>     >   The alarm log is empty.
>     >
>     > r3fc1 #> r3fc1 #> Failed: Unable to switch to admin section
>     > [root at web1 lschule3]#
>
>     I can manually telnet into the FC switch and execute the appropriate
>     commands to enable/disable ports. But the fence_sanbox2 script
>     will not.
>     The fence_sanbox2 code has not changed, however python has been
>     upgraded
>     from 2.6.6-29 to 2.6.6-36.
>
>     Has anyone else seen this? Know of a fix? Am I doing/not doing
>     something
>     stupid? I seem to recall running this command before during setup
>     and it
>     worked just fine then.
>
>     Thanks for any help!
>
>     --
>     Laurence Schuler (Larry)                      
>     Laurence.Schuler at nasa.gov <mailto:Laurence.Schuler at nasa.gov>
>     Systems Support                                       ADNET
>     Systems, Inc
>     Scientific Visualization Studio                
>     http://svs.gsfc.nasa.gov
>
>     --
>     Linux-cluster mailing list
>     Linux-cluster at redhat.com <mailto:Linux-cluster at redhat.com>
>     https://www.redhat.com/mailman/listinfo/linux-cluster
>
>
>
>
> -- 
>
>
>  - jim

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20130324/badff39e/attachment.htm>