19:00:34 <clarkb> #startmeeting infra 19:00:34 <opendevmeet> Meeting started Tue May 13 19:00:34 2025 UTC and is due to finish in 60 minutes. The chair is clarkb. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:00:34 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 19:00:34 <opendevmeet> The meeting name has been set to 'infra' 19:00:50 <clarkb> #link https://lists.opendev.org/archives/list/service-discuss@lists.opendev.org/thread/B3CAZDR2RDA672TKGETT2SAIRISBMKUF/ Our Agenda 19:00:55 <clarkb> #topic Announcements 19:01:17 <clarkb> currently finishing up my last meeting. But wanted to get things started here so there wasn't confusion over whether it was happening 19:01:37 <tonyb> Noted 19:02:44 <clarkb> I dno't have anything else to announce. Were there other announcements? 19:02:57 <tonyb> Not from me 19:03:51 <fungi> summit cfp closes in a month 19:03:57 <clarkb> June 13 is the deadline 19:04:20 <fungi> #link https://summit2025.openinfra.org/cfp/ 19:04:54 <clarkb> #topic Zuul-launcher image builds 19:04:56 <fungi> per the oppeninfra foundation board meeting today, openinfra projects may not need cla enforcement in a month 19:05:02 <clarkb> #undo 19:05:02 <opendevmeet> Removing item from minutes: #topic Zuul-launcher image builds 19:05:16 <clarkb> ya I guess its worth making mention of that as we may be asked to assist with dco enforcement 19:05:32 <fungi> just a heads up there may be some semi-urgent acl changes when that happens 19:05:43 <clarkb> but I think we (opendev) are in a holding pattern until there is stronger direction 19:05:52 <fungi> correct 19:06:00 <fungi> more just early warning 19:06:31 <clarkb> #topic Zuul-launcher image builds 19:06:34 <clarkb> ok lets dive into the agenda 19:06:47 <clarkb> mnasiadka continues to be the zuul launcher port nodepool images wizard 19:06:57 <corvus> yay! 19:06:58 <clarkb> we should have ubuntu and debian images now. CentOS and rocky are in progress 19:07:12 <clarkb> and this covers both x86 and arm images too which is great 19:07:40 <corvus> i think probably most of the major gaps are covered at this point? 19:07:42 <clarkb> I also wanted to mention that corvus added grafana dashboards for zuul-launcher against each of the cloud regions too. So we have better insight into how things are going there without looking in zookeeper directly 19:07:47 <corvus> like, do we have enough to satisfy openstack? 19:08:08 <clarkb> corvus: almost. I think openstack uses a fair bit of centos. There aws a bunch of discussion about that in the TC meeting earlier today 19:08:21 <clarkb> once centos is in then yes I think we're basically there for something like openstack 19:08:36 <corvus> groovy 19:08:40 * tonyb should go read the TC logs 19:09:16 <clarkb> corvus: from an implementation perspective are there any gaps you are concerned about? Do we need better node deletion tooling based on the recent need for that? 19:09:53 <corvus> yeah, we are still missing some manual tasks, so we probably shouldn't switch over more tenants before that 19:10:14 <corvus> (like "nodepool list/delete" equivalents) 19:10:25 <corvus> but with the latest changes, all the known issues resulting in leaked nodes, etc, should be fixed, so if something looks weird, speak up 19:10:54 <clarkb> sounds good. And zuul is dogfooding it 19:11:21 <clarkb> Other than helping mnasiadka and any other image build volunteers is there anything else we should look at doing on the opendev side? 19:11:44 <corvus> don't think so at the moment 19:11:53 <clarkb> great 19:11:56 <clarkb> #topic Gerrit 3.11 Upgrade Planning 19:12:37 <clarkb> Now that we've replaced the Gerrit server the next Gerrit item on my mind is upgrading Gerrit. Gerrit 3.12 is about to release and we've managed to roughly keep up upgrading around when the next release comes out which puts that in the now timeframe 19:12:47 <clarkb> it also works well with the openstack release schedule. 19:12:53 <clarkb> #link https://www.gerritcodereview.com/3.11.html 19:13:15 <clarkb> My rough plan here is to hold some nodes so that I can manually work through upgrade and downgrade steps and make sure that process works as expected 19:13:36 <clarkb> can also use that held node to test some behavior things if anything in the linked release notes concerns us (there is usually a small number of things to double check) 19:14:07 <clarkb> if others have time to look over that 3.11 release page in order to call out concerns that would be great. And I can volunteer to start putting together a plan and do the testing I described 19:14:35 <clarkb> thinking out loud I think this is possible late may/early june. Though early june might be weird for me I'm sure I can make time for it if that ends up being the timeline that works 19:14:56 <clarkb> any thoughts/concerns/input on this item? 19:15:06 <fungi> i think that timing sounds good 19:15:33 <fungi> i did start to go over the release notes a while back, for other reasons, and it's... not brief 19:15:48 <clarkb> yes, I think this upgrade is a little more involved than the last ~3 or 4 we'ev done 19:16:09 <clarkb> the java stuff changes. But shouldn't affect us because we build our own wars 19:16:17 <clarkb> if we were using upstream war builds then we would have to upgrade to java 21 first 19:16:34 <clarkb> intead we'll build a java 17 war and run with javav 17 on bookworm then switch to 21 when trixie becomes an option 19:17:43 <clarkb> #topic Upgrading old servers 19:17:59 <clarkb> The Gerrit server move was related to this but then all the followup kinda threw me off this horse 19:18:20 <clarkb> I'm hoping that I can back to it either this week or next so be on the lookout for more server swaps. Also happy for others to jump in and do some if they are able 19:18:25 <clarkb> there is more than enough work to share here 19:18:53 <clarkb> The noble rsyslog apparmor issue has been accepted by ubuntu 19:19:14 <clarkb> not sure that means we'll get a fix in noble but our workaround is fine and maybe for the 2026 release we won't need a workaround 19:19:44 <clarkb> fungi: we didn't manage to get a definitive answer on refstack shutdown did we? 19:19:56 <clarkb> that is also semi related to this topic (not replacement/upgrade but cleanup)_ 19:19:57 <fungi> no, i need to pick that discussion back up 19:20:09 <fungi> i meant to try to cover it in person during meetings last week with staff 19:20:18 <fungi> (and failed) 19:20:23 <clarkb> ack 19:20:36 <clarkb> Anyone else have server upgrade updates? Otherwise I think we can continue on 19:21:28 <clarkb> #topic Working through our TODO list 19:21:33 <clarkb> #link https://etherpad.opendev.org/p/opendev-january-2025-meetup 19:21:39 <clarkb> reminder we have a high level todo list on this etherpad 19:22:02 <clarkb> if you woudl like to get more involved with opendev pulling up this list is a great place to start. Feel free to reach out to me with any questions you may have about specific tasks. 19:22:19 <clarkb> And for those of us that are regular contributors don't forget to update that list if things get completed or add them if you notice things that need attention 19:22:32 <clarkb> #topic Rotating mailman 3 logs 19:22:39 <clarkb> #link https://review.opendev.org/c/opendev/system-config/+/948478 19:23:00 <clarkb> fungi did end up writing a change for this. I think I convinced myself that we can probably just try it and if it creates problems then backing it out is relatively easy. 19:23:25 <clarkb> email delivery is a process that gets reattempted so if we have to shut stuff down for a short period to back out that is fine 19:24:34 <clarkb> but if other reviewers have additional concerns feel free to raise them on the change and we can sort out more testing ahead of time if necessary 19:24:43 <clarkb> fungi: anything else to add to this one? 19:24:52 <fungi> nope! 19:24:57 <fungi> that sums it up nicely 19:25:20 <clarkb> #topic Open Discussion 19:25:34 <clarkb> We ended up with a fairly short agenda today after I cleaned things up yesterday. 19:25:41 * tonyb has a thing 19:25:46 <clarkb> However, there are two thigns I wanted to call out that came up / was reminded about afterwards 19:26:04 <tonyb> you go first clarkb 19:26:06 <clarkb> The first is I have a change up to upgrade gitea from 1.23.7 to 1.23.8 whcih should be straightforward if we want to proceed with that 19:26:37 <clarkb> and the other is setuptools made a release that broke pbr ~9 days ago. They backed that out, but now pbr is in a position where it has a shot clock to get updates in before setuptools breaks things again 19:27:10 <clarkb> it is probably a good idea for us to help the oslo folks get that sorted out in a reasonable amount of time. But I'm hoping they drive it and we can help with review, ci, testing, etc 19:27:20 <clarkb> tonyb: go for it 19:28:11 <tonyb> Is your point related to frickler's https://meetings.opendev.org/irclogs/%23opendev/%23opendev.2025-05-13.log.html#opendev.2025-05-13.log.html#t2025-05-13T09:06:03 " infra-root: missed to add this to the agenda (and likely won't be at the meering, either) but maybe we should discuss a strategy for dealing with the pbr CI issues? https://review.opendev.org/q/project:openstack/pbr+status:open or should this rather be 19:28:11 <tonyb> discussed in the oslo team context?" ? 19:28:40 <clarkb> tonyb: yes, basically pbr needs some fundamental updates to address setuptools problems but before it can make those updates needs to have working CI again 19:28:51 <tonyb> Okay cool 19:29:10 <tonyb> I wasn't sure and didn't want frickler's thing missed 19:29:17 <fungi> so two separate but connected problems 19:29:23 <tonyb> Okay 19:29:44 <clarkb> my suggestion is that we support oslo (something we've done with pbr for a long time) but not get in the drivers seat unless someone has a lot of time and interest they want to put into that 19:30:05 <clarkb> openstack in particular relies on features in pbr that aren't supported by setuptools scm 19:30:25 <clarkb> opendev and zuul etc currently rely on pbr but don't rely on those features so for us worst case we just change the package management system 19:30:56 <clarkb> so I'm hoping openstack leads the way here :) 19:31:17 <clarkb> tonyb: was that your item? Or was there something else too? 19:31:30 <tonyb> In https://review.opendev.org/c/openstack/project-config/+/948033 (discussion about adding RDO to opendev) clarkb asked where x86_64-v3 is actually available so I wanted to check for objections before I launch a node in each cloud+region to check for cpu-flags to answer that 19:31:54 <clarkb> tonyb: I think that is a great idea and a good way to get concrete info. No objection from me 19:32:00 <fungi> i have no objections to doing that, though you could also probably just check /proc/cpuinfo on the mirrors? 19:32:18 <clarkb> tonyb: I would boot them in the jenkins/zuul tenant just in case nova scheduling by tenant has an impact (I doubt it will but may as well do it that way to be sure) 19:32:25 <fungi> good point 19:32:32 <tonyb> In addition I want to say that RDO is in a ... state of flux ... and after I published that review it became possible that RDO would go to gitlab to follow CentOS 19:32:49 <fungi> or i guess you could log into running ci nodes and cat /proc/cpuinfo 19:33:00 <tonyb> Okay cool that's a good thing to point oit 19:33:22 <tonyb> *out 19:33:26 <clarkb> ya you want to use the same tenant and flavor as much as possible I think 19:33:32 <clarkb> since clouds can do weird scheduling things iirc 19:34:46 <tonyb> fungi: I could ... do that I'd be a little worried about futzing up jobs but `(ssh root@${node} -q cat /proc/cpuinfo) > ${node}.cpuinfo` should be pretty safe 19:34:54 <clarkb> re RDO's home I guess the main thing on our side is to let us know if a decision is made one way or another so that we can either help or standdown on any necessary pre debugging 19:35:11 <tonyb> Yup will do 19:35:48 <tonyb> Okay that's all from me 19:35:53 <clarkb> I'll leave things open until 19:40 if there is anything else. Otherwise I think we can all go grab $meal early. I'm still on texas time so hungry for lunch 19:36:39 <fungi> yeah, i have texas stomach as well, feels like it should be lunchtime still 19:37:17 <corvus> ya'll gonna need a texas-sized lunch? 19:37:38 <clarkb> maybe. We have a pot of beans on that I have had to smell all morning and it has made me very hungry 19:38:52 <clarkb> ok sounding like that really was everything 19:38:55 <clarkb> Thank you everyone! 19:39:12 <clarkb> We will be back at the same time and location next week. Probably with a fuller agenda 19:39:20 <fungi> thankfully i think texas-sized lunches are behind me for a while to come 19:39:29 <fungi> still recovering 19:39:31 <clarkb> and as always feel free to reach out on the mailing list or in #opendev if anything comes up between now and then 19:39:48 <clarkb> #endmeeting