On Sat, 2024-09-21 at 15:20 -0400, Goutham Pacha Ravi wrote:
> Hello Stackers,
>
> The "watcher" project [1] did not have any volunteers to lead the
> project in the 2025.1 election cycle. In the past few years, the
> project's team has slowly disbanded and there have been no significant
> changes proposed/merged in the project's repositories [2]. We noticed
> that there were three deployers that noted their use of Watcher in
> production in the past user survey [3]. It's possible that the project
> failed to gather any significant adoption, and hence there's less
> interest in contributing to, and maintaining the project.
>
> I'm sending this notification pursuant to the OpenStack governance
> recommendations [4] to retire project teams. We can mark a project
> team "inactive", however, an "inactive" project must have some
> contributors aiming to build-back a maintainers team around it - and
> it must pick a project team lead, or adopt a distributed project
> leadership model by nominating liaisons. Since there are no PTL
> nominees, we think marking this project "inactive" serves no purpose.
im on pto for the next week but when i come back i will be moving to a new
team that will still be working on openstack looking at some new usecases.
One of the projects we were evaluating was watcher to address those is watcher.
i may be willing to take on the PTL position or form a distirbuted team
but i need to actually talk to my new team about if that is the direction
we want to take. Watcher has a lot of the infrastructure that would be useful
for workload management and sla enforcement but much of the knowledge of how
the project works has been lost. So one of the questions we need to reflect
on is does the benefit of the exisitng infrastrucrue out weigh the technial
debt. e.g. start from scratch adding only the functionaltiy we need in a new project
vs revive watcher and address thing like removing eventlet ectra.
If marking it inactive but not strating the requirement is an option
that would give us time over the 2025.1 cycle
if we were to pursue using watcher obviously having maintainers
of it would be something we would want to adresss this cycle.
if we end up deciding that watcher is not something we can commit
to investing in or no one else steps forward then we could retire
it at the end of the cycle instead of the start.
>
> Is there any objection to retiring the "openstack/watcher" project
> (and associated deliverables: python-watcherclient, watcher-dashboard
> and watcher-tempest-plugin)? Once the project is retired,
> configuration management and deployment projects can drop their
> integrations as well.
>
> Thanks,
> On behalf of the OpenStack TC,
> Goutham Pacha Ravi
>
> [1]
https://wiki.openstack.org/wiki/Watcher
> [2]
https://review.opendev.org/q/(project:%5Eopenstack/watcher.*+OR+project:openstack/python-watcherclient)
> [3]
https://governance.openstack.org/tc/user_survey/analysis-2023.html
> [4]
https://governance.openstack.org/tc/reference/dropping-projects.html
>