[Gluster-devel] How to handle spurious regression test error

[Zhengping Zhou]

    Thanks a lot !

2016-07-09 4:44 GMT+08:00 Niels de Vos <ndevos at redhat.com>:
> On Sat, Jul 09, 2016 at 03:37:09AM +0800, Zhengping Zhou wrote:
>> Hello all:
>>
>>     I have submitted some patch to glusterfs through gerrit , but there
>> always are spurious regression test errors. I am  wondering what to do when
>> there is no relevant error message around this patch  in  Jenkins console's
>> output. It seems I only have two choice.
>>      First is just waiting for someone to recheck the test in jeekins, the
>> other is rebase the patch. The second way will redo the all things, and I
>> think is a little noisy for the other developers.
>>      So is there any better choice when I encounter this situation.
>
> We try to fix the spurious errors, but that is a very time consuming
> work. If a certain test fails more often, we can blacklist it, but that
> is a last resort solution.
>
> Depending on the Jenkins job that fails, you can write these commands in
> a comment in the review:
>
>   recheck centos
>   recheck netbsd
>   recheck smoke
>
> The Jenkins jobs catch these comments by regular expression. You can
> write them on a signle line, or on multiple lines.
>
> HTH,
> Niels