19:01:02 #startmeeting infra 19:01:03 Meeting started Tue Mar 19 19:01:02 2019 UTC and is due to finish in 60 minutes. The chair is clarkb. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:01:04 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 19:01:06 The meeting name has been set to 'infra' 19:01:11 #link http://lists.openstack.org/pipermail/openstack-infra/2019-March/006296.html 19:01:21 #topic Announcements 19:01:33 \o 19:01:38 I'll be AFK next week from the 25-28th 19:01:55 o/ 19:02:07 Also doesn't look like anyone wanted to run against me for PTL so I get to do that again 19:02:27 clarkb: congratulations! 19:02:29 If you want to volunteer to chair the meeting next week let me know as I won't be able to do it 19:02:54 #topic Actions from last meeting 19:03:07 #link http://eavesdrop.openstack.org/meetings/infra/2019/infra.2019-03-12-19.01.txt minutes from last meeting 19:03:15 ianw: I htink you rotated those backups? 19:03:20 anything interesting to say about that? 19:03:31 yes, please review 19:03:34 #link https://review.openstack.org/644457 19:04:11 that's all :) 19:04:16 the other item was project renames 19:04:25 * jroll lurks 19:04:51 do we want to talk about that now or during the opendev topic? 19:05:22 * dtroyer sneaks in 19:05:37 Dedicated PTL https://usercontent.irccloud-cdn.com/file/87xmHZDF/IMG_20190319_120415.jpg 19:05:47 we can pick it up during the opendev topic 19:05:58 #topic Specs Approval 19:06:06 There are no specs to review 19:06:18 #topic Priority Efforts 19:06:40 We've updated more servers to puppet 4 in the last week. So far the only real gotcha has been that the pip provider doesn't like the warnings that pip prints on stderr 19:07:00 so if pip is emitting warnings we have to fix those (last time was via upgrade of cyptography and pyopenssl 19:07:17 corvus: anything new on the image building front? I guess that its actually in production now? 19:07:21 with documentation and everything? 19:08:06 yep, nothing substantial new there; just cleaning up a few bugs 19:09:07 great. 19:09:12 #topic Opendev 19:09:34 I managed to fixup the bug tracker situation in gitea 19:09:40 that should be all done now 19:10:10 yay! thanks! 19:10:32 Want to talk about project renames now? 19:10:42 i've had the initial redirects change up for review for a while, looking for feedback 19:10:44 #link https://review.openstack.org/642173 Set up legacy git redirect sites 19:10:47 everybody wants to, judging by the ML 19:12:26 #link mailing list thread about openstack namespaces: http://lists.openstack.org/pipermail/openstack-discuss/2019-March/003943.html 19:12:30 jroll: that thread? 19:12:40 sorry, yes, that one 19:13:17 it seems we have some traction, assuming it's died down I'll do a governance change to get the general ack 19:13:20 ya so far I have only heard positive things 19:13:41 when are we looking to try and get renames done? 19:13:44 i haven't really seen any overt objections on that thread, agreed. mostly just questions 19:14:07 jroll: April 19 is the day we pencilled in last week 19:14:14 same time as the switch in domain names, ideally 19:14:16 and we havne't run into any major issues since so that seems to still be on track 19:14:24 cool, that seems doable 19:14:45 jroll: what's your tl;dr read on that thread? 19:14:54 (assuming nobody comes and splatters random colors on the shed) 19:15:07 was it leaning toward one of your 3 options? 19:15:24 i think it was some of option 3 and some of ttx's option 4 19:15:25 corvus: I have to re-read to get a count, sounds like either (2) or ttx's option is the winner 19:15:27 * jroll finds that link 19:15:39 #link http://lists.openstack.org/pipermail/openstack-discuss/2019-March/003971.html 19:15:58 got it, thanks! 19:16:11 i'll be sure to read the whole thread after the meeting; i somehow missed that. sorry. 19:16:34 now that I think about it, both of those have some nebulous naming activities to do, but I think we can meet april 17 19:16:37 no worries :) 19:16:44 * jroll s/17/19/ 19:16:45 there's still some question as to whether team-oriented namespaces make more sense, and also what to do about projects with no representation on the openstack map diagram thingie 19:17:26 cool, those both sound great, and i agree, that sounds plausible 19:17:48 dtroyer: things look good your side? 19:18:19 clarkb: really quiet so far, at this point it looks like we'll just be moving to a starlingx namespace and retiring 3 or so repos 19:18:37 and dropping the stx-prefixes, or have you decided? 19:18:50 yes, dropping stx- too 19:19:46 I expect zuul will be ready 19:20:05 ready to talk about the git:// to https:// transition again? 19:20:08 ++ 19:20:14 I wanted to make sure that we don't get stuck in limbo on that topic 19:20:34 yeah, i think ianw wanted a decision on that at this meeting, and i agree we should do that :) 19:20:49 i did post about it as mentioned in the last meeting, but didn't generate as much discussion 19:20:52 #link http://lists.openstack.org/pipermail/openstack-discuss/2019-March/003825.html 19:21:10 ah it went to that list not the infra list which is probably why i missed it (I'm not great at keeping up there) 19:21:58 sorry i figured everyone would want to have an opinion since it might touch their repos, but it was maybe a little obscure 19:22:16 yeah, and it has a bit of a devstack-focus anyway 19:22:22 fwiw your proposal seems reasonable to me 19:22:24 i think i'm with ianw and on board option 1 19:22:45 same, option 1 for me 19:23:34 in that case it sounds like we can go ahead with 1) as proposed 19:23:57 when should we actually upload the changes? do we want to try to do that at the start of a weekend or something? 19:24:04 given that it looks like from the prior discussion we might be doing some fairly complex namespace swizzles, i think that avoids us having to synchronize 19:24:12 projects are sure to be putting through a ton of other changes after the transition to update domain names and project names and namespaces and stuff in their readmes and contributor docs anyway 19:24:15 corvus: maybe ianw can do it during his monday morning which is sunday for most of us 19:25:27 sure, i can make it drizzle in over 24 hours 19:25:51 ianw: i think if you do it then, you can do it all at once 19:26:09 zuul is pretty idle over the "weekend" 19:26:53 and hopefully that will reduce the window for copycat patches 19:26:58 sounds like a plan 19:27:12 (i still can't believe that's a thing we have to consider) 19:27:34 ok, i will set it all up in a screen on bridge.o.o and we can check it over 19:27:35 (i mean, i believe it, i'm just... you know, wow.) 19:28:02 anything else opendev related? 19:28:58 sounds like now. Onward! 19:29:05 oh 19:29:05 #topic Storyboard 19:29:11 should I undo? 19:29:19 yeah, sorry one more quick thing 19:29:20 oh, yeah, one more thing 19:29:23 #undo 19:29:25 Removing item from minutes: #topic Storyboard 19:29:34 should i send out an email announcing the date now? 19:29:59 corvus: ++ I don't think we have anything coming up that will prevent it at this point 19:30:09 (for me) we agree to force merge outstanding git://->https:// changes too? what date should we do that? 19:30:19 yes, should reduce some further confusion over the anticipated timeframe for things 19:30:29 the announcement i mean 19:31:00 ianw: i'm ambivalent about that -- since i don't expect *functional* breakage, i'm less inclined to force-merge, but i don't object if folks think it's worthwhile. 19:31:21 so, consider me a weak -1 on force-merge but happy to support it if there are other +1s 19:31:25 I think a deadline helps get eyeballs on it 19:32:17 the 12th gives people a couple weeks if we want it done ahead of time 19:32:44 yeah, i'm honestly more concerned about the .gitreview and {.,}zuul{.d/*,}.yaml edits not lingering 19:35:09 ok, ummm i like the idea of a deadline being a call to action -- i'll draft something in the review commit log as part of the change; i'll request comments on in a gerrit review of the scripts 19:35:17 ianw: sounds good 19:35:22 #topic Storyboard 19:35:28 ok anything storyboard to talk about? 19:35:34 diablo_rojo_phon says no 19:35:39 there are changes to review 19:35:52 mordred isn't around today but any idea how his patches for swift things are going? 19:36:56 yeah, lots and lots of sb changes to review 19:37:03 sounds like they are still outstanding due to changes needed in the sdk 19:38:15 Lets move on since diablo_rojo_phon is eating lunch 19:38:20 SotK just pushed up a couple of testing-related improvements as well 19:38:26 Yes please lol 19:38:45 * fungi lets diablo_rojo_phon get back to her lunch. clarkb doesn't actually get to eat though 19:38:54 #topic General Topics 19:39:03 #link https://etherpad.openstack.org/p/201808-infra-server-upgrades-and-cleanup 19:39:06 Server upgrades continue 19:39:15 AFS is completely done now which is nice to get behind us 19:39:20 fungi: how are wiki things? 19:39:48 #link https://review.openstack.org/643476 Switch from PHP5 to default PHP (PHP7 on Xenial) 19:40:08 that's needed to progress further on the wiki-dev02.openstack.org deployment 19:40:34 if any infra-puppet-core is available to review, it's a one-character removal 19:40:52 I can take a look today 19:41:22 The last item on my list is PTG planning 19:41:28 #link https://etherpad.openstack.org/2019-denver-ptg-infra-planning 19:41:37 #link https://www.openstack.org/ptg#tab_schedule Draft schedule. We have Friday and Saturday in a shared room with the QA team. 19:41:42 I added an item after PTG planning, needs refresh :D 19:41:54 The draft schedule has us in a room friday and saturday with the qa team 19:41:56 clarkb: is there any chance we could shift that to thurs-friday? 19:42:06 corvus: we can ask, I don't actually know 19:42:25 I think the saturday is a conflict for fungi too 19:42:36 I'll investigate that 19:43:09 yeah, the tc also wants a full day on saturday 19:43:15 in addition to that, i would be largely idle on thursday 19:43:45 so if we can do something on thursday and possibly friday, it's a better use of time and we dodge fungi's conflict 19:43:48 #action clarkb ask about infra ptg days being thursday friday 19:44:19 if it doesn't work out, maybe we can just find a nice spot on thursday and hack on server container stuff anyway? 19:44:23 ++ 19:44:59 Ok next up is dmsimard's ara github replication item 19:45:08 o/ 19:45:12 I think this is serving as a template for replication to git remotes via zuul? 19:45:33 indeed, I've written a role with a job to test it: https://review.openstack.org/#/q/topic:upload-git-mirror 19:46:11 It should hopefully be flexible enough to push from anywhere to anywhere over SSH 19:46:11 dmsimard: great, thanks! are those ready for review now? 19:46:26 corvus: they are 19:46:58 i will (re-)review them asap 19:47:02 So once we set that up, should we remove ara from the gerrit github replication ? 19:47:17 right now our replication is openstack* 19:47:39 so it might be tricky to pull one out... 19:48:02 Yeah I looked at the replication config: configuration 19:48:05 er 19:48:12 https://opendev.org/openstack-infra/system-config/src/branch/master/modules/openstack_project/manifests/review.pp#L238 19:48:34 dmsimard: but.... 19:48:43 dmsimard: you could *additionally* replicate to another repo pretty easily 19:49:00 dmsimard: like, you could replicate openstack/ara to dmsimard/just-testing-git-pushes 19:49:04 right 19:49:23 i think that would be a sufficient "production test" before we tell everyone to start using that 19:49:46 sounds great to me 19:49:47 The thing is I'd like the ability to *move* that github repo and then once it's moved, keep replicating to the new repo 19:50:09 The reason for the move would be to carry stuff like watchers/stars/forks/pull requests/etc around 19:50:32 basically asking that the namespace moves in gitea also be reflected with org transfers in github? 19:51:07 the tricky bit (perhaps less so for you though) is that last i looked you need a common admin account in both the source and target orgs to make that work 19:51:13 fungi: I don't mind where ara ends up being replicated in gitea 19:51:38 fungi: ya we might have to bootstrap that 19:51:46 fungi: but once its done we should be able to remove ourselves from all the orgs 19:51:56 I'm admin in the destination org, I'm not sure I have admin in the openstack org 19:52:03 do we have an automated way to do that transfer? 19:52:19 not to my knowledge, it's a manual transfer in the github ui 19:52:24 that's the other problem... last i looked, org transfers weren't exposed in the gh api 19:52:49 so if we did this, we may need to perform 1500+ transfers manually through the gh web ui? 19:53:24 i don't think that's something i can commit to. 19:53:44 we should double check that 19:53:49 because ya that will be painful 19:54:03 corvus: where would those 1500+ projects be moved to ? 19:54:07 well, it would if we did it. but i don't think we have said that we would. 19:54:20 it hasn't been in any of the project planning so far 19:55:02 we've made a big point of saying we'll make sure the git urls we've been hosting continue to work and get redirected to the new domain/orgs/repos but nobody's actually asked if the same will be true of repos we're replicating to github. i know a lot of people have remotes and bookmarks to github against our recommendation, and may not realize this is going to impact them as well 19:55:25 its definitely something we should do if feasible 19:55:27 dmsimard: i don't know where they will be moved to; presumably wherever their owners wish? 19:55:54 fungi: we've said that if you aren't in the openstack* namespace, automatic replication will cease. 19:56:09 corvus: fair, I went with the assumption that there was not going to be many projects in a situation similar to ara 19:56:17 corvus: right. most projects are *currently* in the openstack namespace though 19:56:28 dmsimard: you may have missed the opendev topic earlier 19:56:33 ~most of the plans supported so far in the openstack-discuss ml thread will involve evicting a majority of the repositories currently in that namespace 19:56:56 including official openstack projects 19:57:02 yeah that ^ 19:57:31 time check we have ~3 minutes left 19:57:40 what would be left in the openstack namespace would either be just the ~60 service repos, or possibly nothing whatsoever 19:57:41 dmsimard: can you look into whether there's a way to automate that? 19:57:44 fungi: what does eviction imply in practice ? no more replication ? deletion of the project ? moving it elsewhere ? 19:58:06 dmsimard: that hasn't been covered yet exactly 19:58:06 dmsimard: it means a rename in gerrit, and whatever the project wants to do on github. 19:58:10 fungi: ok so github.com/openstack will remain but only with the official official things 19:58:29 likely a very small subset of the official official things 19:58:33 dmsimard: that hasn't been decided yet, but that's the gist of the most likely options. 19:58:34 and perhaps not even those 19:58:34 ack 19:59:19 I'll explore if there is an easy way to automate the transfer 19:59:25 dmsimard: thanks! 19:59:39 And we are basically at time 19:59:42 thank you everyone 19:59:47 and yes I somehow managed to eat lunch too :) 19:59:50 thanks clarkb! 19:59:58 now, dessert 20:00:12 Find us on IRC (#openstack-infra) or on the openstack-infra@lists.openstack.org mailing list if there are things to followup on 20:00:15 #endmeeting