Tuesday, 2026-08-04

ianychoiHi team, I cannot join in upcoming infra meeting in 3 hours.. but would be great if your team can give me / seongsoocho some hints how to move forward https://review.opendev.org/c/openstack/project-config/+/961499 so that he can say during OpenInfra Asia Summit 2026 that I18n SIG "migrated" to Weblate like https://kubecon-cloudnativecon-openinfra-pytorch-2026.sessionize.com/session/1225189 .15:48
fungiianychoi: in the future we can add things to the meeting specifically if it's more than 24 hours ahead of the meeting time, but i'll bring it up during open discussion15:50
fungithough that's more of an openstack topic than an opendev topic. clarkb has been travelling for the past month and i've been buried under other tasks so haven't had a chance to look at it. i think mnasiadka was also talking about reviewing some of that15:51
clarkbyes, there is overlap in people but its likely we need to motivate people from the openstack side to get it over the finish line15:52
ianychoiThank you fungi for the comments and I agree with you to have raised in early manner is encouraged. Note that I pinged 1-2 weeks ago on #openstack-infra channel so I ping here again.. but I suggest seongsoocho to more actively discuss with OpenStack Infra team if there are any issues. Appreciate all the help, honestly.15:53
mnasiadkafungi: thanks for reminding me :)16:21
fungimeeting in ~2 minutes!18:57
fungi#startmeeting infra19:00
opendevmeetMeeting started Tue Aug  4 19:00:00 2026 UTC and is due to finish in 60 minutes.  The chair is fungi. Information about MeetBot at http://wiki.debian.org/MeetBot.19:00
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.19:00
opendevmeetThe meeting name has been set to 'infra'19:00
fungi#link https://lists.opendev.org/archives/list/service-discuss@lists.opendev.org/thread/2NBGQXHA3KHZALTRU3K5LAEVOEYZC3EF/ Our Agenda19:00
fungi#topic Announcements19:00
fungii didn't have any announcements recorded, did anyone else?19:00
clarkbI'm back to a normal timezone19:01
fungiwelcome back!19:01
fungiseems like nothing else for announcements this week, so moving on..19:02
fungi#topic Upgrading Old Servers19:02
fungi#link https://etherpad.opendev.org/p/opendev-server-upgrade-planning Central tracking document which may link to more host specific documents19:02
fungithe change to remove backup02.ca-ymq-1.vexxhost has deployed, but that didn't clean up the old cronjobs on our servers19:03
clarkbya I looked at the ansible code (role install-borg) and it seems to only add cron rules19:03
fungiso it's still getting a redundant copy of backups for now, and having to be pruned periodically19:04
clarkbso I think it should be safe to either manually clear those out or figure out a way to have ansible drop the cron rules (they are atuomatically generated based on group membership in the borg server group19:04
fungiyeah, i can just manually `crontab -e` on them19:04
clarkbwe may need a list that we manually edit in an ansible var that we loop over and ensure absent with19:04
fungiit wouldn't take too long to go through the servers that have been backing up to it19:04
clarkbbut for now manually dropping it is probably fine19:05
fungialso we noticed that the new borg version on backup03.ca-ymq-1.vexxhost no longer compacts backups automatically, causing it to fill up fairly quickly19:05
fungii manually ran a compact cycle across all the backups on it last week just to keep things running for now19:05
corvusinteresting enhancement choice19:05
fungi#link https://review.opendev.org/999465 Compact borg repos after pruning19:06
fungithat's wip for the moment, because the old version of borg on our other server doesn't have that subcommand19:06
fungibut i'll adjust it to be version dependent, unless we're close to being able to replace the other older server in rax ord19:06
fungithat's probably all the updates about server upgrades for the moment, though related to the next topic19:07
fungi#topic Deploying a Prometheus for Server Metrics19:07
fungi#link https://review.opendev.org/c/opendev/system-config/+/999383 deployment automation continues from here19:08
clarkbcorvus: they made that change so you can better schedule high iop activity I guess19:08
fungithe server exists (thanks mnasiadka for creating it)19:08
corvusclarkb: ah interesting.  makes sense i guess19:08
fungithe next step is adding the deployment job19:09
fungiwhich that change does, but needs more eyes19:09
clarkbfungi: mnasiadka are there hosts in the production inventory yet?19:09
clarkbmostly wondering if that change is expected to noop or not19:09
fungiyes19:09
clarkbok so it will not noop19:10
fungiprometheus01 exists and is in inventory and dns19:10
fungibut right now it's not got prometheus installed on it19:10
clarkback I will review it with that in mind thanks. Maybe that means I don't approve it this afternoon before doing router upgrades19:10
fungithe playbooks have previously been exercised by test jobs, just not used to deploy on the prod server yet19:11
fungiwhich is why that change just adds a job to run the existing playbooks19:11
fungione thing i have not double-checked yet is whether any private hostvars are already set up on bridge19:12
fungijust dawned on me19:12
clarkb++ that is a good thing to check before that change gets approved19:12
fungiit's an easy step to forget19:12
fungi#topic Dealing with alien zuul config errors in the openstack tenant19:13
fungiclarkb reached out to starlingx about this19:13
clarkbthe change does lgtm but I didn't approve it due to ^ and needing to upgrade my router after this meeting19:13
clarkbI asked ildiko to ping them about it and scott little responded a few days back asking how urgent this is19:13
fungi#link https://lists.starlingx.io/archives/list/starlingx-discuss@lists.starlingx.io/thread/YQVACUR4OCX74ZULHAJ4AD44MHGY37YI/ Moving StarlingX into a Dedicated Zuul Tenant19:13
clarkbsounds like they are getting into the busy part of their release process? I was planning to respond that this isn't super urgent but that they should work on correcting their current zuul configuration errors in the interim19:14
clarkb(as those should be more straightforward and are less likely to be negatively impactful to their release)19:14
fungithat makes sense19:14
fungi#topic Gitea 1.27.1 Upgrade19:15
clarkbI've got writing that response on my todo list as of this morning. Should get to it eventually19:15
fungi#link https://review.opendev.org/c/opendev/system-config/+/998742 Upgrade Gitea to 1.27.119:15
fungiwe're running that now19:16
fungiwhen i checked yesterday there wasn't a newer version yet19:16
clarkbyup this is the latest version at the moment19:16
clarkbwe can probably drop this item from the agenda. Thank you for getting that landed19:16
fungino problem19:16
fungi#topic Matrix Meetbot Porting19:17
fungihere's an exciting one!19:17
fungicorvus just tested this again a few minutes ago19:17
fungiwhen should we switch to meeting in matrix, and what room (our normal room or a separate meeting room)?19:17
fungiare folks fairly confident this is working and feature-complete now?19:17
mnasiadkaclarkb: there was no deploy job, I think I raised a patch for one, so once it merges we should get prometheus up and running19:18
fungimnasiadka: yes, that was the change we discussed which he's planning to review19:18
clarkbfungi: I think I'm happy to move the meeting to matrix as soon as the test meetings look good19:18
corvusyeah, i think the meetbot is ready for us19:18
clarkbas for whether or not we need a new channel I'm indifferent. Seems like the existing channel should be fine. But if we want to avoid the clutter we can create another room too19:19
corvusi lean slightly toward "new room"19:19
corvusjust because we do have people dropping in in the main room during meetings sometimes19:19
corvusso it's nice not to have to say "pls wait 1 hour" :)19:20
clarkbin that case a new #opendev-meeting:opendev.org room seems like it would work well19:20
fungiand even when we had a non-retired #opendev irc channel we used this #opendev-meeting channel to hold our meetings19:20
fungiso yes, that sounds fine to me, i don't think our reasons for separating them have changed, just the protocol/venue19:21
clarkbso we need to create the room then configure the bot to live there. Then can announce the change of meeting venue?19:21
fungiyes i think so19:21
corvus++19:21
fungiwho wants to create the room and push the one-liner change to add it to the config?19:22
clarkbI'm still building and catching up on my backlog so not sure if I would be able to get to that before next meeting.19:22
clarkbbut I can be the backup/fallback if others don't beat me to it19:22
fungii'm hesitant to pretend it won't fall too far down my list of priorities for this week19:23
corvusi can try, no promises19:23
fungiokay, let's all try to find a few minutes for it and maybe one of us will get to it19:24
clarkbsounds good19:24
fungiand if not, then we'll reconvene here next week until we get to it19:24
fungi#topic Gerrit 3.13.8 upgrade19:24
fungi#link https://review.opendev.org/c/opendev/system-config/+/994938 Upgrade Gerrit to 3.13.819:24
fungiwe talked about maybe doing it tomorrow, is that still feasible?19:25
fungii expect to be around most of the day, though do need to run some errands around lunch19:25
clarkbwe are running 3.13.6. There have been a couple of bugfix updates since. One nice thing is they include a fix for the replication plugin on a tag so we can stop building the branch version there. But otherwise there don't seem to be super critical updates just good to stay up to date generally19:25
clarkbfungi: yes I expect to be around tomorrow as well19:25
clarkbif tomorrow is still good for you then I think it should work for me19:25
fungiwhat time do you like? 20:00 utc?19:25
clarkbyes I think that should work. I can eat lunch while the changes land19:26
fungiperfect, i've put it on my reminders as well19:26
fungiit should be a quick restart so i don't think we need more than a sttatus notice when we're starting19:27
clarkbworks for me19:27
fungiand maybe a warning shot an hour before19:27
clarkb++19:27
fungi#topic Update Anubis to 1.26.219:27
fungithis has deployed, and also the wiki server's package was manually upgraded to 1.26.219:27
clarkbhttps://review.opendev.org/c/opendev/system-config/+/998741 says the change is still open19:28
clarkbwas there a differetn change that landed or is this not deployed yet?19:28
fungioh, wait, i thought i had approved it19:28
fungiah, no you're right19:28
fungi#link https://review.opendev.org/c/opendev/system-config/+/998741 Upgrade Anubis to 1.26.219:29
fungii did upgrade it on wiki, but only revised the change to take it from 1.26.0 to 1.26.219:29
fungiso we can merge this whenever, i think19:29
clarkbyes I would expect that this is something we can land whenever19:29
fungiand i saw they pushed a v1.27.0-pre2 tag too, if we want to start a wip change to test that in advance19:30
fungianyway, 1.26.2 seems to be working fine on wiki so far19:31
fungiit's been runnig since last week with no complaints19:31
clarkbI need to upgrade my home network gear after the meeting but am happy if thatgets approved I think it should be safe enough with a straightforward rollback path19:31
fungiyeah, i need to disappear briefly right after the meeting, but can approve it later when i get back or tomorrow morning19:32
fungi#topic Open discussion19:32
fungiianychoi and seongsoocho[m] were looking for feedback on openstack's translation jobs migration to weblate19:33
fungi#link https://review.opendev.org/c/openstack/openstack-zuul-jobs/+/991432 Add Weblate client Ansible role19:33
fungithat looks like the next step there19:33
seongsoocho[m]Hi fungi.  yes. 991432 is the first patch to be reviewed. 19:34
fungii think clarkb and frickler have looked at some of the other changes in the set, and mnasiadka was talking about taking a look over them too19:35
clarkbone thing I've brought up before but can't recall if it was ever clarified is do we know who can/how to reset the credentials for weblate if we expose them?19:35
clarkbI think that is the biggest risk here. And maybe we should go ahead and accept something we think is close and start working on improvements rather than the big bulk update change19:35
fungilike if a bug in the job causes them to be unexpectedly leaked/disclosed in public logs19:36
clarkbcorrect19:36
seongsoocho[m]Our current approach is to never expose any credentials. They should only be read from the configuration file.19:37
clarkbI'll try to rereview the change. I think if we can clarify that credential reset path then my suggestion would be to proceed with what we've got unless there is something obviously wrong with it and then iterate from there19:37
clarkbseongsoocho[m]: yes, but sometimes ansible doesn't do what you expect etc19:38
corvusdoes weblate support oidc?  if so, using zuul as a federated idp could be an option19:38
seongsoocho[m]Okay. Also, the current patch does not use any credentials.19:39
clarkbah right this change is only adding the role19:39
clarkbso we have to get to the point where we run the role to risk credentials19:39
fungiright it's not a concern for 991432 but in later parts of the implementation the question will still be relevant19:40
clarkbso ya I think my suggestion would be to proceed here if there isn't anything obviously wrong19:40
clarkbthen get to the point we can run something and iterate from there (with a plan for what to do if we expose credentials)19:40
clarkbcorvus: I have no idea if they support that.19:40
clarkbgoogle seems to think it may be possible. Might be a good improvement in the iteration process19:41
fungiyes, i think we discussed it in the past19:42
fungiwell, oidc support i mean, not zuul's ability to do federated authorization19:43
fungianyway, i guess that's all for open discussion unless anybody else had anything?19:43
clarkbI didn't have anything19:44
seongsoocho[m]We use the Weblate API, which requires an API key for authentication. Even if Weblate supports OIDC, I think we still need to keep using an API key for this use case.19:44
clarkbseongsoocho[m]: the idea with OIDC is that zuul can act as a broker for the authentication details and provide short term tokens to the jobs19:45
fungiyes, the weblate api would need oidc support for that to be a possibility19:45
clarkband yes depends on how they have implemented oidc19:45
clarkband whether or not you can configure weblate to trust zuul19:45
fungiright, i.e. not hard-coded to a list of popular providers19:46
seongsoocho[m]ah okay...19:46
fungibut yes down the road we can evaluate that as a potential improvement, then you wouldn't need to maintain any credentials for the jobs that interact with the weblate api19:47
fungiand if a job leaked an oidc token it probably wouldn't be usable by the time someone saw it19:48
fungiokay, seems like discussion is winding down so i'll wrap the meeting up19:49
fungifind us in #opendev:opendev.org matrix, or on the service-discuss@lists.opendev.org mailing list for anything else opendev between now and the next meeting19:50
fungithanks everyone!19:50
fungi#endmeeting19:50
opendevmeetMeeting ended Tue Aug  4 19:50:40 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)19:50
opendevmeetMinutes:        https://meetings.opendev.org/meetings/infra/2026/infra.2026-08-04-19.00.html19:50
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/infra/2026/infra.2026-08-04-19.00.txt19:50
opendevmeetLog:            https://meetings.opendev.org/meetings/infra/2026/infra.2026-08-04-19.00.log.html19:50
clarkbthank you for running the meeting again fungi19:50
fungiany time19:50
seongsoocho[m]thank you for your help!19:50
fungiyou're welcome!19:52

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