Friday, 2026-05-15

*** elodilles_pto is now known as elodilles08:49
*** haleyb is now known as haleyb|away12:38
elodillesreminder: meeting will start in ~20 mins13:39
cardoeelodilles: will you be pushing the hibiscus-1 releases today? 13:57
cardoegtema: hoping you can approve https://review.opendev.org/c/openstack/releases/+/988085 today.13:58
elodillescardoe: friday releases are always risky o:)13:59
elodillesbut if it is needed, then we can discuss that14:00
elodillesbut now, let's do the relmgt weekly meeting14:00
cardoeWell I was just wondering cause the calendar said the releases would be May 11th to May 15th14:00
elodilles#startmeeting releaseteam14:00
opendevmeetMeeting started Fri May 15 14:00:40 2026 UTC and is due to finish in 60 minutes.  The chair is elodilles. Information about MeetBot at http://wiki.debian.org/MeetBot.14:00
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.14:00
opendevmeetThe meeting name has been set to 'releaseteam'14:00
elodillesPing list: release-team14:00
elodilles#link https://etherpad.opendev.org/p/hibiscus-relmgt-tracking14:01
frickler\o14:01
fungiahoy!14:01
elodilleso/14:01
elodilleswe are at line 8114:01
elodillesalready at Hibiscus-1 milestone :-o14:01
elodillestime flies14:01
elodillesso should we14:02
elodilleslet's start with the agenda!14:02
elodilles#topic Review task completion14:02
elodilles- Ensure that all trailing projects have been branched for the previous series. (elod)14:02
elodillesso, i've pinged the PTLs last week,14:02
elodillesand they'd said that they are working on it, but they are not there yet14:03
elodillesKolla has branched14:03
elodillesbut kayobe, Puppet OpenStack and OpenStack Ansible are still haven't branched14:04
fungiearly this week we saw afs syncing the tarballs volume for ~36 hours straight, which i'm guessing was kolla's new images?14:05
elodillesyes, that is possible :)14:05
fungii'll try to remember to keep a closer eye on that next time14:05
elodillesthey their final release is also available, so that should be the reason ;)14:06
elodillesanyway, i'll add a move the task to next week, and will take it on my name14:07
elodilleslet's hope that rest of the cycle-trailing project teams are ready for branching next week14:07
elodilles- Propose autoreleases for cycle-with-intermediary libraries which did not release since the previous release. (elod)14:08
elodillesi've generated the release patches, where they had functional code changes:14:09
elodilleshttps://review.opendev.org/q/topic:hibiscus-milestone-114:09
cardoecan ya ping me during open discussion? (I'm on a work Zoom and tabbing back and forth)14:09
elodillescardoe: ACK, will do14:09
elodillessome of the patches merged after PTL/release liaison Code-Review+114:10
elodillesbut still have quite many without any response :(14:10
elodillesfor example, oslo patches got +1 from stephenfin , but he is not release liaison for oslo :( though he is a long time active maintainer there14:12
elodillesshould we merge them, what do you think?14:12
elodillesmaybe worth mentioning, that however it is advised to release libraries at milestones, in the past cycles we did not force merge milestone-1 releases if there were no responses14:14
elodilles(but with milestone-3 we definitely have to force release, to have the latest state released)14:14
fungioslo is a bit of a weird situation with different people focused on different deliverables14:14
elodillesso maybe part of the question is: should we abandon the release patches that haven't got any response, or let's say we force merge them on Monday?14:15
fungii wonder if they need to just list all their various maintainers as release liaisons so they can ack releases for the deliverables they're maintaining14:17
cardoeSo stuff like python-openstackclient really shouldn't be abandoned.14:18
cardoeThe lack of responses from PTLs is a problem that should be brought to the TC14:18
cardoeBecause this kind of paper work is what the PTL role is signed up to do.14:18
elodillesfungi: maybe that would be too many. in the past that was the reason to add PTL-Approved+1, so that we know that people who care most about the releases, could signal that the release is okay14:19
cardoeSo if you're not filling the role then it's a problem for the health of the project.14:19
elodillesfungi: on the other hand i've already proposed if stephenfin could be release-liaison for oslo, since he is active around oslo and pushes the majority of (non-generated) oslo release patches o:)14:20
fungiyeah, i'm just saying part of why oslo doesn't have a single person interested in releases for all its deliverables is that it's a dumping ground for various libraries anda utilities that didn't fit anywhere better14:20
elodillescardoe: yes, this is i think some kind of a recurring topic in tc14:21
fungiso a lot of the deliverables there are maintained more or less independent of each other, not by a common team across the whole of oslo14:21
stephenfinI'm okay to be a release liason for both SDK and OSC, if governance allows for that14:21
stephenfinsince I'm likely the only person other than tkajinam that regularly touches all of oslo (and probably the only person that regularly works on all of SDK)14:21
elodillesstephenfin: that would be awesome, but the best would be to get an ACK from damani and tkajinam (current release liaisons)14:22
fungithe absent ptls and liaisons problem also has been an issue for the vmt for many years, we finally got the tc to assign volunteer tc liaisons who agree to escalate those when we loop them in14:22
elodillesstephenfin: i mean, ACK from them for you to be a release liaison as well o:)14:23
stephenfinelodilles: is that necessary, given I am already core on the projects and we're using a DPL model (genuine question: governance may well insist on it)14:23
stephenfin?14:23
fungimaybe we could have something similar for the tc and release team to coordinate?14:23
fungistephenfin: "core" is a nebulous concept that doesn't have the same definition from one team to the next14:24
fungiwe need some official list of people for each team published in a common location14:24
elodillesstephenfin: i'm not insisting that it is necessary, but that would be the plainest/cleanest to have an ACK from them for you to be a release liaison too o:)14:25
stephenfinfungi: I thought that would be the answer but always best to ask :)14:25
stephenfinI'll poke damani internally. I suspect tkajinam is (long) gone for the weekend at this point14:26
fungikeep in mind that even just looking up who is authorized to ack a release request takes time, so we have automation to look that up, it just needs a list somewhere to look it up in (which is currently the list of ptls and release liaisons)14:26
elodillesstephenfin: +114:26
stephenfinIs there a YAML file I can propose myself against and have damani vote on?14:26
* stephenfin looks in openstack/governance14:27
elodillesanyway, i don't want to solve this long standing issue now, just raised the question, whether we should abandon the patches,14:27
elodillesbut i see that cardoe has some interest on releasing some of the deliverables,14:27
elodillesso i'm ok with those to even force-merge if needed14:28
cardoewell I'm just wanting some CLI fixes that are hitting me and my team14:28
elodillescardoe: yupp, sounds reasonable14:28
cardoeI'd also like to get some of the typing updates out there in libraries so I can nudge the Outreachy folks to make more progress on typing14:28
elodillesACK, then considering these, i tend to force-merge them on Monday14:29
elodillesany objection?14:29
opendevreviewArtem Goncharov proposed openstack/releases master: Release python-openstackclient for Hibiscus-1 milestone  https://review.opendev.org/c/openstack/releases/+/98808514:29
cardoestephenfin and I have been -1'ing a number of patches telling folks that something is fixed in a library and the dep just needs to be updated.14:30
elodillescardoe: ah, i see14:30
elodillesokay, i don't hear any objection o:)14:34
elodilles#action elod to force-merge remaining open hibiscus-1 release patches14:35
elodillesnext task was then:14:35
elodilles- Catch if there are acl issues in newly created repositories (ttx)14:35
elodilles    No issues detected14:35
elodillesreported on etherpad by ttx ^^^ 14:36
elodillesthat's all about tasks14:36
elodilles#topic Assign R-19 week tasks14:36
elodillessorry, had a network lag14:38
elodillesso, 1 task is related to the remaining hibiscus-1 patches, and the other is about keeping an eye on new-release requirements patches, so...14:39
elodillesall taken14:40
elodilles#topic Review weekly countdown email14:40
elodillesplease review: 14:40
elodilles#link https://etherpad.opendev.org/p/relmgmt-weekly-emails14:41
frickler+1, nice and simple ;)14:41
elodillesthanks frickler o/14:42
elodilles#topic Open Discussion14:43
elodillescardoe: did we talk through your topic already, or do you have anything else to bring up? o:)14:44
*** gibi is now known as gibi_off14:46
elodillesor any other topic from anyone else?14:48
fricklerjust a note that we did have an issue with reno publishing14:50
fricklerfixed by https://review.opendev.org/c/openstack/openstack-zuul-jobs/+/988768, but some reno pages may be lacking recent branches until they get republished14:50
elodillesah, good to know, thanks frickler !14:51
elodillesokay, if nothing else, then thanks everyone for participating and enjoy the weekend o/14:53
elodilles#endmeeting14:53
opendevmeetMeeting ended Fri May 15 14:53:45 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)14:53
opendevmeetMinutes:        https://meetings.opendev.org/meetings/releaseteam/2026/releaseteam.2026-05-15-14.00.html14:53
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/releaseteam/2026/releaseteam.2026-05-15-14.00.txt14:53
opendevmeetLog:            https://meetings.opendev.org/meetings/releaseteam/2026/releaseteam.2026-05-15-14.00.log.html14:53
fricklerthx elodilles 14:53
elodillesmy pleasure :)14:55
elodillescardoe: if you think a release can't wait till monday, then ping me here and i'll try to review it today14:57
cardoeNo it can wait14:58
cardoeSo what I had wanted to bring up was OpenStack Helm and it's container images.14:58
cardoeSo OpenStack Helm targets helm charts that are versioned after the current OpenStack release but support the currently supported releases back. So the 2026.1 charts support back to 2025.114:59
cardoeThere's releases on every commit.14:59
cardoeI'm not sure if we should do some better tie in with the release team or doc improvements.14:59
fungiwhat did you have in mind?14:59
cardoeI'm also trying to make the container images generated a lot more generic and follow the stable branches of each project.14:59
cardoeWell if there's some practices / operations the release team wants projects to follow.15:00
cardoeI'm just wanting to try and figure out how to make it less off in the corner15:01
cardoeWhen I get some cycles I'm gonna try and clean up the skyline release process as well.15:01
fungiare the charts versioned independently? do they follow semver or have any other particular version numbering scheme?15:01
fungijust sequential/monotonically increasing? raw integer version numbers or multi-segment?15:02
cardoeIt'll be '2025.2.17+bb8dd0598' for example for a placement chart15:02
cardoeSo it's 2025.2 base so that's shipping a 2025.2 version of placement by default. It's the 17th version of the placement chart, the monotonic number. And then bb8dd0598 is the git-ish in the repo.15:03
fungiokay, so openstack coordinated release followed by another monotonically-increasing integer15:03
fungiand the latest 2025.2.* versions of all charts are expected to interoperate but their final version components will vary from chart to chart, or are they released in lock-step?15:04
fungiand you don't use distinct git tags for any of this, the versions are just determined on the fly from the repository state from which the chart was built?15:06
cardoeAll 2025.2.* charts will interoperate and the final version varies from chart to chart.15:06
cardoeCorrect no distinct git tags. The versions are determined from repository state at the time.15:06
cardoeWe15:07
cardoeWe've been using reno. Which I'll be honest I don't really understand how to make the notes correctly.15:07
fungipeople getting different version numbers locally when adding their own commits and not having them sort chronologically compared to upstream is also not an issue, i guess15:07
fungireno depends on having git tags15:08
cardoecause there's the general fixes top-level but then there's also the name of the chart top-level in there.15:08
fungiwithout tags, it won't know what version to associate a note with15:08
cardoehttps://docs.openstack.org/releasenotes/openstack-helm/current.html yeah I don't love that output.15:09
fungiif you're not doing tag-based versioning you'll likely need some other mechanism to curate release notes15:09
cardoeUltimately we don't embed changelog entries in a helm standard format. So none of the helm chart tools can show you changelog entries15:10
cardoeAnd the reasoning I'm told is that we're complying with release team release notes.15:10
cardoeBut I don't think the release notes are working correctly.15:10
cardoeSo that led me to just ask are there things that the project should be doing to follow release team process better.15:11
cardoeCause the project isn't listed on https://releases.openstack.org 15:14
cardoeAnyway, it's not really an easy answer question. Just wanting to make the project a better citizen within the umbrella.15:15
fungiopenstack release management is tag-oriented pep-440-compliant 3-segment semver for python packaging, if you're not making git tags and building python packages then you're not covered by release management anyway and not integrated with our tooling, so should probably manage release notes in a way consistent with whatever ecosystem's "releasing" model you're following15:19
fungiinstead15:19
fungiput another way, openstack's release management is automation built on conventions for what's expected in the python packaging ecosystem. there's little reason to integrate with that if you're building artifacts for a different ecosystem15:21
cardoeokay that's fair15:25
cardoeSo the best plan would be for the project to fix the release notes in a way that's compatible with the helm eco system and then publish them to the releasenotes page in a way that'll track the chart better.15:26
cardoeI just want a better experience for users.15:28
fungiyes, i think so15:32
fungikeep in mind that our publication tooling/utilities and automation are similarly built around python packaging ecosystem conventions, so a lot of it may not be reusable and you'll need to find or build more appropriate solutions15:33
elodillescardoe: just a side note, that releases.openstack.org lists the releases of deliverables that are present in the given series in relmgt repositories. e.g. for gazpacho, the https://releases.openstack.org/gazpacho/index.html lists deliverables that are defined here: https://opendev.org/openstack/releases/src/branch/master/deliverables/gazpacho . but i agree with fungi15:36
elodilles* list the releases and their release notes15:37
cardoeSo would it be acceptable to push https://docs.openstack.org/releasenotes/openstack-helm/current.html for example with a tool from the helm eco system to generate the release notes/16:07
cardoeSo like right now for my team, we run our own mirror for the sole purpose of running some tooling to analyze the delta to see what's changed.16:11
-opendevstatus- NOTICE: The Gerrit service on review.opendev.org will be offline momentarily while we restart it onto a new patch release18:04
fungicardoe: yeah, i think you'd probably only need to replace the release notes build job in zuul with one that runs the helm tools to build the html you want and create the resulting content tarball with the same name as the publishing job expects18:55
cardoealright. I'll make some suggestions to the PTL. I just know we cannot be the only ones that cannot follow what's been published.18:56
*** janders2 is now known as janders20:32

Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!