xfs_force_shutdown() called when running xfstests on 2.6.38-rc4

[Chandra Seetharaman]

On Fri, 2011-02-18 at 14:30 +1100, Dave Chinner wrote:
> On Thu, Feb 17, 2011 at 06:09:40PM -0800, Chandra Seetharaman wrote:
> > On Fri, 2011-02-18 at 09:52 +1100, Dave Chinner wrote:
> > > On Thu, Feb 17, 2011 at 10:49:10AM -0800, Chandra Seetharaman wrote:
> > > > Thanks for that info, Eric.
> > > > 
> > > > Now I increased the filesystem size and I see the following failure
> > > > --------------------------------
> > > > QA output created by 180
> > > > file /mnt/xfsScratchMntPt/456 has incorrect size - sync failed
> > > > file /mnt/xfsScratchMntPt/525 has incorrect size - sync failed
> > > > file /mnt/xfsScratchMntPt/624 has incorrect size - sync failed
> > > > ---------------------------------
> > > > 
> > > > I guess that is not good :)
> > > 
> > > No, not good. It passes here on x86_64 and a couple of different
> > > configurations storage back ends (1/2p on h/w raid1 writing
> > > @~90MB/s, 8p + s/w RAID0 @ ~700MB/s), so it doesn't seem like there
> > > is a generic problem.
> > > 
> > > what is the hardware you are testing on? 
> > PPC64
> 
> Yes, I know you are running on PPC64, but that doesn't tell me

It is a 4 processor quad-core, which gets 16 procs. 32G RAM.

Storage is a ipr RAID0 of 2 500G sas disks.

Filesystem size is 20G.

> anything about the hardware you are running on.  How many CPUs, how
> much RAM, what storage you are using, how it is configured, etc, is
> the sort of infomration relevant to understanding the circumstances
> around the failures you are seeing...
> 
> > > Does it happen on every run?
> > 
> > No, it is not consistent. It happened as this
> > 
> > +file /mnt/xfsScratchMntPt/993 has incorrect size - sync failed
> > 
> > and ran 10+ times, didn't fail at all.
> 
> OK. So it's a relatively rare failure, then?

Very rare. I ran it more than 20 times yesterday night, and it didn't
fail.

chandra
> 
> Cheers,
> 
> Dave.