<@tdawson:fedora.im>
21:00:28
!startmeeting EPEL (2023-11-15)
<@meetbot:fedora.im>
21:00:29
Meeting started at 2023-11-29 21:00:28 UTC
<@meetbot:fedora.im>
21:00:29
The Meeting name is 'EPEL (2023-11-15)'
<@tdawson:fedora.im>
21:00:35
!meetingname epel
<@tdawson:fedora.im>
21:00:43
!topic aloha
<@carlwgeorge:matrix.org>
21:00:45
!hi
<@zodbot:fedora.im>
21:00:47
Carl George (carlwgeorge)
<@tdawson:fedora.im>
21:00:56
!hi
<@zodbot:fedora.im>
21:00:57
None (tdawson)
<@pgreco:fedora.im>
21:01:00
!hi
<@zodbot:fedora.im>
21:01:01
Pablo Sebastian Greco (pgreco)
<@tdawson:fedora.im>
21:01:13
Ah ... it still doesn't know me ... oh well.
<@tdawson:fedora.im>
21:01:40
Hi Pablo Greco and Carl George
<@jonathanspw:fedora.im>
21:02:21
!hi ?
<@nirik:matrix.scrye.com>
21:02:22
!hi
<@dherrera:fedora.im>
21:02:23
!hi
<@zodbot:fedora.im>
21:02:24
Kevin Fenzi (kevin) - he / him / his
<@zodbot:fedora.im>
21:02:24
Diego Herrera (dherrera) - he / him / his
<@jonathanspw:fedora.im>
21:02:28
!hi
<@zodbot:fedora.im>
21:02:29
Jonathan Wright (jonathanspw)
<@nirik:matrix.scrye.com>
21:02:52
Troy Dawson: you likely have 'private' set on your account data... it thus can't get your name...
<@tdawson:fedora.im>
21:05:01
!hi
<@zodbot:fedora.im>
21:05:02
None (tdawson)
<@tdawson:fedora.im>
21:05:34
!topic End Of Life (EOL)
<@tdawson:fedora.im>
21:05:41
RHEL 7 / epel-7 will go EOL on 2024-06-30 https://endoflife.date/rhel CentOS Stream 8 / epel-8-next goes EOL in 2024-05-31 CentOS Stream 9 / epel-9-next goes EOL in 2027-05-31 https://endoflife.date/centos-stream
<@davide:cavalca.name>
21:05:56
!hi
<@zodbot:fedora.im>
21:05:57
Davide Cavalca (dcavalca) - he / him / his
<@tdawson:fedora.im>
21:06:24
!topic EPEL Issues https://pagure.io/epel/issues
<@tdawson:fedora.im>
21:06:30
https://pagure.io/epel/issues?tags=meeting&status=Open
<@tdawson:fedora.im>
21:06:52
Wow ... we haven't had a week without any open issues for a while.
<@tdawson:fedora.im>
21:07:23
!topic Old Business
<@nirik:matrix.scrye.com>
21:07:45
👴
<@tdawson:fedora.im>
21:08:03
Does anyone have any old business ... since I wasn't here last week, there might be some that I don't know about.
<@dherrera:fedora.im>
21:09:13
honestly, there wasn't much movement last week either.
<@tdawson:fedora.im>
21:09:14
That looks so different between the web browser and Neochat. The browser shows and old man, and neochat shows what looks like a young kid.
<@jonathanspw:fedora.im>
21:09:41
i just wish the unread message counter would get fixed
<@carlwgeorge:matrix.org>
21:09:53
i guess this one counts as old business, since i filed it 11 days ago
<@carlwgeorge:matrix.org>
21:09:59
<@carlwgeorge:matrix.org>
21:10:37
however i'm not sure i've brought it up at this meeting yet, so maybe not old for the context of this meeting
<@nirik:matrix.scrye.com>
21:11:00
yeah, jednorozec was going to do this tomorrow or soon...
<@nirik:matrix.scrye.com>
21:11:23
but I don't think per minor keys is a good idea really...
<@carlwgeorge:matrix.org>
21:11:41
Conan Kudo: did you want to elaborate more on why you think we should do separate keys per minor version?
<@carlwgeorge:matrix.org>
21:12:24
i agree with nirik, it seems unnecessary, and el users would likely be surprised to have to accept a new epel key every six months
<@conan_kudo:matrix.org>
21:12:43
I wasn't advocating for it one way or another
<@michel:one.ems.host>
21:12:46
!hi
<@carlwgeorge:matrix.org>
21:12:46
i agree with nirik, it seems unnecessary, and el users would likely be surprised to have to accept a new epel key every six months within the same major version
<@jonathanspw:fedora.im>
21:12:47
I'd argue against a key per minor as well
<@zodbot:fedora.im>
21:12:47
Michel Lind (salimma) - he / him / his
<@conan_kudo:matrix.org>
21:12:53
!hi
<@michel:one.ems.host>
21:12:56
I'm still at lunch, will be sporadic
<@zodbot:fedora.im>
21:12:56
Neal Gompa (ngompa) - he / him / his
<@conan_kudo:matrix.org>
21:13:08
The only reason I asked the question was because it would allow reuse of the same SOP as Fedora
<@carlwgeorge:matrix.org>
21:13:48
that sop is going to have to diverge in several other ways anyways, i don't think reusing the fedora one verbatim would work anyways
<@conan_kudo:matrix.org>
21:13:56
okay, that's fine
<@tdawson:fedora.im>
21:13:57
I guess I thought this was the same SOP as Fedora, once per release ... it's just that the releases are very long.
<@conan_kudo:matrix.org>
21:14:21
see, my confusion is whether we want to consider EPEL 10.Y as releases or not
<@carlwgeorge:matrix.org>
21:14:50
in bodhi, yes. for package signing key, no. the nuance of major vs minor.
<@conan_kudo:matrix.org>
21:14:59
I don't think it's actually worth it to do different keys for each minor version
<@tdawson:fedora.im>
21:15:00
Ah ... ok, I see what you were thinking ... for me, I'd say no ... EPEL 10 seems like a release to me.
<@pgreco:fedora.im>
21:15:35
yeah, worth bringing it up so it can be squashed properly
<@conan_kudo:matrix.org>
21:15:37
the only real benefit for it would be making key compromise less painful, but the downside would be that every gpg key would have to be added to the gpgkeys= in the repo file over time
<@conan_kudo:matrix.org>
21:15:40
that would very much suck
<@carlwgeorge:matrix.org>
21:16:04
it was a fair question but i think we're all on board with keeping one key per major epel version
<@conan_kudo:matrix.org>
21:16:11
yeah
<@carlwgeorge:matrix.org>
21:16:16
everything else can change independent of that
<@conan_kudo:matrix.org>
21:16:19
yup
<@conan_kudo:matrix.org>
21:16:29
I just figured it'd be worth asking now rather than later
<@conan_kudo:matrix.org>
21:16:36
since it's one of those "early design decision" things
<@conan_kudo:matrix.org>
21:16:43
it's hard to back out of once it's set
<@carlwgeorge:matrix.org>
21:17:12
once we get the key id, i plan to send the initial epel10 prs to robosignatory and bodhi so they recognize it
<@conan_kudo:matrix.org>
21:17:30
cool
<@carlwgeorge:matrix.org>
21:17:46
i think that's all i had on that one Troy Dawson, we can move on
<@tdawson:fedora.im>
21:17:50
Sounds settled then. Any other Old Business?
<@tdawson:fedora.im>
21:18:14
!topic General Issues / Open Floor
<@tdawson:fedora.im>
21:18:34
Anybody have anything for Open Floor?
<@davide:cavalca.name>
21:18:35
I have one thing for open floor
<@tdawson:fedora.im>
21:18:47
Davide Cavalca Go for it.
<@conan_kudo:matrix.org>
21:19:02
go for it
<@tdawson:fedora.im>
21:19:11
Just don't spill your lunch on your keyboard.
<@carlwgeorge:matrix.org>
21:19:27
nom nom
<@davide:cavalca.name>
21:19:45
https://bugzilla.redhat.com/show_bug.cgi?id=2251820 is another package from HA I'm trying to get branched in epel
<@conan_kudo:matrix.org>
21:20:06
oof
<@davide:cavalca.name>
21:20:23
Just wanted someone to double check that I explained things correctly for the maintainer as they didn't seem familiar with the process
<@conan_kudo:matrix.org>
21:20:51
it seems right
<@carlwgeorge:matrix.org>
21:20:54
lgtm
<@conan_kudo:matrix.org>
21:21:06
I assume the idea is to sync the RHEL/CS version to EPEL9?
<@conan_kudo:matrix.org>
21:21:08
and build it?
<@davide:cavalca.name>
21:21:49
Yep that was my plan
<@conan_kudo:matrix.org>
21:21:58
then that should be all gravy
<@tdawson:fedora.im>
21:22:17
I think you explained it better than I could have ... so LGTM
<@davide:cavalca.name>
21:22:25
Perfect thanks
<@davide:cavalca.name>
21:22:32
That's all for me
<@nirik:matrix.scrye.com>
21:22:38
I had 2 small items.
<@tdawson:fedora.im>
21:22:48
nirik: Go for it
<@nirik:matrix.scrye.com>
21:23:17
first just wanted to note this request: https://pagure.io/releng/issue/11802 hopefully we can help them come up with a good solution.
<@nirik:matrix.scrye.com>
21:23:32
(basically they want to rebuild a thing in rhel with different compile time options)
<@conan_kudo:matrix.org>
21:23:54
this seems like something that doesn't belong in EPEL
<@carlwgeorge:matrix.org>
21:24:07
i spoke with one of the maintainers from intel, ali, and he seems on board with doing it in hyperscale
<@carlwgeorge:matrix.org>
21:24:37
it's the rhel maintainer pushing to do it in epel, and it's not clear to me why
<@conan_kudo:matrix.org>
21:24:39
yeah, Ali is already a member of CentOS Hyperscale and manages the -intel repo where this probably makes sense to be built
<@nirik:matrix.scrye.com>
21:24:39
great. I was hoping we could come up with a solution for them, not just saying no. ;)
<@conan_kudo:matrix.org>
21:24:59
Davide Cavalca and I can reach out to him to get things going there
<@carlwgeorge:matrix.org>
21:25:27
shipping rhel packages rebuild with different configure flags has never been in scope for epel, as far as i'm aware
<@nirik:matrix.scrye.com>
21:25:33
Cool. Glad it's all looking to work out.
<@conan_kudo:matrix.org>
21:26:06
if they can be done in a non-conflicting way, it's probably fine, but I don't know if that's even possible with this in a safe way, since it's loaded by OpenSSL as a plugin
<@conan_kudo:matrix.org>
21:26:17
and plugins are... weird
<@conan_kudo:matrix.org>
21:26:29
especially OpenSSL plugins
<@dherrera:fedora.im>
21:28:02
even if it's fine in that case, it would still would need to go through Fedora instead of requesting a special branch for EPEL IMHO
<@nirik:matrix.scrye.com>
21:28:11
Second item: I created the archives for epel-8.8 and epel-9.2... delayed after the fact. We should try and make sure we track that and make them on the release day or something. Or at least someone files a ticket to track the work or something (I don't think it can easily be automated, but I could be wrong). If it's within 10 days it should be not too hard. Longer requires digging into snapshots and is not much fun.
<@conan_kudo:matrix.org>
21:28:54
doesn't RHEL have some kind of public API or message bus thing that we could subscribe to?
<@conan_kudo:matrix.org>
21:29:09
that way we could automate it and do it at the right moment
<@farribeiro:matrix.org>
21:29:17
!hi
<@zodbot:fedora.im>
21:29:19
Fábio Ribeiro (farribeiro) - he / him / his
<@nirik:matrix.scrye.com>
21:29:23
not that I know of
<@tdawson:fedora.im>
21:29:37
I've tried to create a ticket, when Stephen J Smoogen doesn't beat me to it. But this release caught me by surprise and was at a really busy time.
<@dherrera:fedora.im>
21:29:43
even if it's fine in that case, it would still need to go through Fedora instead of requesting a special branch for EPEL IMHO
<@rcallicotte:fedora.im>
21:29:50
!hi
<@zodbot:fedora.im>
21:29:51
Robby Callicotte (rcallicotte) - he / him / his
<@carlwgeorge:matrix.org>
21:30:05
sounds like we need that epel releng sop sooner than epel10
<@tdawson:fedora.im>
21:30:08
Hi Robby Callicotte
<@rcallicotte:fedora.im>
21:30:18
hello all
<@nirik:matrix.scrye.com>
21:30:43
well, not an sop so much as a 'someone tracking releases and making sure it happens/is requested'
<@tdawson:fedora.im>
21:31:01
Is that a thing we can create? an "EPEL releng SOP"? if so, where?
<@nirik:matrix.scrye.com>
21:31:02
but I guess that could be part of a sop
<@tdawson:fedora.im>
21:31:41
Is there a Fedora SOP?
<@carlwgeorge:matrix.org>
21:31:44
probably alongside https://docs.fedoraproject.org/en-US/infra/release_guide/sop_mass_branching/
<@nirik:matrix.scrye.com>
21:31:52
well, there is a archive sop, but that is useless by itself.
<@nirik:matrix.scrye.com>
21:33:16
the existance of a sop doesn't mean someone knows when to follow it and do those things...
<@nirik:matrix.scrye.com>
21:33:29
especially when release dates are not exactly known in advance
<@conan_kudo:matrix.org>
21:33:38
maybe we should ask the RHELeng folks for some help here?
<@carlwgeorge:matrix.org>
21:33:42
i could probably peek at the rhel schedule and set up calendar reminders, i just can't share those publicly
<@carlwgeorge:matrix.org>
21:33:54
i could probably peek at the rhel schedule and set up calendar reminders to myself, i just can't share those publicly
<@carlwgeorge:matrix.org>
21:34:11
it's not ideal, but better than nothing
<@conan_kudo:matrix.org>
21:34:21
I don't think it's unreasonable to maybe have some kind of message fired into our message bus from RHEL folks to trigger the process
<@conan_kudo:matrix.org>
21:34:48
or have some kind of public message thing we could subscribe to
<@conan_kudo:matrix.org>
21:34:55
something of some sort
<@conan_kudo:matrix.org>
21:35:32
in general, we're trying to have less human-centric processes because we have too many of them as it is :(
<@nirik:matrix.scrye.com>
21:35:40
That might be nice, but I'd settle for just a ticket when a release happens. ;)
<@tdawson:fedora.im>
21:35:50
Have a RHEL machine try to update redhat-release every six hours. And when it succeeds, trigger all of our SOP stuff.
<@conan_kudo:matrix.org>
21:36:02
machine formatted ticket? :P
<@carlwgeorge:matrix.org>
21:36:09
* a rhel machine for each major version
<@conan_kudo:matrix.org>
21:36:26
that is actually not a terrible idea, it could even be done with a container, right? since ubi and stuff is public
<@conan_kudo:matrix.org>
21:37:07
we also have packaged rhel-ubi repo configs in rpmdistro-repoquery that could be used simply to check whether the redhat-release version has changed
<@nirik:matrix.scrye.com>
21:37:14
yes, lets build the best rube roldberg we can. :(
<@nirik:matrix.scrye.com>
21:38:22
anyhow, we don't need to solve it today, but we should try and do better next release(s)...
<@carlwgeorge:matrix.org>
21:38:48
possibly a releng toddler?
<@conan_kudo:matrix.org>
21:39:17
``` ngompa@fedora~> rpmdistro-repoquery rhel-ubi 9 --provides redhat-release Last metadata expiration check: 0:00:36 ago on Wed 29 Nov 2023 04:38:24 PM EST. base-module(platform:el9) config(redhat-release) = 9.3-0.5.el9 redhat-release = 9.3-0.5.el9 redhat-release(x86-64) = 9.3-0.5.el9 redhat-release-client redhat-release-computenode redhat-release-server redhat-release-workstation system-release = 9.3-0.5.el9 system-release(releasever) = 9 ```
<@conan_kudo:matrix.org>
21:40:10
``` ngompa@fedora ~> rpmdistro-repoquery rhel-ubi 9 --qf "%{VERSION}" redhat-release Last metadata expiration check: 0:01:34 ago on Wed 29 Nov 2023 04:38:24 PM EST. 9.3 ```
<@conan_kudo:matrix.org>
21:40:38
``` ngompa@fedora ~> rpmdistro-repoquery rhel-ubi 9 -q --qf "%{VERSION}" redhat-release 9.3 ```
<@conan_kudo:matrix.org>
21:41:12
it's not much of a rube goldberg machine :)
<@pgreco:fedora.im>
21:41:39
it doesn't even need to trigger the task, just send an email to us
<@pgreco:fedora.im>
21:41:59
or ideally, file the issue 🤔
<@davide:cavalca.name>
21:42:11
Yeah this seems reasonable
<@tdawson:fedora.im>
21:42:23
If it can file the issue, that would be best.
<@robert:fedora.im>
21:42:34
!hi
<@zodbot:fedora.im>
21:42:36
Robert Scheck (robert)
<@conan_kudo:matrix.org>
21:42:44
it works for 8 too :)
<@nirik:matrix.scrye.com>
21:43:14
yeah, I'd be ok with it notifying or filing an issue.
<@conan_kudo:matrix.org>
21:43:28
since 8 and 9 don't release at the same time, they'll need to be separate toddlers so things don't get too goofy
<@nirik:matrix.scrye.com>
21:43:34
anyhow, it needs all the work to set it up and check things and such. but we don't need to do it right here. ;)
<@conan_kudo:matrix.org>
21:43:47
at least we know how we can :)
<@tdawson:fedora.im>
21:45:00
I get enough random other indications (all the bugs I opened get closed) that I know the release has happened, it's the remembering to create the issue that I have a hard time remembering. So, scripting that part would be very nice.
<@tdawson:fedora.im>
21:45:40
Before we get too into the weeds of this, does anyone else have anything for Open Floor?
<@tdawson:fedora.im>
21:47:19
We can end early if nobody has anything else.
<@conan_kudo:matrix.org>
21:48:03
I don't have anything.
<@tdawson:fedora.im>
21:48:17
OK, let's call it a meeting.
<@tdawson:fedora.im>
21:48:57
Thank you all for coming and for the good discussion. And thank you all for all your work on EPEL and for the EPEL community.
<@tdawson:fedora.im>
21:49:04
Talk to you next week.
<@pgreco:fedora.im>
21:49:06
see you next week, bye!
<@tdawson:fedora.im>
21:49:24
!endmeeting
<@tdawson:fedora.im>
21:49:55
!endmeeting
<@jonathanspw:fedora.im>
21:50:23
bot says no!
<@nirik:matrix.scrye.com>
21:50:34
thanks Troy Dawson
<@tdawson:fedora.im>
21:50:43
Not again. 🙃
<@pgreco:fedora.im>
21:50:52
!endmeeting