14:00:30 <haleyb> #startmeeting neutron_drivers 14:00:30 <opendevmeet> Meeting started Fri Jan 26 14:00:30 2024 UTC and is due to finish in 60 minutes. The chair is haleyb. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:30 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:30 <opendevmeet> The meeting name has been set to 'neutron_drivers' 14:00:40 <slaweq> o/ 14:00:41 <mlavalle> \o 14:00:41 <haleyb> Ping list: ykarel, mlavalle, mtomaska, slawek, obondarev, tobias-urdin, lajoskatona, amotoki 14:00:46 <obondarev> o/ 14:00:52 <lajoskatona> o/ 14:01:59 <ralonsoh> hi 14:02:22 <haleyb> alright, we have quorum 14:02:46 <mlavalle> \o/ 14:02:59 <mlavalle> first time this year 14:03:01 <haleyb> i forgot to update the wiki, but seems we have 3 things to review 14:03:43 <haleyb> first rfe 14:03:46 <haleyb> https://bugs.launchpad.net/neutron/+bug/2047849 14:03:54 <haleyb> [RFE] Start using oslo messaging namespaces 14:04:03 <haleyb> mtomaska__: are you here? 14:04:22 <haleyb> although i see ralonsoh sent out the patch 14:04:25 <ralonsoh> I've started with a POC: https://review.opendev.org/c/openstack/neutron/+/905309 14:04:45 <ralonsoh> this is duplicating the server RPCs: one with namespace, the other one without it 14:05:07 <ralonsoh> to allow the upgrades (newer servers, older agents) 14:05:15 <ralonsoh> of course, I found some issues... 14:05:29 <ralonsoh> in any case, the question here is if that should be approved or not 14:06:28 <ralonsoh> I'll start: yes, the RPC segregation should have been done since the beginning 14:06:45 <ralonsoh> RPC channels between different agents should have different namespaces and different APIs 14:06:49 <lajoskatona> +1 14:06:55 <ralonsoh> (now we are sharing some of them...) 14:07:06 <mlavalle> it's really technical debt 14:07:11 <mlavalle> so yes 14:07:12 <ralonsoh> exactly 14:07:44 <lajoskatona> but as you mentioned it isi tricky for the upgrade, C is slurp, am I right? 14:07:46 <haleyb> ralonsoh: so what happens during an upgrade? 14:08:03 <slaweq> for me there is no question if we should or not do that - we should BUT we need to have good plan and test well upgrades 14:08:07 <ralonsoh> during the upgrade the server will be listening to two namespaces 14:08:12 <slaweq> with keeping in mind SLURP releases 14:08:36 <ralonsoh> slaweq, right, this is mandatory and probably we'll need something more that grenade jobs 14:08:36 <slaweq> lajoskatona yes, C will be SLURP 14:09:15 <mlavalle> I don't think there is any doubt we should do this. The question is how 14:09:26 <obondarev> so what's the main issue that namespaces are solving? Is it just method names collisions? 14:09:51 <ralonsoh> yes, we had that in the DHCP agent recently 14:10:06 <ralonsoh> so each API should be independent 14:10:18 <obondarev> I see, thanks 14:10:24 <ralonsoh> and each RPC server should be listening only to th specific agents 14:11:30 <haleyb> and to the SLURP question, it seems we should get this done in C to be useful for next SLURP 14:11:51 <ralonsoh> hmmm to be honest, i don't know if I'll be able to finish that in C 14:12:02 <ralonsoh> I know our current release cadence 14:13:03 <lajoskatona> that is my feeling also we are quite in the middle of C for this 14:13:41 <haleyb> ack 14:13:44 <ralonsoh> in any case, this is not a high priority, is not affecting any deployment and should be considered as a tech debt 14:14:34 <haleyb> right, thanks 14:14:56 <lajoskatona> as I see we can prepare it (perhaps discuss during the ptg in detail) and activate it in E 14:16:30 <mlavalle> yes, we don't have a deluge of cases affected by this. so we have time 14:18:11 <slaweq> ++ 14:18:33 <haleyb> ok. and i guess that once it's complete we can decide if we activate in D or E 14:18:40 <haleyb> lets vote 14:18:44 <mlavalle> +1 14:18:49 <haleyb> +1 14:18:50 <obondarev> +1 14:19:00 <ralonsoh> +1 14:19:11 <slaweq> +1 14:19:39 <haleyb> great, and should i assign to you ralonsoh ? 14:19:45 <ralonsoh> yeah 14:19:51 <lajoskatona> +! 14:22:12 <haleyb> ok, there was another RFE filed by ralonsoh this week 14:22:16 <haleyb> https://bugs.launchpad.net/neutron/+bug/2049623 14:22:23 <haleyb> [RFE] Refactor OVS Trunk plugin to have one single port 14:22:37 <ralonsoh> If you don't mind, I'll first try to create a POC 14:22:47 <ralonsoh> if that works, then I'll present it here 14:22:53 <haleyb> ralonsoh: that was going to be my first question 14:22:55 <ralonsoh> if not, I'll close the bug 14:23:11 <opendevreview> morice proposed openstack/neutron master: [ovn] AZs distribution in L3 port scheduler https://review.opendev.org/c/openstack/neutron/+/892604 14:23:29 <ralonsoh> so please, discard this one for now 14:24:57 <haleyb> ok, and the other one we were also going to wait for a POC 14:24:58 <haleyb> https://bugs.launchpad.net/neutron/+bug/2045058 14:25:22 <haleyb> so i guess there is nothing else on the agenda 14:25:44 <haleyb> any other things anyone wants to discuss? 14:25:59 <mlavalle> not from me 14:26:03 <ralonsoh> no thanks 14:26:09 <lajoskatona> nothing from me 14:26:11 <slaweq> nothing from me 14:26:32 <haleyb> ok, thanks for attending and have a nice weekend! 14:26:36 <haleyb> #endmeeting