Failure of xfstest 229

Failure of xfstest 229

[Chandra Seetharaman]

Hi all,

I remember running this test successfully sometime back. 

But, now it is failing on different versions of kernel that I try it on,
like 2.6.34,35,36,37,38 and 39. with the following errors (at the end).

I tried it in 2 filesystem sizes (1TB and 20GB), both of which fail.

I tried to do what the test does externally and found that running
xfs_io -c "extsize ${EXTSIZE}" ${TDIR}

before running 
$here/src/t_holes ${TDIR}/${i}

creates the failures.

Is there any special setup needed for it to run successfully ?

Thanks & Regards,

chandra
---------------------------------------------

229 118s ... [failed, exit status 45] - output mismatch (see 229.out.bad)
--- 229.out	2011-04-20 08:34:36.000000000 -0700
+++ 229.out.bad	2011-05-19 14:51:48.000000000 -0700
@@ -1,4 +1,49 @@
 QA output created by 229
 generating 10 files
 comparing files
-got 0 errors
+/mnt/xfsMntPt//t_holes/0 /mnt/xfsMntPt//t_holes/1 differ: char 282193, line 1
+/mnt/xfsMntPt//t_holes/0 /mnt/xfsMntPt//t_holes/2 differ: char 277873, line 1
+/mnt/xfsMntPt//t_holes/0 /mnt/xfsMntPt//t_holes/3 differ: char 264289, line 1
+/mnt/xfsMntPt//t_holes/0 /mnt/xfsMntPt//t_holes/4 differ: char 282193, line 1
+/mnt/xfsMntPt//t_holes/0 /mnt/xfsMntPt//t_holes/5 differ: char 282193, line 1
+/mnt/xfsMntPt//t_holes/0 /mnt/xfsMntPt//t_holes/6 differ: char 282193, line 1
+/mnt/xfsMntPt//t_holes/0 /mnt/xfsMntPt//t_holes/7 differ: char 282193, line 1
+/mnt/xfsMntPt//t_holes/0 /mnt/xfsMntPt//t_holes/8 differ: char 282193, line 1
+/mnt/xfsMntPt//t_holes/0 /mnt/xfsMntPt//t_holes/9 differ: char 265361, line 1
+/mnt/xfsMntPt//t_holes/1 /mnt/xfsMntPt//t_holes/2 differ: char 277873, line 1
+/mnt/xfsMntPt//t_holes/1 /mnt/xfsMntPt//t_holes/3 differ: char 264289, line 1
+/mnt/xfsMntPt//t_holes/1 /mnt/xfsMntPt//t_holes/4 differ: char 415617, line 1
+/mnt/xfsMntPt//t_holes/1 /mnt/xfsMntPt//t_holes/5 differ: char 304801, line 1
+/mnt/xfsMntPt//t_holes/1 /mnt/xfsMntPt//t_holes/6 differ: char 415617, line 1
+/mnt/xfsMntPt//t_holes/1 /mnt/xfsMntPt//t_holes/7 differ: char 415617, line 1
+/mnt/xfsMntPt//t_holes/1 /mnt/xfsMntPt//t_holes/8 differ: char 402977, line 1
+/mnt/xfsMntPt//t_holes/1 /mnt/xfsMntPt//t_holes/9 differ: char 265361, line 1
+/mnt/xfsMntPt//t_holes/2 /mnt/xfsMntPt//t_holes/3 differ: char 264289, line 1
+/mnt/xfsMntPt//t_holes/2 /mnt/xfsMntPt//t_holes/4 differ: char 277873, line 1
+/mnt/xfsMntPt//t_holes/2 /mnt/xfsMntPt//t_holes/5 differ: char 277873, line 1
+/mnt/xfsMntPt//t_holes/2 /mnt/xfsMntPt//t_holes/6 differ: char 277873, line 1
+/mnt/xfsMntPt//t_holes/2 /mnt/xfsMntPt//t_holes/7 differ: char 277873, line 1
+/mnt/xfsMntPt//t_holes/2 /mnt/xfsMntPt//t_holes/8 differ: char 277873, line 1
+/mnt/xfsMntPt//t_holes/2 /mnt/xfsMntPt//t_holes/9 differ: char 265361, line 1
+/mnt/xfsMntPt//t_holes/3 /mnt/xfsMntPt//t_holes/4 differ: char 264289, line 1
+/mnt/xfsMntPt//t_holes/3 /mnt/xfsMntPt//t_holes/5 differ: char 264289, line 1
+/mnt/xfsMntPt//t_holes/3 /mnt/xfsMntPt//t_holes/6 differ: char 264289, line 1
+/mnt/xfsMntPt//t_holes/3 /mnt/xfsMntPt//t_holes/7 differ: char 264289, line 1
+/mnt/xfsMntPt//t_holes/3 /mnt/xfsMntPt//t_holes/8 differ: char 264289, line 1
+/mnt/xfsMntPt//t_holes/3 /mnt/xfsMntPt//t_holes/9 differ: char 264289, line 1
+/mnt/xfsMntPt//t_holes/4 /mnt/xfsMntPt//t_holes/5 differ: char 304801, line 1
+/mnt/xfsMntPt//t_holes/4 /mnt/xfsMntPt//t_holes/6 differ: char 416689, line 1
+/mnt/xfsMntPt//t_holes/4 /mnt/xfsMntPt//t_holes/7 differ: char 474353, line 1
+/mnt/xfsMntPt//t_holes/4 /mnt/xfsMntPt//t_holes/8 differ: char 402977, line 1
+/mnt/xfsMntPt//t_holes/4 /mnt/xfsMntPt//t_holes/9 differ: char 265361, line 1
+/mnt/xfsMntPt//t_holes/5 /mnt/xfsMntPt//t_holes/6 differ: char 304801, line 1
+/mnt/xfsMntPt//t_holes/5 /mnt/xfsMntPt//t_holes/7 differ: char 304801, line 1
+/mnt/xfsMntPt//t_holes/5 /mnt/xfsMntPt//t_holes/8 differ: char 304801, line 1
+/mnt/xfsMntPt//t_holes/5 /mnt/xfsMntPt//t_holes/9 differ: char 265361, line 1
+/mnt/xfsMntPt//t_holes/6 /mnt/xfsMntPt//t_holes/7 differ: char 416689, line 1
+/mnt/xfsMntPt//t_holes/6 /mnt/xfsMntPt//t_holes/8 differ: char 402977, line 1
+/mnt/xfsMntPt//t_holes/6 /mnt/xfsMntPt//t_holes/9 differ: char 265361, line 1
+/mnt/xfsMntPt//t_holes/7 /mnt/xfsMntPt//t_holes/8 differ: char 402977, line 1
+/mnt/xfsMntPt//t_holes/7 /mnt/xfsMntPt//t_holes/9 differ: char 265361, line 1
+/mnt/xfsMntPt//t_holes/8 /mnt/xfsMntPt//t_holes/9 differ: char 265361, line 1
+got 45 errors
Ran: 229
Failures: 229
Failed 1 of 1 tests
---------------------------

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

Re: Failure of xfstest 229

[Dave Chinner]

On Thu, May 19, 2011 at 05:55:37PM -0700, Chandra Seetharaman wrote:
> Hi all,
> 
> I remember running this test successfully sometime back. 

No. it's never passed. It's a placeholder to remind us how to
reproduce a known problem that is difficult to fix.

Cheers,

Dave.

-- 
Dave Chinner
david@fromorbit.com

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

Re: Failure of xfstest 229

[Chandra Seetharaman]

Can it run successfully by fluke sometimes ? I did have a successful run
on the 20G filesystem yesterday once, but not able to reproduce it again
---------------
FSTYP         -- xfs (non-debug)
PLATFORM      -- Linux/x86_64 test201 2.6.39
MKFS_OPTIONS  -- -f -bsize=4096 /dev/sdd2
MOUNT_OPTIONS -- -o context=system_u:object_r:nfs_t:s0 /dev/sdd2
/mnt/xfsScratchMntPt

229 70s ... 118s
Ran: 229
Passed all 1 tests
-----------------

chandra
On Fri, 2011-05-20 at 12:58 +1000, Dave Chinner wrote:
> On Thu, May 19, 2011 at 05:55:37PM -0700, Chandra Seetharaman wrote:
> > Hi all,
> > 
> > I remember running this test successfully sometime back. 
> 
> No. it's never passed. It's a placeholder to remind us how to
> reproduce a known problem that is difficult to fix.
> 
> Cheers,
> 
> Dave.
> 


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

Re: Failure of xfstest 229

[Christoph Hellwig]

On Fri, May 20, 2011 at 09:35:03AM -0700, Chandra Seetharaman wrote:
> 
> Can it run successfully by fluke sometimes ? I did have a successful run
> on the 20G filesystem yesterday once, but not able to reproduce it again

It doesn't fail on thinly provisioned storage which zeroes out any
blocks it gives to the OS.  Similarly it might not fail if all blocks
that it would normally leak are zeroed underneath for some reason,
but it's rather unlikely that this happens by accident.

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

Re: Failure of xfstest 229

[Chandra Seetharaman]

It is strange then.

It is the same system, and it failed few hours after it succeeded in
2.6.39.

Basically, I saw seeing the failure with 2TB fs. I thought I will try it
with a 20GB fs, and the very first time it succeeded, but after that it
never did even with the same kernel (2.6.39).

On Fri, 2011-05-20 at 13:13 -0400, Christoph Hellwig wrote:
> On Fri, May 20, 2011 at 09:35:03AM -0700, Chandra Seetharaman wrote:
> > 
> > Can it run successfully by fluke sometimes ? I did have a successful run
> > on the 20G filesystem yesterday once, but not able to reproduce it again
> 
> It doesn't fail on thinly provisioned storage which zeroes out any
> blocks it gives to the OS.  Similarly it might not fail if all blocks
> that it would normally leak are zeroed underneath for some reason,
> but it's rather unlikely that this happens by accident.
> 
> _______________________________________________
> xfs mailing list
> xfs@oss.sgi.com
> http://oss.sgi.com/mailman/listinfo/xfs


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