[Gluster-devel] intermittent test failure: tests/basic/afr/sparse-file-self-heal.t

Ravishankar N ravishankar at redhat.com
Wed Dec 9 13:10:21 UTC 2015


I'm able to repro the issue (i.e Failed test #36 of 
sparse-file-self-heal.t) on my ancient rhs-2.1 VM but not on newer 
Fedora 21 machines:

Create a 1x2 replica and from the mount, do : `dd if=/dev/zero of=file 
bs=1024 count=1024`
When both bricks are up, `du /brick/file` = 1024
When one of the bricks is killed and the test repeated, `du /brick/file` 
= 1028

I have no idea why. The issue is reproducible on NFS and fuse mounts on 
the rhs-2.1 VM running  2.6.32 kernel, which is incidentally the same 
version running on slave29.cloud.gluster.org
While I try to figure out the issue, I am adding the test case to bad 
tests for the moment @ http://review.gluster.org/#/c/12925/ . Makes me 
wonder if we can upgrade the build machines to at least centos7 if not 
fedora. 2.6 is really an old kernel!

Thanks,
Ravi


On 12/09/2015 02:40 PM, Ravishankar N wrote:
> I'll take a look at this one.
> -Ravi
>
> On 12/09/2015 01:49 PM, Michael Adam wrote:
>> Another one:
>>
>> https://build.gluster.org/job/rackspace-regression-2GB-triggered/16601/consoleFull
>>
>> by
>>
>> http://review.gluster.org/#/c/12826/
>>
>> Cheers - Michael
>>
>>
>> _______________________________________________
>> Gluster-devel mailing list
>> Gluster-devel at gluster.org
>> http://www.gluster.org/mailman/listinfo/gluster-devel
>
>
> -- 
> Ravishankar N
> work: +91 80 3924 5143
> extension: 8373143
> mobile: +91 96118 43905
> irc nick: itisravi
>
>
> _______________________________________________
> Gluster-devel mailing list
> Gluster-devel at gluster.org
> http://www.gluster.org/mailman/listinfo/gluster-devel


-- 
Ravishankar N
work: +91 80 3924 5143
extension: 8373143
mobile: +91 96118 43905
irc nick: itisravi

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gluster.org/pipermail/gluster-devel/attachments/20151209/3e9d54ee/attachment.html>


More information about the Gluster-devel mailing list