13:00:35 <haleyb> #startmeeting networking 13:00:35 <opendevmeet> Meeting started Tue May 20 13:00:35 2025 UTC and is due to finish in 60 minutes. The chair is haleyb. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:00:35 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:00:35 <opendevmeet> The meeting name has been set to 'networking' 13:00:37 <haleyb> Ping list: bcafarel, elvira, frickler, mlavalle, mtomaska, obondarev, slaweq, tobias-urdin, ykarel, lajoskatona, jlibosva, averdagu, haleyb, ralonsoh 13:00:38 <mlavalle> \o 13:00:46 <mtomaska> o/ 13:00:48 <ralonsoh> hello 13:00:48 <ykarel> o/ 13:00:56 <rubasov> o/ 13:00:58 <obondarev> o/ 13:01:30 <haleyb> alright we can get started 13:01:38 <lajoskatona> o/ 13:01:40 <haleyb> #announcements 13:01:56 <haleyb> We are currently in Week R-18 of Flamingo 13:02:14 <haleyb> we just completed Flamingo-1 13:02:24 <haleyb> Our next milestone in this development cycle will be Flamingo-2, on July 3rd 13:02:51 <haleyb> This milestone is when we freeze the list of deliverables that will be included in the 2025.2 Flamingo final release, so if you plan to introduce new deliverables in this release, please propose a change to add an empty deliverable file in the deliverables/flamingo directory of the openstack/releases repository 13:03:12 <haleyb> Final 2025.2 Flamingo release: October 3rd, 2025 13:03:20 <haleyb> #link https://releases.openstack.org/flamingo/schedule.html 13:03:37 <haleyb> Reminder: If you have a topic for the drivers meeting on Friday, please add it to the wiki @ https://wiki.openstack.org/wiki/Meetings/NeutronDrivers 13:04:01 <haleyb> the meeting last week was canceled, i will look at new rfes for this week 13:04:43 <bcafarel> late o/ 13:04:55 <haleyb> as i was out last week and only worked a few hours, any fires i should be aware of? i didn't see anything on the ML or irc backtrace 13:05:20 <ralonsoh> nothing urgent/critical 13:05:47 <haleyb> ack, thanks for taking care of the place :) 13:06:04 <haleyb> i had no other announcements, anyone else have something? 13:06:31 <haleyb> #topic bugs 13:06:40 <haleyb> bcafarel was the deputy last week, his report is at 13:06:49 <haleyb> #link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/GCVMPXGHVV535VPUNYP2LUDM6PKBKV5W/ 13:06:58 <haleyb> not too many bugs 13:07:09 <bcafarel> I got lucky it seems :) 13:07:51 <haleyb> and it looks like most got an owner and a patch 13:08:04 <haleyb> there was only one CI-related one 13:08:11 <haleyb> #link https://bugs.launchpad.net/neutron/+bug/2110878 13:08:22 <haleyb> only seen once 13:09:03 <ralonsoh> I'll check it today 13:09:25 <haleyb> ack, thanks 13:09:30 <haleyb> any other bugs to discuss? 13:09:32 <mtomaska> I forgot to include one bug in my last week report. https://bugs.launchpad.net/neutron/+bug/2110085 13:10:07 <haleyb> that might be a duplicate? 13:10:23 <haleyb> ralonsoh: that one looks like something you've been working on? 13:10:27 <trident> Any chance there could be some eyes on https://bugs.launchpad.net/ubuntu/+source/neutron/+bug/1924776 - If I have not misunderstood something, this kind of limits the use of address scopes with OVN pretty severely as it doesn't honor address scopes when taking the decision of using NAT or not between two networks connected to a router as I think it should. 13:10:34 <mtomaska> Duplicate? I have to look, but I feel like this is known issue 13:10:44 <ralonsoh> mtomaska, are there any logs? 13:11:13 <mtomaska> no 13:11:26 <slaweq> o/ sorry for being late 13:11:30 <mtomaska> except what is in the bug already 13:12:20 <ralonsoh> ok, let me check that out of this meeting 13:12:24 <ralonsoh> I'll assign this one to me 13:12:50 <haleyb> trident: that seems like an old bug, does this still happen on master branch? 13:13:26 <mtomaska> I feel like this was discussed somewhere that uwsgi is only single threaded.... or I might be confusing with something else. 13:13:41 <haleyb> trident: we can discuss after meeting 13:13:53 <lajoskatona> isn't that related to what gibi posted few hours back ? (https://meetings.opendev.org/irclogs/%23openstack-neutron/%23openstack-neutron.2025-05-20.log.html#openstack-neutron.2025-05-20.log.html#t2025-05-20T11:58:47) 13:14:03 <lajoskatona> I mean the wsgi thing? 13:14:05 <ralonsoh> lajoskatona, no, this one is using eventlet server 13:14:11 <lajoskatona> oh no that is something eventlet 13:14:18 <lajoskatona> ralonsoh: yeah, ack 13:14:21 <ralonsoh> the bug reported is using wsgi 13:14:34 <trident> haleyb: Ok, thanks! 13:14:49 <ralonsoh> trident, haleyb I tried last week to make this feature work 13:14:58 <ralonsoh> without any success 13:15:21 <ralonsoh> so I think this is not supported in ml2/ovn (I'm not saying in OVN, just could be a bad implementation in Neutron) 13:15:33 <ralonsoh> So I think the bug is legit 13:15:38 <haleyb> yes, i remember you doing some address scope testing 13:15:49 <ralonsoh> related to another feature, yes 13:15:55 <ralonsoh> let's talk after the meeting 13:16:04 <haleyb> ack 13:16:06 <opendevreview> Lajos Katona proposed openstack/os-vif master: VS Trunk: Add bridge_name to external_ids https://review.opendev.org/c/openstack/os-vif/+/949736 13:16:29 <haleyb> this week the deputy is elvira, next week is slaweq - does that work for both of you? 13:18:00 <haleyb> ok, i will ping people again later 13:18:17 <haleyb> #topic community goals 13:18:42 <haleyb> lajoskatona: any update on neutronclient changes? 13:18:51 <slaweq> yeap 13:18:58 <lajoskatona> yes, 13:19:34 <lajoskatona> yesterday I found (thanks for reviewers in horizon) that in SDK there is no shared field for sec-groups: https://review.opendev.org/c/openstack/openstacksdk/+/950305 13:19:55 <elvira> (yes, it works, sorry for the late answer o/) 13:20:11 <lajoskatona> and pushed also a patch for QoS: https://review.opendev.org/c/openstack/horizon/+/949764 13:20:12 <haleyb> elvira, slaweq - thanks 13:21:06 <lajoskatona> but no time for fullstack vs SDK patch (the no-auth problem) :https://review.opendev.org/c/openstack/neutron/+/947851 13:21:10 <haleyb> lajoskatona: thanks! i've started watching both 13:21:17 <lajoskatona> that's it for neutronclient 13:21:34 <haleyb> great, thanks for the update 13:21:35 <lajoskatona> some CI is failing I have to go back to see what I broke 13:21:52 <haleyb> ack 13:22:20 <haleyb> ralonsoh: and finally eventlet 13:22:31 <ralonsoh> nothing new in gerrit 13:22:41 <ralonsoh> I've started with the migration of the tests, but locally 13:22:56 <ralonsoh> please check https://review.opendev.org/c/openstack/neutron/+/949135 13:23:07 <ralonsoh> something similar will be needed for the DHCP agent 13:23:11 <ralonsoh> that's all 13:23:31 <haleyb> ack, thanks, will look 13:23:57 <lajoskatona> I work on the dhcp agent 13:24:03 <ralonsoh> lajoskatona, thanks! 13:24:12 <haleyb> ralonsoh: there was one change that Liu was having an issue with, do we need to discuss that further? i'd have to find the link 13:24:21 <ralonsoh> one sec 13:24:31 <ralonsoh> https://review.opendev.org/c/openstack/neutron/+/948200 13:24:39 <ralonsoh> now with functional tests 13:25:05 <ralonsoh> we can talk about this one in the on-demand section 13:25:13 <ralonsoh> actually, I would like to bring it back 13:25:50 <lajoskatona> but for that topic there are patches from Liu also: https://review.opendev.org/q/topic:%22bug/2106463%22 13:26:15 <haleyb> ralonsoh: i can start on-demand 13:26:26 <haleyb> #topic on-demand 13:27:03 <ralonsoh> lajoskatona, yes, I also mentioned that the DB constraint introduced is against what we consider a tunnelled network 13:27:09 <opendevreview> Merged openstack/neutron-lib master: api-ref: Add warning about DRA scheduler https://review.opendev.org/c/openstack/neutron-lib/+/949744 13:27:19 <ralonsoh> for any tunnelled network, the physical network is None, not an empty string 13:27:49 <ralonsoh> and as I commented, my patch is backportable, fixes the problem with multiple workers (this is now executed in the first one only) 13:28:00 <ralonsoh> and works for multiple servers booting at the same time 13:29:28 <haleyb> ralonsoh: is there any merit to also doing the DB constraint going forward? although it's not necessary with all your code 13:29:45 <ralonsoh> it is not necessary this db constraint 13:29:57 <ralonsoh> I also remember trying to introduce something similar years ago 13:30:36 <haleyb> ack 13:30:43 <ralonsoh> in any case, if we want the constraint added, it could be on top of my patch 13:31:01 <haleyb> yes, not instead of, on-top for now and future 13:31:06 <ralonsoh> no problem **IF** we don't add an empty string for tunnelled networks in the physical network 13:31:07 <haleyb> we need to backport 13:31:19 <ralonsoh> yes, my patch is enough and backportable 13:32:22 <haleyb> so i guess we can wait for his response, i also do not understand 13:32:50 <opendevreview> Merged openstack/neutron-lib master: api-ref: Remove crud from conf.py https://review.opendev.org/c/openstack/neutron-lib/+/949745 13:33:33 <ralonsoh> I have another quick topic 13:33:47 <ralonsoh> #link https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/949140 13:34:15 <ralonsoh> I addressed yatin's comment: I'm using ubuntu server only in OVN, due to the issues with IGMP 13:34:32 <ralonsoh> about lajoskatona's comment: the issue is that using Focal VM is a problem in the CI 13:34:46 <lajoskatona> ralonsoh: ack, checking 13:34:50 <ralonsoh> I understand the logic of this comment: do not change what is working 13:35:00 <haleyb> ralonsoh: yes, we should move to jammy or jammy server 13:35:09 <ralonsoh> but the problem is that we are hitting some errors in the CI with focal image 13:35:17 <ralonsoh> haleyb, I've used both 13:35:31 <ralonsoh> minimal by default, server only in ovn (for IGMP) 13:35:39 <haleyb> ah, ok, one is the minimal now 13:35:45 <ralonsoh> with a note to go back to minimal if there is a fix 13:36:12 <ralonsoh> and that's all, thanks! 13:36:42 <haleyb> ack, thanks, i did start watching that bug too 13:37:25 <haleyb> so last on-demand topic is mine 13:37:45 <haleyb> i created backports for a patch that landed in 2025.1 13:37:54 <haleyb> #link https://review.opendev.org/c/openstack/neutron/+/851509 13:38:06 <haleyb> #link https://bugs.launchpad.net/neutron/+bug/1982287 13:38:24 <haleyb> the bug was added as an rfe 13:38:58 <haleyb> since the dependend ovsdbapp code landed in 2024.1 i wanted to backport this neutron fix to there as well 13:39:44 <haleyb> i do understand ralonsoh's point last week that this was a feature, but i feel it was something missed 13:40:32 <ralonsoh> if we accept these backports, I would change the reno from "feature" to "other" in the stable branches 13:40:36 <haleyb> ralonsoh: i am relating this change to the one we did for ovs hybrid driver 13:40:40 <haleyb> #link https://review.opendev.org/c/openstack/neutron/+/913708 13:41:01 <haleyb> that one we backported from 2024.1 to yoga 13:41:15 <haleyb> so i wanted to get other opinions on the OVN one 13:41:34 <haleyb> i did find a tempest patch that never merged to test it all, which i have updated 13:41:56 <haleyb> #link https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/906020 13:42:17 <ralonsoh> I'm ok, but let's change the reno to avoid the wrath of release team. I would add something like "testing feature" 13:42:17 <haleyb> that seems to have failed miserably overnight 13:43:07 <haleyb> ralonsoh: ack, i can update the release note on the backports 13:43:11 <lajoskatona> The RFE label is really a headsup thing but as I understand this patch fixed the OVN backend 13:43:45 <haleyb> i am worried about why the n-t-p change took a turn, but will have to look at logs 13:43:46 <lajoskatona> +1 13:44:09 <ralonsoh> yeah, lets make the n-t-p patch work with the dependencies to the backports 13:44:10 <haleyb> once the backports merge i will update n-t-p to test on those branches 13:44:23 <haleyb> ralonsoh: sure, i could do that as well 13:44:37 <ralonsoh> just to test before merge 13:45:09 <haleyb> i will do that right after meeting, is related to something hot downstream 13:45:40 <haleyb> thanks for the discussion everyone 13:46:03 <haleyb> that was all the topics, anything else to discuss? 13:46:30 <haleyb> ok, thanks for attending everyone, have a great week 13:46:33 <haleyb> #endmeeting