13:02:22 <haleyb> #startmeeting networking 13:02:22 <opendevmeet> Meeting started Tue May 27 13:02:22 2025 UTC and is due to finish in 60 minutes. The chair is haleyb. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:02:22 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:02:22 <opendevmeet> The meeting name has been set to 'networking' 13:02:24 <haleyb> Ping list: bcafarel, elvira, frickler, mlavalle, mtomaska, obondarev, slaweq, tobias-urdin, ykarel, lajoskatona, jlibosva, averdagu, haleyb, ralonsoh 13:02:26 <mlavalle> \o 13:02:30 <elvira> o/ 13:02:38 <ralonsoh> hello 13:02:47 <lajoskatona> o/ 13:02:52 <sahid> o/ 13:03:10 <haleyb> #announcements 13:03:15 <haleyb> We are currently in Week R-17 of Flamingo 13:03:27 <haleyb> Our next milestone in this development cycle will be Flamingo-2, on July 3rd 13:03:46 <haleyb> Final 2025.2 Flamingo release: October 3rd, 2025 13:03:52 <haleyb> #link https://releases.openstack.org/flamingo/schedule.html 13:04:01 <cbuggy> o/ 13:04:01 <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 13:04:35 <haleyb> Plan is to have a meeting this friday to discuss a recent bgp rfe (and any others) 13:05:36 <haleyb> that was all i had for announcements, any others? 13:06:14 <haleyb> #topic bugs 13:06:30 <haleyb> elvira was the deputy last week, her report is at 13:06:32 <rubasov> o/ 13:06:37 <haleyb> #link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/QOLR2TOQZ7PFTZ7ZR7HPKTB7JWNJ4TVB/ 13:07:33 <haleyb> there were quite a few, will go through the unassigned ones 13:09:00 <haleyb> rodolfo has picked up a bunch already i see, but i found one 13:09:03 <haleyb> #link https://bugs.launchpad.net/neutron/+bug/2111572 13:09:14 <haleyb> [OVN] OVN ML2 mech driver forces port update to networking generic switch (NGS) on OVS restart 13:09:52 <ralonsoh> I have a question on this one 13:10:04 <ralonsoh> because as elvira mentioned, this is not affecting anything 13:11:34 <frickler> bbezak might be around to answer? 13:11:40 <haleyb> right, is this a bug in NGS? 13:12:03 <ralonsoh> I don't think this is a bug in NGS, as long as is reacting to the port update 13:12:08 <ralonsoh> we can discuss if that is needed 13:12:24 <ralonsoh> but a vswitchd restart is not in the normal operation 13:12:38 <ralonsoh> (I would like to know what vswitchd service, in what host) 13:12:47 <ralonsoh> compute, most probably 13:13:45 <ralonsoh> anyway, we can ask in the LP bug, I'll do it today (why this restart is needed, if traffic is affected, etc) 13:15:06 <haleyb> ack, thanks 13:16:22 <haleyb> next one 13:16:34 <haleyb> #link https://bugs.launchpad.net/neutron/+bug/2111584 13:16:38 <haleyb> Failed get list of ports due to removed network 13:17:43 <haleyb> guess i would have expected some type of error when doing this operation, but we should be more graceful 13:19:08 <elvira> that is what I thought too, maybe it is low prio and not medium indeed 13:19:15 <haleyb> would agree with the low hanging fruit tag, so if someone wants to pick it up... 13:19:48 <ralonsoh> we have recently optimized this code 13:19:57 <ralonsoh> maybe with the optimizations we can skip it 13:20:28 <bbezak> actually restart of ovs_vswitchd on the controllers 13:20:50 <ralonsoh> --> https://review.opendev.org/q/owner:jimin3.shin@samsung.com+status:merged 13:20:58 <bbezak> invokes the port update 13:21:19 <bbezak> (sorry for the delay) - it is about bug 2111572 13:21:32 <ralonsoh> bbezak, one sec, let's finish the current bug 13:21:49 <ralonsoh> I can try creating 1000 ports and doing the same operation 13:22:05 <ralonsoh> I think we'll need some min qos rules (maybe) 13:22:13 <haleyb> ralonsoh: oh, it might not happen now 13:22:30 <ralonsoh> but with the optimizations we no longer retrieve anything if there are no min-bw rules 13:22:50 <ralonsoh> anyway, I'll add it for next friday morning, to deploy a test env 13:22:59 <haleyb> ack 13:23:10 <ralonsoh> bbezak, hi 13:23:11 <haleyb> we can loop back to other bug 13:23:29 <ralonsoh> yes, we know the restart triggers the update 13:23:33 <ralonsoh> why this restart? 13:24:22 <bbezak> well - during some maintenances - containers upgrades - hosts restart 13:25:05 <ralonsoh> and this update affects the traffic? I guess nothing in this host is running during the restart 13:26:02 <bbezak> it is not affecting traffic on baremetal host, cause ngs applies the same config - but still, in scale - for instance ~500 baremetal hosts, ngs would hammer all switches. which can be problematic 13:27:07 <ralonsoh> bbezak, if you can identify what event is triggered during the vswitchd restart, maybe we can debug this problem 13:27:59 <ralonsoh> (Neutron event, you can enable DEBUG logs) 13:28:47 <bbezak> yeah, I'll reproduce it and update the bug 13:28:47 * slaweq needs to drop for another meeting but I am aware of bug deputy this week :) 13:28:47 <bbezak> thx 13:28:56 <ralonsoh> thansk 13:29:10 <haleyb> slaweq: ack, thanks 13:30:00 <haleyb> ok, so we have some debugging to do on 2111572 13:30:09 <haleyb> moving along to next one 13:30:26 <haleyb> #link https://bugs.launchpad.net/neutron/+bug/2111496 13:30:27 <haleyb> Neutron port stuck in DOWN state after update router external connectivity 13:30:57 <ralonsoh> yes, that is a bug 13:31:09 <ralonsoh> we, most probably, are not detecting the port update 13:31:19 <ralonsoh> and we are not activating again the port 13:32:24 <ralonsoh> (and, just thinking loud, 6 and 7 are not needed) 13:32:41 <ralonsoh> if this is a port in the internal network 13:33:29 <haleyb> so just down/up of a port? ok 13:33:37 <ralonsoh> maybe 13:34:01 <haleyb> i'll ask submitter to try just so we know 13:34:03 <ralonsoh> I don't have an OVS env right now, but I can deploy one 13:34:12 <ralonsoh> but not this week... 13:35:03 <lajoskatona> I can try, I usually have OVS env 13:35:11 <ralonsoh> cool! 13:35:47 <haleyb> lajoskatona: ack, thanks 13:36:09 <haleyb> last bug in list 13:36:21 <haleyb> #link https://bugs.launchpad.net/neutron/+bug/2111660 13:36:22 <haleyb> OVN_AGENT_NEUTRON_SB_CFG_KEY doesn't set when Neutron OVN Agent starts 13:36:55 <ralonsoh> there is a patch proposed 13:37:03 <ralonsoh> and I think a good one, that fixes the problem 13:37:16 <haleyb> #link https://bugs.launchpad.net/neutron/+bug/2111475 13:37:25 <haleyb> oops, that's the rlated bug 13:37:31 <ralonsoh> yes 13:37:41 <ralonsoh> ahhhhh sorry 13:37:43 <haleyb> #link https://review.opendev.org/c/openstack/neutron/+/950673 13:38:04 <ralonsoh> Ok, I asked Min to open a LP bug if he was going to propose a patch only for ovn metadata 13:38:17 <ralonsoh> so this is why we have https://bugs.launchpad.net/neutron/+bug/2111660 13:39:13 <ralonsoh> Ok, I'll take this one, the fix should be similar to https://review.opendev.org/c/openstack/neutron/+/950673 13:39:35 <haleyb> do we need a fix for ovn-agent as well? i'm reading between the lines 13:39:43 <ralonsoh> yes, the issue is the same 13:39:54 <haleyb> ack, thanks 13:40:24 <haleyb> so i think that was all the unassigned bugs 13:40:33 <haleyb> any other bugs to discuss? 13:41:09 <haleyb> slaweq is the deputy this week, i am next week 13:41:48 <haleyb> #topic community goals 13:42:26 <haleyb> lajoskatona: sorry, i haven't been able to review any of your neutronclient patches, can look today 13:42:36 <haleyb> have they made forward progress from other reviews? 13:42:39 <lajoskatona> here they are: https://review.opendev.org/q/topic:%22bug/1999774%22+status:open+project:openstack/python-openstackclient+or+status:open+owner:self+project:openstack/openstacksdk 13:43:40 <lajoskatona> I think these are the ones for sec-groups is_shared field, suggested by ralonsoh and stephenfin (perhaps another depends-on can be in one of the commit-msg but let's wait) 13:44:02 <haleyb> #link https://review.opendev.org/q/topic:%22bug/1999774%22 13:44:03 <lajoskatona> that's all this topic from me, slow progress 13:44:10 <haleyb> that shows them all 13:44:43 <lajoskatona> yes, but the filtred list is the hot this week :-) 13:45:01 <haleyb> it is sometimes hard to move a large boulder :) 13:46:09 <haleyb> ok, finally eventlet 13:46:15 <ralonsoh> hi 13:46:16 <haleyb> i know there were new bugs files 13:46:25 <ralonsoh> yes, related to the test FWs 13:46:38 <ralonsoh> and this going to be a nightmare... in particular fullstack 13:47:07 <ralonsoh> I've proposed https://review.opendev.org/c/openstack/neutron/+/938404 to remove eventlet from the l3 agent 13:47:21 <ralonsoh> and right now I need to skip many FT and fullstack tests 13:47:26 <ralonsoh> this is explained in the docs 13:47:54 <ralonsoh> we also have other patches: https://review.opendev.org/q/topic:%22eventlet-removal%22+project:openstack/neutron+status:open 13:48:07 <ralonsoh> in particular the ones related to unit and functional tests 13:48:19 <lajoskatona> I slowly progress with the DHCP one in the background.... 13:48:22 <ralonsoh> and lajoskatona is working on the DHCP agent: https://review.opendev.org/c/openstack/neutron/+/950498 13:48:24 <ralonsoh> yeah 13:48:51 <ralonsoh> the first step, same as in the l3 agent: to move the RPC initialization out of the init methods 13:49:08 <ralonsoh> this is because oslo.service is spawning several new processes and threads after that 13:49:09 <lajoskatona> yes, I am on it to make it final 13:49:20 <ralonsoh> and the RPC connections do not like that 13:49:35 <ralonsoh> --> please review this: https://review.opendev.org/c/openstack/neutron/+/949135 13:49:42 <ralonsoh> (is ready now) 13:50:01 <ralonsoh> and last topic 13:50:12 <haleyb> ralonsoh: ack, can look later today 13:50:23 <ralonsoh> we have oslo.service 4.2.0, that implements "threading" backend 13:50:36 <ralonsoh> that means: all service FW is written without evnetlet 13:50:41 <ralonsoh> that's all I have 13:52:07 <haleyb> right. so i'm remembering an older change we never pushed for os-ken that i think was related to threading backend, will have to find it 13:52:24 <haleyb> sahid: i think that was yours ^^ just wondering if we need to move it forward now 13:53:05 <lajoskatona> yeah, some patches are hanging for ovs-agent and os-ken 13:53:05 <haleyb> i will look after meeting 13:53:15 <sahid> yes i think that could make sense 13:53:53 <lajoskatona> +1 13:54:38 <haleyb> ok, if we can get those in ready-to-merge state that would be great (if not ther already) 13:54:51 <sahid> i will double check 13:54:57 <haleyb> thanks 13:55:02 <sahid> sure 13:55:11 <haleyb> i will move on since i have a hard stop at top of hour 13:55:15 <haleyb> #topic on-demand 13:55:20 <haleyb> lajoskatona: you had an item 13:55:32 <lajoskatona> yes 13:55:36 <lajoskatona> regarding trunks 13:55:52 <lajoskatona> rubasov already sent out a mail to summarize: https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/W3CNUX2QPKLKWY6UV4ZGR3E65TVE4BLC/ 13:56:35 <lajoskatona> and we have some patches to fix different aspects of the issue which happens when we have many subports for a trunk *actually thats just for easy reproduction) 13:57:11 <lajoskatona> the rootcause is that the status of the trunk set to ACTIVE when the parent goes active not when all the posts goes to active 13:57:27 <lajoskatona> and this was intriduced for a fix with OVN: https://review.opendev.org/c/openstack/neutron/+/853779 13:58:38 <lajoskatona> so this breaks the OVS driver for sure but the OVN also, so one fix from rubasov is to move this status change to OVN only part, and for OVS to fix this in actually os-vif and some extra code doc for code archeologists after us :-) 13:59:08 <lajoskatona> and the patch series: https://review.opendev.org/q/topic:%22bug/2095152%22 13:59:32 <lajoskatona> please check if you have time ,and read the mail I linked with bug report :-) 13:59:42 <lajoskatona> That's it, thanks for the attention 14:00:21 <ralonsoh> qq, the os-vif patch is related to the OVN issue? 14:00:25 <haleyb> lajoskatona: thanks will take a look, and i see you added the right people to the review i think 14:00:33 <lajoskatona> no that is for OVS 14:00:36 <ralonsoh> thanks! 14:00:53 <lajoskatona> thanks 14:01:32 <haleyb> ok, that's all i had, any other topics 14:02:05 <haleyb> ok, thanks for attending everyone, have a good week 14:02:07 <haleyb> #endmeeting