13:02:49 #startmeeting deb_packaging 13:02:50 Meeting started Mon Sep 5 13:02:49 2016 UTC and is due to finish in 60 minutes. The chair is IgorYozhikov. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:02:51 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:02:54 The meeting name has been set to 'deb_packaging' 13:03:04 #chair IgorYozhikov zigo tlbr mordred 13:03:05 Warning: Nick not in channel: mordred 13:03:06 Current chairs: IgorYozhikov mordred tlbr zigo 13:03:38 #link https://etherpad.openstack.org/p/openstack-deb-packaging 13:04:02 #topic Stackalitycs status 13:04:22 I have no news from Ilya about it. 13:04:31 Though he knows already what needs to be done. 13:04:41 It'd be nice to have it fixed before the elections for PTL start. 13:04:52 Did he provide any ETA ? 13:04:54 He's in Moscow office, right? 13:05:00 No, Saratov. 13:05:01 He said last week ... 13:05:04 Ah... 13:05:10 Daniil is where? 13:05:20 Do we have anyone near him to beat him hard? :) 13:05:41 He doesn't feel well today. =\ 13:05:51 :/ 13:05:53 zigo, Daniil on sick leave 13:05:55 mivanov is also away 13:06:15 So we don't anyone from there for now. 13:06:18 So, the only way is to ping Ilya by email then. 13:06:28 No the only way. 13:06:28 I'll do that. 13:06:32 Slack is ok. 13:06:56 #action zigo to ping Ilya about Stackalytics needed update for packaging-deb 13:07:07 Next topic? 13:07:21 #topic project status 13:07:39 We got almost all of Oslo packages built on infra. 13:07:56 zigo, it is awesome 13:08:28 where can these packages be found? 13:08:34 oslo.cache and oslo.db are missing, because we need deb-alembic to be approved as new gerrit repo by the infra team, and we need tag push ACL for pushing new versions of dogpile.cache 13:08:49 The rest should build ok (a few are remaining to trigger) 13:08:50 https://review.openstack.org/#/c/365477/1 13:09:05 Yup, that's the one. 13:09:21 #link https://review.openstack.org/365477 13:09:39 strigazi, http://mirror.regionone.osic-cloud1.openstack.org/debian-openstack/ 13:09:44 thx 13:10:01 strigazi: That's *one* of the URLs, there's in fact one repo per cloud that infra uses. 13:10:21 ok 13:10:22 zigo, are we going to push tags manually or through existing in ifra mechanism? 13:10:23 You also shall remember that these are only for testing purposes, what counts at the end is what we upload to distros. 13:10:47 IgorYozhikov: The thing that I am not clear with is if we can push non-signed upstream tags. 13:11:10 IgorYozhikov: For example, alembic has rel_0_6_2, we need to push a 0.6.2 label on top. 13:11:17 None of them are in gerrit. 13:11:29 I'm not sure how this will work, but it's very important we clear that up. 13:11:40 IgorYozhikov: Any suggestion / thoughts about it? 13:11:48 zigo, u r now about git-buildpackage structure right? 13:11:57 Yup. 13:12:04 Using git tag workflow. 13:12:06 i c, now it's clear 13:12:16 No package uploaded to gerrit is using pristine tar. 13:12:33 Probably, if we have the ACL, that's enough. 13:12:38 I'd say, we'll see, no? :) 13:13:00 we could try 13:13:45 Also, nearly all of the packages in deb-auto-backports are built too. 13:13:55 We can just build the remaining as we need them. 13:14:31 #topic canonical server team contribution 13:14:55 It looks like it should be possible to have James Page change his mind, according to the discussion on the Alioth list. 13:14:56 wow, really, canonical will join us? 13:15:21 Let me find the link of the thread... 13:15:38 #link http://lists.alioth.debian.org/pipermail/openstack-devel/2016-September/018098.html 13:16:26 We need to define a clear workflow for uploads in Debian, and explicit the fact that we consider infra deb repo as non-official testing only repo. 13:16:31 ah, I've seen this 13:16:35 Meaning someone needs to write in the install-guide. 13:16:53 So far, I've been the only contributor to the install-guide in the team. 13:16:59 Anyone up for the task? 13:17:15 It'd be really nice if we could show to the doc guys that I wont be the only contributor. 13:17:41 They've been very vocal over the years that it shouldn't be the case. 13:17:50 IgorYozhikov: tlbr: Do you have time for that? 13:18:01 I can do that, but we need to discuss it. 13:18:12 To have a clear vision. 13:18:32 tlbr: What do you need to know? 13:18:36 zigo, good point, also we need to start with filling up of wiki space 13:19:05 zigo: you mean the debconf install-guide? 13:19:07 zigo, for example get commits from Alioth, not everyone uses Gerrit for it. 13:19:14 #action tlbr to commit in the install-guide about the fact OpenStack infra repo are *not* for production use, but testing only. 13:19:27 strigazi: Both the debconf one and the non-interactive one. 13:19:49 tlbr: This has nothing to do with Alioth, as as far as I'm concerned, Alioth git repo will die. 13:20:35 tlbr: It's about OpenStack infra debian repository vs distribution (ie: Debian Sid/testing & Ubuntu Xenial + UCA or the latest Ubuntu). 13:20:40 zigo, well, we still need ongoing changes from it. 13:20:40 zigo, let's tlbr prepare a draft 13:20:47 after that we will review it 13:20:52 and post on wiki 13:21:00 tlbr: Feel free to start a WIP patch first... 13:21:26 zigo, okay 13:21:42 IgorYozhikov, ok, will do 13:21:46 Do you guys see anything else that could ease Canonical's team to join our effort? 13:21:48 cool 13:22:02 +decision 13:22:13 What do they actually need ? 13:22:48 So far, James only expressed 2 concerns: 1/ that we should explicitely be vocal that users should use distribution repositories 13:22:53 I very much agree with this 13:23:11 and 2/ that we should have all the PKG OpenStack team to agree on the move to Gerrit 13:23:23 I really think both are workable. 13:23:35 I don't see any blockers. 13:23:37 For 2/, I believe it's already the case that everyone agrees. 13:23:50 yep 13:24:02 But maybe there are more things we could do to make them comfortable? 13:24:07 that was initially decided 13:24:11 I think 2 will increase contributions from projects 13:24:12 If you have any idea, please shoot ... 13:24:36 strigazi: That's indeed my hope as well. 13:24:37 Provide them with documentation if required. 13:24:55 Shall we set the topic to building contributor's guide? :) 13:24:56 zigo, without any input from them - we can only guess 13:25:08 #topic documenting contribution to packaging-deb 13:25:10 I guess so :) 13:25:10 agree 13:25:24 I'll stop typing, IgorYozhikov, you have the con ... :P 13:25:42 zigo, ok 13:26:22 I suppose to prepare a contributions guide and post it here - https://wiki.openstack.org/wiki/DEB-packaging 13:26:55 I believe that this is a good place to store it 13:27:14 What content shall we set there? 13:27:15 I'll prepare initial page for it 13:27:28 Things are too obvious for me, I don't think I'd be the good person to write in there. 13:27:45 Though ideas could be taken from: http://openstack.alioth.debian.org/ 13:27:46 I think, that at 1st - how to update already exiting projects 13:27:47 #link http://openstack.alioth.debian.org/ 13:28:13 I don't think that is very obvious that users should have enough rights to push merge commits :) 13:28:19 2nd - style 13:28:28 with examples 13:28:45 tlbr: I agree, we should document the fact that only packaging core group has this right. 13:29:04 ie - how should resulting package specification looks like 13:29:17 zigo, at leat we may specify that it could be changed in future. 13:29:23 *least 13:29:55 It should be all but a guide on how to do actual packaging. There's enough docs about that in Debian. We could point to existing packages already in production within the packaging-deb project, as well as the packaging-tutorial package in Debian, if contributors need to know. 13:30:34 Maybe we should also explain what the git-buildpackage tag based workflow is, no? 13:30:49 That page here http://openstack.alioth.debian.org/ has such details. 13:30:54 so future doc need to be filled up with links to already existing documentation 13:31:00 But it's outdated, and should be reviewed. 13:31:07 Yup. 13:31:12 Just specify that we use git-buildpackage, it should be enough. 13:31:51 tlbr: And the fact we generate tarballs out of upstream git tags too (ie: we don't use pristine-tar). 13:32:04 zigo, yeap, good idea 13:32:16 IgorYozhikov: You're up for the task, right? 13:32:28 I can try 13:33:11 #action IgorYozhikov to write contributor guide, explaining the git-buildpackage tag based workflow, how pushing tags & merge commit work (ie: only from core group), plus links to existing Debian packaging docs 13:33:12 zigo, assign an AI on me 13:33:16 ah, ok 13:33:41 Next item? 13:33:45 IgorYozhikov If you need feedback from a new contributor count me in 13:34:08 strigazi, thanx 13:34:08 strigazi: Nice, you're very much welcome to help! 13:34:28 #topic todo list for next week 13:34:51 So, remaining to do before we build services, is building all the python*client 13:35:12 agree here, same we are going to do in downstream 13:35:19 I attempted building python-keystoneclient, and it needs tempest, which is currently stuck with a failed download 13:35:32 Though it should normally work... 13:35:54 Then we'll need to build cross-project libs like os-brick, neutron-lib and such. 13:35:57 I saw, and about tempest 13:36:08 Yeah, that's *very* annoying. 13:36:13 do we really need it for unit tests? 13:36:23 It's due to the fact that the Debian mirror from infra is unsigned. 13:36:31 Meaning we can't use it with sbuild-createchroot. 13:36:42 Hopefully, they will fix it and use a *real* ftp-sync script. 13:36:48 Though currently, we have to deal with what we have. 13:37:12 tempest is a build dependency of *many* things, since it now replaces tempest-list. 13:37:28 It's possible we'll need to remove temporarily some build-dependency of tempest so that it can build... 13:37:31 zigo, I saw a similar test reqs in *murano* 13:37:32 The usual drill you know. 13:37:44 I'm really not worried. 13:38:05 Anything else to discuss ? 13:38:17 and dev team told me that tempest needed only 4 functional testing 13:38:19 Once we've build everything, we should attempt to upgrade all to Newton b3 (all is currently Newton b2...). 13:38:19 not unit 13:38:31 that's why I'm asking 13:38:35 Yeah, another topic! :) 13:38:41 :) 13:38:42 #topic removal of duplicate repo 13:39:28 #link https://review.openstack.org/362732 13:39:53 I don't think that we should remove it at this moment 13:39:54 tlbr: Can you take care of that one? 13:39:58 Why? 13:40:35 I'm not sure, if python- prefixes could be an issue when we will use zuul-cloner 13:40:53 Because Debian name differs =\ 13:40:57 It isn't, as much as I can tell. 13:41:06 Well, I'm not sure 13:41:14 What issue do you forsee exactly? 13:41:15 It need further investigation 13:41:40 tlbr: So are you suggesting we just wait, and deal with the deletion later? 13:41:55 I don't really mind if we do that... 13:42:21 zigo, lets postpone it for now, please. :) 13:42:25 Ok 13:42:54 #action We will *not* remove duplicate repositories for now, it's postpone for at least a week 13:43:06 #topic open discussion 13:43:31 strigazi: IgorYozhikov: tlbr: Anything else you'd like to discuss? 13:43:41 zigo, are you picking new changes from Alioth ? 13:43:53 or we don't need it at all at this moment 13:43:58 and can discuss it later 13:44:01 tlbr: I'll do that manually, that's implicitely part of Newton b3 move. 13:44:18 tlbr: I'm not sure how to detect these changes though. 13:44:27 zigo, by squashing commits into one ? 13:44:29 tlbr: Are you suggesting some kind of automation ? :) 13:44:42 zigo, yes, sure we need it 13:44:51 tlbr: Yeah, no choice, because we must do "merge commit" with the new upstream version ... 13:45:07 About the build of clients, is python-magnumclient going to be included? 13:45:12 zigo, that is my only concern for now :) 13:45:18 The merge commit must contain the merge of the new tag, AND the necessary packaging change in a single commit. 13:45:28 strigazi: Yes, it will. 13:46:02 strigazi: Everything that is over here will be in upstream OpenStack infra at some point: https://qa.debian.org/developer.php?login=openstack-devel@lists.alioth.debian.org 13:46:29 tlbr: Will you write the automation script to detect changes between alioth and Gerrit? 13:46:42 IMO, it should only detect changes in the debian folder, that's enough. 13:47:17 strigazi, Heat requires it https://github.com/openstack/heat/blob/master/requirements.txt#L42. We will do it for sure 13:47:30 zigo, at least I may try to do that 13:47:38 Cool. 13:47:39 zigo, as a new job ? 13:47:50 I mean, periodic job 13:47:55 tlbr: Just a shell script, which you can drop anywhere, that's IMO fine. 13:48:03 zigo, ok, will do 13:48:05 We will only need it for the time of the migration, IMO. 13:48:19 Agree :) 13:48:27 #action tlbr to write a script to check inconsistency between alioth and gerrit, so we wont forget changes pushed to alioth 13:48:35 zigo, what are your plans for summit, r u going to be in Barcelona? 13:48:36 I think we're done here, no? 13:48:53 yeap, thank you for your time! 13:48:57 IgorYozhikov: Definitively going yes. I need to open a ticket @mirantis for that. 13:49:21 It's close from where I live anyway. We may even drive to it, not sure about it yet. 13:49:51 ok, rpm-packaging colleagues want to discuss "build machine" 13:49:54 Ok, let's close the meeting then, let's chat in #openstack-pkg. strigazi you're of course welcome there. 13:50:03 IgorYozhikov: At the summit? 13:50:07 yep 13:50:12 cool 13:50:15 What does "build machine" means ? 13:50:28 I asked him about that on previous week 13:50:38 package builder on infra 13:50:46 Ah, right. 13:51:01 Well, they can look how we've done with openstack-pkg-tools and what's in project-config. 13:51:09 I'm not sure how we can advise better than this. 13:51:26 :) 13:51:29 Also, probably we'll move pkgos-infra-install-sbuild to diskimage-builder somehow. 13:51:40 so, we have 9 mins to go 13:51:42 It really takes too much time to setup sbuild on each build. 13:51:50 IgorYozhikov: Yes, but we're done, no? :) 13:52:06 looks like it there are no objections 13:52:14 or topics :) 13:52:24 s/it/if/ 13:52:26 Thanks everyone! 13:52:28 #endmeeting