18:00:54 #startmeeting EPEL (2019-05-01) 18:00:54 Meeting started Wed May 8 18:00:54 2019 UTC. 18:00:54 This meeting is logged and archived in a public location. 18:00:54 The chair is smooge. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:00:54 Useful Commands: #action #agreed #halp #info #idea #link #topic. 18:00:54 The meeting name has been set to 'epel_(2019-05-01)' 18:00:54 #meetingname epel 18:00:54 The meeting name has been set to 'epel' 18:00:54 #topic Chair and Introductions 18:00:54 #chair bstinson Evolution nirik smooge pgreco tdawson 18:00:54 Current chairs: Evolution bstinson nirik pgreco smooge tdawson 18:01:03 morning 18:01:07 oh wednesday already 18:01:10 hello 18:01:16 Due to RH Summit I think we may have bare quorum 18:01:24 bstinson: time flies when you are having fun? :) 18:01:25 yes, quiet week, not much to talk about :D 18:01:36 * tdawson is here 18:01:58 s/having fun/didn't sleep on monday night/ 18:02:01 #topic Agenda 18:02:01 #info tdawson cleaned up -testing 18:02:01 #info RHEL-8 Released 18:02:11 Any other items for the agenda? 18:02:23 oh probably 18:02:33 #info Should we do an EPEL-8 18:03:07 * smooge is a bit punchy after 4 days of 4-6 hours sleep and getting up at 0300 to 0400 18:03:17 naw, it's over man... KIDDING! 18:03:27 #topic tdawson cleaned up -testing 18:03:49 Thank you very much tdawson for cleaning out the testing agenda 18:03:59 s/agenda/backlog/ 18:03:59 Very welcome 18:04:23 did you get much reply from that? 18:04:23 I think it did point out one thing.. where some people are using -testing as a rawhide 18:04:26 any comments, etc? 18:04:36 True 18:04:53 I don't think the number of people doing that was very high, but there was some doing that. 18:05:10 can we count those who complain about it as karma? 18:05:16 and release those? 18:05:18 when we have minor releases we could drop all updates from the previous minor at that time if we wanted... 18:05:33 pgreco: They didn't want them released 18:05:45 nirik, my plan was to put the minor releases in archives and start fresh 18:05:52 ok 18:06:12 smooge: right, but we would need to decide if we do anything with in flight updates... 18:06:21 ah yeah.. 18:06:41 * smooge gets that 40 yard stare 18:06:43 it would be anoying to push a security fix or something the day before a minor release and it gets flushed... but on the other hand... 18:08:08 yeah.. I think that we would need to do minor branching 18:08:31 and I realize we are back to why we never did this before 18:09:11 probibly best to just leave it up to maintainers and clean out really old ones from time to time. Or nag them more. 18:09:18 yeah 18:09:31 ok next up 18:09:38 #info RHEL-8 Released 18:09:44 #topic RHEL-8 Released 18:10:20 #info RHEL-8 looks to be doing minor releases on a regular schedule. 18:10:43 #info RHEL-8 has modules which are fun, tasty and good for you like brocolli 18:11:10 * tdawson wonders if there is any dip for the brocolli 18:12:02 spicy ranch! 18:13:14 #info CentOS team is examing the RHEL and will announce their schedule as they work out which Elder God is used 18:13:21 anyhow, I think we have a plan for moving forward on epel8 now? 18:13:50 #topic Should we do an EPEL-8 (OK really we will do an EPEL-8) 18:14:26 Alright I have a general plan on how to do an EPEL-8... 18:15:54 https://etherpad.gnome.org/p/EPEL-8-maybe 18:16:17 * smooge is having document meltdown issues at the moment 18:16:58 yeah, it's really slow here, but it did load 18:17:32 there were a couple of others I worked on.. but I think this was the shortest 18:17:41 I think we should skip to the making repos koji can use... doing a pass with no modular content seems a waste to me 18:18:27 but I guess that says that I just cant read 18:19:18 I think that was in an earlier version 18:19:43 How is koji going to consume modular content? 18:20:17 anyhow, (for irc logs) IMHO its: mirror content, setup script to split modules into seperate repos, get list of default modules, add to staging koji, try and build a base of things, later rinse repeat 18:20:47 orionp: we are going to split modules into their own repos and only add as external repos in koji the default or buildonly ones. 18:20:47 nirik: In short, we just pretend they are small repo's, correct? 18:20:59 or ... what you just said :) 18:21:02 no, they will really be modules... the next part is: 18:21:27 koji patch to how it merges external repos + mergerepos_c patch that merges module repodata 18:21:51 so then dnf in koji will hopefully do the right thing with what we give it 18:22:23 so one thing we won't be able to do in the past was download and then do a createrepo on the downloads.. it tends to lose the moduleinfo 18:23:30 smooge: right, we need the full metadata... 18:23:53 once we get it working in stg, we can move to prod with a small pool of packages and open the gates... 18:24:03 so one thing I was thinking of was what if instead of building against the always updated version, we rebuild against what is shipped in the ISO only. 18:24:06 (and modules would come later... this is all just non modular content) 18:24:56 thats going to mean we might miss security updates and bugfixes tho right? 18:25:07 well security updates we aren't shipping 18:25:15 but rhel is 18:25:55 the api for the base should stay the same, so that would be ok, but appstream could change? or only by adding new modules? 18:26:48 at the moment i am not sure how the appstream will change or how often.. or if they will change defaults in the middle of a minor version 18:26:49 and when do they change defaults? only at minor releases? or anytime? 18:26:54 yeah 18:27:19 but if they change defaults during a minor we will have to change what we are building against anyway 18:27:33 yep 18:27:45 because we will be hardcoding which micro-repos we are building against 18:27:53 * nirik nods 18:28:39 do we have/want to send an announcement with the plan? and/or blog post? or does anyone find the plan bad ? :) 18:28:41 so anyway.. I need to write out a couple more lines.. but is there any questions on the doc in general to send to the mailing list? I have some other content to add but it was being reviewed 18:28:46 hah 18:29:06 jinx. 18:29:30 smooge: if you like I can write something up too... you have done so many of these things... but if you want to go for it. 18:29:33 When we branch in stage, with those be epel8 branches, or epel8.0, or epel8-stage ? 18:29:43 Or is that too minor a detail right now? 18:30:10 IMHO epel8 18:30:17 I was going to say epel8.0 18:30:40 and we do a branch set for epel8.1 when it goes into beta 18:30:43 per minor branches are going to annoy maintainers a lot 18:31:02 having to request every minor release... 18:31:02 I gotta run - but I'd like to be involved in building/testing/helping out 18:31:25 unless you can build for older point releases, what's the point of epel8.X? 18:31:29 I think we need to do the minor branches, or the new EPEL scheme will never work. 18:31:35 nirik, so for minor releases I was thinking we just auto branch from 8.0 -> 8.1 and if a maintainer wants to stop they dead.package 8.1 18:31:49 but they could just dead.package epel8 ? 18:32:34 orionp: yeah 18:32:47 I was figuring it would make pungi/etc easier if the branches they pointed to a minor 18:33:08 I typed what I did before I ready orionp's comment. He has a good point. If you can't build on older releases ... 18:33:09 pungi doesn't know about branches, it gets everything from koji tags. 18:33:27 orionp, so what I am seeing is that during the beta for 8.1 you would build a newer version there and keep the older one in 8.0 18:33:41 if you build the newer one then you are oging ot publish in the older tree 18:34:28 so there is a period where you are 'building' for both a new and old. If someone has a better way of dealing with that I am ok with it. I just didn't come up with anything 18:35:15 or does the problem I am seeing even exist? 18:35:42 well, I am not sure what the problem is... wanting to push newer to beta, while still being able to push fixes/security to older? 18:36:02 I don't see how we will be able to build with beta's ... I thought Red Hat doesn't publically release them. 18:36:05 well, it could be epel8-stable and epel8-testing 18:36:08 or whatever 18:36:13 minor beta's I mean 18:36:23 tdawson: yeah, true 18:36:24 just not a number linked to point releases 18:36:34 well, we have that now... updates and updates-testing 18:37:00 clearly there's lots of corner cases we will have to deal with. ;) 18:37:24 so I wasn't looking to build with the beta, just announce that the beta is there and if you want a major jump you can test it in the 8.N branch. 18:37:52 but yeah.. I am missing the fact that we would need to rebuild all that with 8.N when it came out 18:38:22 ok so let us just go with epel8 branch 18:38:30 8.x still gets updates after 8.x+1 beta is out right? so thats another case... 18:38:32 if we end up needing more we can amke them later 18:38:55 epel8-rawhide 18:39:09 actually that might make sense.. 18:39:31 indeed. it well could. 18:40:03 it solves my problem and could be a place where people who dont give a F about stability on their production systems could put things 18:40:08 "update as you like, users keep all parts when it breaks" 18:40:42 I will put that as a proposal and people can dwell on it for the coming week 18:40:56 You know what ... I really like that, and not just for epel8 ... there might be some epel7 people who want that. 18:40:59 so when we have 8.0 and 8.1 beta comes out, we tell people to test in epel8-rawhide, then when 8.1 comes out we archive 8.0, tell people they can land their big 8.1 changes for a week, move on? 18:41:24 smooge: yeah, it's a pretty nice idea actually. 18:41:29 tdawson: indeed. 18:42:04 and epel8-rawhide acts like rawhide (nightly compose of all builds) 18:43:32 I'm reallying liking that idea... 18:43:37 ok cool 18:44:22 let us write it up, send it out ot the lists and vote on it. 18:44:28 once we have modular builds people can really go nuts there... 18:44:55 I am going ot need to get some food.. my brain has turned on the big E sign 18:45:04 #topic Open Floor 18:45:12 Anything else for this week? 18:45:13 red smooge needs food badly! 18:45:42 blue smooge shot the food 18:46:28 ok if not we are going to call this a week and try to have blog posts and plans put to the list by end of day tomorrow 18:46:50 Sounds good. 18:46:59 nothing from me here 18:47:49 thank you all for coming and see you next week 18:47:52 #endmeeting