15:00:11 #startmeeting kolla 15:00:11 Meeting started Wed Nov 17 15:00:11 2021 UTC and is due to finish in 60 minutes. The chair is mnasiadka. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:11 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:11 The meeting name has been set to 'kolla' 15:00:21 #topic rollcall 15:01:19 \o 15:01:35 \o 15:01:36 o/ 15:02:18 o/ 15:02:25 o/ 15:03:08 [°][o] 15:05:21 #topic agenda 15:05:21 * Announcements 15:05:21 * Review action items from the last meeting 15:05:21 * CI status 15:05:21 * Release tasks 15:05:22 * Backports & stable releases 15:05:22 * Yoga cycle planning 15:05:24 * [infra] Need to fix Zuul errors 15:05:24 * CentOS plans 15:05:26 * Open discussion 15:05:39 #topic Announcements 15:06:46 No announcements today. 15:06:52 #topic Review action items from the last meeting 15:08:11 action on mnasiadka to action security bugs - did nada - because I was on vacation :) 15:08:24 action on yoctozepto to hide properly init-runonce 15:08:26 I have not done mine either 15:08:38 #action mnasiadka to triage security bugs and update them with resolution plan (if needed) 15:08:45 #action yoctozepto hide properly init-runonce 15:08:55 #topic CI Status 15:09:18 I see all green 15:09:40 (we'll cover the zuul config errors later) 15:09:44 #topic Release tasks 15:10:00 we should release xena 15:10:27 There are a few backports still waiting merge for kayobe 15:10:27 Yes, but we need an RC3 for Kayobe first 15:10:50 https://review.opendev.org/c/openstack/kayobe/+/818217/ 15:11:11 we can although release finals for Kolla and Kolla-Ansible, right? 15:12:11 yeah, that's my point 15:12:15 let's do it 15:12:20 priteau: let's get a list of Kayobe RC3 pending changes on the whiteboard 15:12:48 It's R-19 week for Yoga, we have some R-17 items on our list - but I think there are some changes raised by mgoddard (looking at previous meeting logs) 15:12:50 mnasiadka: make kaybobe-release-blockers topic? 15:12:52 So we should be fine 15:13:09 hrw: yes, that also works, we'll work it out :) 15:13:19 that way one link to put ;D 15:14:05 ok then, let's continue 15:14:14 #topic Backports & stable releases 15:14:30 the whiteboard says: "We are probably overdue a backport session (see L248 for last checked commits)." 15:14:37 After backports merge, we should release stable branches. 15:15:00 So - any volunteer to go through backports and make sure those are done? 15:15:03 so many backports to be seen then :D 15:15:28 not me, /me has too much on his plate atm 15:16:39 I have a lot on my plate as well, so would prefer someone else to take it ;) 15:17:06 need more plates :D 15:17:20 to break ;) 15:17:30 hopefully not 15:18:36 Well, if there are no volunteers, it will just need to wait a bit 15:19:12 #action not forget to go through backports for stable branches (L248 on Whiteboard) and do stable releases afterwards. 15:19:34 #topic Yoga cycle planning 15:20:13 I think we need to go through the prereqs for deprecating binary - if we're going to deprecate it early this cycle. 15:21:32 L349 in https://etherpad.opendev.org/p/kolla-yoga-ptg 15:23:30 Any volunteers for the tasks there? 15:25:27 tough crowd today 15:25:44 I guess 15:26:07 Anybody wants to write the reno about deprecating? 15:26:09 I want to stay in quiet corner and wait for end of the world. 15:26:45 Ok then, so let's wait until binary deprecates by itself :) 15:27:36 Let's continue if there are no volunteers 15:27:38 #topic [infra] Need to fix Zuul errors 15:27:55 for stein and rocky 15:27:57 seems we have bad configs in kolla-ansible stable/rocky and stable/stein 15:28:08 so I guess it's best to eol them 15:28:13 We had a discussion earlier today about it, and I think we EOL 15:28:15 as train is still unhappy ;-) 15:28:27 all right, I agree 15:28:32 anyone disagrees? 15:28:33 priteau: is on a mission to save the train, but maybe not now ;) 15:29:01 I made some progress during train upgrades, but once all our customers are beyond train I don't know if there will be much motivation :P 15:29:19 then we can EOL Train and forget about Python2 :) 15:29:28 what is a name of Z cycle? 15:29:42 not decided yet 15:29:45 ok 15:30:19 ok, no protests, I'll raise EOL changes for rocky and stein 15:30:36 for both k-a and kolla, right? 15:31:29 Marcin Juszkiewicz proposed openstack/kolla master: reno: binary images are now deprecated https://review.opendev.org/c/openstack/kolla/+/818240 15:31:42 #action mnasiadka to EOL rocky and stein for both kolla and kolla-ansible 15:31:52 patch landed. message needs rewriting probably but patch exists 15:31:59 ok, that's a start 15:32:00 thanks hrw 15:32:15 #topic CentOS plans 15:32:39 binary images gets deprecated in Y, dropped in Z 15:32:44 that was the plan I guess 15:32:50 but 15:32:53 RDO moves from CS8 to CS9 in Y cycle 15:32:53 packages for both will be provided to simplify migration 15:33:16 that means Yoga is the last cycle with cs8+binary for Kolla? Since we opted-out of cs9 migration? 15:33:43 so if we decide to go CS8 to the end of CentOS in Kolla then Y is cs8/binary+source and Z is cs8/source only 15:34:32 and as we migrate to "latest binary" in R-late then it looks like sane plan 15:35:18 seems sane to me 15:35:22 anybody disagrees? 15:35:50 makes sense 15:35:55 just need to deprecate it properly 15:36:02 and glad we can support it until then 15:36:33 looks like quick track: Yoga: cs8/binary+source, u2004/binary+source, d11/binary+source. Z: cs8/source, u2004/source, d11/source, A: d11/source is still possible 15:37:39 Yoga has binary deprecation, Z has binary drop and !debian deprecation, A has only one set of images 15:37:48 Yes, but let's focus on getting rid of binary for now. 15:37:58 yes, just wanted to mention 15:38:16 ok, no protesters again 15:38:33 so cs8 to the end 15:39:11 #agreed to not pursue cs9 (as agreed on the PTG) and Z release will use cs8+source (for centos) 15:39:37 I guess that concludes the CentOS discussion 15:39:43 #topic Open discussion 15:39:50 now #action inform openstack ml about binary deprecation and centos plans 15:39:53 ? 15:40:05 that's already in agreed PTG action items 15:40:11 ah, right 15:40:16 So it's on my list :) 15:40:35 so - open discussion - anybody has any points? 15:40:41 kevko goes for core - vote on ML if not done yet 15:41:34 have we done the necessary user feedback gathering to start deprecating things? 15:42:15 looking at the PTG etherpad - we wanted first to deprecate binary, and then send a mail - should we do it other way around? 15:42:30 we not merged it yet 15:43:05 I think we wanted to deprecate in Xena, but it didn't happen 15:43:12 in which case I'd say discuss first 15:44:21 Well, I think it will finish like the kolla-ceph code removal, but I can send a mail to openstack-discuss and kolla-klub around a plan to deprecate binary and if anyone wants to step up and keep it maintained 15:44:38 Is that what we want to do? 15:46:22 I'd suggest including the single distro plan too 15:46:24 "Dear (potential and present) users. Due to size of our team we decided to drop support for building binary images. What do you think about it? Your beloved Kolla team" 15:46:44 sorry, too much [sarcasm emoji] recently 15:46:45 I'm not in favor of sending two bombs at the same time. 15:47:31 you'd rather send another one next week, saying 'oh, by the way, you can't use your current distro either?' 15:47:33 it's one plan 15:48:44 I don't believe it's one plan, it's not as in we do both or nothing. 15:49:17 it is more or less one plan. just split on 3-4 cycles 15:49:19 I can include both in one mail, but I think it will bring more confusion than real user feedback. 15:49:35 from what is now" 15:49:43 from "what is now" to "one set of images" 15:49:58 we are reducing 6 sets of images to one 15:50:19 that's long way to go 15:50:25 but that's one plan 15:50:52 here is a plan. we split it to parts. this cycle gets part1, then part2, then part3 15:50:54 yes, and people need to know how their setup will end up. Why let them think they can use centos/source? 15:51:20 especially when they jump releases 15:51:44 most people would want to switch straight to the chosen image config, rather than via another type 15:52:20 wallaby centos/binary -> xena centos/binary -> yoga centos/binary -> zeus centos/source -> athena debian/source would require planning 15:53:17 One Set to rule them all 15:55:31 Honestly, we all know it's going to be shooting contest in the user feedback - but let me agree to send it out in one mail. 15:56:45 ok, can I get other topic? 15:56:49 Sure 15:57:00 did someone use docker squash support? 15:57:17 I have a patch to towards system scope in keystone ready for review: https://review.opendev.org/c/openstack/kolla-ansible/+/815577 15:57:26 can we kill it? 15:58:14 what's the problem with it? 15:58:52 headphoneJames: I reviewed today 15:59:24 mgoddard: thanks! 15:59:54 mnasiadka: just wondering how much it is used. if no one uses it then we do not even know does it work 16:00:28 well, I would say it might be useful for some people, but if it's not tested - then probably it doesn't work ;) 16:01:24 What's the usual number of Kolla image layers nowadays? Does it make sense to squash them if you add some things on top? 16:02:45 mnasiadka: test2/nova-libvirt:13.1.0 has 29 layers 16:02:52 ok, we're over time, let's continue after the meeting 16:02:53 hrw: I don't think it's hurting anyone 16:02:53 #endmeeting