[Ocfs2-users] OCFS2 Filesystem inconsistency across nodes

[Nohez]

> Nohez, thanks for testing with those patches and letting us know how things
> went. I look forward to seeing how your multi-node bonnie++ test goes,
> though I have a feeling you might hit the same issue Claudio hit below.

Thanks 2 u 2 for helping me out.

Bonnie++ running on 2 nodes and writing on the same volume works. Executed
bonnie++ on each node concurrently for 20 times (each bonnie++ writes
8GB of files). Bonnie++ ran for roughly 7 hours before both the nodes
panicked. The panic happened because qlogic driver started giving the
following error message:

qla2xxx 0000:23:01.0: scsi(1:3:1): Abort command issued -- b8b753 2002

I suspect some SAN problem and not Qlogic driver problem.

I also want to remind u that I am not using ocfs2 which has been
integrated into latest kernel. I am using the one provided by SuSE
in OpenSuSE10.1 Beta3 which is ocfs2 1.1.7-SLES. I requested OpenSuSE
maintainers to upgrade it to the latest available. 

Reply:
-----------------------------------------------------------------
>From Chris L Mason 2006-02-09 07:25 MST
(https://bugzilla.novell.com/show_bug.cgi?id=149500)

This is intentional.  ocfs2 was put into the mainline kernel very late in the
SLES10/SUSE 10.1 cycle, and oracle asked us to keep the older rev level for
certifications purposes.  We are mirroring bug fixes from the mainline kernel
into our code base (with help from oracle).

So, the version number is different, but our ocfs2 will be updated regularly
for fixes and improvements.
-----------------------------------------------------------------

Will redo the multi-node tests and post the results.

Thanks

Nohez