[openstack-dev] [nova] gate is fubar (again)

Matt Riedemann mriedem at linux.vnet.ibm.com
Fri Feb 20 15:51:02 UTC 2015



On 2/20/2015 12:08 AM, melanie witt wrote:
> On Feb 19, 2015, at 17:54, Matt Riedemann <mriedem at linux.vnet.ibm.com> wrote:
>
>> Regarding the novaclient change, the plan is to get either the devstack cells exercise job voting on novaclient changes again (which it was at some point prior to the 2.21 release and should have caught this problem), or get a functional test in novaclient that hits the bash completion problem - then we can revert the revert and make that work with the current code on master.
>
> I have put up a review to reinstate the devstack cells exercise job on novaclient: https://review.openstack.org/#/c/157661/
>
> And I think jogo is already working on functional tests that would cover this, if not the bash completion then the volume-attach functionality (technically the volume-attach succeeds but the call made by the bash completion cache write fails).
>
>> I might have missed something but figured it'd be good to get this out so people are aware that things are busted, again.
>
> Thanks for writing this up.
>
> melanie (melwitt)
>
>
>
>
>
>
> __________________________________________________________________________
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>

sdague pointed out that the devstack cells job, even if running on 
python-novaclient again, will not test the changes in novaclient, it 
will still run against the latest released novaclient version on pypi, 
so that won't help.

We need the functional tests that will recreate the problem, then 
probably replay https://review.openstack.org/#/c/157591/ on top of that 
until we can see that it's passing with the revert.

Given the gate on master is OK again and we have some time to fix this, 
we should probably hold off on https://review.openstack.org/#/c/157591/ 
until we have a test in place that hits the problem and verifies that 
https://review.openstack.org/#/c/157591/ fixes it.

The other stable/icehouse stuff is being unwedged also this morning, so 
there is a path forward there it looks like with some workarounds from 
the stable maint team (Ihar and Adam G).

-- 

Thanks,

Matt Riedemann




More information about the OpenStack-dev mailing list