14:03:47 <mnasiadka> #startmeeting kolla
14:03:47 <opendevmeet> Meeting started Wed Jan 15 14:03:47 2025 UTC and is due to finish in 60 minutes.  The chair is mnasiadka. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:03:47 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:03:47 <opendevmeet> The meeting name has been set to 'kolla'
14:03:50 <mnasiadka> #topic rollcall
14:03:50 <mnasiadka> o/
14:03:53 <jovial> 0/
14:03:57 <koperg[m]> o/
14:04:30 <bbezak> o/
14:04:44 <mmalchuk> o/
14:04:46 <mnasiadka> this one is going to be short, I have 16 minutes left :)
14:04:56 <mnasiadka> #topic agenda
14:05:03 <mnasiadka> * Current cycle planning
14:05:07 <mnasiadka> * Additional agenda (from whiteboard)
14:05:13 <mnasiadka> * Open discussion
14:05:20 <mnasiadka> #topic Current cycle planning
14:05:48 <mnasiadka> I'm working on OVN relays, unfortunately it will be a third patch, but I'll add other patch authors as co-authors ;-)
14:06:10 <mnasiadka> fluentd crusade finished - all logs should now be properly parsed and readable
14:06:17 <mnasiadka> at least those that we have in fluentd config
14:06:20 <bbezak> hurray
14:06:27 <mnasiadka> if something gets broken - CI will be red
14:06:33 <mnasiadka> so keep an eye on that :)
14:06:44 <mnasiadka> Anybody else working on something from the priority list?
14:06:56 <jovial> very nice
14:06:58 <mnasiadka> darmach: how's it going with Ubuntu 24.04 in Caracal?
14:08:38 <mnasiadka> Ok then, let's move forward
14:09:13 <mnasiadka> I think we'll skip the whiteboard list, because that's only request for reviews, and we don't have the majority of core reviewers - so I'll leave that for next time
14:09:38 <mnasiadka> #topic Open discussion
14:09:41 <mnasiadka> Anybody anything?
14:10:14 <jovial> I was looking at adding support for the pxe-filter service so that we have feature parity with standalone inspector
14:10:33 <jovial> https://review.opendev.org/c/openstack/kolla/+/939256
14:10:46 <jovial> https://review.opendev.org/c/openstack/kolla-ansible/+/939231
14:10:59 <jovial> One question one the implementation..
14:11:34 <jovial> I was trying to recreate the cleaning up of the hostsdir from: https://docs.openstack.org/ironic/latest/admin/inspection/pxe_filter.html
14:12:08 <jovial> and can see two ways: send the sighup to the dnsmasq container via systemd unit
14:12:14 <jovial> or a shared PID namespace
14:12:21 <jovial> what would be preferable?
14:12:55 <mmalchuk> I have question about ovn-bgp-agent. do we have plans  on it? there review in gerrit
14:13:00 <mnasiadka> sending sighup to systemd might be considered sort of unsafe and elevated privileges?
14:13:23 <mnasiadka> shared PID namespace probably would be the way to go that is more elegant
14:13:32 <mmalchuk> https://review.opendev.org/c/openstack/kolla-ansible/+/937066
14:13:57 <mnasiadka> mmalchuk: I'm aware of them, I don't have time to look in them - and probably won't have this cycle
14:14:14 <mmalchuk> may add PR+1 ?
14:14:28 <mmalchuk> RP+1
14:14:41 <jovial> I was thinking the ironic-pxe-filter container could have something like ExecStartPre=docker kill -s HUP ironic_dnsmasq, but I think i'd also need to declare a requires relationship so could get messy
14:15:07 <mnasiadka> This would need to be added by a core reviewer that thinks the patch is ready for in-depth reviews - I haven't looked at the patch
14:16:13 <mnasiadka> mmalchuk: I'm sorry I don't have better answer - but you can reach out to other core reviewers if they have time to look at that
14:16:43 <jovial> Currently I've gone with the shared pid namespace approach as it seemed like fewer changes - just wondered if that seemed reasonable really
14:16:46 <mmalchuk> ok, thanks
14:16:56 <mnasiadka> jovial: well, docker kill is another problematic thing - if they would share a PID namespace - you could just send a HUP signal without weird approaches?
14:17:37 <darmach> mnasiadka - Noble backport is in progress, running into issues and solving them one by one. Will likely take few more days.
14:17:50 <mnasiadka> Hard to say really, but as long as we can not use docker kill or systemctl restart or something like that - that's a good solution ;)
14:18:04 <jovial> sure, that was my thinking too. My only worry is that the PID file might be stable and we would ending up killing something random, but I could check that pid was dnsmasq before killing it
14:18:20 <jovial> stale*
14:18:24 <mnasiadka> well, that pid namespace really would be two containers, right?
14:18:43 <mnasiadka> so it would be hard to shoot something other than dnsmasq or the process that is reloading dnsmasq :)
14:18:59 <jovial> Do we use --pid container:some-container anywhere else? Or do we only have support for pid_mode: host?
14:19:16 <mnasiadka> I don't think we do
14:19:34 <yuval> did I missed the weekly meeting?
14:19:39 <mnasiadka> so might need some systemd units dependency to be introduced in systemd worker
14:20:05 <mnasiadka> yuval: we're finishing now, I need to run - sorry. But feel free to ask questions on the channel and we'll respond later.
14:20:11 <mnasiadka> sorry, need to run :(
14:20:13 <mnasiadka> #endmeeting