14:00:58 <lajoskatona> #startmeeting networking 14:00:58 <opendevmeet> Meeting started Tue Dec 7 14:00:58 2021 UTC and is due to finish in 60 minutes. The chair is lajoskatona. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:58 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:58 <opendevmeet> The meeting name has been set to 'networking' 14:01:00 <mlavalle> o/ 14:01:03 <lajoskatona> Hi 14:01:09 <slaweq> hi 14:01:13 <ralonsoh> hi 14:02:17 <lajoskatona> I think we can start 14:02:19 <bcafarel> o/ 14:02:28 <obondarev> hi 14:02:32 <lajoskatona> #topic Announcements 14:02:48 <mlavalle> with bcafarel and obondarev we have completed the ususal suspects bunch for this meeting 14:03:15 <lajoskatona> mlavalle: exactly 14:03:35 <lajoskatona> The usual cycle calendar: https://releases.openstack.org/yoga/schedule.html 14:03:47 <lajoskatona> This is Week R-16: http://lists.openstack.org/pipermail/openstack-discuss/2021-December/026173.html 14:04:13 <rubasov> o/ 14:05:37 <lajoskatona> I think we should cut new release for n-lib at least (have to check other libs if we have enough things merged 14:06:07 <slaweq> for neutron-lib I think we are ok as we already released it this cycle 14:06:14 <slaweq> same for os-ken iirc 14:06:21 <slaweq> I'm not sure about ovsdbapp 14:06:41 <slaweq> but of course we can release new versions around X-2 milestone 14:07:01 <opendevreview> Merged openstack/neutron stable/xena: [OVN] Fix gateway_mtu option should not always be set https://review.opendev.org/c/openstack/neutron/+/819578 14:07:03 <lajoskatona> for neutron-lib just for the regularity we should have one, and last time it was this group we waited for: https://review.opendev.org/q/topic:%2522bug/1950454%2522+(status:open+OR+status:merged)+update_network 14:07:06 <opendevreview> Merged openstack/neutron stable/wallaby: [OVN] Fix gateway_mtu option should not always be set https://review.opendev.org/c/openstack/neutron/+/819579 14:07:40 <lajoskatona> slaweq: thanks 14:08:21 <lajoskatona> Some news for the testing runtime: 14:08:27 <lajoskatona> full discussion: http://lists.openstack.org/pipermail/openstack-discuss/2021-December/026175.html 14:08:46 <lajoskatona> and the summary from last week TCmeeting: "We have an agreement now on Yoga testing runtime and keeping python3.6 support in Yoga." 14:09:09 <lajoskatona> so we keep py36 in this cycle 14:10:06 <slaweq> ++ 14:10:50 <lajoskatona> another announcement: "Pain Points" discussion continues: http://lists.openstack.org/pipermail/openstack-discuss/2021-December/026169.html 14:11:08 <lajoskatona> There's an etherpad: https://etherpad.opendev.org/p/pain-point-elimination 14:11:23 <lajoskatona> and there will be a meeting: Dec. 8th, 15:00 UTC 14:12:29 <lajoskatona> As I see OSC is the main pain point, and we are on the good side 14:12:33 <slaweq> lajoskatona do You know if tomorrow it will be discussion about Neutron maybe? 14:12:52 <slaweq> I'm not sure if I should join there or maybe it's not needed at all 14:12:56 <lajoskatona> slaweq: it will be openstack wide if I understand well 14:13:07 <slaweq> ok 14:13:12 <slaweq> I will try to be there 14:13:22 <mlavalle> yeah, looking at the email, it is not specific to Neutron 14:13:23 <lajoskatona> I can join tomorrow 14:14:21 <amotoki> lajoskatona: where does it happen? 14:14:33 <mlavalle> It's a video conference 14:14:35 <lajoskatona> in the etherpad there are a few neutron related things (hostname, osc/neutronclient compatibility) 14:14:37 <amotoki> I don't see the URL for it 14:14:53 <lajoskatona> amotoki: #link https://meet.google.com/xsx-inbw-mos 14:15:13 <amotoki> lajoskatona: thanks. I also found the link in the mailing list thread 14:15:49 <lajoskatona> so I have hopes that not Neutron will be the bad guy :-) 14:16:18 <slaweq> lajoskatona it's always networking issue so it will be ;P 14:16:44 <lajoskatona> slaweq: that's true :-) 14:16:46 <bcafarel> :) 14:16:51 <amotoki> everyone talks over networks :) 14:17:07 <mlavalle> poor Neutron guys, they are always blamed 14:17:32 <slaweq> LOL 14:17:47 <lajoskatona> Any comments for announcements? 14:18:21 <lajoskatona> #topic Bugs 14:18:46 <lajoskatona> lajoskatona (it's me :-)) was bug deputy last week: http://lists.openstack.org/pipermail/openstack-discuss/2021-December/026193.html 14:19:07 <lajoskatona> We have 3 unassigned bugs: 14:19:15 <lajoskatona> #link https://bugs.launchpad.net/neutron/+bug/1952675 14:19:47 <lajoskatona> this one is about a db upgrade issue 14:20:03 <lajoskatona> #link https://bugs.launchpad.net/neutron/+bug/1944779 14:20:22 <lajoskatona> This one is tough: VirtualInterfaceCreateException due to "Timeout waiting for [('network-vif-plugged'..." 14:20:42 <lajoskatona> As I remember there was a long mail thread for this 14:21:15 <lajoskatona> I think this mail started it: #link http://lists.openstack.org/pipermail/openstack-discuss/2021-November/025892.html 14:21:35 <ralonsoh> there are several mixed discussions there 14:21:43 <ralonsoh> including several backend types 14:22:00 <ralonsoh> so when investigating those kind of errors, focus on one backend 14:23:21 <lajoskatona> from nova perspective that is one of the problems that nova has to act based on network backend and that is not elegant, as I remember 14:24:13 <slaweq> yes, and the problem is that it behaves differently in each neutron backend :/ 14:24:15 <lajoskatona> I mean that nova has to use big if-else blocks to wait or not wait for event based on something that is Neutron internal thing 14:24:27 <lajoskatona> slaweq: exactly 14:25:10 <lajoskatona> I think in our downstream CI the issue was that odl and ovs send vif-pluged differently for example 14:25:33 <lajoskatona> so ti was hard to debug even what happens or not happens 14:27:27 <lajoskatona> ok we had another unassigned bug from last week: #link https://bugs.launchpad.net/neutron/+bug/1953165 14:27:52 <lajoskatona> and actually rubasov started to check it 14:28:29 <rubasov> it may be a duplicate of https://bugs.launchpad.net/neutron/+bug/1930414 14:29:26 <lajoskatona> rubasov: thanks, I was looking for this 14:29:27 <slaweq> rubasov so do You think we are leaking IPv6 traffic between isolated networks? 14:29:35 <haleyb> I would also guess it's possible that there is a second namespace with that IPv6 address, like the qrouter 14:29:38 <rubasov> yes I believe so 14:29:58 <haleyb> rubasov: oh, that would be worse :( 14:30:28 <rubasov> I think it's only a short window when ports belong to the DEAD_VLAN 14:31:49 <rubasov> I'll update that bug report soon 14:31:49 <slaweq> but priteau said there that even after restarting neutron-dhcp-agent it wasn't fixed in many cases 14:32:09 <slaweq> shouldn't that dead vlan tag be already removed in such case? 14:32:21 <mlavalle> good point 14:32:21 <rubasov> slaweq: nobody will clear dadfailed state 14:32:38 <slaweq> rubasov ahh, ok 14:32:41 <rubasov> so if dadfailed happens during the short windows it stays 14:33:03 <slaweq> dhcp agent will just check that interface is there and will do nothing with it after restart 14:33:37 <rubasov> I also have a patch, which I'll upload, but that's at most a partial fix and not fixing the full root cause which I still don't get 14:34:10 <priteau> (I am here in case you need further details) 14:35:16 <rubasov> priteau: feel free to try the reproduction in bug 1930414 to see if it's the same as you see 14:35:29 <rubasov> priteau: also after the update I'll add soon 14:36:48 <priteau> I started setting up a dev environment to try and reproduce it. So far I've only seen it on a prod system 14:38:22 <lajoskatona> thanks rubasov for the explanation and for working on this issue 14:39:03 <lajoskatona> This week amotoki is the deputy, and next week mlavalle will be. 14:39:10 <mlavalle> ack 14:39:14 <amotoki> ack 14:39:15 <mlavalle> Thanks for the reminder 14:39:19 <lajoskatona> cool 14:39:27 <mlavalle> I still require a formal email.... 14:39:39 <lajoskatona> mlavalle: I just wanted to ask :-) 14:39:39 <mlavalle> no, just kdding. slaweq spoiled me 14:40:18 <slaweq> LOL 14:41:17 <lajoskatona> advertisement: slaweq prepared something for the "reduce rechecks" topic: http://lists.openstack.org/pipermail/openstack-discuss/2021-December/026208.html 14:41:49 <lajoskatona> so if you have ideas, or interested join please to the CI meeting from 15:00 UTC :-) 14:42:08 <slaweq> ++ 14:42:15 <slaweq> anyone's welcome there 14:43:19 <lajoskatona> thanks slaweq 14:43:24 <lajoskatona> we have nothing for L3, and no changes in ryu 14:43:46 <lajoskatona> #topic On Demand Agenda 14:44:50 <opendevreview> Merged openstack/neutron stable/victoria: [OVN] Fix gateway_mtu option should not always be set https://review.opendev.org/c/openstack/neutron/+/819580 14:45:15 <lajoskatona> If there is nothing to discuss, let's close the meeting 14:45:24 <mlavalle> sounds good to me 14:45:31 <lajoskatona> #endmeeting