20:01:27 #startmeeting EPEL (2022-10-26) 20:01:27 Meeting started Wed Oct 26 20:01:27 2022 UTC. 20:01:27 This meeting is logged and archived in a public location. 20:01:27 The chair is tdawson. Information about MeetBot at https://fedoraproject.org/wiki/Zodbot#Meeting_Functions. 20:01:27 Useful Commands: #action #agreed #halp #info #idea #link #topic. 20:01:27 The meeting name has been set to 'epel_(2022-10-26)' 20:01:28 #meetingname epel 20:01:28 The meeting name has been set to 'epel' 20:01:30 #chair nirik tdawson pgreco carlwgeorge salimma dcavalca dherrera gotmax[m] smooge 20:01:30 Current chairs: carlwgeorge dcavalca dherrera gotmax[m] nirik pgreco salimma smooge tdawson 20:01:31 #topic aloha 20:01:33 .hello robert 20:01:34 rsc: robert 'Robert Scheck' 20:01:38 morning 20:01:40 .hi 20:01:41 dherrera: dherrera 'Diego Herrera' 20:01:49 .hi 20:01:50 michel-slm: Sorry, but user 'michel-slm' does not exist 20:01:51 Hello rsc 20:01:57 Morning nirik 20:01:58 .hi michel 20:02:00 michel-slm: Sorry, but user 'michel-slm' does not exist 20:02:06 Hi dherrera and michel-slm 20:02:07 .hello salimma 20:02:08 michel-slm: salimma 'Michel Alexandre Salim' 20:02:10 .hello dcavalca 20:02:11 davide: dcavalca 'Davide Cavalca' 20:02:17 Hello davide 20:02:19 .hi 20:02:20 carlwgeorge: carlwgeorge 'Carl George' 20:02:42 Hi carlwgeorge 20:03:13 now to find which of my Matrix I'm salimma on in case I need to add something to the log :) 20:03:25 test 20:03:31 found it 20:03:57 Hi salimma on the other Matrix 20:04:30 hello 20:04:47 .hi 20:04:48 rcallicotte: rcallicotte 'Robby Callicotte' 20:05:11 #chair michel-slm 20:05:11 Current chairs: carlwgeorge dcavalca dherrera gotmax[m] michel-slm nirik pgreco salimma smooge tdawson 20:05:38 #topic End Of Life (EOL) 20:05:40 RHEL 7 will go EOL on 2024-06-30 20:05:41 CentOS Stream 8 goes EOL in 2024-05-31 20:05:43 CentOS Stream 9 goes EOL in 2027-05-31 20:06:11 #topic EPEL Issues https://pagure.io/epel/issues 20:06:12 https://pagure.io/epel/issues?tags=meeting&status=Open 20:07:04 We only have two issues left, the rest are on "back-burner" 20:07:28 I just re-read #200, and realized that it was waiting on an action from me. 20:08:14 .hi 20:08:15 pgreco: pgreco 'Pablo Sebastian Greco' 20:08:37 I was supposed to contact the subscription manager team and see how we could get access to some type of API, and I totally dropped that ball. 20:08:39 Hi pgreco 20:09:08 So that leaves us with #39 20:09:11 .epel 39 20:09:12 tdawson: Issue #39: Retiring EPEL Packages (End of lifing CEPH in EL7 and process improvement) - epel - Pagure.io - https://pagure.io/epel/issue/39 20:10:15 This ticket keeps going down the "Orphan" rabbit hole. 20:10:55 I'm wondering if, for now, we want to just do something fast with what we have, and say you need to send an email if you are retiring a package. 20:11:15 * gotmax[m] nods 20:11:41 Create a completely seperate ticket to figure out the orphaning stuff. 20:12:07 .hello ngompa 20:12:09 Eighth_Doctor: ngompa 'Neal Gompa' 20:12:14 i am ok with that 20:12:15 Hello Eighth_Doctor 20:12:22 I'm not sure people will know that they need to do that tho... depending on users to know some process we just made is not likely to work too great 20:13:33 nirik: True, but right now the policy is that you can retire a package with no warning. Although some might not see the new policy, at least some might look to see if there is a retirement policy. 20:13:41 tdawson: Yeah, I think we should separate this out. Require an email to epel-devel and/or epel-announce before retiring and we can discuss other procedural improvements later 20:14:14 we can give it a try, sure... just saying it might be better in the long run if we can automate more. 20:14:34 but one step at a time I suppose 20:14:50 Yep, I would love to automate more, but that's why this one is totally stalled. 20:15:10 Yup 20:15:49 announce emails for epel branch retirements sound like a great first step 20:16:58 So, salimma already has some of it done with this pull request - https://pagure.io/epel/pull-request/182 20:18:47 I think some tweaking needs to be done, which started this whole debate, but at least it has the framework for the documentation. 20:20:16 salimma: Are you ok if we continue making suggestions for your pull request, or would you rather we start fresh? 20:21:14 tdawson: I have not had time to take a look, either way works 20:21:48 michel-slm: You wouldn't feel bad if I made my own pull request with a complete re-write? 20:22:03 tdawson: go for it! 20:22:10 that's probably going to end up faster anyway 20:22:15 :) 20:22:18 OK, thank you. 20:23:04 Just a sec while I put that on my to-do list ... 20:24:10 OK,, I'll take it. I'll see if I can have a draft done in time for next weeks meeting. 20:25:00 Which means ... that's the last issue . 20:25:40 I don't have anything written down from last meeting for Old Business ... which puts us in the Open Floor already ... wow ... 20:25:46 #topic General Issues / Open Floor 20:27:34 Do we really not have that much EPEL stuff now that we're getting rid of modules? 20:27:40 hehe 20:27:50 lol 20:28:04 ha 20:28:13 I can ask for something that could potentially break as many things... 20:28:19 occasionally i see confusion about bugzilla components that don't match the package name. i'm working on a pr to the epel request guide for how to locate the component (srpm) name. 20:28:37 really, just a question, tbf 20:29:02 carlwgeorge: Ya ... another one came up today where they just put the request on epel-release. 20:29:16 carlwgeorge: So if you could figure out better wording, that would be great. 20:29:17 I saw some more fallout from the xmlstarlet thing in bugzilla 20:29:25 pgreco: Go for it. 20:29:31 I remember that building multilib was a problem, can anyone clarify why was that? 20:29:50 I have only one package in mind, but it has repercussions 20:30:13 currently wine in epel only builds (and runs) x86_64 which makes it basically useless :( 20:30:30 pgreco: I thought we tested and found out that we can do that. I belive we already have a couple packages where we have EPEL packages for just one or two arches. 20:30:45 * gotmax[m] wishes this was a more friendly process. Many seem confused by Bugzilla and the current process. 20:31:11 tdawson: you're talking about limiting, I'm talking about adding i686 20:31:28 which I know is a problem, just can't remember what problem... 20:31:42 pgreco: Ohh ... ok. Well ... we don't have an i686 build target that I know of. 20:32:10 Yeah, there's no multilib for EL 20:32:55 carlwgeorge: nice. though that's also a problem on the Fedora side right 20:32:56 But don't have RHEL clones the required build artifacts anyway? 20:33:07 (sorry, was distracted by my company's earnings) 20:33:33 tdawson: we don't, that's correct, and I know it was hard to do for "$reasons", just trying to remember which were those reasons 20:33:54 as in... should we document 'how to find component name' universally for both Fedora and EPEL? It's the same methodology anyway 20:34:06 pgreco: Maybe nirik or Ebeneezer_Smooge might know. 20:34:26 michel-slm: that's a good point, i'll have to check if it's mentioned in the fedora docs anywhere and maybe consolidate and link to it 20:34:55 whats the question? 20:34:58 got distracted. ;) 20:35:02 why does multilib suck 20:35:25 i believe why don't epel build targets in koji include i686 20:35:50 because RHEL8 also doesn't have i686 ... or does it? 20:35:52 because there are not generally enough i686 items to build things 20:35:57 right, we would need to build everything (or at least everything thats mulltilib) i686 too. 20:36:04 it has just enough to keep some i686 binaries to work 20:36:13 basically, unpublished artifacts? 20:36:13 it does have some I am pretty sure. 20:36:37 I believe in RHEL the list of i686 packages release has been getting smaller with each release. 20:36:38 yes, it definitely does 20:36:41 many are published, i think just not for every package 20:37:02 and we don't have enough to build, ok 20:37:05 there are a lot of missing -devel and some other items if I remember from trying a couple of things 20:37:17 basically confirming what I suspected 20:37:17 basically if you wanted to build i686 you need to build a working clone of i686 20:37:37 I know for RHEL 10 there is already discussions on if this is the RHEL release that we can finally get rid of all i686 packages. (I'm betting the answer is no) 20:37:51 i doubt very much i686 will be kept around after 9 because there are too many 2038 issues 20:37:54 carlwgeorge: The command is `dnf repoquery --source PKGNAME`, but that might also cause confusion 20:37:58 the only i686 package that would be useful for me is wine (and its deps), I guess I'll keep maintaining it on the side (with arrfab) 20:38:05 packages.fp.o is also a place to look 20:38:20 Hahaha, RHEL without i686 while there are still i686-only proprietary tools by Red Hat partners that refuse today to provide x86_64 additionally? :> 20:38:31 gotmax[m]: i know, i'm mentioning both methods 20:38:48 👍️ 20:38:53 wine and steam are the big holdouts the last time it was discussed for fedora 20:39:01 pgreco: +1 to wine 20:39:17 The problem is that 2036 (or is it 2038) is coming up into the lifetime of the next RHEL release, and we dont' want to have 32 bit binaries when that hits. 20:39:19 rsc: they probably won't change until red hat forces them to i suspect 20:40:08 carlwgeorge: right. But if Red Hat can force IBM, HPE and Dell? 20:40:09 2038 20:40:11 sadly true 20:40:30 Well, 2038 is far away. And some glibc folks is working on that, too? 20:40:31 gotmax[m]: for installed packages I just do 'rpm -qi' 20:40:34 it's faster :) 20:40:41 nope but most of those tools will say they need to have been run on versions of the OS no longer supported anyway 20:40:48 rhel 10 els will hit 2038 i believe 20:40:55 anyway this is out of scope 20:41:07 Yep 20:41:46 wine.i686 still would be helpful to me, just to say it. 20:42:04 carlwgeorge: I'll put your idea for rewording the package request on "Old Business" for next week. 20:42:15 * carlwgeorge nods 20:42:18 rsc: yeah, I can't think of anything else where I'd want i686 20:43:03 pgreco: (Steam on Fedora, I know users, not using it myself) 20:43:17 wine.i686 would be helpful to a lot of people but I expect that it is going to end up being someone's flatpak 20:43:31 anyway.. not something that can be fixed here 20:43:37 yeah, I run steam happily from flatpak 20:43:42 but moving on 20:43:52 I don't have anything else. 20:44:36 Eighth_Doctor: I noticed that there was movement on the wayland update in RHEL 9.2 ... that makes me happy. It will help KDE be able to continue to move forward there. 20:44:42 yay 20:45:19 Anybody have anything else? I'm ok ending the meeting early. 20:45:25 yeah I have one thing 20:45:31 we ran over last time so lets end early today 20:45:38 Eighth_Doctor is our Steve Jobs 20:45:46 lol 20:46:16 Hyperscale needs https://src.fedoraproject.org/rpms/koji/pull-request/15 merged in and released to Rawhide and EPEL for our work on CentOS images 20:46:31 would appreciate it if someone could take a look at this and get it released soonish 20:46:54 Eighth_Doctor: yeah, can do. 20:46:58 thanks! 20:47:09 actually having it in Fedora would be good too 20:47:16 * nirik was on pto all last week and been living with covid since I got back, so I am not caught up. 20:47:19 because otherwise I can't actually use the kiwi-build cli plugin 20:47:32 which makes trying to run builds in CBS kind of difficult :/ 20:48:07 nirik: ouch, feel better soon 20:48:17 Eighth_Doctor: Does that patch affect the client, the server, or both? 20:48:26 both 20:48:32 it adds a subcommand to koji cli 20:48:40 and extends koji server to be able to build images using kiwi 20:48:52 OK. cool. 20:49:32 will get it this afternoon hopefully. 20:49:37 this will also be useful for alt images sig :) 20:49:43 nirik: awesome, thanks 20:49:52 I hope your recovery is going well too 20:49:55 Yep, that's what I'm hoping. 20:50:06 Anything else before I close? 20:50:07 one day we might use it in fedora too :P 20:50:13 but that's it for me 20:51:32 Thank you all for coming, and for the good discussions. 20:51:49 You all are great, and thank you very much for all you do for the EPEL community. 20:51:55 Thanks! 20:52:06 I'll talk to ya'll next week if not sooner. 20:52:10 Thank you. 20:52:20 #endmeeting