Open Stack

Jeremy Stanley <fungi at yuggoth.org> wrote:

> On 2015-12-16 13:59:55 -0700 (-0700), Carl Baldwin wrote:
>> Is someone from Neutron actively helping out here?  Need more?
> [...]
>
> I believe all of the jobs currently voting on changes proposed to
> the master branch of the openstack/neutron repo are using centrally
> constrained requirements (they all either have "constraints" or
> "dsvm" in their job names).
>
> I'll let Robert and Sachi, who have been spearheading this effort up
> to this point, comment on whether additional assistance is needed
> and where they see next steps leading.

I will update on Neutron side of things since it seems I have some  
knowledge from the fields.

So first, on resources: we have Sachi and lifeless from infra side + Ihar  
and pc_m looking into expanding constrained jobs in Neutron world.

Current status is:
- all devstack based jobs are already constrained thanks to infra;
- openstack/neutron pep8/unit/doc/cover jobs are already constrained in  
Liberty and master;
- for *aas repos, pc_m is currently on top of it, following up on my  
original patches to introduce tox targets and make them voting in master  
gate; we also look into backporting targets and gate setup into Liberty.

Note that the new alembic release still broke openstack/neutron repo  
(functional job). This is because even though the job uses devstack to  
bootstrap the environment, it still calls tox to start tests, which makes  
the new venv with no constraints applied prepared and used. Same problem  
probably affects fullstack and api jobs since they use similar setup  
approach.

So I have closing the remaining loop hole on my personal agenda; but note  
that the agenda is really packed pretty much all the time, hence no time  
guarantees; so if someone has cycles to help making the neutron gate more  
fault proof, don’t hesitate to volunteer, I am extremely happy to help with  
it.

Ihar