A corruption that seems to span a few kernels

[Joe Landman]

On 02/03/2012 04:27 PM, Christoph Hellwig wrote:
> On Fri, Feb 03, 2012 at 03:02:14PM -0500, Joe Landman wrote:
>> XFS file system, built using a 2.6.32.41 kernel.  External journals.
>>
>> repair the file system, mount, turn on gluster (3.2.5), do a find,
>> and get a corruption.  Actually it looks like xfs_repair doesn't fix
>> the problem.
>>
>> Rinse repeat, reproduce.  Works (procedure to generate the
>> corruption) in 2.6.32.41.scalable, 2.6.39.1, 3.2.2
>
> [PATCH v2] repair: update extent count after zapping duplicate blocks
>
> which was sent to the list yesterday should make xfs_repair handle this
> case.
>
> How do you reproduce the corruption?  Just start gluster on a fresh fs
> an do a find?  Any chance you could share command lines used?

It looks like the corruption was just never repaired by xfs_repair. I 
just followed the notes here: 
http://xfs.org/index.php/XFS_FAQ#Q:_How_to_get_around_a_bad_inode_repair_is_unable_to_clean_up 
and got it operational again.  I first blockget -n -i ###  to figure out 
the mapping and let the user know.

I'll grab the patch and apply it.  If this comes up again, I'll report.

To summarize, this was a corruption that xfs_repair didn't fix, so 
fixing it by hand (and rerunning xfs_repair) did.

Thanks!

Joe

-- 
Joseph Landman, Ph.D
Founder and CEO
Scalable Informatics Inc.
email: landman at scalableinformatics.com
web  : http://scalableinformatics.com
        http://scalableinformatics.com/sicluster
phone: +1 734 786 8423 x121
fax  : +1 866 888 3112
cell : +1 734 612 4615