Dear release team,

tl;dr: I'd like to push OpenStack Epoxy (aka: 2025.1) into Trixie, but I need the release team approval before uploading to Unstable.

0/ Release names

OpenStack, like Debian and Ubuntu, is using named releases. If you get lost, just read this page:
https://releases.openstack.org/

So, the current release in Unstable/Testing is Dalmatian. The one I'm currently working on is Epoxy. The OpenStack release in Bookworm is Zed.
etc.

1/ Topic

As the freeze dates have shifted 1.5 month compared to the last 2 releases, this gives me the opportunity to have enough time to polish one more OpenStack release and get it into the next Debian. There are several reasons why I would like to do that, and I'd like to discuss this with the release team.

2/ Current upstream OpenStack Schedule & state in Debian

The schedule for the OpenStack release is here:
https://releases.openstack.org/epoxy/schedule.html

This means that Epoxy final release will be, at most, on the 4th of April. However, there's usually very little difference between the RC releases, which I have already packaged (all is in Experimental).

3/ Functional testing

Before declaring an OpenStack release validated, I always run functional testing. This is using the upstream "tempest" project, that helps me to run more than 1500 functional tests, including spawning networks, VMs, objects, block storage, VM images, etc. With it, I am always making sure everything works together in a nice way.

4/ Current state of packaging

At this point, all of Epoxy is packaged and backported to Bookworm (automatically, on the unofficial debian.net repository). I'm at the start of functional testing. This usually takes me about a week to pass, so I feel like on schedule. If everything goes well, I could be ready to upload Epoxy RC1 to Unstable next week, so that it'd be ready before the official release. When upstream releases it, with usually almost no change, I'll do these few uploads that will mainly be changing version numbers to remove "~rc1".

5/ Experience from Ubuntu team

They also have freeze exceptions from the Ubuntu release team so that they can push OpenStack later than the rest of Ubuntu. I believe it worked well for them.

6/ SLURP (skipable OpenStack releases)

Every 2 OpenStack release is skipable. That's nice, but Debian is miss-alligned. This means that if Dalmatian is going into Trixie, Bookworm users will only be able to skip Bobcat:

Zed (bookworm)
  -> Antelope
    -> Caracal
      -> Dalmatian (bookworm)
        -> Dalmatian (Trixie)

If I get Epoxy in Trixie, Debian users will be able to skip both Bobcat and Dalmatian, doing:

Zed (bookworm)
  -> Antelope
    -> Caracal
      -> Epoxy (bookworm)
        -> Epoxy (Trixie)

This means that with the same amount of efforts, Debian users will be able to upgrade one more OpenStack release.

Also, if Forky freeze is in schedule, there's going also to be one less upgrade to do, as OpenStack "F" and "H" will be skipable, and Forxky should be released with OpenStack "G", because Debian and OpenStack will be aligned.

7/ Limited security support upstream

OpenStack Zed (currently in Bookworm) is already unmaintained. If there's security issues, it's a pain to do backports. The closer to latest upstream release, the better, even if I often get upstream help to backport security patches to earlier releases.

8/ Nice feature I'd like to have

I'd love to have Manial (OpenStack file share as a service) support for viriofs. That's a huge improvement, without which I don't consider Manila production ready with HA. That's on top of other features I haven't discovered yet... :)

9/ Drawback to that plan

Of course, if I push Epoxy from Experimental to Unstable, in the hope to make it in time for Trixie, I'll have to ask for a few more unblocks, meaning probably a little more work from the release team. Though IMO, that's for the best. Given at least a month or 2 of freeze, I'm confident I can make it to Trixie with OpenStack in good enough shape.

10/ Point release

I'll try to do more OpenStack point release packaging, as this tend to remove a lot of bugs. Hopefully, the stable release team will agree.

Dear release team,

Please let me know what you think about this plan.

Cheers,

Thomas Goirand (zigo)

Reply via email to