[OpenStack-Infra] Re-using OpenStack CI - Stuck with Jenkins job trigger

Noorul Islam K M noorul at noorul.com
Sun Mar 13 00:08:25 UTC 2016


Noorul Islam K M <noorul at noorul.com> writes:

> Jeremy Stanley <fungi at yuggoth.org> writes:
>
>> On 2016-03-13 00:32:35 +0530 (+0530), Noorul Islam K M wrote:
>> [...]
>>
>>> On zuul node, when I start zuul using service command, it is not working
>>> even though the process there is running process. I get Service
>>> Unavailable message in UI.
>>> 
>>>    $ sudo service zuul start
>> [...]
>>
>> Did you also start the zuul-merger service? Based on the debug log
>> it looks like you probably have, but just wanted to double check.
>>
>>> Zuul debug log [5] shows that when a patch is submitted check
>>> queue gets updated and also gearman job is submitted.
>> [...]
>>
>> I see:
>>
>>     2016-03-12 17:47:58,365 DEBUG zuul.MergeClient: Submitting job
>>         <gear.Job 0x7f0a00347050 handle: None name: merger:merge
>>         unique: dbda5985f9b24d4cb59b806fedfc77c4> ...
>>
>> And then:
>>
>>     2016-03-12 17:51:16,119 INFO zuul.MergeClient: Merge <gear.Job
>>         0x7f0a00347050 handle: H:10.29.82.214:1 name: merger:merge
>>         unique: dbda5985f9b24d4cb59b806fedfc77c4> complete, merged:
>>         False, updated: False, commit: None
>>
>> So I think something is wrong with the merger, but if that's the
>> case I don't know why the change isn't getting kicked back as unable
>> to merge. At any rate, until that's working it's not going to send
>> any work requests for the jobs for that change.
>>
>
> Indeed! Thanks a lot! It worked after running zuul-merger on same zuul
> node. The documentation
> (http://docs.openstack.org/infra/system-config/running-your-own.html)
> does not mention anything about merger. And it looks like merger puppet
> module assumes zuul-merger will be running on separate node.

I have three executors on the slave. But when I submit the patch, zuul
is only running one jenkins job [1] at a time.

Also when the first one fails, gearman made the slave offline, keeping
others still in queue [1]. No luck even if I bring back the slave online
manually.

Regards,
Noorul

[1] http://picpaste.com/Screen_Shot_2016-03-13_at_5.34.33_AM-GLVSDNJl.png



More information about the OpenStack-Infra mailing list