[openstack-dev] [Tempest][qa] : backporting "Test Security Groups Basic Ops" in stable/havana ?

Matthew Treinish mtreinish at kortar.org
Mon Apr 14 15:26:35 UTC 2014


On Mon, Apr 14, 2014 at 07:48:57AM -0400, Yair Fried wrote:
> Hi Julien
> I've tried to backport it in the past, and it was rejected, as it needs quite a lot of work.

The specific case you referenced I -2'd because of the all the required changes
just to make that test work were far too large and risky when weighed against
the value of the test being added. We have accepted other test backports before
and after that were more self contained. This honestly would probably be the
case for most neutron tests that were trying to be backported to Havana. The
state of neutron testing during Havana wasn't very good and we've made tons of
progress this cycle on improving that.

> 
> If you are able to convince everyone otherwise, I will gladly help you with the code.
> 
> However, I think the better solution would be to somehow create a gate that runs the scenarios in the Master branch against Havana, because, AFAIK, most those scenarios should work against it.

There is ongoing work on doing this moving forward see:
http://lists.openstack.org/pipermail/openstack-dev/2014-April/032588.html

> 
> ----- Original Message -----
> > From: "LELOUP Julien" <Julien.LELOUP at 3ds.com>
> > To: openstack-dev at lists.openstack.org
> > Sent: Wednesday, April 2, 2014 12:48:31 PM
> > Subject: [openstack-dev] [Tempest][qa] : backporting "Test Security Groups Basic Ops" in stable/havana ?
> > 
> > Hi everyone,
> > 
> > I'm interested in having the Tempest scenario "Test Security Groups
> > Basic Ops" available in the stable/Havana branch .
> > This scenario is nice for acceptance tests and it's running fine with
> > an Havana deployment.
> > 
> > Can  someone in the Stable-maint team can backport it from master to
> > stable/Havana ? If it's OK for you of course :)

Yes, please go ahead and propose the backport. Adding additional tests is a
valid backport to tempest as long as it doesn't require making tons of changes
to the common code and is relatively self contained it should be fine. For each
proposed change we have to weigh the change vs the risk of regression especially
because we're not constantly watching the stable branches like we are on master.
Take a look at: 

https://wiki.openstack.org/wiki/StableBranch

-Matt Treinish



More information about the OpenStack-dev mailing list