Tuesday, 2025-11-11

clarkbmeeting time!19:00
clarkb#startmeeting infra19:00
opendevmeetMeeting started Tue Nov 11 19:00:16 2025 UTC and is due to finish in 60 minutes.  The chair is clarkb. 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
clarkb#link https://lists.opendev.org/archives/list/service-discuss@lists.opendev.org/thread/XEGEBPR2GSFB5UDOI5WGMTUIXHKQAEAP/ Our Agenda19:00
clarkb#topic Announcements19:01
clarkbwe are just over 2 weeks away from a major holiday in the US. I expect to be around tuesday and probably wednesday that week but not thursday and friday19:01
clarkbAll that to say I don't think it will affect our meeting schedule, but it probably will affect when people are around and active19:02
clarkbWas there anything else to announce?19:02
fungii have nothing19:04
clarkb#topic Gerrit 3.11 Upgrade Planning19:05
clarkbGerrit 3.13 has released19:05
clarkbthis means the pressure to upgrade to 3.11 is increasing19:06
clarkbBefore we do that there are new bugfix releases19:06
clarkb#link https://review.opendev.org/c/opendev/system-config/+/966084 Update to Gerrit 3.10.9 and 3.11.719:06
clarkband before we update to address the bugfix relases we have a docker compose bug to fix19:06
clarkb#link https://review.opendev.org/c/opendev/system-config/+/966083 Fix container bind mounts for Gerrit19:06
clarkbLanding these two changes and restarting Gerrit is going to be a big goal for me this week. I'm still catching up on stuff after being out yesterday but expect to be able to merge these changes and restart Gerrit sometime this week. Maybe friday if we are trying to cut down on impacts but possibly sooner19:07
clarkbWe also heard back from vexxhost on the gerrit server and it was a memory issue which should be mitigated now19:08
clarkb(which makes updating gerrit and restarting things safer19:08
clarkbAny other questions or concerns about Gerrit?19:08
tonybif we can schedule the restart while I'm around I'd like to be a second set of eyes19:09
tonybmostly to confirm what a normal start looks like19:09
clarkboh yes we should do that. So maybe thursday afternoon (for me)/friday morning for you19:09
tonybsounds good19:10
clarkbtonyb: feel free to propose some time blocks. I'm generally pretty flexible late week19:10
clarkb#topic Upgrading old servers19:10
clarkbtonyb has the wiki change stack been updated for quay and/or noble?19:11
tonybnoble yes quay no19:11
clarkback thanks. I think updating the image build change to do that is the next step for this effort.19:12
tonybalso the ansible changes are going to be restructured a little to move ansible-nextvto jammy!+3.1119:12
tonybI'll get them updated this week19:13
clarkbthanks19:13
clarkbany other server upgrade updates? (I don't think so but want to double check before we move on)19:14
tonyb(sorry about the typos, speed and accuracy are low on my phone)19:14
tonybnothing more from me19:15
clarkb#topic Matrix for OpenDev comms19:15
clarkbtonyb offered to look into creating the new room last week. Not sure if that happened19:15
clarkbthat is step -2 of many to get this moving forward but it is an important step19:15
tonybnope.   today!19:15
clarkbthanks!19:16
clarkb#topic Upgrade Zuul Zookeeper Cluster to 3.919:16
clarkb#link https://review.opendev.org/c/opendev/system-config/+/96661219:16
tonybI was thinking I might also make a tooling test room .... to target with tools ... for testing 19:16
corvuswe have one19:16
tonyboh!  never mind then19:16
clarkbthe zookeeper cluster is running 3.8 which is the stable release19:16
clarkb3.9 is the current release and has existed for enough time now to probably also be considered stable19:17
clarkbthe normal upgrade process is to upgrade each of the non leaders first then the leader which our ansible is not smart enough to do19:17
corvusi think it's very likely that zuul is going to make 3.9 a requirement for zuul-launcher19:17
clarkbcorvus: what specific feature(s) make 3.9 useful for thel auncher?19:18
corvusso getting ahead of that would be beneficial19:18
corvusthe watch event returns the zk transaction id starting with 3.9, so we can tell our current position in cache replays19:18
clarkbgot it19:19
corvushttps://review.opendev.org/966501 is the zuul change that takes advantage of it19:19
corvusi've written a fallback change for zuul19:19
clarkbas far as upgrading goes I have no objections to moving to 3.9. i think I have a slight preference for manually doing the upgrade to employ the correct expected process19:19
clarkbnote you have to check the status of each member after each restart beacuse sometimes the leader moves19:19
corvusso this doesn't have to be in the critical path, we can upgrade whenever, but i'd like soon to increase our confidence19:19
clarkbbut I'm happy to help with the process which is something like put servers in emergency file, edit docker compose.yaml by hand and upgrade the first follower, repeat on the second follower after checking which node is leader, then finally do the last node19:20
clarkbthe release notes for 3.9 say no special steps are required to upgrade from 3.8 to 3.9 so it should be striaghtforward if we use the normal process19:20
corvusi could do it this saturday morning (my time)19:21
corvusyeah, i also went over the notes and didn't see anything19:21
corvusalso, a lot of our zuul tests have already been using 3.919:21
clarkband when done we can merge that change and pull the nodes out of the emergency file19:21
clarkbso I guess heads up, review the upgrade change but don't approve it and if you have any concerns please raise them19:21
corvusthat process sounds good to me, and it sounds like if no one objects we could do it saturday19:22
corvuswe should make sure to take a zuul zk backup before starting too, just in case19:22
clarkb++19:22
corvus(with zuul-client)19:22
clarkbI just approved the test fix that the zk upgrade is a child of19:23
tonybsounds good to me19:23
clarkb#topic Gitea 1.25.1 Upgrade19:24
clarkb#link https://review.opendev.org/c/opendev/system-config/+/965960 Upgrade Gitea to 1.25.119:24
clarkbhttps://158.69.67.86/opendev/system-config is a held node you can interact with to check this upgrade19:24
clarkbgerrit bug fix upgrades, gitea new release upgrade, and zookeeper upgrades all on tap this week19:25
clarkbI'd appreciate reviews of the change itself to make sure I haven't done anything silly when updating templates, but also read over the release notes and make sure there aren't new features we need to enable/disable/configure19:25
clarkbThis release seemed to avoid big changes like that so I think it should be easy but let me know19:25
clarkbmostly just trying to keep up so we don't fall behind19:26
clarkb#topic Gitea Performance19:26
clarkbThen related to that I spot checked giteas today and they all look busy but not to the point where they are slow19:26
clarkbboth the memcached memory increase and the "force everything through the load balancer" changes merged19:27
clarkbprobably a bit early to claim improvement, but not having evidence of problems is something19:27
clarkbfungi: related I noticed this morning when prepping for the meeting that the lists server seems sad again19:27
clarkbI think mariadb is busy so we may have something crawling apis again and maybe we need to double check iops look reasonable still19:28
fungimmm19:28
clarkbbut wanted to call that out if we're discussing general performance issues related to crawlers19:28
fungiload average is hovering around 10 at the moment, yeah19:29
clarkbI suspect its the same story just hitting us in new and exciting ways as we continue to improve bottlenecks19:29
clarkbevery fixed bottleneck is an opportunity to find a new one19:29
clarkbPlease say something if you notice problems in gitea (or any other service).19:30
clarkb#topic Raxflex DFW3 Disabled19:31
clarkbI don't think this server has been fixed or replaced yet19:31
clarkblast week we basically said if after a week it wasn't fixed we'd boot a new one19:31
clarkbI think we can probably proceed with that plan now if anyone has time19:31
clarkb(my focus is probably on gerrit and gitea and whatever lists needs to be performant, but I'm happy to help if you point me to specific actions that are needed)19:31
clarkb#topic Open Discussion19:33
tonybI'll try but if someone else has cycles don't let me stop you19:33
clarkbThat was all I had on the agenda. I cut out afs stuff since trixie is mirrored now. I cut out launcher things because the major bug there was fixed. We also got vexxhost to address the gerrit vm issues. We upgraded etherpad too19:34
clarkball that to say we got a lot done last week and I was able to trim the agenda as a result. Thank you everyone for making that happen19:34
tonybyeah well done!19:35
fungigreat work everyone!19:36
clarkbmaybe we can upgrade gitea tomorrow and plan for gerrit thursday. tonyb we can sync up outside of the meeting on timing for gerrit19:37
clarkband with that I think we can probably end early if there is nothing else19:37
clarkbI have some zuul launcher bug fix code reviews I need to do then lunch19:37
clarkbthanks everyone. We'll be back here at the same time and location next week19:38
clarkb#endmeeting19:38
opendevmeetMeeting ended Tue Nov 11 19:38:27 2025 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)19:38
opendevmeetMinutes:        https://meetings.opendev.org/meetings/infra/2025/infra.2025-11-11-19.00.html19:38
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/infra/2025/infra.2025-11-11-19.00.txt19:38
opendevmeetLog:            https://meetings.opendev.org/meetings/infra/2025/infra.2025-11-11-19.00.log.html19:38

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