20:00:22 <tdawson> #startmeeting EPEL (2022-09-21)
20:00:22 <zodbot> Meeting started Wed Sep 21 20:00:22 2022 UTC.
20:00:22 <zodbot> This meeting is logged and archived in a public location.
20:00:22 <zodbot> The chair is tdawson. Information about MeetBot at https://fedoraproject.org/wiki/Zodbot#Meeting_Functions.
20:00:22 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
20:00:22 <zodbot> The meeting name has been set to 'epel_(2022-09-21)'
20:00:24 <tdawson> #meetingname epel
20:00:24 <zodbot> The meeting name has been set to 'epel'
20:00:25 <tdawson> #chair nirik tdawson pgreco carlwgeorge salimma dcavalca dherrera gotmax[m]
20:00:25 <zodbot> Current chairs: carlwgeorge dcavalca dherrera gotmax[m] nirik pgreco salimma tdawson
20:00:27 <tdawson> #topic aloha
20:00:31 <pgreco> .hi
20:00:32 <zodbot> pgreco: pgreco 'Pablo Sebastian Greco' <pablo@fliagreco.com.ar>
20:00:41 <salimma> .hi
20:00:42 <rcallicotte> .hi
20:00:42 <zodbot> salimma: salimma 'Michel Alexandre Salim' <michel@michel-slm.name>
20:00:43 <tdawson> Hi pgreco
20:00:45 <zodbot> rcallicotte: rcallicotte 'Robby Callicotte' <rcallicotte@mailbox.org>
20:00:46 <dherrera> .h
20:00:49 <dherrera> .hi
20:00:51 <smooge> hello
20:00:51 <tdawson> Hi salimma
20:00:52 <zodbot> dherrera: dherrera 'Diego Herrera' <dherrera@redhat.com>
20:00:53 <gotmax[m]> .hello gotmax23
20:00:54 <pgreco> sorry for being MIA last week, I had a tooth removed and completely forgot about this...
20:00:55 <zodbot> gotmax[m]: gotmax23 'Maxwell G' <gotmax@e.email>
20:01:01 <tdawson> Hi rcallicotte and dherrera
20:01:09 <tdawson> Hello smooge
20:01:15 * salimma hopes pgreco is feeling better
20:01:17 <tdawson> Hello gotmax[m]
20:01:31 <salimma> I missed last week due to timeshifting 6 hours to attend conferences in Dublin. Worse than actually being there!
20:01:33 * gotmax[m] appreciates being part of the chair list so I get pinged and can't forget to come :)
20:01:36 <gotmax[m]> Hi tdawson
20:01:39 <pgreco> yeah, much better now
20:01:44 <gotmax[m]> Great!
20:02:35 <carlwgeorge> .hi
20:02:36 <zodbot> carlwgeorge: carlwgeorge 'Carl George' <carl@redhat.com>
20:02:42 <tdawson> Hi carlwgeorge
20:02:57 * gotmax[m] waves to carlwgeorge
20:04:02 <smooge> we are all chairs here
20:04:05 <tdawson> gotmax[m]  That's the downside of running the meeting.  Nobody pings me. :(
20:04:12 <smooge> wait wait.. I lost lost my chair ???
20:04:15 <tdawson> smooge: Actually ... you aren't on the list.
20:04:29 <smooge> oh yeah I said I retired
20:04:34 <tdawson> :)
20:04:40 <tdawson> smooge: I'll put you back if you want.
20:05:07 <smooge> sure.
20:05:10 <gotmax[m]> tdawson: :)
20:05:33 <smooge> i guess i am a glutton for epel
20:05:42 <tdawson> smooge: OK, you're back on ... well, you'll be on next week.
20:05:52 <tdawson> #topic End Of Life (EOL)
20:05:53 <tdawson> RHEL 7 will go EOL on 2024-06-30
20:05:55 <tdawson> CentOS Stream 8 goes EOL in 2024-05-31
20:05:56 <tdawson> CentOS Stream 9 goes EOL in 2027-05-31
20:06:14 <tdawson> Getting closer ...
20:06:22 <tdawson> #topic EPEL Issues https://pagure.io/epel/issues
20:06:24 <tdawson> https://pagure.io/epel/issues?tags=meeting&status=Open
20:06:51 <tdawson> I don't see dcavalca again this week.
20:07:02 <gotmax[m]> There was a follow up on the buildroot thing...
20:07:16 <tdawson> Yep
20:07:28 <tdawson> .epel 200
20:07:32 <zodbot> tdawson: Issue #200: Investigate whether EPEL packagers can get access to the BR repo - epel - Pagure.io - https://pagure.io/epel/issue/200
20:07:47 <salimma> tdawson: dcavalca is on vacation in Italy
20:08:05 <salimma> so it's probably his bedtime now :)
20:08:26 <tdawson> salimma: Ah, ok.  I hope he has fun ... we'll let issue 199 go for another week.
20:08:26 <gotmax[m]> 10:07 CEST
20:09:26 <gotmax[m]> Anyways... the update on that ticket is that Red Hat said we can give access to the buildroot repo to anyone with a RHEL subscription
20:09:37 <gotmax[m]> But the question is how do we do that
20:09:39 <salimma> should we document that?
20:09:57 <salimma> worst case, everyone who doesn't have one and need to work on EPEL just sign up for the developer subscription, right
20:10:07 <salimma> oh... but right. how do we keep track of who has one
20:10:28 <gotmax[m]> Correct about the developer subscription
20:10:57 <tdawson> To me, the problem seems to be how do we get koji (or wherever it is) to recognize that we have a RHEL subscription.
20:11:45 <tdawson> I have no idea how to pass an "I have a RHEL subscription" token to a service.
20:12:03 <gotmax[m]> Yeah, the proxys that currently blocks access to https://infrastructure.fedoraproject.org/repo/rhel would need some kind of auth
20:12:36 <gotmax[m]> Even if we did it manually checked people, how would that work?
20:13:26 * nirik arrives late
20:13:43 <gotmax[m]> Hi nirik
20:13:45 <tdawson> Hi nirik
20:13:54 <nirik> the only way it could work is with the rhn certs, but I bet it would be a mess to setup
20:14:04 <smooge> yeah
20:15:07 <smooge> I wonder if it would be easier to just make grobisplitter used locally
20:15:40 <salimma> would that mean you have to be running RHEL locally?
20:15:50 <gotmax[m]> And have a full mirror?
20:16:05 <smooge> well I was going to say just make it also work with alma/rocky/whatever-floats-your-boat
20:16:57 <tdawson> salimma: I'm betting you would need to be running it locally.
20:18:04 <gotmax[m]> The buildroot repodata currently points directly to https://infrastructure.fedoraproject.org/repo/rhel AFAIK, which would present problems when trying to use a local grobisplit mirror with a different URL.
20:18:35 <salimma> depending on how many people need this... one workable solution is to have a web app where you prove you have a license and is a given Fedora username, and get a signed proof back you can use
20:19:25 <gotmax[m]> Yeah, but how do you prove it in the first place?
20:20:22 <nirik> the yummy to trouble ration is not great here, IMHO.
20:20:34 <tdawson> :)
20:20:42 <smooge> it isn't easy to prove. our systems are not tied together in any way that would allow that and I expect that tieing them together would be a multi-year battle to keep working
20:21:23 <gotmax[m]> Maybe we should toss it back to RH and ask how they suggest we do that?
20:21:47 <smooge> well the people who say to do it aren't probably the ones who would implement it.
20:22:06 <tdawson> I was just thinking that.  Contacting the subscription-manager team and ask them if there is API call somehwere we can use.
20:22:25 <gotmax[m]> Yeah, I was wondering if there was an API
20:22:48 <salimma> ah. then FAS just need a field to say 'what's the email you use for your RHEL subscription'
20:23:17 <tdawson> Or something to that effect.
20:23:39 <nirik> just
20:24:06 <tdawson> I guess I did the last thing ... I'll try to contact the correct team and find out if there is some API or something we can use.
20:25:16 <smooge> 'just need a field'. We also just need developers to write the code, etc
20:25:20 <tdawson> That might take more than a week for it, cuz I'm expecting to be passed around several time.
20:25:49 <tdawson> smooge: Yep :) ... After / If we find an API like thing, then the hard part comes.
20:26:22 <tdawson> Anything else for this before I move on?
20:27:10 <tdawson> smooge: How is the module stuff coming along?  Are we ready to decide yet?
20:27:30 <tdawson> Or do you really want me to wait until Oct. 15th ?
20:27:44 <gotmax[m]> It could be a manual process where you provide the information to infra and they use the API or whatever to verify it, but I don't want to sign you all up for that :)
20:28:40 <salimma> smooge: oh, sure, I just meant the additional information we need to record
20:29:20 <smooge> nothing has been added since last time so I think its a 'lets decide this thing and watch what crawls out of the woodwork'
20:29:26 <salimma> yeah, if not many people need it, just filing a ticket somewhere and manual verification + updating a whitelist might not be onerous enough. or... just say, heck if you're building for EPEL we trust you :)
20:29:28 <nirik> it would still need some kind of auth too tho.
20:30:15 <gotmax[m]> Yup, that too
20:30:16 <carlwgeorge> does the RHEL buildroot stuff get easier once grobispiltter goes away?
20:30:21 <smooge> no
20:30:26 <smooge> it has never been easy
20:30:35 <tdawson> .epel 198
20:30:36 <zodbot> tdawson: Issue #198: Drop modularity from EPEL-8. Do not enable modularity for EPEL-9 - epel - Pagure.io - https://pagure.io/epel/issue/198
20:30:37 <smooge> this is something we have wrestled with since EPEL started
20:30:48 <smooge> ok changing to 198
20:30:59 <carlwgeorge> easier, relative to doing it with grobispiltter
20:31:40 <smooge> carlwgeorge, on the letting people look at the buildroot, it has never been easy and won't be any easier with/without grobisplitter
20:32:19 <smooge> on ticket 198. I am good with saying we vote on this, put the vote in the ticket and start doing whatever is needed
20:32:56 <gotmax[m]> For EPEL 7/9, I'd guess not having to deal with grobisplitter creates less work, but the buildroot access issue is orthogonal
20:34:00 <tdawson> Just to summarize ticket 198.  It is to get rid of modularity in epel8.  Although the steps are in there, the steps aren't what we are voting on.  We are voting on whether we should do it or not.
20:34:26 <tdawson> Does anyone need any more information before we vote?
20:34:27 <smooge> I am strangely for my own proposal
20:34:55 <tdawson> I think that's a good thing.
20:35:05 <carlwgeorge> +1 from me
20:35:05 <nirik> +1 here. I am a bit surprised there was not more of an outcry, but... ok. ;)
20:35:13 <tdawson> +1
20:35:26 <pgreco> +1
20:35:33 <salimma> +1
20:35:36 <dherrera> +1
20:35:38 <pgreco> I hate that we have to do this, but I have to agree
20:35:43 <salimma> yeah the mailing list is surprisingly quiet last I checked
20:35:53 <gotmax[m]> +1 on principle
20:35:53 <smooge> nirik, my guess is that the outcry will occur ater it has been dropped AND several months go by
20:36:00 * nirik nods.
20:36:09 <gotmax[m]> I think how we go about this is also important
20:37:28 <tdawson> #info Motion passed.  We will remove modularity from epel8.  For: 8  Against: 0  NoVote: 0
20:38:38 <carlwgeorge> i'll set up a pr that implements step 3 (epel-repos-modular)
20:38:57 <tdawson> carlwgeorge: Thanks.
20:39:07 <tdawson> I think either way we needed to do that.
20:39:31 <tdawson> smooge: Do you want to send the email, or do you want me to?
20:39:46 <smooge> I can send the email.
20:39:56 <tdawson> Sounds good.  Thanks.
20:40:06 <smooge> since I started this mess in more ways than one
20:41:12 <tdawson> Since we're moving the modules to archives, and pointing the mirrormanager to that ... in theory, that won't break anyones machine ... right?
20:41:37 <carlwgeorge> in theory
20:42:29 <smooge> in most theory anymore than they might be broken now
20:42:30 <gotmax[m]> I kind of liked the idea of pushing one last update to change the descriptions of the modules to say they're deprecated
20:42:46 <pgreco> well, it can be both
20:42:54 <pgreco> point to archive and change the title
20:42:58 <smooge> if people want to do that, it can be done.
20:44:01 <smooge> Hmm.. I never sent this to Fedora devel but just epel-devel.
20:44:26 <tdawson> smooge: Are you going to do the relen work?  Or do you still have permissions to do that?
20:44:30 <gotmax[m]> I'm not super set on this, but I like the idea of making clear that the content is unsupported
20:45:28 <smooge> gotmax[m], I have always thought that it would be good if we drop things from EPEL, that the last build is done with a statement that says it has reached EOL. However that has been a lot more work than I or anyone else has done.
20:46:00 <smooge> tdawson, I can start looking at it. I was going to wait to do most of it in November after F37 ships
20:46:20 <smooge> and nirik is back from various PTO so I can make sure I am not blowing things up
20:46:26 <tdawson> smooge: November sounds good.  No real rush.
20:46:50 <gotmax[m]> I have some experience with mass packaging changes from the rebuilds I've done, but I've never built a module
20:47:01 <gotmax[m]> so not sure how much I can help
20:48:03 <smooge> i have never successfully built a module either
20:48:52 <smooge> and modules usually get built for multiple releases so I would be worried I pushed to F36 that foobaz is no longer supported when I just meant EPEL
20:48:53 <gotmax[m]> It feels a bit ironic to learn how to build a module just for the sake of getting rid of modules. But I digress...
20:49:14 <tdawson> So, one thing I was thinking was put the info that it's deprecated in the /etc/yum.repos.d/epel-modular.repo
20:49:16 <gotmax[m]> Yeah, I was thinking that
20:50:08 <tdawson> So, these are all details.  I'm going to move on incase we have some other things on the open floor.
20:50:38 <tdawson> #topic General Issues / Open Floor
20:50:53 <smooge> The floor is lava
20:51:28 * tdawson gingerly steps on the "General Issues" instead of the "Open Floor" which is lava
20:52:18 * michel has to step out. Baby fussy
20:53:37 <tdawson> So, the "uninstallable packages" stuff is moving forward, although slowly.  I've got dates on will-it-install for when a package first was uninstallable.
20:54:11 <tdawson> One big note is that it only goes back to when that code was put in, so only a couple days ago.
20:55:03 <tdawson> carlwgeorge also has been working on getting a check into bodhi.
20:55:43 <gotmax[m]> Yeah. It shouldn't even be possible to push uninstallable packages.
20:55:50 <carlwgeorge> currently bodhi runs rpminspect on fedora updates, but not epel updates.  i'm trying to get that enabled, then separately looking into an installability check for all bodhi updates.
20:55:50 <rcallicotte> agreed
20:56:09 <tdawson> I don't agree, but I feel you should be warned.
20:56:20 <carlwgeorge> like the rpminspect checks for fedora now, these would be waivable
20:56:23 <nirik> its likely not as easy as you might think. ;)
20:57:31 <tdawson> nirik: Which part, getting the test in there?
20:58:06 <rcallicotte> would it be running a mock --postinstall or something to test installability?
20:58:15 <nirik> well, installability... has to check against other builds in the update, other updates going stable at the same time, etc
20:58:24 <nirik> and has to be waiveable
20:58:50 <carlwgeorge> my understanding is the team that works on that wants to use https://github.com/fedora-ci/mini-tps
20:59:25 <gotmax[m]> I think the currently installability gating test only works on rawhide
21:00:05 <nirik> epel8 also is likely a mess... this is installable, but only if you have this module enabled.
21:00:21 <rcallicotte> oh boy
21:00:29 <tdawson> yipes
21:00:49 <carlwgeorge> epel8 packages can only build against default module streams
21:01:00 <tdawson> But it's best to get the tests setup, and work out the bugs as we go.
21:01:20 <nirik> yeah, but I mean the tests could be anoying with modules in the mix.
21:01:54 <tdawson> Looks like our time is up.  Thank you all for coming and having good discussions.  And thank you all for the work you do on EPEL.
21:02:03 <tdawson> Talk to you next week, if not sooner.
21:02:15 <tdawson> #endmeeting