21:00:33 <tdawson> #startmeeting EPEL (2020-08-07)
21:00:33 <zodbot> Meeting started Fri Aug  7 21:00:33 2020 UTC.
21:00:33 <zodbot> This meeting is logged and archived in a public location.
21:00:33 <zodbot> The chair is tdawson. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:34 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
21:00:34 <zodbot> The meeting name has been set to 'epel_(2020-08-07)'
21:00:35 <tdawson> #meetingname epel
21:00:35 <zodbot> The meeting name has been set to 'epel'
21:00:36 <tdawson> #chair nirik tdawson bstinson Evolution pgreco carlwgeorge
21:00:36 <zodbot> Current chairs: Evolution bstinson carlwgeorge nirik pgreco tdawson
21:00:38 <tdawson> #topic aloha
21:00:45 <pgreco> hi!
21:01:05 <carlwgeorge> howdy y'all
21:01:21 <tdawson> Hi pgreco
21:01:23 <tdawson> Hi carlwgeorge
21:02:38 * nirik is only sort of here, dealing with fires
21:02:58 <pgreco> nirik, "this is fine" ;)
21:03:07 * tdawson hands nirik a fire extinguisher, hoping it helps.
21:05:32 <tdawson> #topic Old Business
21:05:33 <tdawson> .epel 105
21:05:34 <zodbot> tdawson: Issue #105: Implement rhel module name/stream avoidance - epel - Pagure.io - https://pagure.io/epel/issue/105
21:05:35 <tdawson> Status on investigation of creating bodhi tests
21:05:51 <tdawson> I don't have much of a status this week.
21:06:09 <tdawson> I wasn't able to work on it at all.
21:07:13 <tdawson> I looked at my schedule and it doesn't look like I will be able to next week eaither.  But I have marked off August 17 to devote as much time as I can to getting it progressed.
21:08:33 <tdawson> Other old business is epel-playground
21:08:39 <pgreco> well, we've been months without it, and we're aware of the issue
21:09:12 <tdawson> Yep, so I think we'll be ok, for the time being.
21:10:20 <tdawson> I sent out an email about the epel-playground discussion ... but didn't get very many responses.
21:11:40 <tdawson> Did we want to have more discussion here?  Or via email?
21:12:00 <tdawson> Since nirik is fighting fires, I don't know if he'll be able to contribute much during this meeting.
21:12:58 <pgreco> sorry, phone rang (fires here too)
21:13:18 <tdawson> Are we ok tabling the epel-playground discussion for another couple weeks, and having the disucssion via email?
21:13:38 <pgreco> yeah, I linked nirik's idea, sort of agreed last week
21:13:51 <pgreco> but we need info on what it would imply work-wise
21:13:58 <tdawson> True
21:14:18 * carlwgeorge checks his epel-devel folder in email
21:15:00 <pgreco> my current thought is, playground inherits from epel8, and can be used to build extra (newer) things, and ideally, uses stream instead of 8.x
21:15:27 <tdawson> pgreco: That sorta sounds like a merge of both proposals.
21:15:33 <pgreco> yes
21:16:19 <carlwgeorge> the problem i fear is that if someone is using rhel but wants to consume playground content, that doesn't install without a library rebase coming in the next point release (stream)
21:16:49 <pgreco> yeap, it is not a silver bullet
21:17:39 <carlwgeorge> i'll finish reading the email and offer a more substantial reply on list
21:17:40 <pgreco> what I'm trying to avoid is the "epel package X can't be used in stream"
21:18:23 <pgreco> for RHEL, if you can't use playground, you can use epel8
21:18:57 <pgreco> so everybody has something they can use
21:19:34 <carlwgeorge> but that leaves out the advertised use case of "add this thing to playground that will never be stable enough for main epel"
21:20:22 <pgreco> yes, I said it and I'm not even convinced
21:20:36 <tdawson> *laughs*
21:20:38 <tdawson> :)
21:20:50 <carlwgeorge> does anyone use it like that that we know of?
21:21:22 <pgreco> I'm not aware of anybody using it that way
21:22:01 <pgreco> it could still be used that way, but with caveats...
21:22:46 <pgreco> I don't know, tdawson please move on or I'll continue rambling :P
21:22:50 <carlwgeorge> where did we land on the "breaking changes at point release time" rule for epel?  that could factor into this.
21:22:58 <carlwgeorge> s/rule/idea/
21:23:53 <tdawson> carlwgeorge: Meaning RHEL has breaking changes?  or EPEL has breaking changes?
21:23:54 <carlwgeorge> i feel like that would make the stream on playground use case a lot more attractive, i.e. at point release time packages are allowed to merge epel8-playground to epel8
21:24:22 <carlwgeorge> tdawson: epel breaking changes, but only at point release time
21:24:38 <carlwgeorge> i remember something about point release archives that smooge was doing
21:25:12 <tdawson> I don't think we've had much of a discussion about that.  I thought it was still the same as before ... make sure you tell everyone about it and unless someone screams no, do it carefully.
21:26:46 <tdawson> Basically the same policy EPEL has always had.
21:27:16 <carlwgeorge> we can move on, i'll work up a list reply with the rest of my thoughts
21:27:38 <tdawson> OK, moving on.
21:27:46 <tdawson> #info EPEL-6 is End of Life in 2020-11. It will be moved to archives in 2020-12
21:27:48 <tdawson> #info THIS IS NOT A DRILL.
21:28:04 <tdawson> Three months.
21:28:20 <tdawson> #topic EPEL-7
21:28:26 <tdawson> Anything for EPEL7 ?
21:29:12 <tdawson> I saw an email about SCL usage on EPEL7.  I didn't touch it, cuz I don't know enough about it.
21:29:26 <tdawson> but other than that, I don't remember anything for EPEL7
21:30:06 <pgreco> what about scl?
21:30:21 <nirik> well, I ran into something fun related there
21:30:28 <pgreco> I thought we had them all enabled, at least all that we can
21:31:15 <tdawson> I looks like someone wanted to build a package in EPEL7, using the SCL packages from CentOS7
21:31:45 <tdawson> Although actually ... now that I re-read it ... they really just wanted to use a newer nodejs
21:32:13 <nirik> newer mock has a 'bootstrap mode' where it makes a small chroot with the target os and then uses that to install / create the chroot for building in. We don't currently use it in fedora koji, but I was trying to enable it with the upgrades this last week. It fails on epel7 because... we have devtoolset enabled and it somehow pucks up a library it needs from that to run yum, and then when it tries to run yum it fails because the library is in
21:32:13 <nirik> /opt/whatever
21:32:55 <tdawson> nirik: Oh ... fun
21:33:49 <tdawson> But, I would think that it would pick up that library during normal mock mode as well ... that's a bit strange
21:34:04 <pgreco> yeah, the bootstrap mode is useful when you want to build fedora or el8 packages with newer spec tags in el7
21:34:32 <nirik> yeah, and after we upgrade builders to f33... they will be using a rpm that has sqlite for rpm db.
21:34:39 <nirik> which epel6/7 will... not work with
21:34:57 <pgreco> is there an overlap there?
21:35:03 <nirik> tdawson: it uses dnf from the host (f33) so it never calls yum...
21:35:16 <tdawson> Ohhh
21:35:46 <nirik> also right now all the fedora images have a rpm bdb... and get converted by rpm on first use, which is not ideal.
21:35:53 <nirik> (rawhide images)
21:38:04 <tdawson> Anything else for EPEL7?
21:38:19 <tdawson> Not that I don't like to talk some, but I do have one thing for the open floor
21:38:36 <nirik> nothing more from me. I might have a workaround for the above, need to test
21:38:41 <tdawson> #topic EPEL-8
21:39:37 <tdawson> I don't have anything for EPEL8 ... does anyone else?
21:40:38 <pgreco> nope
21:40:51 <tdawson> #topic General Issues / Open Floor
21:41:18 <tdawson> I'm not going to be available next week.
21:41:46 <tdawson> I know a few others won't be either.
21:42:05 <tdawson> But I won't have any internet access (camping) so it's not even an option for me.
21:42:19 <tdawson> Does someone else want to run the meeting?  Or should we cancel next weeks meeting?
21:42:47 <nirik> lets cancel
21:42:52 * nirik wont be here either
21:43:17 <pgreco> yeap, 2 out is more than enough to cancel
21:43:37 <tdawson> Sounds good.  Meeting canceled.
21:43:49 <tdawson> Next meeting will be August 21.
21:44:28 <tdawson> That's all I had for open floor.  Does anyone have anything else?
21:45:13 <pgreco> did we run out of time last week and carlwgeorge wanted something for open floor
21:45:15 <pgreco> ?
21:46:09 <tdawson> That's right ... I forgot about that.  carlwgeorge do you have anything?
21:46:09 <carlwgeorge> my thoughts are on epel9
21:47:02 <carlwgeorge> i'd like to get epel9 started once c9s comes out, and not wait until rhel9 ga
21:47:22 <tdawson> Ahh ... good ponit.
21:47:31 <pgreco> do you have a timeline for it? even if you can't say it?
21:48:47 <tdawson> I know the ending time, and I'm pretty sure carlwgeorge knows that too.
21:49:24 <tdawson> But more important than the timeline, is the workflow of RHEL9.
21:49:54 <pgreco> how so?
21:50:00 <tdawson> It will go Fedora -> ELN (already happening, though not complete) -> CentOS 9 Stream -> RHEL9
21:50:12 <pgreco> oh, that, yes
21:50:17 <tdawson> Fedora / ELN and CentOS 9 Stream, will all be completely open
21:50:26 <nirik> I would expect...
21:50:29 <pgreco> we can start bootstrapping things at any point in that cycle
21:50:33 <carlwgeorge> i think i can say that c9s will be publically available after f34 release, and before rhel9 ga release
21:50:38 <nirik> sometime before rhel9 that ELN will repoint to RHEL10
21:50:54 <tdawson> So, in theory, we could take CentOS 9 Stream, and start getting EPEL9 ready.
21:51:17 <carlwgeorge> we all know that epel availability is a big deal for adoption
21:51:20 <pgreco> or even a good enough snapshot of eln
21:52:01 <tdawson> pgreco: ELN will have too much ... cruft in it.  I'd feel better with the CentOS 9 Stream.
21:52:20 <tdawson> But ... ELN is built on the same servers as EPEL9 will be ... so that would always be nice. :)
21:52:56 <pgreco> yeah, it will be a nice experiment
21:53:19 <tdawson> Plus, ELN is updated constantly.  Just last week we implemented the rebuild scripts, so that whenever a rawhide build finishes successfully, an ELN build kicks off, if that package is in the list.
21:53:48 <pgreco> that is why I said snapshot
21:53:50 <tdawson> So ELN will have alot of churn ... not something you want to point EPEL at.
21:54:26 <pgreco> so when RH branches off ELN, we can save that and use it to star bootstrapping things we may need
21:54:31 <nirik> part of the problem is that EPEL is not a defined set
21:54:48 <pgreco> that will be a few days/months until c9s
21:55:00 <tdawson> nirik: I can think of two definitions of what you said, what do you mean?
21:55:12 <tdawson> pgreco: months
21:55:57 <nirik> while we can setup/branch/build/work on some packages, we can't simply mass branch everything from epel8 and blindly rebuild it... people need to sign up to maintain. So the best we can do is setup things so those people who want to can build what they want.
21:56:26 <nirik> but having that base early would be nice... we didn't have it for 8 due to all the changes.
21:56:39 <tdawson> nirik: Yep ... and Yep
21:56:51 <pgreco> that's what I was going for with the snapshot of eln
21:57:31 <pgreco> we can start seeing the picture, even before stream
21:57:38 <pgreco> *c9s
21:58:00 <tdawson> carlwgeorge: Let's find out how much we can talk about dates, and then bring this up next meeting.  I know I can talk about workflows, just don't know about dates of things.
21:58:27 <pgreco> I don't know how much of that you can share
21:58:56 <tdawson> pgreco: Well, this time, it's totally different than RHEL8.  We've been told everything out in the open.
21:59:14 <pgreco> that's good news
21:59:28 <tdawson> pgreco: But, we're talking still over a month before we even have a list of what goes in RHEL9 figured out.
21:59:51 <pgreco> yeah, absolutely, but it is nice to have an idea
22:00:08 <tdawson> And luckily ... this time, I don't have to create the initial list
22:00:42 <tdawson> OK ... so technically, I did ... but we just pulled that list out, and it's all on the maintainers shoulders now.
22:00:51 <tdawson> Oh ... times up.
22:01:00 <pgreco> nice meeting everybody!
22:01:05 <pgreco> thanks tdawson
22:01:07 <nirik> thanks everyone
22:01:13 <tdawson> I'll put this on the agenda for two weeks from now.
22:01:19 <tdawson> Thank you everyone for coming.
22:01:26 <tdawson> #endmeeting