Re: XFS internal error xfs_trans_cancel at line 1163 of file fs/xfs/xfs_trans.c

[Miquel van Smoorenburg <mikevs@xs4all.net>]

On Fri, 2008-06-13 at 13:08 +1000, Dave Chinner wrote: 
> On Fri, Jun 13, 2008 at 12:27:09AM +0200, Miquel van Smoorenburg wrote:
> > On Thu, 2008-06-12 at 16:33 +0200, Oliver Pinter wrote:
> > > add CC's
> > > 
> > > On 6/12/08, lists@ku-gbr.de <lists@ku-gbr.de> wrote:
> > > > Hi!
> > > >
> > > > Today morning my server at home bailed out two times (reboot between):
> > 
> > > > Jun 12 07:23:40 zappa
> > > > Jun 12 07:23:40 zappa xfs_force_shutdown(sda7,0x8) called from line
> > > > 1164 of file fs/xfs/xfs_trans.c.  Return address = 0xffffffff802f4d8a
> > > > Jun 12 07:23:40 zappa Filesystem "sda7": Corruption of in-memory data
> > > > detected.  Shutting down filesystem: sda7
> > > > Jun 12 07:23:40 zappa Please umount the filesystem, and rectify the
> > > > problem(s)
> > 
> > Hmm, interesting. I'm seeing the same thing on one of my servers since I
> > upgraded from 2.6.ancient (14 or so) to 2.6.25, while XFS otherwise has
> > been very stable for me over the years:
> > 
> > Linux transit5.news.xs4all.nl 2.6.25.6 #1 SMP Wed Jun 11 10:59:10 CEST
> > 2008 x86_64 GNU/Linux
> > 
> > Filesystem "sda4": XFS internal error xfs_trans_cancel at line 1163 of
> > file fs/xfs/xfs_trans.c.  Caller 0xffffffff880f1315
> > Pid: 3402, comm: diablo Not tainted 2.6.25.6 #1
> 
> This commit in 2.6.26 will probably fix it.
> 
> http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=75de2a91c98a6f486f261c1367fe59f5583e15a3

"At ENOSPC, we can get a filesystem shutdown due to a cancelling a dirty
transaction in xfs_mkdir or xfs_create."

But:

$ df -h
Filesystem            Size  Used Avail Use% Mounted on
/dev/sda4              15G  5.1G  9.6G  35% /news

The filesystem is only used for 35%. It might have hit 100% somewhere in
the recent past though (a few reboots ago).

I've applied the patch to 2.6.25.6 just in case, I'll let it run over
the weekend to see what happens.

Mike.