16:00:10 #startmeeting Octavia 16:00:11 Meeting started Wed Jul 8 16:00:10 2020 UTC and is due to finish in 60 minutes. The chair is johnsom. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:13 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:15 The meeting name has been set to 'octavia' 16:00:16 Hi everyone! 16:00:26 Hi! 16:00:32 hi 16:00:34 hi 16:00:36 hi 16:00:45 I am having a slow start to the morning with irccloud down and chrome blowing up on me this morning. 16:01:12 Nothing like having four desktops worth of chrome windows all pile on the main desktop. lol 16:01:33 #topic Announcements 16:01:53 HAProxy 2.2 LTS released yesterday. 16:02:14 #link https://www.haproxy.com/blog/announcing-haproxy-2-2/ 16:02:25 Those are the high level details. 16:02:57 Mostly bug fixes, performance stuff. There are some nice certificate update enhancements we could pull in once we have a path to 2.2 16:03:31 Just another reminder, the MS2 milestone is at the end of this month. 16:05:41 The summit program committee is looking for people to review summit presentations: 16:05:43 #link https://openstackfoundation.formstack.com/forms/programmingcommitteenom_summit2020 16:05:56 o/ 16:06:05 It's nice that this is being announced this round, but I don't think I can do it this time. 16:06:25 Any other announcements this weekk? 16:06:27 Summit as in the Berlin Summit later this year? 16:06:45 As far as I am aware, yes 16:06:47 #link https://www.openstack.org/summit/2020/ 16:06:56 I was pretty sure that wasn't gonna happen? 16:07:26 especially since the US is a shitshow right now with no signs of looking up until after early November <_< 16:07:47 I know they were talking about delaying it, but that was a while ago. I haven't heard anything new. 16:08:05 Yeah, I think travel is going to be an issue 16:08:29 Probably not for cgoncalves though.... lol 16:08:58 it would still a 5h30 journey :) 16:10:03 Ok, anything else? 16:10:23 #topic Brief progress reports / bugs needing review 16:11:24 I created a section with some fixes for amphorav2 in https://etherpad.opendev.org/p/octavia-priority-reviews 16:11:59 Reviews, various little bug fixes, and progress on failover for amphorav2. The tasks should all be converted now, I just need to work on the flows. 16:12:16 Got the first green test run last night (though stuff is still broken) 16:12:54 Ah, yeah, that has been on my todo list. I need to update the priority review list for Victoria. Thanks for the start! 16:13:07 I proposed backport of the failover refactor to ussuri and train. ussuri is ready, train fixing py2 issues. worked on a path in the amphorav2 provider to not use jobboard and so not require extra infrastructure like redis/zookeeper and additional mysql database 16:14:28 Someone almost deleted the priority review list... (there is a history so it can be recovered) 16:14:43 I proposed patches for SCTP support in the Amphora 16:14:51 johnsom: ahem 16:14:54 Yeah, cool 16:15:56 I have a handful of patches to add to the priority review list... 16:16:38 Please do. I will try to go through it sometime this week. 16:17:08 ataraday_, thanks for reviewing https://review.opendev.org/#/c/739053/. be assured we want jobboard enabled by default 16:18:57 We are working on refactoring the statistics collection driver layer -- per johnsom's comments on the patch, it looks like we're going to have to do a ton more work, and actually move it up a level, outside of the amphora driver and as a top level interface (since it will be usable be all providers, not just amphora -- and already is being used that way actually) 16:19:39 Yeah, I just want to make sure we include/consider the non-amphora drivers in that effort. 16:21:57 rm_you, it came to my attention you -2 Backport-Candidate the failover refactor patch. why do you think it is not backport material? 16:22:31 Oh, I also found the issue with our IPv6 only gate job, so that is back to passing once the patch merges. 16:22:33 it's a complete rewrite of our failover flows and is a major and very complex change 16:22:37 the commit title may be misleading but it fixes a ton of issues we've been seeing people getting into due to nova issues 16:22:42 I in no way would have expected it to be backport material 16:22:50 which is why i said i was surprised to see you backporting it 16:23:20 I went to some lengths to make sure there are no "non-backportable" changes, but it is a large change. 16:23:26 I'm not ... strictly opposed to VERY careful work backporting it.... 16:23:32 but GREAT CARE must be taken 16:23:48 That said, it does fix some nasty issues we have seen people run into. Most notably the nova bugs 16:23:55 I am hesitant to actually merge a change of that magnitude in such a critical path, as a backport 16:23:57 yeah, for sure it's a massive patch 16:24:36 and we definitely haven't had much time-testing on this yet 16:24:58 I am ... *relatively* confident it works, but ... 16:25:01 we've lately seen outages in different clouds due to these nova bugs, causing us to spend time go fix it manually 16:25:16 For personal reasons, I would really like to see it make it to train 16:25:45 I understand it's a balance between trying not to backport things that could upset the stability of *stable* releases, and also fixing things in stable releases that aren't so "stable: 16:25:47 " 16:26:14 yep 16:26:25 #topic Open Discussion 16:26:39 so if you are willing to take the responsibility for making sure it's all working... and you are VERY confident that it's going to be an improvement and not cause any other issues... I won't stand in your way 16:26:53 I would much rather encourage people to UPGRADE to get it 16:27:08 rather than backport it into existing people's deployments as a minor version increment 16:28:11 What do others think? 16:28:46 it's not that trivial to upgrade on commercial offerings :) 16:28:54 heh 16:29:05 can't just change the VERSION flag to "master"? :D 16:29:09 it is also not even yet released in victoria 16:29:41 I thought it actually merged in time for our ussuri release? 16:29:49 did it not 16:29:50 I'm for backporting this as it not to get customer upgraded to latest version 16:29:53 I proposed the backport to ussuri so... 16:29:54 did I very much lose track of time 16:30:01 I've been running it since mid-ussuri lol 16:30:21 No, it sat and didn't get into Ussuri 16:30:25 :( 16:30:25 *not so easy 16:30:32 alright 16:30:36 ya'll go for it :D 16:30:41 I wish you the best of luck 16:31:15 and i suppose I can slap a +A on there if you've very confident in it, though my ability to test on older versions is limited 16:31:17 rm_you, the fact you run it in production gives us an extra boost of convidence that it is stable and worth backporting :) 16:31:33 Let's move forward with the backports, but maybe put extra testing cycles into those patches 16:31:38 *confidence 16:31:39 yes, I imagine my environment is like 95% of the testing it has so far, lol 16:31:54 *real world testing, at least 16:31:59 possibly 100% <_< 16:32:59 ataraday_, you're also supportive of the backport? cool, that makes at least 2 vendors interested 16:34:54 rm_you, FWIW I'm not going to propose backports to stein or older releases 16:35:11 We should all commit to doing some extra testing on the backports. 16:35:43 cgoncalves, yep, I'm for backports 16:35:51 i'll test the frontports ;) 16:36:30 I reviewed the master patch line-by-line and proposed the backports. internally we will be testing this on Python 2 and 3 and on Queens and Train clouds 16:36:37 rm_you, lol 16:37:12 Ok, any other topics for today? Good discussion, it's an important patch. 16:37:45 thank you for all your hard work on the patch, johnsom! 16:38:18 That was a doozy, thanks! 16:39:00 the single haproxy patch was also something :) and it was backported! 16:39:15 True 16:41:06 I don't think I have any other topics today. 16:41:51 There was a discussion this morning that the barbican ACL call may need to be updated for some of these alternate keystone token types. I expect similar to what we did with the neutron subnet call. 16:42:04 If someone has some cycles there should be a story open for it 16:43:38 Ok, if there aren't any more topics I will call the meeting for the week. 16:43:46 Thank you all for your work! 16:44:13 thanks! 16:44:27 o/ 16:44:37 #endmeeting