[OpenStack-Infra] Issues starting centos7 job

Jeremy Stanley fungi at yuggoth.org
Thu Jul 31 12:25:28 UTC 2014


On 2014-07-31 15:34:07 +1000 (+1000), Ian Wienand wrote:
[...]
> See review [2] where I'm seeing
> 
>   check-tempest-dsvm-centos7 NOT_REGISTERED (non-voting)

Definitely odd, since before you retried I was seeing it registered
in zuul:

    fungi at zuul:~$ echo status|nc -q 3 localhost 4730|grep centos7
    build:check-tempest-dsvm-centos7:devstack-centos7   0   0   2
    build:check-tempest-dsvm-centos7    0   0   2

> I can't really see from the zuul web-page where it tried to send
> this job, but from discussions jenkins2 & jenkins4 have centos7
> slaves, and I'm presuming it got sent somewhere other than that.

It's up to a worker who registered the job to pick it up from the
queue, zuul doesn't assign it to a specific worker. But in this case
zuul's reporting that it didn't see the job registered at all. I'll
check the logs, but I suspect this is another case of our worker
filter parameter function for the trusty transition causing it to
look for check-tempest-dsvm-centos7:devstack-trusty instead. If so,
(relatively) simple zuul configuration change to correct it.

> Any tips on debugging this? Is it possible zuul is not taking
> account that not all jenkins hosts have the right nodes to run
> this job?

I think you've taken it as far as you can without access to zuul's
debug logs (or setting up your own replica of our infrastructure).
I'll follow up shortly with details and hopefully a fix.
-- 
Jeremy Stanley



More information about the OpenStack-Infra mailing list