[Openstack-operators] [Large Deployments Team] [nova] Cells v1 attach/detach blueprint

Matt Riedemann mriedem at linux.vnet.ibm.com
Thu Nov 12 17:35:07 UTC 2015

On 11/11/2015 4:12 PM, Matt Riedemann wrote:
> On 11/11/2015 11:28 AM, Andrew Laski wrote:
>> On 11/10/15 at 06:50pm, Matt Riedemann wrote:
>>> On 11/10/2015 4:42 PM, Mathieu Gagné wrote:
>>>> On 2015-11-10 5:18 PM, Matt Riedemann wrote:
>>>>> It should be noted that cells v1 doesn't support the os-server-events
>>>>> API which is key for stability with nova + neutron in gate testing,
>>>>> so I
>>>>> think that would be a barrier to cells + neutron as a voting job in
>>>>> nova
>>>>> (eventually).
>>>> I think Sam Morrison proposed a change [1] to support external instance
>>>> events. Is it was you are referring to?
>>>> [1] https://review.openstack.org/#/c/184155/
>>> Yeah the external instance events is what needs plumbing for a
>>> stable(ish) cells + neutron CI job.
>> I have updated and revived the patch at
>> https://review.openstack.org/#/c/184155/.
>>> --
>>> Thanks,
>>> Matt Riedemann
>>> _______________________________________________
>>> OpenStack-operators mailing list
>>> OpenStack-operators at lists.openstack.org
>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
> Here is the project-config change to add a cells + neutron job to nova's
> experimental queue:
> https://review.openstack.org/#/c/244330/

This came up in the nova meeting today [1] and it sounds like of these 
three options:

1) drop (existing) cells v1 job, and be clear its dead
2) add a few missing bits from cells v1 (like os-attach-interfaces support)
3) feature freeze v1 hard, but keep on top of regressions (where 
regressions does not mean latent bugs that are exposed by new tests, 
like bug 1489581).

Most people are in favor of option 3, which would mean we wouldn't 
approve adding support for os-attach-interfaces or 

The main issue is the distraction from working on cells v2, since there 
are a very limited number of people that work on cells so we have to 
prioritize their time.

I'm not exactly sure what this means for a cells + neutron job in the 
experimental queue. The reason I started working on that was in case it 
could be tested against the os-attach-interfaces change, but if that's 
blocked now then there might not be much of a point in a new job, unless 
it can be re-used later for the cells v2 effort. I'm guessing the job 
would just bitrot by the time that came up though, so maybe it's not 
worth it right now.




Matt Riedemann

More information about the OpenStack-operators mailing list