13:00:56 <haleyb> #startmeeting networking 13:00:56 <opendevmeet> Meeting started Tue Sep 16 13:00:56 2025 UTC and is due to finish in 60 minutes. The chair is haleyb. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:00:56 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:00:56 <opendevmeet> The meeting name has been set to 'networking' 13:01:06 <haleyb> hi everyone o/ 13:01:10 <opendevreview> Merged openstack/os-vif stable/2024.1: OVS Trunk: Add bridge_name to external_ids https://review.opendev.org/c/openstack/os-vif/+/957491 13:01:11 <opendevreview> Merged openstack/os-vif stable/2025.2: Remove remaining job with Ubuntu Jammy (22.04) https://review.opendev.org/c/openstack/os-vif/+/959855 13:01:30 <haleyb> Ping list: bcafarel, elvira, frickler, mlavalle, mtomaska, slaweq, tobias-urdin, ykarel, lajoskatona, jlibosva, averdagu, haleyb, ralonsoh, sahid 13:01:40 <ralonsoh> hello 13:01:40 <slaweq> o/ 13:01:43 <jlibosva> o/ 13:02:05 <mlavalle> \o 13:02:06 <lajoskatona> o/ 13:02:08 <bcafarel> o/ 13:02:22 <ykarel> o/ 13:02:39 <haleyb> #announcements 13:02:42 <haleyb> alright we can get started 13:02:57 <haleyb> We are currently in Week R-2 of Flamingo 13:03:07 <haleyb> Final 2025.2 Flamingo release: October 3rd, 2025 13:03:16 <frickler> \o 13:03:16 <haleyb> We just passed the RC1 milestone the week of September 12th, so RC1 is tagged 13:03:28 <haleyb> Reminder: Release blockers will need to be merged to master, then cherry-picked to stable/2025.2 13:03:51 <haleyb> If nothing else merges RC1 will be Flamingo 13:04:43 <haleyb> So master branch is now open for Gazpacho/2026.1 changes 13:05:14 <lajoskatona> We are over it (again) :-) 13:05:15 <haleyb> I'll try and get the usual assortment of zuul job updates out soon 13:05:31 <haleyb> i.e. skip-level voting, etc 13:05:51 <haleyb> lajoskatona: yes, in the final stretch 13:07:01 <slaweq> also we should probably update neutron-tempest-plugin jobs as well 13:07:18 <lajoskatona> +1 13:07:19 <ralonsoh> right 13:07:19 <haleyb> Having the bulk of the eventlet work behind us is great, thanks for everyone that contributed. We have 6 months to fix any issues and update testing 13:07:59 <ralonsoh> yeah, I'm going to say the same later in this meeting 13:08:03 <haleyb> slaweq: yes, i usually find the changes from last cycle or cycle-2 and propose similar 13:09:00 <haleyb> The next OpenInfra PTG will take place October 27-31, 2025 and registration for the event is now open 13:09:08 <haleyb> #link https://ptg.openinfra.dev/ 13:09:37 <haleyb> i did not see an etherpad yet, i will create if i can't find it 13:09:47 <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 13:12:10 <haleyb> Oh, and continue to use the priorities dashboard for important changes or things ready to merge 13:12:40 <haleyb> that was all i had for announcements, anything else? 13:13:24 <haleyb> #topic bugs 13:13:36 <haleyb> i was the deputy last week, my report is at 13:13:42 <haleyb> #link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/FYDKFH4TEUUJBFSLLCBVGD4TRSSYUFBY/ 13:14:15 <haleyb> We did great assigning bugs, 7/9 and all with reviews 13:14:49 <haleyb> there were just two functional test failures that we can discuss 13:14:58 <haleyb> #link https://bugs.launchpad.net/neutron/+bug/2122524 13:15:43 <haleyb> was this failure just coming from the eventlet removal patch? i couldn't tell from the search 13:16:04 <ralonsoh> yes, that could be caused by this patch 13:16:17 <opendevreview> Merged openstack/neutron master: setup: Remove pbr's wsgi_scripts https://review.opendev.org/c/openstack/neutron/+/960834 13:16:31 <ralonsoh> without eventlet, the FTs are not stable using sqlite 13:17:06 <ralonsoh> we can talk about this later 13:17:10 <haleyb> ralonsoh: ack 13:17:26 <haleyb> so maybe the other is related 13:17:30 <haleyb> #link https://bugs.launchpad.net/neutron/+bug/2122525 13:17:49 <haleyb> could not find subnet in postcommit almost doesn't seem possible 13:17:55 <ralonsoh> yes, that too 13:18:27 <haleyb> ok thanks 13:19:21 <haleyb> any other bugs to discuss, those were the only two i had 13:19:42 <haleyb> This week rubasov is the bug deputy, next week will be mlavalle - good for both of you? 13:19:49 <rubasov> yes, on it 13:20:16 <mlavalle> fine with me as well 13:20:22 <haleyb> +1 great! 13:20:55 <haleyb> #topic specs 13:21:01 <haleyb> #link https://review.opendev.org/q/project:openstack/neutron-specs+status:open 13:21:24 <haleyb> just a reminder there might be specs in need of review 13:21:36 <haleyb> i keep forgetting to ping about old ones will do this week 13:22:19 <haleyb> #topic community goals 13:22:33 <haleyb> lajoskatona: i think i saw some new neutronclient patches? 13:23:35 <haleyb> we can loop back 13:23:44 <lajoskatona> not much 13:23:50 <lajoskatona> I try to work on the nova patch 13:24:02 <lajoskatona> to fix the latest failures with it in tempest 13:24:29 <haleyb> ack, thanks for making progress 13:24:33 <lajoskatona> the horizon ones I think can be finished when horizon team starts to work on Gazpacho 13:25:47 <haleyb> and i saw the patch to move Taas code into neutronclient - did we want to merge that, then discuss moving it all into OSC? 13:26:08 <haleyb> or just skip the middle man? 13:26:30 <ralonsoh> did we say to move everything to OSC? 13:26:44 <ralonsoh> didn't* 13:26:45 <lajoskatona> that was my feeling that we said :-) 13:26:53 <haleyb> i need to look at notes, but we were giong to discuss at PTG 13:27:27 <haleyb> yeah, i was surprised to see the patch to tell the truth 13:27:32 <lajoskatona> I think for now we can merge the taas client code to n-client and perhaps after a discussion on the PTG we can do the move to OSC 13:27:39 <ralonsoh> ^ agree 13:28:05 <lajoskatona> perhaps stephenfin will rethink and just run away from so many exotic networking things :-) 13:28:40 <haleyb> yes, maybe i was confusing - but if we have all in one place might make it all easier is what i was thinking 13:29:04 <haleyb> #link https://review.opendev.org/c/openstack/python-neutronclient/+/960846 13:30:30 <haleyb> ok, and finally eventlet. Looks like great progress on the UT/FT patch 13:30:47 <ralonsoh> #link https://review.opendev.org/c/openstack/neutron/+/952258 13:31:00 <ralonsoh> yes, the patch is finally passing the CI, but at what cost 13:31:10 <ralonsoh> I've documented that in https://review.opendev.org/c/openstack/neutron/+/952258/55/doc/source/eventlet_deprecation/index.rst 13:31:18 <ralonsoh> in a nutshell: 13:31:33 <ralonsoh> unit tests: there are some that need to be refactored, not big deal 13:31:49 <ralonsoh> functional tests: this is a drama. Nothing related to ml2/ovs is working 13:31:55 <ralonsoh> I've commented that in this document 13:32:05 <ralonsoh> that would need, if possible, a huge refactor 13:32:16 <ralonsoh> and of course, this is not stable 13:32:26 <ralonsoh> for example: py310 is stable with concurrency=7 13:32:43 <ralonsoh> and functional job will need constant attention during the next cycle 13:33:02 <ralonsoh> we, maybe, will disable it temporarily (but not for now) 13:33:12 <ralonsoh> that's all, please check and comment the patch 13:33:34 <lajoskatona> I have also seen such instabilities with stadiums (where I tried to remove eventlet) 13:33:51 <ralonsoh> lajoskatona, but also related to the ml2/ovs stuff 13:33:59 <lajoskatona> yes 13:34:01 <ralonsoh> we can't test, right now, anything in FTs with the OVS agent 13:34:49 <haleyb> not being able to kill the threads is a major change 13:35:12 <ralonsoh> and how python deal with the signals (only in main thread) 13:35:25 <ralonsoh> and os-ken, that doesn't work well at all in FT 13:35:52 <ralonsoh> the communication is usually broken between the client and the OF server attending the requests 13:37:19 * haleyb is reading the deprecation note 13:37:43 <ralonsoh> ^^ please feel free to add any other comment needed in that document 13:38:01 <lajoskatona> thanks, I will check it 13:39:02 <haleyb> related to this is the etherpad you created 13:39:07 <haleyb> #link https://etherpad.opendev.org/p/migrate_Neutron_fullstack_to_Tempest 13:39:44 <ralonsoh> lajoskatona created it 13:39:53 <ralonsoh> yes, we also need to address that 13:40:23 <lajoskatona> not easy, I pushed a test patch, but didn't have time recently to go back to that topic 13:40:48 <lajoskatona> https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/953900 13:40:52 <ralonsoh> just to be clear, with the eventlet removal, the testing has been reduced 13:41:41 <haleyb> yes, understood, and it especially impacts ml2/ovs 13:43:15 <haleyb> let's get reviews on the patch since it is a large change and will probably start having merge conflicts otherwise 13:43:45 <haleyb> thanks for the updates 13:44:12 <haleyb> #topic on-demand 13:44:17 <haleyb> i saw one topic from jlibosva 13:44:23 <haleyb> #link https://review.opendev.org/c/openstack/neutron/+/960892 13:44:40 <jlibosva> yeah, maybe could have been discussed on ML or normally in the channel but here we go :) 13:45:31 <ralonsoh> I would go a step forward: same as we create a fixture for NB and SB for each OVN test, we can do the same for the OVS DB, in order to isolate the running OVS DB 13:45:36 <jlibosva> tl;dr: while working on some functional tests for OVN agent BGP extension I noticed the functional test for the OVN agent use the system ovsdb-server instead of a custom one, and since I needed to test interaction between Open_vSwitch table, that is unique, and the agent, I found it out 13:47:22 <jlibosva> later I found out that with the dedicated ovsdb-server per test there is a problem when we want to test interaction between the agent and openflow rules, so in this case the system ovsdb would be better. so there are tradeoffs - I sent the patch out just because I thought everyone would benefit of it but now I'm not sure, so this is a call for opinions :) I'd be fine 13:47:24 <jlibosva> abandoning the patch and use the dedicated ovsdb server only per the test interactig with open_vswitch table 13:47:54 <jlibosva> ralonsoh: thanks, I thought the same, that's exactly what the patch does 13:48:15 <ralonsoh> but only for the OVN agent, I'm saying for all tests importing TestOVNFunctionalBase 13:48:21 <jlibosva> ah, i see 13:48:50 <jlibosva> I'm not convinced it's needed even for the OVN agent, as maybe isolation on the bridge level seems to have been sufficient so far 13:49:52 <ralonsoh> this is why br-int can be renamed 13:53:50 <haleyb> ok, i guess we can discuss more in the review 13:54:12 <jlibosva> ack, thanks! 13:54:41 <haleyb> sure, and anything that can make tests more stable with isolation is a good thing 13:54:56 <haleyb> any other topics to discuss? 13:55:37 <haleyb> oh, i will be out this Thursday if you're looking for me, will set my nick accordingly. Back on Friday 13:55:50 <ralonsoh> sorry yes 13:55:56 <ralonsoh> one quick heads-up 13:55:58 <haleyb> sure 13:56:12 <ralonsoh> the patches are self-explanatory 13:56:14 <ralonsoh> setup: Remove pbr's wsgi_scripts: https://review.opendev.org/c/openstack/neutron/+/960834 13:56:14 <ralonsoh> Migrate setup configuration to pyproject.toml: https://review.opendev.org/c/openstack/neutron/+/960835 13:56:14 <ralonsoh> Move api-paste.ini, rootwrap.conf files: https://review.opendev.org/c/openstack/neutron/+/961130 13:56:20 <ralonsoh> (sorry for the indentation) 13:56:28 <ralonsoh> that's all 13:56:57 <haleyb> i think i approved the one first in the series today and they are in the gate :) 13:57:48 <haleyb> thanks for attending everyone, and have a good week! 13:57:51 <haleyb> #endmeeting