19:00:02 <smooge> #startmeeting epel
19:00:02 <zodbot> Meeting started Wed Feb 17 19:00:02 2016 UTC.  The chair is smooge. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:00:02 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
19:00:02 <zodbot> The meeting name has been set to 'epel'
19:00:06 <smooge> #meetingname epel
19:00:06 <zodbot> The meeting name has been set to 'epel'
19:00:31 <smooge> #chair smooge dgilmore nirik avij bstinson
19:00:31 <zodbot> Current chairs: avij bstinson dgilmore nirik smooge
19:00:43 <bstinson> woo epel meeting!
19:01:06 <smooge> hi guys.
19:01:08 <avij> hello
19:01:24 <yselkowitz> .hello yselkowitz
19:01:25 <zodbot> yselkowitz: yselkowitz 'Yaakov Selkowitz' <yselkowi@redhat.com>
19:01:42 <smooge> this may be a short meeting due to the fact that several of us may be dealing with the glibc and kernel issues
19:01:56 <smooge> #topic meet and greet
19:02:00 <smooge> hello yselkowitz
19:02:05 <yselkowitz> hi smooge
19:02:06 <smooge> hello orionp
19:02:12 <nirik> morning
19:02:14 <dgilmore> hi
19:02:26 <dgilmore> I am still way way behind from the last 3 weeks
19:04:17 <smooge> ok I didn't put out an agenda last night but there will be one for next week.
19:04:57 <smooge> my current tasks are to finish writing up various proposals that came up at FOSDEM so we can either talk about them on the mailing list and discuss them at tnext meeting
19:05:12 <smooge> #topic limited agenda
19:05:38 <smooge> does anyone have anything that needs to be on the agenda this week?
19:07:01 <orionp> hello
19:07:03 <bstinson> none from me. for this week, we should probably communicate back our decision from last week to the nginx maintainer
19:07:42 <smooge> I have one item: alternative architecture buildout for arm32
19:08:23 <smooge> #topic nginx
19:08:43 <smooge> I believe in the last meeting we looked over the nginx proposal and approved it.
19:08:48 <smooge> bstinson, is that what you remember?
19:10:33 <yselkowitz> that's what my logs show
19:10:47 <bstinson> yeah, just skimmed through last week's minutes
19:10:57 <yselkowitz> <smooge> #agreed nginx to update to latest version
19:11:20 <smooge> ok I will let him know after this meeting.
19:11:42 <smooge> #topic altarch work for arm32
19:12:43 <smooge> so one of the things that came up a couple of times at FOSDEM was having EPEL for arm32. This is probably the hardest of the arches to put into our 'buildsystem' even if we wanted to do so..
19:13:53 <smooge> but it wouldn't be hard in the CentOS build system where this alternative architecture lives already.
19:15:08 <smooge> I was wondering if we could work a one-off build in cbs with either the EPEL name or something similar so that various people can get their projects done and then if we can do it permanently with the other EPEL stuff stand it up later
19:15:09 <nirik> if we can use different urls for the different arches in koji ( as I think dgilmore said at one point ), it should be easy to bring it up just the same way we did ppc64le.
19:16:33 <nirik> well, as easy as ppc64le was... ie, releng time from peter.
19:16:36 <yselkowitz> imo the question here is if epel is based on top of rhel or centos
19:16:42 <bstinson> the arm32 buildsystem is actually separate from CBS in the CentOS space
19:16:49 <yselkowitz> arm32 is a centos-only arch
19:17:00 <bstinson> we could probably work on a throwaway build for bug-filing purposes
19:17:03 <yselkowitz> ppc64le is a rhel arch
19:17:29 <nirik> right, thats why I was saying if we can use different repos for the different arches in koji.
19:17:39 <nirik> ie, use centos for i686/armv7
19:17:56 <nirik> I don't know how that would work, but was told it could at one point.
19:18:14 <smooge> nirik, actually I think he is asking a deeper question. Does EPEL only build on RHEL?
19:18:46 <nirik> right now yes, but it may not need to for arches rhel doesn't exist for
19:19:05 <yselkowitz> has anyone built epel on centos/armv7?  have bugs been filed to fix any issues?  (I'm all in favour of fixing any issues in epel, just not sure if we should be doing the actual builds.)
19:19:05 * nirik waits for dgilmore to actually bring facts. :)
19:20:42 <nirik> I thought there was more interest in i686, but I could be wrong.
19:20:53 <bstinson> yselkowitz: afaik there hasn't yet been a build of EPEL on armv7
19:21:10 <bstinson> nirik: there's quite a bit of interest from the CentOS userbase there as well
19:21:29 * nirik nods.
19:21:32 <smooge> there is probably more interest in i686 but the arm came up as a "some school system has a want to do this in N weeks and could we help"
19:21:43 <nirik> if we can build those arches against centos and the rest against rhel I would be fine with it personally.
19:21:44 <yselkowitz> well it would be helpful if they undertook a build of epel themselves and file bugs, like the other arches have done
19:22:40 <dgilmore> sorry got pulled away
19:22:47 <nirik> yselkowitz: +1
19:23:39 <dgilmore> the main issue we need to solve with altarches is dealling with rhel point releases and the time inbetween rhel releasing a point release and centos catching up
19:23:58 <yselkowitz> ouch good point
19:24:15 <nirik> yeah, if we build some against centos we would need to wait for centos release.
19:24:46 <dgilmore> we can build i686 and armv7hl against centos easy enough
19:24:54 <dgilmore> and the rest against rhel
19:25:07 <smooge> pbrobinson thought it would be best if the altarches were done with a shadow koji so that we could 'turn' off the chain to them while CentOS caught up
19:25:17 <yselkowitz> smooge++
19:25:17 <zodbot> yselkowitz: Karma for smooge changed to 8 (for the f23 release cycle):  https://badges.fedoraproject.org/tags/cookie/any
19:25:33 <smooge> well that should be pbrobinson++
19:25:36 <dgilmore> smooge: it is an option, but requires someone to do a lot of tooling work
19:25:42 <yselkowitz> right, forgot about that bot
19:26:13 <dgilmore> any option to support will mean quite a bit of extra work
19:26:27 <dgilmore> so for me step one is finding resources to do the work
19:26:29 <yselkowitz> pbrobinson is pretty busy as it is
19:26:53 <smooge> yeah it is all going to take some tooling work. what I was looking for was a way that we could work on a "quick" fix while that tooling occured.
19:27:05 <dgilmore> smooge: there is no quick fix
19:27:18 <dgilmore> what is needed is people
19:27:21 <yselkowitz> the "quick" fix is they rebuild it themselves
19:27:29 <smooge> My quick fix was to "franchise" the EPEL name so that the stuff could be built as a 1time thing in the CentOS infrastructure.
19:27:30 <dgilmore> there is 3 or 4 ways we can deal with it
19:28:00 <dgilmore> yselkowitz: thats not a quick fix either
19:28:15 <yselkowitz> quicker than setting up koji to do it!
19:28:25 <dgilmore> yselkowitz: it is not
19:28:33 <dgilmore> i can do it in koji in about 5 minutes
19:28:44 <dgilmore> the cost is after it is in koji
19:29:22 <bstinson> smooge: (with my CentOS hat on) doing a build in our infra would be mostly limited to an exploratory run for filing bugs i think
19:29:31 <smooge> which few people know about.. because they get fixated on the "5 minutes" quote you just said
19:29:33 <nirik> dgilmore: perhaps you could mail the list the 3 or 4 ways and tradeoffs and we can see if people are willing to work on some one?
19:29:58 <smooge> I will send out a blog-post/email on this and dgilmore can answer?
19:30:13 <yselkowitz> bstinson, that would be a very good start
19:30:15 <smooge> or what nirik said
19:30:24 <dgilmore> nirik: at this point I am nearly ready to say we do nothing unless we get people to do the work
19:30:27 <dgilmore> but sure
19:30:57 <smooge> dgilmore, well you know it would be useful to know what the work is
19:31:40 <smooge> because "do the work" means everything from "patch this line" to "boil the ocean" :)
19:32:00 <dgilmore> smooge: the work depends on the way we do it
19:32:25 <nirik> yeah, thats why it would be nice to see the options...
19:32:34 <nirik> some people may be willing to work on some of the paths
19:34:50 <smooge> ok I think we aren't getting very far on this today. I will put out a post with the proposal and we can try and get some short views on the options on the list
19:34:56 <smooge> does that sound ok?
19:35:09 <dgilmore> smooge: with what proposal?
19:36:03 <smooge> well I can start with the general proposal I listed above: Doing alternative architectures with CentOS. Then I can put in the one with the shadow-koji as one alternative with questions on what would need to do this
19:36:16 <smooge> then someone can say "well that might work but we could do it..." and we work from there.
19:36:42 <dgilmore> smooge: I would rather not
19:37:00 <dgilmore> I think we should put out all the options with costs/benefits of each
19:37:16 <smooge> but I only know 1 option.. I want to know what the other options are
19:40:03 <smooge> ok I am not sure we are getting anywhere on this so ok to table til later?
19:41:56 <avij> right, looks like we'll need some more information before discussing further
19:41:56 <bstinson> smooge: i think we can table for now and start work on a bug-filing build
19:42:13 <smooge> ok going to table
19:42:16 <smooge> #open floor
19:42:25 <smooge> #topic open floor
19:42:33 <smooge> anything from people on the openfloor?
19:43:27 <avij> thanks to smooge for writing the reports. I'll need some more time to digest it all.
19:43:43 * nirik needs time to read them, but yeah, very appreciated. :)
19:43:59 <bstinson> +1
19:44:09 <smooge> I still have 2 more to write up. I will get those done by Friday. WHat are are odds we can look at them next week?
19:45:29 <smooge> OK I have one item from tibbs_
19:46:11 <smooge> #info Please test the epel-rpm-macros-5 in EL-5 so that it can be added to the buildroot
19:46:59 <nirik> that reminds me, should we start making noise about epel5 retirement?
19:47:15 <smooge> yes.
19:47:34 <dgilmore> it is a year away?
19:47:42 <nirik> I guess it's about a year yeah
19:47:47 <nirik> March 31, 2017
19:48:58 <smooge> I will add that to the blog list
19:49:08 <tibbs|w> Heh.
19:49:16 <tibbs|w> Just saw the ping.
19:49:49 <tibbs|w> I'd like to get epel-rpm-macros-5 into the buildroot soonish.
19:50:16 <tibbs|w> It's been in stable for a while and I've tried to test it as much as I can but more eyes always help.
19:51:23 <smooge> tibbs you ok with waiting until next Wednesday?
19:51:33 <smooge> or would you like it sooner?
19:51:49 <tibbs|w> I'm not really in any hurry.  I'd just like to be able to check it off of my list.
19:52:32 <bstinson> i try to fit in some testing for next week
19:53:14 <smooge> tibbs_, I will put it on my list for next weeks agenda and then it can be off the list then
19:53:17 <tibbs|w> If you have mock set up, it's one line in a file and some fedpkg mockbuild runs.
19:53:21 <smooge> thank you very much for the work
19:54:56 <smooge> ok anything else?
19:55:09 <tibbs|w> No problem.  It's been a learning experience.  I still have more to do.
19:55:25 <smooge> well that epel-4 macro fix :)
19:55:50 <tibbs|w> DOn't make me hurt myself.
19:56:13 <smooge> ok will close out the meeting before youy do that.
19:56:19 <smooge> thank you everyone for coming and helping out
19:56:24 <smooge> see you next wednesday
19:56:28 <smooge> #endmeeting