17:15:14 #startmeeting ovn_community_development_discussion 17:15:15 Meeting started Thu Aug 6 17:15:14 2020 UTC and is due to finish in 60 minutes. The chair is mmichelson. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:15:16 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:15:19 The meeting name has been set to 'ovn_community_development_discussion' 17:15:24 Hi 17:16:02 Hi everyone. I'll go first real quick 17:17:02 I've got a small improvement to be able to gather pinctrl stats. You can find my WIP here: https://github.com/putnopvut/ovn/tree/pinctrl_stats 17:17:40 Right now it just keeps track of the count of packet-in operations that occur per port binding. 17:18:45 I emailed Jeremy Kerr again earlier this week about getting a separate OVN project for patchwork, but I never heard back from him. So I'll need to ping him again to see what needs to be done on our end to facilitate it. 17:19:02 Other than that, it's mostly been helping to put out fires for Red Hat customer issues. 17:19:04 And that's it for me. 17:21:35 hm, quiet day. 17:22:36 I can go next 17:24:06 I continued working on ovn scale test, comparing between versions. The basic findings so far is that 20.03/06's northd spent much more time than 2.12. Second finding is 20.06's ovn-controller I-P has reduced number of lflow_runs. 17:24:53 I still didn't got time to dig out the reason why 20.03 started to spent almost doubled time in northd. That will be my next target. 17:26:02 zhouhan, no regressions in ovn-controller due to the I-P patches right ? 17:27:05 For the e2e latency, it seems the I-P improvement recently done in 20.06 doesn't help much. I.e. 20.03 and 20.06 shows similar result. I think it is due to the flow comparing part. I will work on incrementally install the flows without full comparison, and see how it improves. 17:27:15 numans: No regression seen on my testing :) 17:27:42 ok. good to hear about that 17:28:01 given that there were quite a few regressions earlier 17:28:29 apart from this, there were lots of questions and discussions in the ML, I tried my best to answer some of them. I think it is a good sign that OVN is getting more and more popular :) 17:29:10 However, it seems some documentation needs to be further improved, especially for the OVN DB clustering. 17:29:35 agree. 17:30:00 Oh, finally, thanks dceara for fixing the RAFT bug I introduced, which caused the inconsistancy problem. 17:30:19 that's it from me 17:30:43 zhouhan: no problem, it's good that we figured it out. 17:30:45 I can go real quick next. 17:30:56 Not much on the OVN patches this week. 17:31:09 I was busy with mmichelson on some issues. 17:31:34 not much reviews this week. But I'll review zhouhan's ARP patches after this meeting. 17:31:44 thanks numans 17:32:06 And request review these patches - https://patchwork.ozlabs.org/project/openvswitch/list/?series=194004 17:32:11 That's it from me. 17:32:27 numans: I'm looking at them as we speak :) 17:33:18 I forgot to mention that it would be appreciated if someone could help review this: #link https://patchwork.ozlabs.org/project/openvswitch/patch/1596217711-11945-1-git-send-email-hzhou@ovn.org/ 17:33:38 zhouhan: ack, i'll put it on my list for tomorrow 17:33:45 dceara: thanks! 17:34:26 dceara, thanks. 17:36:33 I have a quick update too: except for the RAFT issue I found a bug in the IGMP code that was uncovered when ovn-k8s moved to ovn-monitor-all. #link https://patchwork.ozlabs.org/project/openvswitch/list/?series=194449 17:37:01 And I was planning to do a round of reviews today/tomorrow. That's it on my side, thanks! 17:37:22 One last point from me, I need some attention on this patch too https://patchwork.ozlabs.org/project/openvswitch/patch/20200709091156.773306-1-numans@ovn.org/. Ankur and I are having discussions there. It would be nice to get comments from others too. 17:37:36 dceara, I'll take a look. 17:37:52 numans: thanks! 17:38:49 Anybody else? 17:39:30 No much updates from me. I'm just looking at DB inconsistency fix from dceara. Will likely apply it today or tomorrow. 17:39:41 imaximets: thanks! 17:40:20 imaximets: I assume it's ok to backport it to 2.13, right? 17:40:41 dceara, it should go down to 2.12, I think. 17:40:56 imaximets: great, thanks 17:41:19 dceara, np. thanks for fixing! 17:42:13 All right everyone thanks for the updates. Talk to you next week 17:42:18 Bye 17:42:20 #endmeeting