Open Stack

Friction between PTL's/projects and cross project work is not a new thing. It has been a major issue for OpenStack almost since inception.

There were some hurtles I had when trying to make any progress on cross project work that ultimately caused me to give up. (I'll make some generalizations here... not saying everyone has these issues)

* PTL's (and the whole project really) tend only to look at their corner of the OpenStack and not so much how their project fits into the greater picture.
* PTL's have been good about discussing cross project issues when directly raised, but push back that their project should be part of the solution. The other project is always the better place to solve it, or a new project should solve it. That needs to stop being the go to answer.
* PTL's don't go out and actively look where their project does not play so well in the grater OpenStack. They focus on solving their own problems. This leads to architecture that work for a project but fails for the overall OpenStack.

* The way code gets into a project mostly involves getting enough capital (reviews) on a given project to get attention when needed. But cross project work doesn't get you enough capital in enough specific projects to ever really progress. There needs to be a cross project review score or something that says, this dev is working on cross project goals and submitting work of that type. give attention to their pr's because their work is important too! PTL's could prioritize this kind of work.

So, I kind of like the idea that PTL's need to focus more on the whole, not just their own project, so that they can help lead in the right directions. Right now each project is going in its own directions without much thought to the whole.

I'm not trying to blame the folks I've interacted with in the PTL role. They have by and large  been very helpful. But the Role of PTL, as well as the TC, have not enabled significant progress on the cross project issues due to these gaps. Maybe part of the solution is tweaking the responsibilities of the PTL's to focus more on cross project issues.

Thanks,
Kevin
________________________________________
From: Thierry Carrez [thierry at openstack.org]
Sent: Tuesday, January 15, 2019 1:49 AM
To: openstack-discuss at lists.openstack.org
Subject: Re: [tc] [all] Please help verify the role of the TC

Chris Dent wrote:
> [...]
> Since there is a significant and friction creating division of power
> and leadership between the TC and PTLs,

I'm not sure I follow you there... the division of power is between
keeping an eye on the big picture and caring for OpenStack as a whole
(TC) vs. rubber-hits-the-road, being responsible for a specific set of
deliverables (PTL). The same individuals can care for both concerns, but
those are different tasks... I think the division is clear.

The only friction I've observed recently is when it comes to driving
cross-project work -- an area that TC members and affected PTLs care
about. We need more people driving that type of work, and as we've said
in other threads, TC members (as well as other respected members of our
community) are in a good position to help drive that work.

> [...] what would it be like if we
> required half or more of the TC be elected from PTLs? Then the
> "providing the technical leadership" aspect of the TC mission [3]
> would be vested with the people who also have some responsibility for
> executing on that leadership.
>
> That would be like something we had before, but now there are many
> more PTLs.

I don't think PTLs have any difficulty getting elected when they run, so
I'm not sure a provision that the TC must have reserved seats for PTLs
would have a significant impact, beyond complicating the election process.

In terms of increasing TC efficiency... As others said, being a PTL for
a large project is already a lot of work and that leaves little to do
"TC work". And if the goal is to get more TC members to drive
cross-project work, the main reason TC members don't drive (more)
cross-project work is generally that they don't have enough bandwidth to
do so. Mandating more PTLs to be TC members is unlikely to result in a
TC membership with more available cross-project work bandwidth...

I agree that it is important to have representation of classic developer
teams on the TC, but I feel like today's TC membership is a good mix
between horizontal teams and vertical teams, including deployment
concerns and adjacent communities perspective. We should definitely
continue to encourage TC candidacies from vertical/classic project
teams, but I don't think that should be reduced to PTLs, and I don't
think that should be reserved seats.

--
Thierry Carrez (ttx)