[Gluster-devel] weighted-rebalance.t failure

[Shyam]

On 06/19/2015 09:33 AM, Shyam wrote:
> On 06/19/2015 12:35 AM, Atin Mukherjee wrote:
>> One of my patch failed NetBSD regression [1] in the test case mentioned
>> in $Subject.
>>
>> [1]
>> http://build.gluster.org/job/rackspace-netbsd7-regression-triggered/7025/consoleFull
>>
>
> The failure is as follows,
>
> [06:37:09] ./tests/features/weighted-rebalance.t ..
> not ok 6
> not ok 7
> not ok 9
> not ok 10
> not ok 20
>
> The steps failing are MKFS_LOOP and MOUNT_LOOP (6/7 and 9/10), as a
> result the backend brick are created equal and distribution is equal,
> which is why test 20 is failing to find an uneven distribution count.
>
> Gluster logs may not help resolving this at all, as the is even before
> creating the gluster volume, where the failure is noted.
>
> Digging through to see what else I can glean from other logs available
> in the failed test cases.

FWIW, this test failed thrice on NetBSD slave 7c, both manu and myself 
tried to reproduce the failure (on 7c) but it did not occur.

Currently out of ideas and no log messages available. If this test 
continues to fail for anyone, shout, and maybe we can add some runs with 
set +x to debug this further.

>
> @Manu, any thoughts? TC failing: tests/features/weighted-rebalance.t
>
> Shyam
> _______________________________________________
> Gluster-devel mailing list
> Gluster-devel at gluster.org
> http://www.gluster.org/mailman/listinfo/gluster-devel