Ocfs2 users - Dec 2007 - message to node 2 fails with error -75

I'm in the process of trying to upgrade the kernel on my OCFS2 cluster from
2.6.18 to 2.6.22.  I've downloaded and compiled the source for kernel
2.6.22.14 on my machines, and I've restarted each of the machines with the
new kernel.  Two of the machines are x86_64 (Intel core2) machines, the other
one is an older Xeon/i686 machine.  The cluster worked fine in 2.6.18.8.  With
the new kernel, the first machine is able to mount the filesystem correctly, but
when the second machine goes to mount the filesystem I get the following error
in the /var/log/messages file and dmesg output:

Dec  4 15:19:51 vmdevel2 kernel: o2net: accepted connection from node vmdevel0
(num 2) at 192.168.30.20:7777
Dec  4 15:19:55 vmdevel2 kernel: ocfs2_dlm: Node 2 joins domain
9F27F6D72A954780B76A4FD9CD97A0AF
Dec  4 15:19:55 vmdevel2 kernel: ocfs2_dlm: Nodes in domain
("9F27F6D72A954780B76A4FD9CD97A0AF"): 1 2
Dec  4 15:19:55 vmdevel2 kernel: (31952,0):ocfs2_process_vote:206 ERROR: message
to node 2 fails with error -75!

One the second machine, the machine trying to join the cluster and mount the
filesystem, this is the output:

Dec  4 15:18:11 vmdevel0 kernel: o2net: connected to node vmdevel2 (num 1) at
192.168.30.22:7777
Dec  4 15:18:15 vmdevel0 kernel: OCFS2 1.3.3 
Dec  4 15:18:15 vmdevel0 kernel: ocfs2_dlm: Nodes in domain
("9F27F6D72A954780B76A4FD9CD97A0AF"): 1 2
Dec  4 15:18:15 vmdevel0 kernel: kjournald starting.  Commit interval 5 seconds 

Seems the second machine believes it has joined okay, but then just hangs.  Once
it actually started to heartbeat, but then locked hard and I had to power cycle
it to get it back.  Anyone have any clues on where I might go to start?  Again,
kernel is 2.6.22.14 vanilla; tools version is 1.2.2.

Thanks, 
Nick 
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://oss.oracle.com/pipermail/ocfs2-users/attachments/20071204/7f5839e5/attachment.html

You are running into bugzilla#913.
http://oss.oracle.com/bugzilla/show_bug.cgi?id=913

The fix is here:
http://oss.oracle.com/bugzilla/attachment.cgi?id=535

Also, do remember to apply relevant patches collected here.
http://www.kernel.org/pub/linux/kernel/people/mfasheh/ocfs2/backports/

Nick Couchman wrote:
>
> I'm in the process of trying to upgrade the kernel on my OCFS2 cluster 
> from 2.6.18 to 2.6.22.  I've downloaded and compiled the source for 
> kernel 2.6.22.14 on my machines, and I've restarted each of the 
> machines with the new kernel.  Two of the machines are x86_64 (Intel 
> core2) machines, the other one is an older Xeon/i686 machine.  The 
> cluster worked fine in 2.6.18.8.  With the new kernel, the first 
> machine is able to mount the filesystem correctly, but when the second 
> machine goes to mount the filesystem I get the following error in the 
> /var/log/messages file and dmesg output:
>
>
> Dec  4 15:19:51 vmdevel2 kernel: o2net: accepted connection from node 
> vmdevel0 (num 2) at 192.168.30.20:7777
>
> Dec  4 15:19:55 vmdevel2 kernel: ocfs2_dlm: Node 2 joins domain 
> 9F27F6D72A954780B76A4FD9CD97A0AF
>
> Dec  4 15:19:55 vmdevel2 kernel: ocfs2_dlm: Nodes in domain 
> ("9F27F6D72A954780B76A4FD9CD97A0AF"): 1 2
>
> Dec  4 15:19:55 vmdevel2 kernel: (31952,0):ocfs2_process_vote:206 
> ERROR: message to node 2 fails with error -75!
>
>
> One the second machine, the machine trying to join the cluster and 
> mount the filesystem, this is the output:
>
>
> Dec  4 15:18:11 vmdevel0 kernel: o2net: connected to node vmdevel2 
> (num 1) at 192.168.30.22:7777
>
> Dec  4 15:18:15 vmdevel0 kernel: OCFS2 1.3.3
>
> Dec  4 15:18:15 vmdevel0 kernel: ocfs2_dlm: Nodes in domain 
> ("9F27F6D72A954780B76A4FD9CD97A0AF"): 1 2
>
> Dec  4 15:18:15 vmdevel0 kernel: kjournald starting.  Commit interval 
> 5 seconds
>
>
> Seems the second machine believes it has joined okay, but then just 
> hangs.  Once it actually started to heartbeat, but then locked hard 
> and I had to power cycle it to get it back.  Anyone have any clues on 
> where I might go to start?  Again, kernel is 2.6.22.14 vanilla; tools 
> version is 1.2.2.
>
>
> Thanks,
>
> Nick
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> Ocfs2-users mailing list
> Ocfs2-users@oss.oracle.com
> http://oss.oracle.com/mailman/listinfo/ocfs2-users