Friday, 2026-02-06

opendevreviewMerged openstack/releases master: Release taskflow 6.1.1  https://review.opendev.org/c/openstack/releases/+/97576111:19
ttxo/14:00
elodilleso/14:00
ttx#startmeeting releaseteam14:00
opendevmeetMeeting started Fri Feb  6 14:00:24 2026 UTC and is due to finish in 60 minutes.  The chair is ttx. 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
PavloKostianov[m]o/14:00
ttxPing list: release-team elod14:00
ttxOur agenda is at:14:00
ttx#link https://etherpad.opendev.org/p/gazpacho-relmgt-tracking#L24614:00
elodilleso/14:01
ttx#topic Review R-8 week task completion14:01
ttx- Make sure the next development series name has been added to the data/series_status.yaml file. (elod)14:01
elodilles(elod) done, 2026.2 Hibiscus proposed release schedule patch has merged14:02
ttx- Send weekly email (ttx)14:03
ttxWill do after meeting!14:03
elodilles+114:03
ttx#topic Assign R-7 week tasks14:03
ttxall assigned!14:03
elodillesyepp14:03
ttx#topic Review weekly countdown email14:03
elodillesstill not that many tasks14:03
ttx#link https://etherpad.opendev.org/p/relmgmt-weekly-emails14:03
* elodilles clicks14:04
ttxFor RC1 deadline and final-RC deadline I set the date to the Fridays, since technically it is a "week"14:04
ttxbut I know we have been going back and forth on that14:04
elodillesyeah14:06
ttxok let's keep it that way then14:06
fungifor cycle statistics i've been going thursday to thursday of rc1 week, for a rough approximation of when most stable branches are created, but making friday the deadline doesn't really change that14:06
ttxah good point14:07
* ttx moves to all Thursdays14:07
fungiin part because the release team normally avoids approving non-emergency changes on fridays14:08
elodillesfinal rc deadline should be March 26th, am i wrong?14:08
elodillesin the mail it is March 19th14:08
elodillesbut i think that is not correct14:09
ttxchecking14:09
ttxnice catch!14:09
ttxanything else?14:10
elodillesi don't see any other issue14:11
elodillesLGTM14:11
ttxok will send shortly after meeting14:11
elodilles+114:11
ttx#topic Open Discussion14:11
ttxLooks like something slipped past tests and reviews14:12
ttxGazpacho M2 release was tagged 31.0.0.0b2 while Flamingo was 31.0.0, so after14:12
elodilles /o\14:12
ttxSo to answer rosmaita 's questions...14:13
ttxYes it should have been 32.0.0.0b214:13
rosmaitaok14:13
ttxHow could it happen? I'm looking at the test that should have failed...14:13
elodillesmaybe not all cases are covered in case of beta :/14:14
rosmaitawhat i noticed was the last beta was 30.0.0.0b2, we didn't have one in flamingo14:14
elodillessince there wasn't any 31.0.0.0b2 in Flamingo, there were no overlap14:14
elodillesyepp14:14
fungiyeah, but the version rolled backwards14:14
rosmaitayeah, so i guess that was a weird case14:15
* ttx deepdives into validation code14:15
ttxplease hold :)14:15
ttx(or beat me to it)14:15
fungi(31.0.0.0b2 tagged after 31.0.0 even though 31.0.0.0b2 sorts before 31.0.0)14:15
fungiwhich repository was this in?14:16
ttxLooks like it should have failed https://opendev.org/openstack/releases/src/branch/master/openstack_releases/cmds/validate.py#L31514:17
rosmaitafungi: glance14:17
fungiaha14:17
ttxThat was for openstack/glance, release requested in openstack/releases14:17
ttxoh14:18
ttxmaybe it does not test the progression from last series to current, only progression in the same series?14:18
rosmaitathat seems to be the case for that function14:20
ttxsurprised it would not have failed earlier if that was the case14:21
ttxok so we have a hole in our validate testing14:23
ttxFirst priority is... what can we do about it14:23
rosmaitawell, i think since 31.0.0 has been released, no one in their right mind would use the 31 beta14:24
fungiaugment the `len(releases) == 1` case to check the last release in the previous series? would require loading that in though14:24
ttxyeah, I don't think it's an issue really, but we might want to clean up14:24
ttxfungi: yes I'm under the impression we actually do not look at the context, so unless you pick an existing tag it just passes14:25
elodillesmaybe we could re-tag that hash with 32.0.0.0b2?14:26
ttxneed to pay specvial attention to those "first in series" releases then14:26
ttxelodilles: yes that would be a good idea14:26
fungiright, it also doesn't guarantee version numbers go forward, just that e.g. you don't tag a beta after rc14:26
elodilleswe usually don't delete tags, that are already applied. the question how much confusion that could cause.14:26
ttxthat would clean up the deliverable file, at least14:26
rosmaitathis is interesting: https://pypi.org/project/glance/#history14:27
ttxlol14:27
elodilleswell, and our validator might catch THAT case when we try to re-tag a hash, so it might not work o:)14:27
fungii wouldn't worry about the existence of the stray beta tag, pip installing that only happens if the version number is explicitly requested or `--pre` is specified and it's the most recent version number14:27
ttxOK let's just fix the deliverable file, tagging 32.0.0.0b2 instead14:28
elodillescan we mark 31.0.0.0b2 in pypi as 'yanked' or something?14:28
ttxthat won;t remove the old one but we can leave with it14:28
ttxlive*14:28
fungii can yank it, which causes it not to appear in the index, but doesn't stop it from being installable14:28
ttxthat's ok14:29
elodillesfungi: but as you said, pip won't pick it up anyway14:29
fungibut it'll still be in the list of git tags regardless14:29
elodillesfungi: so probably that's okay, and if someone wonders then they can see that it is yanked on pypi, right?14:29
fungiright, someone would need a very unusual set of circumstances to end up with it installed even now14:29
fungilike `pip install --pre glance<31`14:30
rosmaitayeah, because i guess --pre would get you the rc1 release?14:30
fungioh, right14:30
fungilike `pip install --pre glance<31.0.0.0rc1` then14:30
fungiin practice it won't ever happen14:31
opendevreviewThierry Carrez proposed openstack/releases master: Re-release glance gazpacho-2 as 32.0.0.0b2  https://review.opendev.org/c/openstack/releases/+/97591114:31
ttxthat ^ and yanking the file from Pypi14:31
ttxdoes that sound ok as mitigation?14:32
fungii'll mark it yanked on pypi in a few minutes when i break out the mfa for that account14:32
ttxNow we need to avoid that in the future, but I'm not sure the validate framework can easily test for that14:33
rosmaitai think that sounds good ... so the 31.0.0.0b2 tag will still be there, but won't be considered a release because it's not in the deliverables glance.yaml file?14:33
ttxsince it basically looks at files in isolation afaict14:33
elodillesttx: LGTM, but let's wait if validator likes it as well14:33
ttxrosmaita: yes, won;t be listed in the releases page14:33
rosmaitathat works for me14:33
ttx\o/14:34
elodilleswith running validator locally it passed me14:34
elodillesso should be fine then14:34
ttxI wonder how many more we have fumbled in the past, but never noticed14:34
elodilleshope the post jobs are doing the tagging as well o:)14:34
ttxafter all those years I thought the validator was infaillible14:35
rosmaita:D14:35
ttxI'll list it under "things to change" so that we know we have to work on it. But I'll likely not be able to find time to fix it myself14:37
elodilleseven list-changes job shows the latest (highest) version, so i somehow missed that too :( but yeah, i thought validator to be catching this :/14:37
elodillesttx: +114:37
elodilles(it's strange that i think we should have a log message 'checking progression from {} to {}' at the 'Pre-release versions [..]' test, but there isn't...)14:38
fungibasically two holes: doesn't check against the prior series at start of a new one, and doesn't check that version numbers roll forward14:40
ttxIt looks like all validation is within the context of the changed file only, so we only rely on tag collision to test numbers between series14:40
ttxrosmaita: good that you caught that one before the end release :))14:41
rosmaitattx: elodilles: fungi: thanks for taking care of this so quickly14:41
ttxwould have been fun times around rc114:41
elodillesyeah, we would have been quite surprised at rc1 release o:)14:42
rosmaitait was just luck, i was looking at the latest tags for the maintained branches, and that looked weird14:42
fungigood eye14:43
ttxmeanwhile we should probably pay extra attention to first releases in a series, especially when not autogenerated14:43
elodilles+114:43
ttx(we tend to avoid that with generated patches which actually take previous releases context into account)14:44
fungichecking that versions roll forward is relatively trivial if you have a proper pep 440 version sort (e.g. from the packaging library)14:44
fungibut this also needed to check outside the current series14:44
ttxI mean, who does write those by hand in 2026? :)14:46
ttxOK anything else to discuss?14:46
elodilles-14:47
elodillesdid we lost ttx ?14:59
ttxoops15:00
ttx#endmeeting15:00
opendevmeetMeeting ended Fri Feb  6 15:00:56 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)15:00
opendevmeetMinutes:        https://meetings.opendev.org/meetings/releaseteam/2026/releaseteam.2026-02-06-14.00.html15:00
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/releaseteam/2026/releaseteam.2026-02-06-14.00.txt15:00
opendevmeetLog:            https://meetings.opendev.org/meetings/releaseteam/2026/releaseteam.2026-02-06-14.00.log.html15:00
elodilles:)15:01
elodillesthanks ttx o/15:01
ttxsorry I thought I had closed it :)15:01
fungiokay, glance 31.0.0.0b2 has been yanked from pypi now15:25
elodillesthanks o/15:29
-opendevstatus- NOTICE: Gerrit on review.opendev.org will experience a short outage while we upgrade it to 3.11.817:50
*** gmaan is now known as gmaan_afk22:46

Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!