Dexter Filmore wrote:
> Am Dienstag, 10. Oktober 2006 05:23 schrieb James G. Sack (jim):
>> Dexter Filmore wrote:
>>> When I copy a lot of small files on a file server, it eventually shuts
>>> down the fs.
>>> Looks like this in dmesg:
>>>
>>> [ 1734.674678] xfs_force_shutdown(dm-1,0x8) called from line 1139 of file
>>> fs/xfs/xfs_trans.c.  Return address = 0xc0297f83
>>> [ 1734.694671] Filesystem "dm-1": Corruption of in-memory data detected.
>>> Shutting down filesystem: dm-1
>>> [ 1734.694689] Please umount the filesystem, and rectify the problem(s)
>>>
>>> Is this an xfs issue or a hardware problem?
>>>
>>> Said xfs sits on top of a LVM2 volume which sits on a software raid5 with
>>> sATA disks on a SIL3112 controller.
>>> Slackware 11.0, but occured on 10.2 as well.
>> I've seen <similar> messages tracable to LVM snapshots. Are you using
>> any lvm snapshots? Do you have any 'smart' backup operations doing
>>   dmsetup suspend/resume
>> or
>>   xfs_freeze/xfs_freeze -u
> 
> At least none which I am aware of. I didn't install any such.
> 
>> Of course, maybe there _really_ is something bad on one of the disks,
>> and raid5 might be failing a read. On a write, the disk might be doing
>> auto remapping -- but failed reads are actually nastier!
> 
> raid is fine, smartctl looks good.

My reference to read failures above is likely a red herring, since a
read failure would lead to failing a raid member disk.

Searching on
  Corruption of in-memory data detected. Shutting down filesystem: dm

does lead to some reading material, including, for example
  http://oss.sgi.com/projects/xfs/faq.html#dir2

Since the error is memory-related, you might consider an extended
memtest86 run -- by which I mean multiple passes, say overnight.

Regards,
..jim


-- 
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list

Reply via email to