[Linux-cluster] Problems after gfs_resize

[Bob Peterson]

On Thu, 2008-08-07 at 16:01 -0400, Shawn Hood wrote:
> Yesterday, I resized (+300GB) a clustered logical volume using
> lvresize.  I then executed gfs_grow.  All went well.  Today, however,
> I'm getting the error message 'attempt to access beyond end of
> device.'  See below.  I wanted to give some cursory information in
> case this is a known issue with a quick fix.  I'll follow up with
> additional information.  Any ideas??

Hi Shawn,

This is most likely bugzilla bug #436383.  Assuming it is,
there is a fix to gfs-kernel that resolves the problem so it can't
happen again.  Unfortunately, the bug causes damage to the gfs
file system.  The latest version of gfs_fsck knows how to repair
the damage.  The fix to gfs_fsck was done for bugzilla bug 440896.
My apologies if you can't view either of these bug records, but
they may be marked private to protect the customer who reported the
problem.

Of course, without actually seeing your file system, I can't be
sure that it's the same problem.  It's likely it though.

The fix to both gfs and gfs_fsck were posted to RHN.  Here is a
link to the gfs_fsck information:
http://rhn.redhat.com/errata/RHBA-2008-0804.html

(There are RHEL4 and RHEL5 fixes for both of these, but I'm posting
the RHEL4 version because your kernel indicated you were using RHEL4.)

Regards,

Bob Peterson
Red Hat Clustering & GFS