Re: kernel: Access to block zero: fs: <sdg> inode: 231137029....

[Dave Chinner <david@xxxxxxxxxxxxx>]

On Mon, Feb 09, 2009 at 03:33:15PM +0100, Ralf Gross wrote:
> Hello,
> 
> this is the second time I got hit by this problem in the last 2 months.
> 
> The server did respond to ping and all processes were still running, but the
> load went up to >30. Neither a shutdown nor a sync command could be executed,
> in the end I had to reset the server because nobody could access the samba
> shares anymore. 
> 
> System: Debian Etch linux-image-2.6.18-6-amd64 (since today 2.6.24 from
> etch-n-half). sdg is a fibre attached (QLA2422) RAID array.
> 
> Any idea what the reason for this could be? xfs_check of /dev/sdg didn't find 
> anything.

Btree corruption in memory (not yet written to disk). I'd suggest
that you should upgrade to at least 2.6.27 which is when we fixed
the last of the known btree corruption problems...

Cheers,

Dave.
-- 
Dave Chinner
david@xxxxxxxxxxxxx