Tuesday, 2024-02-20

clarkbJust about meeting time18:59
clarkb#startmeeting infra19:01
opendevmeetMeeting started Tue Feb 20 19:01:37 2024 UTC and is due to finish in 60 minutes.  The chair is clarkb. Information about MeetBot at http://wiki.debian.org/MeetBot.19:01
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.19:01
opendevmeetThe meeting name has been set to 'infra'19:01
clarkb#link https://lists.opendev.org/archives/list/service-discuss@lists.opendev.org/thread/DHYKPAV5EU5SAIYRYPR6YUHNTZIIKJSD/ Our Agenda19:01
clarkb#topic Announcements19:01
clarkbService coordinator nominations open February 6, 2024 - February 20, 202419:01
clarkb#link https://lists.opendev.org/archives/list/service-discuss@lists.opendev.org/thread/TB2OFBIGWZEYC7L4MCYA46EXIX5T47TY/19:01
clarkbit is ~19:01 UTC now and nominations end at the end of day today UTC time19:02
clarkbI suspect this is a case of people being happy with the status quo and I should make that official?19:03
clarkbor do we have any last minute voluinteers?19:03
* frickler needs 10min break after tc meeting ran long, bbl19:03
clarkbIf no one indicates they will volunteer by the end of the meeting I guess I'll make it official19:04
clarkb#topic Server Upgrades19:05
clarkb#link https://review.opendev.org/c/opendev/system-config/+/905510 Upgrading meetpad service to jammy19:05
clarkbstill waiting on this stack.19:05
clarkbThe change to filter out nodes with insufficient cpu flags did merge though19:05
clarkbDon't really have anything else to add here. tonyb I guess let me or others know if we can help with the meetpad stuff at all19:07
clarkband with that we can dive into the prePTG summary. But I'll pause for a few minutes here since frickler may be interested in this19:08
clarkb#topic PrePTG Summary19:08
clarkb#link https://lists.opendev.org/archives/list/service-discuss@lists.opendev.org/thread/GJFCZJ5QECXYNIIF62RCOT2DDKRGTAC6/ PrePTG Summary19:08
fungithanks for sending that out!19:08
corvusi see a mention of hound on the agenda19:09
corvusi think that's basically a completely stateless server?  and a redeploy from scratch with new os/versions should be fine?19:09
clarkbcorvus: oh yup I think I need to prune that as we did end up resolving that19:09
fungii vaguely recall discussing that recently, and such was the determination19:10
clarkbcorvus: the tldr is yes that server can just be replaced without issue19:10
corvusthen yes i agree :)19:10
corvus(i probably just didn't say that last time cause someone else did :)19:10
clarkbWe had our Pre PTG event last week and I think it went well and was useful.19:10
clarkbIf you disagree on that or have feedback to make it more useful that would be useful info if we plan another. Feel free to PM me that info if you prefer19:11
clarkbOne thing I wanted to call out is that we seemed to have general agreement on review policies and practices which is good. I'm glad there were no surprises there19:11
clarkbtl;dr is that reviewers should use judgement when single core approving changes to ensure they are low impact or safe (based on testing etc)19:12
clarkband that we should continue to do our best to approve things when we acn watch them land and deploy. Holding off on approvals is appropriate if you can't do that19:12
corvusas a part-time contributor, i found it very useful to ensure everyone was up to speed on what's going on, who's working on what, what could use attention, sharing what we're interested and working on, other expectations, etc.  i didn't leave with a lot of work assigned to me (some, but not a huge amount), but nonetheless found the whole time useful.19:13
corvusand i think the summary reflects that19:13
clarkbthat is good to hear. It wasn't a ton of effort to put together but it was non zero effort so I'm happy people foudn it useful19:14
clarkbI did identify three topics I wanted to followup on here in our meeting, but afterwards we can dive into any others if there is interest19:14
fungii found it useful for talking through some stuff that i may have pushed to the back burner and forgotten about19:15
clarkb#topic MariaDB Upgrades19:15
clarkbfungi: agreed I found it useful to have a refresher and also sitrep on whether or not stuff ahs changed since we lst considered them19:15
clarkbPaste, Etherpad, Gitea, Gerrit, Refstack, and Mailman databases could use upgrades.19:15
clarkbDuring the preptg we agreed that we would go ahead and try this with the container image's MARIADB_AUTO_UPGRADE flag19:16
clarkbI'ev since held a paste node to test this, but haven't done any testing yet.19:16
clarkbMy rough plan there is to add some pastes of varying size to the held node, then manually attempt an upgrade from 10.4 to 10.11 setting that flag19:16
clarkbif the data survives then I think we can proceed with upgrading production this way19:17
fungialso unicode snowman ;)19:17
clarkb++19:17
clarkblook for more info this week and maybe even an upgrade or two19:17
clarkb#topic AFS Mirror cleanups19:18
clarkbThis is sort of the evolution of the prior meetings' afs quotas topic19:18
clarkbDuring the PTG we agreed that OpenSUSE Leap, Debian Buster, CentOS 7 and Ubuntu Xenial could be cleaned up from nodepool and our mirrors19:18
clarkbthis should free up quite a bit of disk for more modern mirroring that we may want to perform (for example ubuntu has a release coming up and rocky etc have asked about mirroring too)19:19
clarkbThe other idea (I think from fungi) related to this was that we can process our apache logs on the mirrors to determine which mirror content is most used19:19
clarkbthen using that data determine if there are things we are hosting that we probably don't need to host any longer19:19
clarkbfor example usage data on python wheel cache/mirror content would be useful to determien if we can just claen all that stuff up19:20
clarkb#topic OpenDev Email Hosting19:21
clarkbthis is the last one I wanted to bring up today. And the reason is that we didn't reach consensus during the pre ptg19:21
clarkbwe did agree that having email hosting for opendev.org is something we should address. We can either host it ourselves or pay a reasonable amount for a hosted service to do it for us19:21
clarkbfor example I personally pay $45/year for fastmail service that I think would be sufficient here19:22
clarkbit would be great if everyone can think about that over the next week or two then we can see if we've got a preference for one or the other19:22
fricklerwould the foundation pay that?19:22
clarkbfrickler: I haven't confirmed that yet, but I suspect they would19:23
fungiit would definitely be one of the first things we'll check if interest is leaning that direction19:23
fricklerwe also might want to consider bot accounts like for bounce handling of lists19:23
frickleror are those local on lists.o.o anyway?19:24
clarkbthe way my $45/year fastmail service works is all email to my domain ends up in one account's inbox. But I can filter it from there19:24
fungibot accounts for bounce handling of which lists?19:24
fungifor mailman, it's integrated into how mailman works19:24
clarkbso we probably wouldn't try to have proper accounts for everything but setup a single inbox that can handle the traffic to opendev.org and organize it appropriately19:24
fungiit creates verp addresses on the fly for bounce checking19:24
fungiand also dynamic reply addresses for certain kinds of dmarc mitigation features19:25
fungiall of that is internal to mailman itself19:25
fricklerah, ok19:25
fungibut yes, for things like gerrit notification reply addresses, we currently funnel those into a sub-inbox of infra-root19:26
clarkbAnd that was all I had on the agenda19:26
clarkb#topic Open Discussion19:26
fungimay need something similar for keycloak if we turn on e-mail validation features19:26
clarkbAnything else? I'm happy to dig more into the preptg as well19:26
fungi30 minutes to prepare for my next meeting, i guess ;)19:28
fricklerI didn't get to read all of the pad and mail yet. and sorry again for now showing up, just didn't work out19:28
fricklers/now/not/19:28
clarkbthat said given this was useful I think it may be good to try and do them more often. Maybe twice a year or quarterly?19:28
clarkbif we increase the regularity we may not need as much time either which hopefully makes it easier to attend19:29
clarkbI'll have to think about this and look at calendars19:29
clarkbfrickler: when you do read the etherpad and email feel free to reach out with questions etc. Happy to do my best to recount19:30
fricklertwice a year in between ptgs would sound good I think19:30
fricklerclarkb: ack, will do19:31
clarkbI'll leave the meeting open a few more minutes for any other items. But ya maybe we can get a break between meetings fungi  :)19:31
fungia rare treat ;)19:32
clarkbsounds like that is it. Thanks everyone. We'll be back here next week at the same time and location19:34
clarkb#endmeeting19:34
opendevmeetMeeting ended Tue Feb 20 19:34:42 2024 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)19:34
opendevmeetMinutes:        https://meetings.opendev.org/meetings/infra/2024/infra.2024-02-20-19.01.html19:34
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/infra/2024/infra.2024-02-20-19.01.txt19:34
opendevmeetLog:            https://meetings.opendev.org/meetings/infra/2024/infra.2024-02-20-19.01.log.html19:34
fungithanks clarkb!19:34
clarkbre service coordinator nominations I'll make it official after my next meeting and take advantage of this break between meetings for find something to drink19:35

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!