15:35:38 <dgilmore> #startmeeting RELENG (2015-09-28)
15:35:38 <zodbot> Meeting started Mon Sep 28 15:35:38 2015 UTC.  The chair is dgilmore. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:35:38 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:35:47 <dgilmore> #meetingname releng
15:35:47 <zodbot> The meeting name has been set to 'releng'
15:35:47 <dgilmore> #chair dgilmore nirik tyll sharkcz bochecha masta pbrobinson pingou maxamillion
15:35:47 <zodbot> Current chairs: bochecha dgilmore masta maxamillion nirik pbrobinson pingou sharkcz tyll
15:35:50 <dgilmore> #topic init process
15:36:39 <nirik> morning
15:36:43 <pbrobinson> morning
15:37:21 <dgilmore> pbrobinson: night :)
15:37:47 <pbrobinson> dgilmore: in 23 mins it'll be morning ;-)
15:38:19 <dgilmore> pbrobinson: gotta get it in while I can
15:39:18 <maxamillion> .hello maxamillion
15:39:19 <zodbot> maxamillion: maxamillion 'Adam Miller' <maxamillion@gmail.com>
15:39:28 <dgilmore> hey maxamillion
15:39:31 <dgilmore> lets get started
15:39:37 <dgilmore> #topic #6262 drop rawhide-stable tag and consider master branch to be always stable
15:39:46 <dgilmore> https://fedorahosted.org/rel-eng/ticket/6262
15:39:56 <dgilmore> not sure I really care too much
15:40:06 * nirik is fine with whatever .
15:40:16 <dgilmore> but I think before doing this we should lay out a new development workflow
15:40:34 <dgilmore> set clear expectations on how things work
15:41:04 <maxamillion> we could adopt the development workflow that the Fedora QA team uses gitflow so the develop branch is dev and master is stable
15:41:20 <dgilmore> maxamillion: I have no idea what workflow they use
15:41:40 <maxamillion> dgilmore: I do mostly just because I've been mucking around in taskotron bits recently
15:41:57 <tflink> we use part of gitflow: http://nvie.com/posts/a-successful-git-branching-model/
15:42:25 <nirik> right, so you develop/commit against dev and merge to master when you are ready for stable
15:42:28 <maxamillion> also -> http://jeffkreeftmeijer.com/2010/why-arent-you-using-git-flow/
15:42:31 <tflink> we don't use all of it because we're small enough not to need all of that but it's been helpful for us so far
15:42:34 <pbrobinson> I would think we'd want to land on master and then merge back to stable once it's ready to go?
15:42:36 <dgilmore> there is a twoweek-stable tag also
15:42:41 <pbrobinson> ie master is dev
15:42:55 <maxamillion> dgilmore: we couuld probably get rid of that one also and just run on master
15:42:58 <maxamillion> could*
15:43:03 <maxamillion> if master is stable
15:43:18 <maxamillion> pbrobinson: either way, it's effectively just namespacing at that point
15:43:26 <pbrobinson> yep
15:43:39 <tflink> pbrobinson: fwiw, that concern was brought up before we switched and there's not been any confusion as near as I can tell
15:43:42 <pbrobinson> I was more thinking it reflects other branches as used in Fedora
15:43:45 <dgilmore> as I said at the start, if we are going to change it we should change the entire workflow and make it clear to all
15:43:48 <tflink> any significant confusion, rather
15:44:06 <pbrobinson> ie master is rawhide/Fedora next release for most stuff
15:45:01 <maxamillion> pbrobinson: I think the Fedora Apps group actually follows a similar model as the QA team with "develop" as dev and "master" as stable ... at least I'm pretty sure I've seen some of lmacken's stuff like that .... so it's kind of where you look, but I'm pretty impartial ... I've worked with both "styles" and was happy with either
15:45:31 <pbrobinson> who are the "Fedora Apps group"?
15:45:46 <dgilmore> pbrobinson: the people reporting to Paul Frields
15:46:01 <pbrobinson> ah
15:46:07 * pbrobinson goes back to sleep
15:46:08 <dgilmore> ralph, lmacken, pingou etc
15:46:40 <maxamillion> #link http://nvie.com/posts/a-successful-git-branching-model/
15:46:47 <maxamillion> #link http://jeffkreeftmeijer.com/2010/why-arent-you-using-git-flow/
15:46:51 <dgilmore> so I guess we should sit down, workout how things will work and make the change
15:47:14 <dgilmore> maxamillion: they show up without th #link if you paste just a url
15:47:17 * nirik largely doesn't care.
15:47:20 <maxamillion> dgilmore: oh
15:47:47 <dgilmore> nirik: I don't either
15:47:49 <maxamillion> If anyone cares, I'll take an action item to start an email thread about it and see what people think at large and we can decide next week based on responses
15:47:53 <dgilmore> and tyll is not here
15:48:14 <dgilmore> maxamillion: I think tyll is the only one that really cares since he filed the ticket
15:48:19 <maxamillion> dgilmore: ah ok
15:48:53 <dgilmore> I wouldn't say how we do it is ideal
15:49:03 <dgilmore> but it's not really broken
15:49:14 <dgilmore> but doing things simpler would not hurt
15:49:25 <dgilmore> we just need to be very clear about it
15:49:29 <maxamillion> yeah
15:49:45 <nirik> it's not like we gate things based on the tag or anything now...
15:49:57 <nirik> so the current method is pretty useless. ;)
15:50:10 <maxamillion> nirik: different cron jobs run from different tags
15:50:48 <nirik> sure, but I mean, you make changes to master and you move the tag...
15:50:54 <nirik> there's not like some kind of testing process or ack
15:52:00 <maxamillion> nirik: oh, yeah
15:52:25 <dgilmore> nirik: there kinda is
15:52:31 <dgilmore> but not in the way designed
15:52:39 <nirik> the tagged master has been broken before in the past. ;)
15:52:42 <dgilmore> I will run things in a test sometimes
15:52:52 <dgilmore> nirik: it has
15:53:48 <dgilmore> I am not opposed to changing things
15:53:54 <nirik> anyhow, fine with whatever changes people want
15:54:10 <dgilmore> I just think we should step back and look at it all and change the workflows
15:54:35 <maxamillion> on that note, it would be nice if we could test things
15:54:45 <dgilmore> #action maxamillion to start a discussion on code review and development workflows
15:55:07 <dgilmore> maxamillion: it would be. not sure how possible that is for all things
15:55:16 <dgilmore> though maybe its easier with stg working
15:55:47 <nirik> we could do composes in stg now I think...
15:57:00 <dgilmore> we can try
15:57:41 <dgilmore> there is not really many changes in buildbranched and buildrawhide
15:57:57 <dgilmore> lets move on
15:58:25 <dgilmore> #topic Secondary Architectures updates
15:58:26 <dgilmore> #topic Secondary Architectures update - ppc
15:58:59 <dgilmore> pbrobinson: other than we shipped ppc beta anything to go over?
15:59:27 <pbrobinson> not really, trying to smack imagefactory around enough to get docker images working
15:59:50 <pbrobinson> and planning to get some cycles into the Power 8 HW
16:00:25 <nirik> pbrobinson: on the power8 stuff you want to try out 7.2 beta?
16:01:15 <dgilmore> pbrobinson: speaking of power8
16:01:16 <pbrobinson> nirik: it might be easier in the short term, I think there's some assumptions on EL for the virt hosts so it might be easier for maint wise moving forward
16:01:36 <nirik> sure, whatever you think best
16:01:39 <dgilmore> if we are to add ppc64le to epel primary koji will need two builders capable of building it
16:01:41 <nirik> I can sync the beta content
16:02:13 <pbrobinson> dgilmore: yes, all planned :)
16:02:37 <pbrobinson> dgilmore: synced all the plans with nirik @ Flock and we worked it out
16:02:42 <dgilmore> adding aarch64 i686 and ppc64le came up in the epel meeting friday
16:03:00 <dgilmore> okay
16:03:17 <nirik> yeah, the power8 stuff has multiple interfaces we can hopefully add one to the right net and hook guests to it
16:03:33 <pbrobinson> dgilmore: yea, I have the first and last on my list, the later being highest, the first it depends on a few bits
16:03:43 <nirik> then hopefully retire the current buildppc ?
16:04:06 <pbrobinson> nirik: yup
16:04:21 <dgilmore> nirik: likely, but would depend on getting be and le builders
16:04:35 <nirik> really would be nice to nuke them as they are el6 still and on an old ppc hw
16:04:41 <nirik> yeah
16:04:49 <pbrobinson> dgilmore: the plan is to do resilient VMs across the HW
16:04:59 <dgilmore> okay
16:05:10 * nirik nods.
16:05:24 <dgilmore> anything else ppc?
16:05:31 <pbrobinson> not from me
16:05:34 <dgilmore> #topic Secondary Architectures update - s390
16:05:43 <dgilmore> no dan in here
16:06:21 <dgilmore> pbrobinson: any idea where s390 stands?
16:06:48 <pbrobinson> I know he's prepping Beta
16:07:00 <dgilmore> cool
16:07:25 <dgilmore> #info s390 beta is in preperation stages
16:07:28 <dgilmore> #topic Secondary Architectures update - arm
16:08:13 <pbrobinson> nothing really here from me
16:08:25 <nirik> pbrobinson: oh, I noticed something on the aarch64 builders.. they were getting stuck on ansible runs.
16:08:52 <nirik> It seems we still had a default nfs variable they were using, so they were trying nfs v4 mounts and that was hanging.
16:08:59 <nirik> I changed them to v3 and that cleared that up...
16:09:05 <pbrobinson> nirik: they sometimes lock up if they've been up a while
16:09:09 <nirik> but some of them still have high load due to the stuck nfs
16:09:13 <pbrobinson> I've got a possible patch to test
16:09:16 <nirik> and some of them I canot reach. ;(
16:09:16 <nirik> yeah
16:09:26 <pbrobinson> nirik: I rebooted most of them this morning
16:09:41 <nirik> ok. I can make a pass and see what ones are now up
16:09:56 <maxamillion> need to run, apologies
16:09:57 <nirik> I commented some in ansible inventory the other day
16:10:13 <nirik> f5baba3d9aa8035e877f16ec0f3bc2ddc7034e39
16:10:59 * pbrobinson looks
16:11:37 <nirik> can sort it out of meeting, just wanted to mention it
16:11:45 <pbrobinson> nirik: 3-10 should be live, not sure what happened to 11/12 (they were still being shipped)
16:11:47 <dgilmore> speaking of ansible. lets try ansibilise the arm hub. we should be able to bring up more vms on the host the hub and db are on
16:12:13 <dgilmore> what are the remaining issues with the s390 hub?
16:12:15 <pbrobinson> dgilmore: I was awaiting the last bits of the s390 one to be cleaned up
16:12:44 <dgilmore> pbrobinson: what exactly are they?
16:12:45 <pbrobinson> but I'm basically happy to have it done when ever nirik is happy to do it, just need to back up some bits first
16:13:01 <nirik> dgilmore: there's a ticket.
16:13:05 <pbrobinson> dgilmore: they're documented in the ticket
16:13:06 <nirik> I was hoping to work on that this week. ;)
16:13:09 <dgilmore> we should have plenty of space to leave the old hub and db disks intact
16:13:24 <nirik> yep.
16:13:24 <nirik> I can make new ones and we can look them over then migrate.
16:13:47 * nirik finds the ticket
16:14:05 <nirik> https://fedorahosted.org/fedora-infrastructure/ticket/4840
16:15:28 <dgilmore> I will try and fix the issues this week
16:16:05 <nirik> dgilmore: cool. I had it on my list, but if you want to, fine with me. ;)
16:16:24 <nirik> I think it's mostly just tweaking
16:16:38 <pbrobinson> agreed
16:17:27 <dgilmore> https://fedorahosted.org/fedora-infrastructure/ticket/4840#comment:3
16:17:35 <dgilmore> I want to bring that up
16:17:50 <dgilmore> on primary we do not do what was done there
16:18:20 <dgilmore> we have all the directories in /mnt/koji explictly exported under /
16:19:00 <dgilmore> https://kojipkgs.fedoraproject.org/packages/ https://kojipkgs.fedoraproject.org/repos/ https://kojipkgs.fedoraproject.org/mash/ etc
16:19:12 <dgilmore> I would rather we keep things consistent
16:19:25 <nirik> sure.
16:20:37 <dgilmore> http://s390pkgs.fedoraproject.org/packages/ redirects to kojipkgs.fp.o/packages
16:21:50 <dgilmore> and they all seem to be on s390 other than http:// for packages redirecting incorrectly
16:23:10 <dgilmore> So i think we just need to udpate the ticket to make expectations clear
16:23:21 <dgilmore> okay lets move on
16:23:28 <dgilmore> #topic open floor
16:23:37 <dgilmore> does anyone have anything?
16:23:44 <nirik> I had one item...
16:24:16 <nirik> I'd like to update docs and then announce this week the new pkgdb scm process and retire the old in bugzilla one. We have had this ready for a long time, we should move to it.
16:24:33 <nirik> unless anyone knows of some reason it's not done?
16:24:57 * maxamillion is back ... apologies for anything I missed
16:25:00 <dgilmore> lets do it
16:25:10 <dgilmore> maxamillion: it is okay, we gave you work
16:25:14 <nirik> cool. I will talk to folks, update docs and send an announcement.
16:25:20 <maxamillion> dgilmore: sounds good
16:25:22 <maxamillion> nirik: +1
16:25:35 <nirik> also likely we should make a script to find bugs with the old thing and tell them it's not used anymore and run that for a few weeks at least
16:25:43 <dgilmore> #info nirik to update docs and officially retire branch requests via bugzilla
16:26:10 <dgilmore> nirik: probably a good idea
16:27:20 <dgilmore> shoudl be able to modify the one that processes the queue
16:27:24 <nirik> yep
16:28:12 <dgilmore> anything else?
16:28:29 <nirik> not from me.
16:28:37 <dgilmore> #endmeeting