20:00:30 #startmeeting EPEL (2021-05-19) 20:00:30 Meeting started Wed May 19 20:00:30 2021 UTC. 20:00:30 This meeting is logged and archived in a public location. 20:00:30 The chair is tdawson. Information about MeetBot at http://wiki.debian.org/MeetBot. 20:00:30 Useful Commands: #action #agreed #halp #info #idea #link #topic. 20:00:30 The meeting name has been set to 'epel_(2021-05-19)' 20:00:32 #meetingname epel 20:00:32 The meeting name has been set to 'epel' 20:00:33 #chair nirik tdawson bstinson pgreco carlwgeorge michel_slm 20:00:33 Current chairs: bstinson carlwgeorge michel_slm nirik pgreco tdawson 20:00:35 #topic aloha 20:00:47 howdy 20:00:51 Hi carlwgeorge 20:00:57 hey, everybody, goodbye old friend.... 20:01:01 morning 20:01:04 Hi pgreco 20:01:12 Hi nirik 20:02:05 .hi 20:02:06 dcavalca: dcavalca 'Davide Cavalca' 20:02:25 Hi dcavalca 20:03:31 .hello salimma 20:03:32 michel_slm: salimma 'Michel Alexandre Salim' 20:03:37 Hi michel_slm 20:04:11 We've got almost everyone already ... this hasn't happened for a while. 20:05:24 I'm going to break with the usual agenda to start with what is probrubly on everyones minds anyway. 20:05:32 #topic IRC Stuff 20:05:50 :( 20:06:05 Hopefully we can be quick and get back to the usual stuff. 20:06:44 Has everyone heard of the Freenode / Libra.Chat proposal for Fedora's IRC ? 20:07:02 yup 20:07:14 #info https://pagure.io/Fedora-Council/tickets/issue/371 and https://pagure.io/Fedora-Council/tickets/issue/372 20:07:16 I have a new LiberaChat account, but I can report the servers are slammed enough even getting a channel listing is ratelimited 20:07:22 easier question is probably who hasn't heard 20:07:25 fyim CentOS has already announced it's moving to Libera too 20:08:24 Yes, the Libera servers are currently overrun as it seems 20:08:35 My proposal is to follow Fedora to whichever IRC they go to. Which currently looks like Libra.Chat 20:08:56 * nirik nods 20:09:05 i mean, epel is part of fedora, so i assumed that was the forgone conclusion 20:09:38 carlwgeorge: That's what I feel too, but I wanted to say we'd talked about it and voted, so it's all official and so forth. 20:10:11 works for me 20:10:27 yeah 20:10:32 someone described as a "really bad name change" 20:10:40 *laughs* 20:10:49 but eventually the hope is that it will be all that it is 20:11:04 I sorta like the new name actually, but changing names is always a pain. 20:11:39 same people, and likely same servers 20:11:46 but with a few days of mess 20:11:51 Ready for a vote. All in favor of following Fedora to whichever IRC they choose. 20:12:16 +1 20:12:17 +1 20:12:18 +1 20:12:27 +1 20:12:43 +1 20:12:57 +1 20:13:08 ps tdawson, the new name is Libera, not Libra, so don't get attached ;) 20:13:49 #info Proposal passed. We (EPEL) will follow Fedora to whichever IRC they choose: 6 For, 0 Against 20:13:55 my phone autocorrected libera to liberal earlier, so this will be fun 20:14:07 Ugg ... libera ... who thew that extra e in there. :( 20:14:11 oof. in Texas of all places 20:14:21 haha 20:14:32 Oh well, I cut and paste everything anyway. 20:14:46 FB created that crypto consortium previously called Libra ... I lost track of their name changes 20:14:52 Diem? 20:15:02 nirik: You said we already have the epel channel on libera.chat, correct? 20:15:21 yep. 20:15:31 Sounds good. 20:15:56 just joined it 20:16:04 As soon as the Fedora council passes their issue, we can send out an email saying we're moving IRC over. 20:16:38 fingers crossed that also involves re-bridging Matrix 20:17:07 yes, it will 20:17:35 Anything else before we move to the next subject? 20:18:03 #topic Old Business 20:18:14 EPEL Packaging SIG 20:18:37 How did the bodhi update go? 20:18:41 it's all done. 20:18:43 please test. 20:18:43 it works! 20:18:49 sweet 20:18:52 I was able to submit an update as a collaborator the other day 20:18:52 excellent. 20:19:00 cool 20:19:12 Now for the hard part ... documentation. :) 20:19:13 * nirik wonders if we need a old solviet poster style badge for collaborator 20:19:15 so now we can work on a template to request branches, that will contain a standard boilerplate text 20:19:40 and a process for escalating when it gets ignored 20:20:08 yeah. who do we need to talk to for the escalation process? fesco? or council? 20:20:34 or even FPC? (but that's more about packaging standards, right) 20:20:54 * bcotton_ pokes head in 20:20:59 escalating what, exactly? 20:21:28 bcotton_: if we request an epel branch and the primary maintainer(s) ignore it, right now there's no recourse short of triggering the non-responsive maintainer process 20:21:35 It would be through whoever we asked when we were working through the non-responsive maintainer policy. 20:21:56 ideally we have a less ... escalated option 20:22:17 tdawson: oh, I mean, if we create a new process, who can approve it 20:22:17 I still feel going the non-responsive route is a bit too forceful for this scenario 20:22:21 yeah 20:22:42 michel_slm: ah, yeah, i think FESCo would be the group to approve the process 20:23:13 and i'll be doing this soon for a bunch of epel8-next branches :D 20:23:28 i.e. who can pass a proposal that says "for EPEL branch requests by epel-packagers-sig, this escalation path can be used". ok, I can work on filing a ticket with FESCo 20:23:54 there is an option 20:23:58 * nirik looks 20:24:05 I think we discussed the details before. are we OK with still waiting for two weeks (tdawson mentioned people might be on vacation) but then we want the automated process to kick in? 20:24:08 nirik: ooh 20:24:44 https://fedoraproject.org/wiki/EPEL/GuidelinesAndPolicies#Stalled_EPEL_Requests 20:24:58 unless thats missing some cases now? 20:25:13 Ah, that's what I was looking for. 20:26:28 oh nice 20:26:38 So it's filing a rel-eng ticket after two weeks with two requests. 20:26:57 I think the only difference is we'd want to request the epel-sig user, instead of ourselves. 20:27:11 so we still need to have a nice template to make it more likely people accept this :) 20:27:19 the epel-packagers-sig group, yeah 20:27:32 I think that's fine, the policy says "appropriate privileges" 20:27:44 That's a good thing, yep. That way it's easier for rel-eng to know what it's about. 20:28:14 I'll have something ready for next time. er, speaking of docs, is anyone looking at asciidoc? 20:28:32 Ohh ... that fell off my radar 20:28:41 same 20:28:53 I'll talk to Petr this week, see if he now has time to do the mass conversion. 20:29:16 mass converting our wiki pages to asciidoc? or something else 20:29:34 asciidoc, and put it in the main Fedora docs page. 20:30:36 He was busy with F34, RHEL 8.4, and CentOS Stream 9 pages ... but I believe all those are pretty much done, so hopefully he has some time now. 20:31:02 nice 20:31:30 nice. it'll still be based on the epel pagure repo, right? 20:31:49 That's were it will all go, yes, in the doc (docs?) directory. 20:31:57 in which case I'll put up a PR there with the proposed text. and once approved, figure out how to make a Bugzilla template 20:32:13 Sounds good. 20:32:46 Anything else before we move to the -next topic? 20:33:18 not from me 20:33:28 Ha ... that one sounded good in my head ... and even looks good when typed. :) 20:33:49 epel-next 20:33:54 carlwgeorge: How are things going? 20:34:34 i did a live test of the fedpkg changes and requested an epel8-next branch for a qt thing https://pagure.io/releng/fedora-scm-requests/issue/34092 20:34:55 it was rejected because i'm not on that package, but the request was created correctly 20:35:04 cool 20:35:41 me, mboddu, nirik, and smooge are meeting tomorrow to look at bodhi stuff and see what else is left 20:35:55 * nirik nods. 20:36:19 and i think we're still waiting for the bodhi backend syncs script pr to get merged 20:36:31 carlwgeorge: ^ hopefully, I might take the day off tomorrow :(, but I will try to attend the meeting 20:37:11 What about the repo for the missing devel packages ... do you think the proposal on the centos-devel mailling list will work? (grabbing the buildroot repo) 20:39:00 before thats merged, we need empty repos created for it to sync 20:39:17 tdawson: I think the problem there is modules... 20:39:26 originally we said that the reason for not allowing artifact download out of the centos koji was that we didn't want to put it under that load. 20:39:38 That's a valid problem. 20:39:49 however, brian says he changed things up and now it's fine to mirror the buildroots for cs8 and cs9. 20:40:04 something about how that path was served, i don't recall the exact details. 20:40:20 I saw something about load balancers. 20:41:50 This is something I'll need to look at, but when koji is doing a build, using the buildroot, what does it do about modules? Are they already in the buildroot? 20:42:31 some are, not all 20:42:34 I think I used to know that, and then forgot, cuz it seems like something I would know. 20:42:54 https://koji.mbox.centos.org/koji/taginfo?tagID=666 20:43:05 those get inherited 20:43:42 are you using ursa-major there for that? 20:43:58 nope, just manually managed tag 20:44:10 ah, ok 20:44:38 i like the approach of using centos koji buildroot for epel-next builds. only problem is we don't have an equivalent for epel builds. 20:45:05 True, and it's possible that there might be packages that build in -next, but not in regular EPEL 20:45:07 equivalent? 20:45:25 rhel buildroot 20:45:25 equivalent==missing devel packages 20:45:52 we could set it up with the cl8 buildroot but that only gets us through the end of the year 20:45:55 things present in the centos koji buildroot will not be present in the real epel buildroot 20:46:54 maybe I'm not following right, but couldn't we populate the Devel repo with the missing stuff from the buildroot, and then point epel-next builds to that? 20:47:20 (or make another repo if Devel isn't appropriate) 20:47:22 that's been suggested several times, and the rhel bu won't allow it 20:47:47 * carlwgeorge is getting real tired of not giving people the credit they deserve 20:48:35 We could create our own EPEL Devel repo, doing what dcavalca said. 20:48:46 I was just about to suggest that 20:48:57 for -next, sure, but that won't translate to real epel 20:49:01 like, this doesn't necessarily need to be a CentOS / RHEL thing 20:49:10 Instead of syncing over the whole repo, just those that are in Devel. 20:49:26 pgreco: But there already is a Devel repo for regular EPEL. 20:49:54 for -next you may need to tweak some things for modules but it should work 20:50:03 * nirik is tired, but also loosing track of the goal(s) here. 20:50:31 i think the goal is for epel packages to not be held hostage by what rhel won't ship or allow centos to ship 20:50:49 for next we have to build against stream, thats the point... so building against the cs8 buildroot seems ok as long as we don't hit modules that we flatten in the epel8 buildroot and people use 20:51:05 and if we end up with a way to make these packages also usable by regular installs, I'd call that a bonus :) 20:51:48 and i'm fine with solving the problem that way, but we need to be clear that it only solves half the problem 20:52:18 e.g. you have a potential epel package that buildrequires libuser-devel. you'd be able to build it in epel8-next, but not epel8. 20:52:33 we need a spreadsheet of repo/missing/modules/arches/whatever 20:53:09 I'm for trying the buildroot way. 20:53:24 nirik: https://pagure.io/centos-infra/issue/316 has the ones I've specifically hit so far, but there's definitely more 20:53:37 nirik: smooge made one back when RHEL8 released, but I think things have changed. 20:54:02 fun. 20:54:17 it's such a mess 20:54:43 Anyway ... for now, we go for the buildroot way, correct? Deal with the epel vs epel-next problems when they start coming up? 20:54:45 epel is allowed to mirror rhel...could we ask for permission to mirror the rhel buildroot also for epel builds? 20:56:32 Just looked at the time, anything else that needs to get discussed before we go? 20:56:37 I'm game to try it, but I suspect it's going to result in things failing due to missing modules that we currently flatten 20:57:06 has the epel8 buildroot been synced up with rhel 8.4 yet? 20:57:22 it syncs once a day, so I think it will sync tonight. 20:57:58 nice 20:58:56 nirik: Would you, or smooge, be able to do the snapshot to the archives? 20:59:18 I can try. I'm pretty swamped. :) Can you file a releng ticket on it? 20:59:31 Yep, Yep I can, and I will. 20:59:59 What time is the meeting? 21:00:27 Sicnus: what meeting? the epel meeting is right now, you are in it! :) 21:00:36 Looks like our time is up. Thanks everyone for coming this week and for the good discussion. 21:00:46 thanks tdawson 21:00:51 See you next week, quite possibly, not here, but still at the same time. 21:00:55 later yall 21:00:57 #endmeeting