scratch filesystem corruptions

scratch filesystem corruptions

[alexander.tsvetkov]

Hello,

I've noted that the enabling of scratch device post checks after 
xfstests commit
"common: Check the file system consistency on SCRATCH_DEV" has started 
to show
the scratch filesystem corruptions for the following tests runs: 
xfs/050,134,187,244,287.
I have these corruptions reproduced with upstream linux kernel of 
version 3.15 and
the latest xfsprogs/xfstests. Are these known issues?

Thanks,
Alexander Tsvetkov

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: scratch filesystem corruptions

[Dave Chinner]

On Thu, Nov 06, 2014 at 01:57:30PM +0300, alexander.tsvetkov@oracle.com wrote:
> Hello,
> 
> I've noted that the enabling of scratch device post checks after
> xfstests commit
> "common: Check the file system consistency on SCRATCH_DEV" has
> started to show
> the scratch filesystem corruptions for the following tests runs:
> xfs/050,134,187,244,287.
> I have these corruptions reproduced with upstream linux kernel of
> version 3.15 and
> the latest xfsprogs/xfstests. Are these known issues?

# ./check xfs/050 xfs/134 xfs/187 xfs/244 xfs/287
FSTYP         -- xfs (non-debug)
PLATFORM      -- Linux/x86_64 test2 3.18.0-rc2-dgc+
MKFS_OPTIONS  -- -f -bsize=4096 /dev/vdb
MOUNT_OPTIONS -- /dev/vdb /mnt/scratch

xfs/050 3s ... 4s
xfs/134 0s ... 0s
xfs/187 1s ... 1s
_check_xfs_filesystem: filesystem on /dev/vdb is inconsistent (c) (see /home/dave/src/xfstests-dev/results//xfs/187.full)
_check_xfs_filesystem: filesystem on /dev/vdb is inconsistent (r) (see /home/dave/src/xfstests-dev/results//xfs/187.full)
xfs/244 0s ... 1s
xfs/287 1s ... 0s
Ran: xfs/050 xfs/134 xfs/187 xfs/244 xfs/287
Passed all 5 tests

# cat  ~/results/xfs/187.full
....
_check_xfs_filesystem: filesystem on /dev/vdb is inconsistent
*** xfs_check output ***
sb_features2 (0x2) not same as sb_bad_features2 (0xa)
*** end xfs_check output
.....

The xfs/187 failure needs investigation - that's a bad_features2
mismatch, but it's been way down my list of priorities...

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs