[openstack-dev] [all] [infra] check-osc-plugins job breaks clients

Jeremy Stanley fungi at yuggoth.org
Wed Aug 17 15:57:40 UTC 2016

On 2016-08-17 12:16:57 +0200 (+0200), Dmitry Tantsur wrote:
> This is probably a known problem, just bringing the attention to it: the
> check-osc-plugins is pretty unstable right now. It feels like on more than a
> half of runs it times out. For example, ironicclient: http://logs.openstack.org/91/328191/10/check/check-osc-plugins/f85d602/console.html,
> tripleoclient: http://logs.openstack.org/94/332694/7/check/check-osc-plugins/e04fdec/console.html,
> openstackclient: http://logs.openstack.org/46/355746/1/check/check-osc-plugins/23c05df/console.html.

Both of those seem to have run in OSIC and exceeded the 30-minute
timeout configured for the job.

> Is there any ideas how to fix it? We're approaching the client release
> deadline in two weeks, so inability to land client changes is pretty
> annoying.

Earlier today we temporarily removed OSIC from our CI because this
seemed to be happening frequently on Ubuntu Trusty nodes there.
Kevin Carter (one of the operators for that environment) determined
that at least one of the symptoms we were seeing that could be
connected to this problem seemed to be fixed by disabling IPv6
privacy extensions. We've got a change merged to do that in the base
images currently building, and will reenable running jobs in OSIC
once it has a fresh compliment of images uploaded successfully.

Sorry this impacted everyone for as long as it did, but we were
trying to weigh the merits of reduced throughput from random job
failures on Ubuntu Trusty vs. reduced throughput from turning off
~1/3 of our test capacity.
Jeremy Stanley

More information about the OpenStack-dev mailing list