Hi Jon,

we encountered the same problem on four of our 64-bit machines in Aachen. They
are setup as homedir servers and quite loaded. On one machine we had to do the
xfs_repair after access to the filesystem resulted in Input/Output errors.

The XFS is on top of a sofware RAID-5 consisting of 4 HDDs. The filesystem is
exported via NFS3 to our desktop cluster. Before the kernel update no problems
occured. We decided to step back to the old kernel version with the XFS
modules not included in the kernel rpms. Until now everything seems to be
quiet again.

We hope to find some time next week to test a similar machine with NFS4 and
software RAID-5 with XFS on the newest 64-bit SL5 kernel.

Cheers,

Michael.

Jon Peatfield schrieb:
> On wednesday morning we updated most of our sl53 machines to the current
> 2.6.18-164.6.1.el5 kernel.
> 
> Since then we have had two machines (both x86_64 of course) using xfs
> report xfs corruption problems, e.g. on one of the machines today:
> 
<snip>
>
> running xfs_repair(*) shows no obvious problems, and the fs then appears
> to be ok for a while at least.  On one of the machines the problem came
> back after a few more hours, but since then hasn't happened again - yet.
> 
> Is anyone else seeing this?  So far we have only noticed this on two
> machines which happen to be also having the xfs volume used quite
> heavily over NFS but that may be co-incidence.
> 
> Until the new kernel these machines were apparently working ok with the
> previous sl53 kernel, so maybe this is caused by how TUV happen to have
> built their xfs modules - as compared to the ones which SL made before...
> 
> (*) xfs_repair (and xfs_check) refused to touch the file-system claiming
> it was mounted even though we had unmounted it.  We ended up needing to
> reboot with the fs commented out of fstab to get xfs_repair to run. 
> I've never needed to run xfs_repair before so I don't know if that is
> normal or not but it seems odd - though probably not related to the real
> problem.
> 
> Is there a way we can do any tests with xfs built like it was for the
> older sl kernels?
> 
>  -- Jon
> 

-- 
-----------------------------------------------------------------------
 Dipl.-Phys. Michael Bontenackels

 III. Physikalisches Institut A                 Phone +49 241 80 27285
 Office 28A221                                  Fax   +49 241 80 22189

 III. Physikalisches Institut B                 Phone +49 241 80 27281
 Office 28A206                                  Fax   +49 241 80 22244

 RWTH Aachen Physikzentrum
 Otto-Blumenthal-Straße
 52074 Aachen
 Germany                            [log in to unmask]

-- private ------------------------------------------------------------

 Michael Bontenackels                      Phone  +49 241 4459858
 Haßlerstraße 7-9                          Mobile +49 178 1488219
 52066 Aachen                              Fax    +49 1212 515369028
 Germany                                   [log in to unmask]
-----------------------------------------------------------------------