18:01:17 <smooge> #startmeeting EPEL
18:01:17 <zodbot> Meeting started Wed Aug 17 18:01:17 2016 UTC.  The chair is smooge. Information about MeetBot at http://wiki.debian.org/MeetBot.
18:01:17 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
18:01:17 <zodbot> The meeting name has been set to 'epel'
18:01:22 <smooge> #meetingname EPEL
18:01:22 <zodbot> The meeting name has been set to 'epel'
18:02:27 <smooge> #chair smooge nirik Evolution avij bstinson
18:02:27 <zodbot> Current chairs: Evolution avij bstinson nirik smooge
18:03:10 <smooge> I am not sure we have any form of quorum today and Fedocal has decided meeting is later
18:03:15 <bstinson> \o/
18:03:18 <avij> greetings
18:03:44 <smooge> And I honestly thought today was Tuesday and was going to send out the mail for the meeting this evening
18:04:25 <smooge> #topic EPEL outstanding issues
18:04:55 <smooge> OK we have quorum. And I see builds for i386 and aarch64 in the CentOS build system
18:05:15 <smooge> the aarch64 ones look to be a long list of "not on this arch" type things :/
18:05:34 <smooge> the i386 ones are getting to that point
18:06:16 <avij> still building all those
18:06:39 <smooge> the nodejs and some others
18:07:11 <smooge> Anything else on the list?
18:08:31 <smooge> bstinson, Evolution any other news in CentOS land?
18:09:03 <smooge> and I see we are competing with a CentOS meeting right now.
18:09:08 <Evolution> mostly waiting for the go-ahead on aarch64.
18:09:36 <Evolution> once epel has that in place we can stop our epel builds and use actual-epel.
18:09:44 <bstinson> I'm hoping to work on some sort of proposal for how to collaborate on non-RHEL arches
18:09:48 <smooge> ok will see with pbrobinson what is needed
18:09:57 <bstinson> it's on one of my lists from FLOCK
18:10:01 <Evolution> afaik it's a firmware patch from HP
18:10:07 <smooge> yeah..
18:10:13 <smooge> any day now
18:10:17 <smooge> ...
18:10:20 <Evolution> heh
18:10:37 <smooge> what aarch64 ? we have this lovely itanium which emulates it. you want that?
18:11:22 <smooge> ok I still have a backlog of packages to make also for EL7
18:11:48 <smooge> #EPEL New Business
18:11:56 <smooge> Does anyone have any open/new business?
18:12:23 <Evolution> smooge: at one point you and bstinson were talking about amending/rewriting the epel charter
18:12:24 <smooge> I saw that mock builds are 'broke' for people due to EPEL configs pointing to centos keys
18:12:49 <Evolution> hmm? bz/url?
18:13:10 <smooge> https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/thread/IBOEOVPWNEDXEJI3ZIF26LAAVAFS253G/
18:13:22 <smooge> Evolution, the EPEL charter stuff needs to be dealt with.
18:14:09 <smooge> I am declaring round tuit bankruptcy and hoping that a refinance will give me time to start on it this weekend.
18:14:30 <smooge> Evolution, the link was for the mock borkness. it should be fixed soon
18:14:33 <Evolution> ah, that didn't go to epel-devel. that's why I missed it
18:14:35 <Evolution> got it.
18:14:41 <Evolution> yeah.
18:16:58 <smooge> yeah I got it because someone found it some other way this morning
18:17:36 <smooge> bstinson, Evolution while i have a blank paper for Charter here.. is there anything you think should be in the first draft?
18:18:08 <Evolution> smooge: well, I did notice spot's request for devtools scl bits.
18:18:30 <smooge> how are those available in CentOS?
18:18:41 <smooge> because the mock would need to have them also
18:18:42 <Evolution> as a sig. additional repo for someone to enable.
18:18:55 <Evolution> so reasonably similar to rhel in that regard.
18:19:39 <Evolution> and at the risk of sysadmin related heresy.. but with an eye to the future, we might consider containers in some form
18:19:49 <pbrobinson> smooge: bstinson: Evolution: I think we're looking OK for aarch64 once we have moonshot in prod
18:20:14 <pbrobinson> smooge: bstinson: Evolution: from the non RHEL arches
18:20:40 <smooge> pbrobinson, cool
18:20:44 <Evolution> pbrobinson: I'm assuming that's the generic "two weeks" placeholder timeframe until further notice?
18:20:48 <pbrobinson> I think it would be useful to take it to the list, I have some ideas but I suspect I'm treading on a meeting :)
18:21:07 <pbrobinson> but if there's time I can do some brief brain dump here
18:21:23 <smooge> brain dump on EPEL or Architectures :)?
18:21:26 <pbrobinson> Evolution: :-D
18:21:46 <pbrobinson> on support for non RHEL in EPEL (ie i686/armv7hl
18:21:52 <pbrobinson> )
18:22:01 <smooge> pbrobinson, ah ok
18:22:11 <smooge> pbrobinson, we have 20 minutes so dump
18:22:24 <Evolution> I'm on smooge's schedule. lets do it
18:22:48 <pbrobinson> so the major issue is we get all RHEL arches at the same time. RHEL-7 is moving quick
18:23:16 <pbrobinson> the ppc644le bootstrap was terrible because I had to fix a crap ton of issues where RHEL bumped stuff
18:23:39 <pbrobinson> so I had to do huge amount of fixes to deal with that even to just build on x86_64
18:24:01 <pbrobinson> so chatting with bstinson at flock about the problem
18:24:26 <pbrobinson> the major issue is that do we hold and have broken primary support arches until the others catch up
18:24:59 <pbrobinson> or do we push and have an issue where koji pulls in external deps and have some arches build with one soname and others with the old one
18:25:31 <pbrobinson> so bstinson mentioned maybe we should basically branch with each el7 release and do 7.x branches
18:26:12 <pbrobinson> which kind of makes sense as we'll have a bunch of el7 x86_64 customers that want to stick on a .x long term branch too
18:26:30 <Evolution> can you convince a maintainer to build to .x branch?
18:26:31 <pbrobinson> sadly I think it's a process that's going to add a bunch of work
18:26:44 <pbrobinson> well maybe you don't need to
18:26:57 <pbrobinson> if they don't it sticks on the last known good
18:27:09 <pbrobinson> that doesn't deal with CVEs etc
18:27:20 <pbrobinson> but it also doesn't actively break shit
18:27:39 <pbrobinson> those that want it can do so, those that don't have branches stuck at a point in time
18:27:59 <pbrobinson> it also doesn't 100% deal with the non RHEL arches
18:28:27 <pbrobinson> but nor does the current situation
18:28:34 <Evolution> there's a fair bit of matrixing and engineering for that.
18:28:40 <pbrobinson> yep
18:28:57 <bstinson> the only thing that gets us is a well-known point for 'breaking' things
18:29:16 <Evolution> that's a valuable thing. I'm not discounting the idea
18:29:19 <pbrobinson> but I think we're going to need to do something like that because of the way el7 is changing compared to old releases
18:29:22 <bstinson> which would be handy in the case of arch work
18:29:29 <Evolution> I'm trying to sort out mentally if it's worth the pain it incurs
18:29:46 <pbrobinson> I'm not convinced it's the solution
18:30:19 <smooge> does it feel like it gets us closer to a solution (even if we don't know what it is)?
18:30:31 <pbrobinson> but after the pain of 7.2 and ppc64le bootstrap I know it's something we're going to need to address at some point
18:30:31 <bstinson> the other option is to have a 'breakage window' around RHEL releases, and start doing mass rebuilds
18:30:49 <bstinson> but that again doesn't solve the non-RHEL arch problem directly
18:30:51 <smooge> well that was my take on it
18:31:24 <pbrobinson> basically I know I'm doing aarch64 before 7.3 lands because I know epel isn't too broken given the work I did for ppc64le so it will be less painful for me
18:31:43 <pbrobinson> I estimated 2-3 days work, it ended up being like 10 days of 12 hours
18:32:27 <smooge> pbrobinson, looking at the builds the main aarch64 stuff looks currently like nodejs and similar all the world is x86_64 isn't it?
18:32:45 <smooge> s/stuff/bad stuff/
18:33:18 <Evolution> some of that is just needlessly exclusive-arch'd
18:33:42 <smooge> so I am trying to see how this branch issue is different from 'breakage window' around RHEL releases
18:34:24 <smooge> since the amount of things changing in 7.x looks to be bigish until 7.<2nd stage of support starts>
18:34:26 <pbrobinson> smooge: yes, but I know sgallagh has plans for nodejs 4 for EPEL and that will add both ppc64le and aarch64 and people will be happy
18:34:43 <smooge> ah ok
18:34:50 <smooge> well as long as people are happy
18:35:04 <bstinson> the 'breakage window' gives room for packagers to update their packages, and for us to schedule a mass rebuild. which feeds into the arch discussion because we don't have CentOS versions for non-RHEL arches for a certain period of time
18:35:08 <pbrobinson> and there's an issue with haskell but the maintainer also has a plan to fix that (time @ flock was awesome!)
18:35:13 <sgallagh> pbrobinson: We're actually considering Node.js 6 for EPEL, but we're going to hold off a couple months until it enters LTE phase.
18:35:42 <pbrobinson> sgallagh: awesome! so basically I meant nodejs 4+ ;-)
18:35:45 <sgallagh> But the rest of that (arch support) is accurate
18:36:19 <smooge> There is a 3rd option. We just have regular breakage windows in EPEL.
18:36:45 <smooge> not linked to RHEL releases but just its been 9 months and the CHaos monkey has awoken
18:37:10 <pbrobinson> smooge: big problem there is enterprise users
18:37:29 <pbrobinson> the issues with non EL arches is hard to articulate on irc
18:37:34 <smooge> pbrobinson, bstinson if you guys could write up the hypothesis plan to the mailin glist
18:37:42 <pbrobinson> happy to have a voice chat with those interested
18:38:00 <Evolution> mailing list discussion is probably better
18:38:03 * pbrobinson throws that to bstinson as I did at flock :-P
18:38:20 <smooge> pbrobinson, when the enterprise customers start paying for both stability and we want the latest (since they are also the ones asking that)... I will be happy to try and keep them happy :)
18:38:22 <bstinson> yeah, i promised a write-up :P
18:38:27 <pbrobinson> which is what I asked bstinson to do :)
18:38:48 <smooge> bstinson, I will find wet fish and mail them to you to slap yourself to remind you
18:38:52 <pbrobinson> smooge: what is the point of doing it if not for the enterprise customers?
18:38:54 <Evolution> he owes so many writeups....
18:39:01 * Evolution taps foot in bstinson's general direction
18:39:05 <pbrobinson> smooge++
18:39:05 <zodbot> pbrobinson: Karma for smooge changed to 8 (for the f24 release cycle):  https://badges.fedoraproject.org/tags/cookie/any
18:39:26 <smooge> pbrobinson, the point is that the Enterprise customers want both never changing and always changing
18:39:36 <smooge> from a volunteer group.
18:40:04 <smooge> and they rarely show up except when they didn't get one of the two things above.
18:40:22 <pbrobinson> of course! all the stability none of the contribution
18:40:53 <smooge> so if we tell them that every 12 months we are looking to change things (they may not change because alpine is just as good as it was before.. but they may change a lot because nethack is now written in JS)
18:41:14 <smooge> it gives them a window to say "no we really want the C version of nethack."
18:41:49 <bstinson> a window that's tied to a schedule that they already know (point release time)
18:42:06 <smooge> well the problem is that no one knows what the point release time is
18:42:09 <smooge> no one
18:42:30 <pbrobinson> TBH not even internally via the PM :-D
18:42:44 * pbrobinson will burn in hell at some point!!
18:43:11 <smooge> so we can tie it to the point release or to a set date.
18:43:26 <smooge> each has its bonus and each has its problems.
18:43:50 <smooge> I will write up the breakage and set date ones as proposals.
18:44:03 <smooge> [although the breakage one might be already done.]
18:44:21 <smooge> ok meeting time is coming to a close.
18:44:39 <smooge> #topic open floor
18:44:44 <smooge> does anyone have any other items?
18:44:57 <smooge> I can close the meeting otherwise and give you all back 6 minutes
18:45:07 <bstinson> none here
18:45:34 <avij> none here either
18:45:36 <Evolution> none here
18:46:46 <smooge> I love exits like:
18:46:49 <smooge> var/tmp/rpm-tmp.IE05K8: line 30: fg: no job control
18:46:49 <smooge> error: Bad exit status from /var/tmp/rpm-tmp.IE05K8 (%build)
18:46:49 <smooge> Bad exit status from /var/tmp/rpm-tmp.IE05K8 (%build)
18:46:49 <smooge> Child return code was: 1
18:46:50 <smooge> EXCEPTION: Command failed. See logs for output.
18:47:05 <smooge> the logs of course say nothing
18:47:15 <smooge> ok with that enlightening tidbit
18:47:19 <smooge> good night everyone
18:47:23 <smooge> #endmeeting