13:01:21 <mnasiadka> #startmeeting kolla 13:01:21 <opendevmeet> Meeting started Wed Apr 30 13:01:21 2025 UTC and is due to finish in 60 minutes. The chair is mnasiadka. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:01:21 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:01:21 <opendevmeet> The meeting name has been set to 'kolla' 13:01:24 <mnasiadka> #topic rollcall 13:01:26 <mnasiadka> o/ 13:01:29 <frickler> \o 13:01:54 <opendevreview> Michal Nasiadka proposed openstack/kolla master: neutron: Add agents wrappers in the neutron-base image https://review.opendev.org/c/openstack/kolla/+/946215 13:01:54 <frickler> mnasiadka: do a courtesy ping since you didn't give advance warning? 13:02:01 <mnasiadka> right 13:02:28 <mnasiadka> courtesy ping: evko SvenKieske mmalchuk gkoper jangutter jsuazo jovial osmanlicilegi mattcrees dougszu darmach pabloclsn ravlew amir58118 13:02:51 <MattCrees[m]> o/ 13:03:54 <frickler> kevko 13:04:25 <frickler> I guess we should also do a spring cleaning for that list. certainly sven is no longer around 13:06:00 <mnasiadka> #topic Agenda 13:06:03 <mnasiadka> * CI status 13:06:09 <mnasiadka> * Release tasks 13:06:15 <mnasiadka> * Additional agenda (from whiteboard) 13:06:19 <mnasiadka> * Open discussion 13:06:22 <mnasiadka> #topic CI status 13:06:35 <frickler> green? 13:06:47 <mnasiadka> CI is overall green, but I noticed in the ML2/OVS patch that upgrade is seriously unstable 13:06:51 <mnasiadka> as in upgrade jobs 13:07:11 <mnasiadka> I'll have a look, but the most problematic thing is some RMQ message timeouts 13:07:34 <mnasiadka> #topic Release tasks 13:07:44 <mnasiadka> Is anything stopping us from doing the release today? 13:08:09 <mnasiadka> No RP+2 in Kolla 13:08:17 <frickler> there are a couple of RP+2 changes, but I'm not sure we need all of them 13:08:17 <mnasiadka> Two LE related patches which are features 13:08:23 <mnasiadka> and one around Thales 13:08:56 <mnasiadka> frickler: I think we need to update Gerrit acls to be able to remove RP - because it's a bit problematic now 13:09:07 <frickler> "doing the release" = create rc1 and branch? 13:09:32 <frickler> mnasiadka: ah, you could override with RP-1 or -2 and amend your filter 13:09:38 <mnasiadka> yes, cutting rc1 13:10:07 <mnasiadka> I don't think docker to podman migration is ready, and it's a bit late to merge it I guess 13:10:13 <frickler> but we can also look into amending gerrit acl 13:10:56 <frickler> IMO the migration needs more testing, we will get that only when we merge it 13:11:13 <frickler> maybe add some bigger dragons warning to it for now 13:11:33 <mnasiadka> I'd be happy to get the ML2/OVS improvement in as a beta feature/technology preview - but we can delay it as well to F/2025.2 13:12:15 <mnasiadka> so let's try to get docker to podman migration in 13:12:21 <mnasiadka> with a big dragon warning 13:12:32 <mnasiadka> and the same with ML2/OVS improvement? 13:12:39 <mnasiadka> And then cut rc1 next week? 13:12:58 <frickler> sounds like a plan, ack 13:13:15 <mnasiadka> good, we have an agreement 13:13:31 <mnasiadka> #topic Additional agenda 13:13:41 <mnasiadka> There is this: https://review.opendev.org/c/openstack/kolla-ansible/+/934696 13:14:36 <frickler> I'm not convinced we need that upstream 13:14:49 <frickler> but I'm biased since I always use ceph anyway 13:14:57 <mnasiadka> I don't think it's ready, so let's skip it 13:15:07 <mnasiadka> I commented if Cinder Backup needs the same 13:15:32 <mnasiadka> But basically as proposed for the PTG - I'm planning to do a big cleanup of ansible/group_vars/all and other things to get a bit smaller CI footprint 13:15:52 <mnasiadka> #topic Open discussion 13:16:26 <mnasiadka> Anybody anything? 13:16:29 <frickler> do we want to schedule a post-ptg meetpad session? if yes, when 13:17:37 <frickler> like maybe just do the weekly meeting on meetpad? in, say, 2 weeks? 13:17:39 <mnasiadka> Yes, I think we should. I guess after the 2025.1 release would be a good plan 13:18:01 <mnasiadka> Makes sense, I'll send a mail that's the plan. 13:18:18 <mnasiadka> 13th of May? 13:18:27 <mnasiadka> or 21st? 13:18:47 <frickler> 14th in my calender 13:18:55 <mmalchuk> o/ 13:19:20 <mnasiadka> Yes, 14th 13:19:22 <frickler> maybe plan for 2h though, just in case? 13:19:25 <mnasiadka> I'm still blind :) 13:19:33 <frickler> so 13-15 UTC? 13:19:34 <mnasiadka> Yeah, 2h should be enough - 1h might not be. 13:19:54 <mnasiadka> So 14th of May, 13-15 UTC - post-ptg meetpad session 13:20:09 <frickler> +1 13:20:22 <mmalchuk> have a question about monitoring stack 13:20:29 <frickler> do we need a new etherpad or just continue on the ptg one? 13:20:37 <mmalchuk> we planned to deprecate it now 13:20:52 <mmalchuk> its related all including exporters? 13:21:02 <mnasiadka> frickler: I think let's continue on the PTG one 13:21:07 <mnasiadka> deprecate monitoring stack? where? 13:21:28 <frickler> that sounds like news to me, too 13:21:39 <frickler> +1 to same pad 13:21:48 <mnasiadka> We maybe wanted to deprecate collectd and co - because I don't think anybody uses it 13:21:57 <mnasiadka> But I lacked time to do it, so maybe next time. 13:22:03 <mmalchuk> https://etherpad.opendev.org/p/kolla-epoxy-ptg 13:22:07 <mmalchuk> line 120 13:22:33 <mmalchuk> (mnasiadka): Monitoring stack - Prometheus, TICK (Telegraf and others), Collectd, etc - should we stick to one? 13:22:52 <mmalchuk> Actions: Deprecate in E, drop in F (if nobody wants to step in and maintain, like at least some basic CI jobs) +1+1 13:23:01 <mnasiadka> Yes, stick to one - so get rid of TICK and collectd 13:23:04 <mnasiadka> Leave only Prometheus 13:23:26 <mnasiadka> That was at least my plan 13:23:37 <mnasiadka> Let's rediscuss on 14th of May 13:23:42 <mmalchuk> so get rid of TICK and Collected only? yes? 13:23:53 <frickler> that's something to discuss then, sure. but +1 from me already 13:23:55 <mnasiadka> Yes, that was my plan, two people agreed :) 13:24:39 <mmalchuk> what about to switch to Victoria Metrics ? its more mature than Prometheus 13:24:53 <frickler> never heard about that 13:25:17 <mmalchuk> https://victoriametrics.com/ 13:25:22 <mnasiadka> Prometheus is more widely adopted, has a community - are you going to facilitate the migration from one to the other? 13:25:34 <mmalchuk> https://github.com/VictoriaMetrics/VictoriaMetrics 13:26:01 <mmalchuk> it cat be used as an alternative 13:26:07 <mnasiadka> I think we've been into that discussion in the past 13:26:15 <mnasiadka> I don't want to maintain two different monitoring stacks 13:26:20 <mmalchuk> prometheus exporters work with it 13:26:30 <mnasiadka> Or in a different way - do we have enough maintainers? 13:26:33 <frickler> sounds like yet another open core solution 13:27:11 <mmalchuk> ok. anyway I've got an answers, thanks. 13:27:23 <mnasiadka> 320 contributors vs 1k contributors for Prometheus, looks like open source but there's probably one company behind it 13:28:02 <mnasiadka> I'd say we clean up the other monitoring stacks and stick with Prometheus for now, we can have a discussion once that happens if we want to support two similar solutions 13:29:03 <mnasiadka> Although we don't have any long term metrics storage solution, so we could use Victoria Metrics for that 13:29:16 <mnasiadka> If there's a contributor wanting to do that 13:29:49 <mnasiadka> Anyway, let's focus on getting these two last features in - and let's try to cut rc1 next week 13:29:53 <mnasiadka> Thank you all for coming :) 13:29:56 <mnasiadka> #endmeeting