18:01:48 <smooge> #startmeeting EPEL (2018-12-05)
18:01:48 <zodbot> Meeting started Wed Dec  5 18:01:48 2018 UTC.
18:01:48 <zodbot> This meeting is logged and archived in a public location.
18:01:48 <zodbot> The chair is smooge. Information about MeetBot at http://wiki.debian.org/MeetBot.
18:01:48 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
18:01:48 <zodbot> The meeting name has been set to 'epel_(2018-12-05)'
18:01:48 <smooge> #meetingname epel
18:01:48 <zodbot> The meeting name has been set to 'epel'
18:01:48 <smooge> #topic Chair and Introductions
18:01:49 <smooge> #chair avij bstinson Evolution nirik smooge
18:01:49 <zodbot> Current chairs: Evolution avij bstinson nirik smooge
18:02:01 <nirik> morning...
18:02:07 <smooge> Gekki ri WPWK
18:02:15 <smooge> Hello to EPEL
18:02:40 <pgreco> hello
18:03:18 <smooge> hello nirik pgreco .. will see if Evolution and bstinson are around
18:03:28 <smooge> I think avij is in a vehicle this meeting
18:03:36 * bstinson is here
18:03:42 <smooge> hello bstinson
18:04:38 <smooge> so this meeting is pretty short. I would like to get ideas on a proposal for having EPEL i386/arm32/risc-v built in CentOS networks
18:05:57 <smooge> also
18:06:03 <smooge> #topic Announcements
18:06:25 <smooge> #info CentOS-7.6 was released to mirrors on Monday.
18:06:47 <smooge> #info DTS-8 is in the koji builders so can be used in packages
18:07:07 <smooge> #info smooge still hasn't gotten a document written for EPEL-8
18:07:20 <smooge> any other announcements?
18:07:37 <pgreco> none here
18:09:23 <nirik> oh, one...
18:09:50 <nirik> #info smooge and nirik will be giving a epel8/future of epel type talk at devconf
18:10:19 <smooge> which requires me to write the document that we give the talk on
18:10:36 <smooge> so no excuses.. get writing after I get offcall tomorrow
18:10:42 <smooge> ok offcall on the 8th
18:10:46 <nirik> yeah. ;) but I can help or write parts for sure
18:11:13 <smooge> well I expect that my first draft will be flowery vomit and then you will clean it up
18:11:26 <pgreco> that sounds reasonable
18:12:03 <smooge> ok so I have a sort of idea/proposal
18:12:15 <pgreco> great, I'm all ears/eyes
18:12:17 <smooge> #topic Secondary Architectures Building in CentOS Land
18:13:05 <smooge> so I don't think there is going to be a way for us to build EPEL for i386/arm32/etc in our koji without breaking things
18:13:51 <smooge> we either have to make the branches for things something like epel7-centos and then have koji only build against CentOS for that (but it also means maintainers have to know about it and put packages there)
18:14:07 <smooge> OR we move all of the build system to just CentOS
18:14:38 <pgreco> none of which is easy
18:15:07 <smooge> OR some other trisection of the angle/squaring the circle I haven't found
18:15:24 <nirik> keeping things in sync might be difficult... but sure.
18:15:54 <smooge> so what I think could be done is that we work out how to build those in the CentOS infrastructure and sign the gpg key which is used to sign those packages
18:16:12 <tdawson> Do the CentOS builders have enough access to the Fedora infrastructure to be able to keep things in sync?
18:16:36 <pgreco> tdawson, yes, or at least they will
18:16:41 <pgreco> after the git unification
18:16:46 <smooge> so with the combined git between fedora and centos they should
18:17:22 <tdawson> That was my only concern ... and it sounds like it's being taken care of
18:17:51 <nirik> true... on that end...
18:18:06 <bstinson> it won't be right away either, because we won't have access to the fedora lookaside
18:18:08 <nirik> but there still can be lag or whatever between builds, etc
18:18:24 <smooge> I expect there will be but these aren't major ones
18:18:42 <nirik> ie, epel builds something it's synced but then until someone fires off a build for the other arches it's not there.
18:18:46 <smooge> nirik, also the next part .. we don't compose EPEL for i386/arm
18:18:48 <smooge> they do
18:19:09 <nirik> sure, just saying that sometimes they could be out of sync by a little
18:19:43 <smooge> well we are looking at a N+ week out of sync every minor release so they will be probably used to it
18:20:13 <pgreco> and another thing to take care of is epel-release
18:20:44 <pgreco> that should point somewhere else, unless we move everything back to fedora servers
18:21:05 <nirik> right, should build differently for those arches.
18:21:11 <nirik> should be doable via conditionals.
18:21:37 <pgreco> epel-release-altarch, obsoleting epel-release
18:21:40 <pgreco> or something like that
18:22:20 <bstinson> or just contribute patches back to epel-release in fedora, and be sure we pull it in for release to CentOS Extras
18:22:23 <smooge> Anyway.. that is my starting point for an idea. I know it needs to be formally proposed and documented
18:22:23 <nirik> I think it would be better to just keep epel-release, it just has different repo files and gpg keys for those other arches.
18:22:42 <smooge> but it would move this forward
18:23:27 <pgreco> well, epel-release is noarch, so it uses the same repo for all arches
18:23:59 <nirik> we could make it archfull...
18:23:59 <smooge> pgreco, bstinson nirik does that for a 10k foot plan look possible?
18:24:26 <pgreco> bstinson, wrt patches, I'm trying to push to everything back so src.f.o, some of those are there already
18:24:47 <pgreco> so the idea is that we should never need separate branches
18:25:01 <nirik> smooge: I think so yeah...
18:25:14 <pgreco> unless a maintainer really doesn't want what we patch
18:25:38 <bstinson> smooge: at 10k foot, yes. it's going to break horribly on the runway though
18:26:03 <smooge> I expect so.
18:26:10 <pgreco> it looks possible on my end
18:26:16 <smooge> the goal here is to finish building the plane before it hits
18:26:16 <bstinson> pgreco: we have a bit of influence on the maintainers :) /me looks at nirik
18:26:37 <pgreco> from doing this manually, to doing it this way, it is a lot better :)
18:26:39 <bstinson> pgreco: there are limitations in CBS that we'll need to work on. and that work may not be fast
18:26:41 <pgreco> bstinson ;)
18:27:17 <pgreco> I don't mind waiting, all I'm doing now is executing a script manually every morning
18:27:17 <nirik> we could also look at doing this with 8 from the start... might be easier than shoehorning into 7... but perhaps not
18:27:18 <smooge> my hope at that point is that we can focus on helping fix CBS or ReimzulBuildSystem2 or whaterver
18:27:47 <smooge> nirik, well the reason I want to do this is that 7 is going to be around for another 6 years
18:27:56 <pgreco> nirik, that was my idea, ironing out everything during 7, so 8 is easier
18:28:16 <smooge> just because 8 comes out, we don't see people moving to it for probably 3-4 years
18:28:49 * nirik nods
18:29:02 <smooge> so I figure we are probably needing to push things back to 7 at times because well it all sucks no matter what
18:29:11 <smooge> 6 I will say no
18:29:21 <smooge> Oh speaking of which
18:29:31 <smooge> fedbot: @eol6rhel
18:29:40 <bstinson> (EPEL hat on) I think this is definitely worth putting some work into, especially for the armhfp userbase
18:29:49 <bstinson> (CentOS Infra hat on) help please
18:30:50 <smooge> fedbot: @rhel6eol
18:30:52 <pgreco> bstinson, anything I can do to help
18:32:02 <smooge> @rhel6eol
18:32:02 <fedbot> RHEL6 goes end of life on 2020-11-30, which is 1 year, 51 weeks, 3 days, 13 hours, 27 minutes, and 58 seconds from now
18:32:06 <smooge> thank you
18:32:20 <smooge> #info RHEL6 goes end of life on 2020-11-30, which is 1 year, 51 weeks, 3 days, 13 hours, 27 minutes, and 58 seconds from now
18:32:23 <bstinson> pgreco: it might help for us to figure out the compose process on the CentOS side. we have a bunch of stuff left to do for the SIGs and if we can get the same benefit out of that work, it might help
18:32:34 <bstinson> pgreco: other things will just require patience (like the lookaside problem)
18:33:00 <nirik> short term you could just pull from the fedora lookaside (but might be slowish)
18:33:19 <pgreco> bstinson, I still need to finish with hughesjr the i386 rebuild
18:33:23 <bstinson> nirik: builders don't have internet access
18:33:30 <pgreco> to take it to the same level as armhfp
18:33:42 <smooge> so I have an idea
18:33:45 <pgreco> so by the time everything is ready, all we need is a rebuild
18:33:52 <nirik> well, they do to your lookaside right? they could to ours too... but sure
18:34:07 <bstinson> they don't have routes at all
18:34:21 <smooge> how do they get the lookaside in cbs?
18:34:25 <bstinson> the lookasides are on the same network
18:34:34 <nirik> huh, ok
18:34:55 <bstinson> well, as of now they don't because we haven't enabled SCM builds in CBS :P
18:35:08 <bstinson> but the theory was to have the lookasides on the same network :)
18:35:14 <bstinson> it's the same in fedora, no?
18:35:34 <nirik> it is, but there's an explicit firewall rule to allow it
18:35:42 <nirik> (and git access to check things out)
18:36:35 <smooge> the lookaside is on the netapp right?
18:36:53 <smooge> so we just need to get them a netapp and snapmirror it to them
18:36:57 <smooge> easy peasy
18:37:19 * smooge solves all the world's problems like a pro
18:37:56 <smooge> ok I have derailed this far enough
18:38:00 <smooge> #topic Open Floor
18:39:07 <tdawson> I was wondering about your epel8 talk at devcon ... well actually had just a question or two.
18:40:15 <tdawson> Were we going to be able to get some type of epel8-beta going?
18:40:16 <nirik> fire away
18:40:23 <nirik> hopefully
18:40:50 <tdawson> Cool ... then the next question was ... what's a rough timeline for that?
18:41:18 <tdawson> (rough as in months) ... one month, two months ... 5 months?
18:41:24 <nirik> not sure. I can work on it, but I am pretty swamped right now... so it might be over the holidays if things are quiet...
18:42:00 <tdawson> So, possibly two months it sounds like
18:42:04 <nirik> also I am not sure how useful things will be without ursa major...
18:42:24 <nirik> but I have been told it's not needed, but not sure how it will work in practice
18:42:46 <tdawson> I heard that someone was going to try to package that for Fedora ... that might make it easier to put into epel8
18:43:28 <nirik> well, its not the packaging, it's the setup and fedora is still deciding if it's going to be allowed and under what conditions....
18:44:45 <tdawson> True ... it's more than packaging ... we'll see how the other stuff goes with getting it in Fedora.
18:45:36 <nirik> yeah.
18:45:58 <smooge> tdawson, so the first thing that needs to be done is "how do we want to make EPEL-8 work?" with modularity, restricted package sets, etc etc
18:46:23 <smooge> that was needing some proposals and it also needs some items where we may need ursa-major or not
18:46:50 <smooge> I expect that we aren't going to see an EPEL-8-beta until March
18:47:31 <smooge> we might see it at devconf but it would be a 'here is what we slapped together but don't touch it'
18:47:45 <tdawson> That makes sense ... cuz even if we have infrastructure, all the other decisions need to be debated/agreed upon.
18:47:50 <tdawson> :)
18:48:54 <smooge> ok anything else for this meeting? Our next meeting will be next week and I will have a written proposal to critique OR I will eat this hat
18:49:12 <nirik> I think we can have something sooner, but I'm an optimist. ;)
18:50:27 <smooge> yeah.. I should be more glass is half full here
18:50:53 <smooge> ok with that. thank you all and see you next week
18:50:58 <pgreco> thanks!
18:51:00 <smooge> #endmeeting