[Users] VM Disk corruption

[Usman Aslam]

I have a few VM's that have their root LVM file system corrupted because of
improper shutdown. I experienced an issue described here which cause a node
to restart
http://lists.ovirt.org/pipermail/users/2013-July/015247.html

But now I'm experiencing the same problem with other VM's. A DB VM was in
single user mode when it was powered off. I created snapshot to clone it.
Powered the CentOS 6 VM back on and it cannot execute the mount command on
boot and dropped me to a root maintenance prompt.

Running fsck comes back with way too many errors, many of them about files
that shouldnt have even been open. Countless Innode issues and clone
multiply-claimed blocks. Running fsck -y seems to fix the file system and
it comes back as clean but upon restart, the VM load the files system as
read only and trying to remount it as rw, the mount command throws a
segmentation fault.

I dont know if its the new version of ovirt engine. I'm afraid to shutdown
any VM using the ovirt UI as they dont always come backup. I have tried
repairing the file system with the live CD and such to no avail. Given the
fact the alot of the corrupt files are plain static html, or archived tar
files, I assume it has something to do with ovirt. The only corrupting for
data should be and live application data (open files).

Please advise on how to proceed. I can provide whatever logs you may
requires.

Thanks,
-- 
Usman Aslam
401.6.99.66.55
usman at linkshift.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/users/attachments/20131007/5d9ae398/attachment-0001.html>