On Fri, Jul 7, 2017 at 12:33 PM, Krutika Dhananjay <[email protected]>
wrote:

> The patch[1] that introduced tests/basic/stats-dump.t was merged in
> October 2015 and
> my patch underwent (and passed too![2]) centos regression tests, including
> stats-dump.t on 05 June, 2017.
> The only change that the test script underwent during this time was this
> line in 2016, which is harmless:
>
> a4f84d78 (Kaleb S KEITHLEY 2016-03-15 06:16:31 -0400 17) EXPECT_WITHIN
> $NFS_EXPORT_TIMEOUT "1" is_nfs_export_available
>
> So there was NO change that went between 5th June and the time my patch
> was merged, which could have broken the test suite, that could have been
> caught easily with a mere rebase? Or am I missing something here? The
> problem is simply that the test script didn't fail on my patch.
>

It's not the test but probably something in the code which has gone in to
make this test fail now. So no way it's something the developer/maintainer
has to take a blame for.


>
>
> [1] - https://review.gluster.org/12209
> [2] - https://build.gluster.org/job/centos6-regression/4897/consoleFull
>
> -Krutika
>
> On Fri, Jul 7, 2017 at 9:51 AM, Atin Mukherjee <[email protected]>
> wrote:
>
>> Krutika,
>>
>> tests/basis/stats-dump.t is failing all the time and as per my initial
>> analysis after https://review.gluster.org/#/c/17709/ got into the
>> mainline the failures are seen and reverting this patch makes the test to
>> run successfully. I do understand that the centos vote for this patch was
>> green but the last run was on 5th June which was 1 month back. So some
>> other changes have gone into in between which is now causing this patch to
>> break the test.
>>
>> This makes me think as a maintainer we do need to ensure the if the
>> regression vote on the patch is quite old, a rebase of the patch is must to
>> be on the safer side?
>>
>> ~Atin
>>
>
>
_______________________________________________
Gluster-devel mailing list
[email protected]
http://lists.gluster.org/mailman/listinfo/gluster-devel

Reply via email to