[openstack-dev] [tc] Re: [kolla] A new kolla-salt deliverable
Michał Jastrzębski
inc007 at gmail.com
Wed Dec 28 19:12:23 UTC 2016
I agree this would make good PTG discussion for us, and maybe someone
from TC could join in. We are somehow special beast as Kolla itself is
just meant to be consumed. Kolla to me is similar to Oslo in that
matter. But still we don't have oslo-compute projects, we have nova,
but we would have kolla-salt, even tho core teams would be totally
different potentially. As Britt said, growing pains.
Cheers,
Michal
On 24 December 2016 at 07:36, Britt Houser (bhouser) <bhouser at cisco.com> wrote:
> Seems like these are some of the same growing pains (cores can’t be experts on all technologies) neutron went through. Maybe at the PTG we could pick their brain and see if the path they have chosen would work well for Kolla.
>
> Thx,
> britt
>
> On 12/24/16, 10:31 AM, "Steven Dake (stdake)" <stdake at cisco.com> wrote:
>
> My response wasn’t clear and I’ve also thought more about your proposal. I’d be highly in favor of the approach you mentioned as long as 2 was modified in your proposal to include some larger number then 2 individuals. One option that comes to mind is a majority of each core review sub-team for point 2 taking into account some of our core reviewers have issues that temporarily prevent them from fulfilling their core reviewer duties (although they plan to re-engage).
>
> I agree we don’t want to duplicate the TC – that would be super heavy – not that the TC is heavy – but rather that Kolla doesn’t need its own governance structure as the technical governance of OpenStack is directed by the technical committee. I for one, don’t want to have a full-blown governance structure for Kolla, although I can’t speak for other core reviewers.
>
> Regards
> -steve
>
>
> -----Original Message-----
> From: "Steven Dake (stdake)" <stdake at cisco.com>
> Date: Friday, December 23, 2016 at 3:53 PM
> To: "OpenStack Development Mailing List (not for usage questions)" <openstack-dev at lists.openstack.org>
> Subject: Re: [openstack-dev] [tc] Re: [kolla] A new kolla-salt deliverable
>
> WFM – I’d be partly in favor of such an approach although I can’t speak for others. I think we should require some larger set then 2 individuals from the kolla-policy-core; perhaps a majority of active reviewers for some definition of active reviewers.
>
> Regards
> -steve
>
>
> -----Original Message-----
> From: Michał Jastrzębski <inc007 at gmail.com>
> Reply-To: "OpenStack Development Mailing List (not for usage questions)" <openstack-dev at lists.openstack.org>
> Date: Friday, December 23, 2016 at 3:38 PM
> To: "OpenStack Development Mailing List (not for usage questions)" <openstack-dev at lists.openstack.org>
> Subject: Re: [openstack-dev] [tc] Re: [kolla] A new kolla-salt deliverable
>
> So I agree with you that we need policy established here. What I'm
> getting at - which core teams will vote on inclusion of new
> deliverable? All of them? Just Kolla? This is grey area I'm referring
> to. What's kolla-k8s core team's business if somebody would like to
> add saltstack support? What I wouldn't want to have is to establish
> new semi-tc in form of our core team that will decide what is and
> isn't good orchiestration engine for Kolla. That would seriously
> hinder our ability to innovate, experiment. What if we find out this
> new orchiestration engine and just want to play with it? But keep it
> community from start?
>
> So let me throw an idea there, one which we should vote on:
>
> Prep:
> 1. We create kolla-policy-core-team which is sum of all core teams of
> kolla supported projects
> 2. We create list of kolla supported projects - today it's kolla,
> kolla-ansible and kolla-k8s
>
> Add new project:
> 1. Everyone is free to create kolla-* deliverable as long as they
> follow certain documented standards (action: document these)
> 2. We create lightweight process to include new deliverable to Kolla,
> like just 2* +2 from kolla-policy-core-team to include project like
> that
> 3. If project gets traction, interest and is successful, we vote on
> including it's core team to kolla-policy-core-team
>
> This way it would be easy to try and fail fast to run kolla with
> whatever. We need this kind of flexibility for people to innovate.
>
> Thoughts?
> Michal
>
> On 23 December 2016 at 13:11, Steven Dake (stdake) <stdake at cisco.com> wrote:
> > Michal,
> >
> > Really what I was getting at was placing in the governance repository as a kolla deliverable. In days past we *always* voted on additions and removals of deliverables. It doesn’t seem like a gray area to me – we have always followed a voting pattern for adding and removal of deliverables. This repo could be added to the git openstack namespace but then not have it as a kolla deliverable without a vote I think; this is sort of what Fuel did with Fuel-ccp – that proposal is a gray area. I found when Fuel did that to be extremely odd personally ☺ I’m not sure if there is a trademark policy or something similar that affects the use of Kolla and OpenStack together. I’ve included the [tc] in the topic so they can provide guidance on the route you suggested (incubation for new kolla deliverables that are not actually deliverables).
> >
> > I think we really don’t need the tc to intervene here though, we can just make new policies on our own via the typical policy voting process we have followed in the past. Before we make any decisions about that though, I think we need a vote on the topic. ☺
> >
> > Regards
> > -steve
> >
> > -----Original Message-----
> > From: Michał Jastrzębski <inc007 at gmail.com>
> > Reply-To: "OpenStack Development Mailing List (not for usage questions)" <openstack-dev at lists.openstack.org>
> > Date: Friday, December 23, 2016 at 10:08 AM
> > To: "OpenStack Development Mailing List (not for usage questions)" <openstack-dev at lists.openstack.org>
> > Subject: Re: [openstack-dev] [kolla] A new kolla-salt deliverable
> >
> > Hello,
> >
> > Ok this is grey area we haven't had proper discussion yet, I agree.
> > Since we decided to have separate core teams, I personally don't
> > really see *why* we should have any form of vote for projects to use
> > kolla containers.
> > Things change when we talk about it being kolla deliverable, but what
> > exactly does that mean? They can use kolla name? Everybody can. They
> > follow Kolla policies, use Kolla irc channel and have Kolla PTL to
> > represent them? That's also a choice everybody can make.
> >
> > In the spirit of inclusiveness, I'd say keep it free and open. I would
> > rather have people use kolla name, be open about using kolla
> > containers and be part of our community. Maybe some sort of "free to
> > add, but incubate" would be in order, but I personally think that
> > would be overkill at this stage.
> >
> > Thoughts?
> >
> > Cheers Michal
> >
> > On 23 December 2016 at 05:34, Steven Dake (stdake) <stdake at cisco.com> wrote:
> > > Michal,
> > >
> > >
> > >
> > > I was thinking about kolla-salt and our Wednesday team meeting and the
> > > declaration you made about how it should be done. I personally feel it is
> > > mandatory we hold a vote of the core review teams to add a new deliverable.
> > > We have voted on the addition of every deliverable we have ever added to
> > > kolla including application initially to the big tent. I’m in favor of the
> > > idea of kolla-salt and it would have my +1 vote. I am not attempting to
> > > block the addition. It’s more a matter of policies we have established over
> > > the last several years. We have also voted to retire deliverables from the
> > > Kolla project as well (kolla-mesos and the cli).
> > >
> > >
> > >
> > > This was easier when there was one core review team. Perhaps a solution to
> > > that problem is to make a global core team in gerrit which includes everyone
> > > just for policy decisions (such as adding a deliverable). Another option to
> > > count whether consensus was reached is to count the core reviewers in each
> > > deliverable, divide by two, and determine if consensus is reached.
> > >
> > >
> > >
> > > If we don’t hold a vote, it looks like a BDFL model that PTLs don’t operate
> > > under. Rather PTLs operate under a service model.
> > >
> > >
> > >
> > > Regards
> > >
> > > -steve
> > >
> > >
> > >
> > >
> > > __________________________________________________________________________
> > > 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
> > >
> >
> > __________________________________________________________________________
> > 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
> >
> >
> > __________________________________________________________________________
> > 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
>
> __________________________________________________________________________
> 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
>
>
>
>
> __________________________________________________________________________
> 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
>
>
> __________________________________________________________________________
> 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
More information about the OpenStack-dev
mailing list