17:59:43 <smooge> #startmeeting EPEL (2020-02-05)
17:59:43 <smooge> #meetingname epel
17:59:43 <smooge> #topic aloha
17:59:43 <smooge> #chair nirik smooge tdawson bstinson Evolution pgreco sgallagh merlinm
17:59:43 <zodbot> Meeting started Wed Feb  5 17:59:43 2020 UTC.
17:59:43 <zodbot> This meeting is logged and archived in a public location.
17:59:43 <zodbot> The chair is smooge. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:59:43 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
17:59:43 <zodbot> The meeting name has been set to 'epel_(2020-02-05)'
17:59:43 <zodbot> The meeting name has been set to 'epel'
17:59:43 <zodbot> Current chairs: Evolution bstinson merlinm nirik pgreco sgallagh smooge tdawson
17:59:43 <smooge> #info Meeting is run from https://board.net/p/epel
18:00:40 <pgreco> hey everybody!
18:00:50 <tdawson> Howdy
18:00:58 <sgallagh> .hello2
18:00:58 <zodbot> sgallagh: sgallagh 'Stephen Gallagher' <sgallagh@redhat.com>
18:01:20 <smooge> hello
18:01:21 * sgallagh really needs to put this meeting on his calendar; it keeps being a surprise for me
18:01:40 <smooge> sgallagh, <knock><knock>
18:01:50 <sgallagh> Who's there?
18:01:58 <smooge> LAND SHARK
18:02:11 <sgallagh> Land shark... who?
18:02:54 <smooge> I think that was the punch line but it has been 30 years since I watched Saturday Night Live
18:03:14 <nirik> morning
18:03:32 <nirik> candygram
18:03:45 <smooge> #topic EPEL-6
18:03:45 <smooge> #info EPEL-6 is End of Life in 2020-11. It will be moved to archives in 2020-12
18:03:45 <smooge> #info THIS IS NOT A DRILL.
18:04:14 <smooge> any other EPEL-6 items?
18:04:50 <smooge> #topic EPEL-7
18:05:16 <smooge> ok anyone have epel7 items?
18:06:21 <smooge> alright then..
18:06:25 <smooge> #topic EPEL-8
18:06:25 <smooge> #info modularity is enabled.
18:06:58 <sgallagh> I've been on other stuff for a few weeks, but I was just sitting down to try to solve the missing -devel problem when I got pinged for this meeting
18:07:08 <smooge> understood
18:07:14 <sgallagh> (Weird bit of synchronicity there)
18:08:04 <nirik> we should decide and document what epel agrees to not overlap with. ;)
18:08:07 <smooge> so I would like to get some feedback from the CentOS people bstinson and such as I believe they have a solution they are looking at also
18:08:08 <nirik> there's a thread on the list
18:09:29 <smooge> so going from contyk's(?) email... the libssh2 will have to be a module so it can install over the libssh2 items in RHEL-8
18:09:36 <sgallagh> I've been operating on the assumption that modular EPEL 8 is allowed to overlap with anything so long as it's not a default stream
18:09:46 <sgallagh> (And we don't allow default streams in EPEL8 presently)
18:10:10 <bstinson> we're starting with the non-modular case in CentOS
18:10:11 <smooge> the problem seems that the libssh2 items were in a default stream but aren't(?) now
18:11:05 <sgallagh> smooge: That shouldn't be permissible in RHEL. What happened there?
18:11:05 <tdawson> No, it still is, from everything I've tested.
18:12:27 <smooge> so I don't know.. the module (virt people) author says it shouldn't be there anymore.
18:13:02 <smooge> it was only there because it was needed for one thing and it was causing all kinds of headaches so it was supposed to be removed from 8.0 -> 8.1
18:13:13 <smooge> however as tdawson says.. it is still there
18:13:47 <smooge> so ¯\_(ツ)_/¯
18:14:52 <tdawson> Maybe they (RHEL release) need to work on their module dropping a package skills.  Maybe they tried to remove it.
18:15:27 <sgallagh> Well, right now there's no mechanism in DNF for allowing that, so it may have just been postponed anyway.
18:15:44 <sgallagh> Or at least for dropping it out of a module and into the non-modular content
18:15:47 * sgallagh shrugs
18:17:03 <smooge> so looking at our systems
18:17:33 <smooge> it is in  ./virt-devel:rhel:8000020190828150510:f8e95b4e:x86_64/libssh2-1.8.0-8.module+el8.0.0+4084+cceb9f44.1.i686.rpm but it is not in any later module
18:18:57 <smooge> so the way grobisplitter works.. it is not available for koji to use
18:19:34 <smooge> tdawson, so it looks like it is removed but because of dnf and the fact that rhn sees all old modules.. it is still there?
18:20:00 <tdawson> OK, so it really has been removed from the module, so epel8 can't see it, but if you are running RHEL8 .. you still see it.  ugg ... weird.
18:20:15 <smooge> yep.
18:20:30 <sgallagh> The fact that RHEL 8 sees it is a bug
18:20:37 <tdawson> OK, well, that explains why I can still see it in my tests.
18:21:06 <smooge> contyk(?) explained it in his email that it is how dnf deals with modules and it is on a roadmap to fix someday
18:21:30 <sgallagh> "Someday" is "targeted at 8.3", last I heard.
18:21:45 <sgallagh> (The previous statement should not be construed as making any promises)
18:21:59 <smooge> so for the time being.. we could come up with some sort of fix where we build this somewhere?
18:23:03 <nirik> so, the proposed rule would be: no overlapping bare rpms, no overlapping things in default modules?
18:23:05 <smooge> nirik, if we figure out how to override pdc?/src?/whatever? to allow a branch we could try a build in EPEL-8 and see what else might not work
18:23:32 <nirik> smooge: well, we need to know the rules to adjust the script that makes that json file.
18:23:37 <smooge> agreed
18:23:52 <nirik> right now it includes all rpms from everything modular and not
18:24:18 <smooge> I would go with no overlapping bare rpms, no overlapping things in default 8.latest modules
18:24:42 <smooge> since we really only care about X.latest for any RHEL release
18:25:09 <smooge> does that sound workable?
18:25:18 <nirik> so it would have to somehow know what modules are 'default'. ;(
18:26:37 <smooge> and I realized that virt-devel was not a default module (don't know if it is now)
18:26:39 <sgallagh> nirik: That's easily retrieved from the metadata
18:27:05 <nirik> sure, but that script would have to grow that ability.
18:27:09 <smooge> what is the syntax for a proposal in zodbot? #propose #proposal
18:27:15 <nirik> right now it just gets all the rpms from the metadata and puts them in a json file
18:27:28 <nirik> smooge: whatever you like, it doesn't have any proposal function. ;)
18:28:32 <smooge> #propose EPEL does not allow for EPEL packages to overlap rpms which are either bare or in enabled modules in the latest 8 sub release.
18:29:32 <smooge> the metadata it would use to determine this would be in the metadata from rhel/rhel8/koji/latest/x86_64/
18:29:48 <smooge> which just has the packages which we would build against
18:30:38 <nirik> enabled? you mean default?
18:31:13 <smooge> well I realized that querying that data you would not see any modular data
18:31:40 <smooge> you would just see the rpms that we build against and wouldn't want to conflict with
18:32:03 <nirik> but some of those are in non default modules that we do want to allow overlaping with?
18:32:34 <smooge> I was trying to come up with something we could do without too much 'growing new things' we have little time to do
18:33:01 <smooge> if we can grow something to learn what is default and isn't etc.. then we could go with just the default modules
18:33:14 <sgallagh> FTR, "list all the default streams" is a single API call in libmodulemd once the data is loaded
18:34:06 <smooge> sgallagh, I currently would not be able to look at adding any functionality until July with my current schedule... so it would require someone else finding time to do so.
18:34:31 <smooge> I am guessing Kevin is in a similar boat
18:34:37 * sgallagh nods
18:34:43 <nirik> well, this is the rhel2json script pingou wrote... so probibly he would be the one to check with on changes..
18:34:57 <sgallagh> If one of you can point me in that direction, I could try to do it.
18:36:03 <nirik> https://infrastructure.fedoraproject.org/cgit/ansible.git/tree/roles/repo2json in ansible...
18:36:08 <nirik> not sure if it has another upstream.
18:36:55 <nirik> it looks like it uses the modular repos directly before grobisplitting?
18:36:59 <smooge> thanks sgallagh I also didn't want to commit you to working on it :)
18:37:36 <sgallagh> I'm not making promises, but I'll have a look
18:38:00 <smooge> understood completely
18:38:17 <smooge> ok I have one more item from the discussions if we are done here?
18:39:12 <nirik> sure
18:39:14 <smooge> #topic Un-responsive maintainers in EPEL.
18:39:53 <sgallagh> Grab the pitchforks!
18:40:03 <smooge> this came up on devel list from someone who got tickets from EPEL assigned to them because they weren't getting dealt with
18:40:38 <smooge> part of this comes from the pkgdb changes of 3-4 years ago, and some others come from lack of rules on our part
18:40:55 <nirik> well, it's hard to know with the info we have sadly
18:41:16 <smooge> I am not sure exactly what to do about it at the moment (brain empty) but wanted to say it was at least brought up here versus ignored
18:41:47 <smooge> I think we will need a policy that can be followed by security and other groups when dealing with packages not getting fixes
18:41:59 <nirik> take package foo... has 3 co-maintainers and fedora and epel branches. maintainers 1 and 2 work on fedora, 3 on epel. Maintainer 3 disappears... how can we know the epel branches are not being looked after? ;(
18:42:25 <smooge> I do not have time to write said policy and as nirik says.. this is complicated and not easy to deal with.
18:42:54 <nirik> ideally we can bring back seperate epel/fedora maintainers at least... that would help...
18:43:23 <smooge> so input from the community is wanted. I will put more in a mail later today and if we can figure out what and where that would live I would love it
18:44:33 <smooge> any other input here because someone asked about a related topic when I started on this
18:45:28 <smooge> #topic What is the policy and steps for someone on request-for-branch tickets not getting responses?
18:45:49 <smooge> skywalker, the floor is yours
18:46:54 <skywalker> sure, thanks smooge
18:47:12 <skywalker> so, i have a few tickets on releng regarding claming ownership of packages from people that don't respond. the tickets have been sitting there for 10 days now
18:47:31 <skywalker> is there anything we could do about them? am i doing anything wrong?
18:47:55 <nirik> skywalker: releng is basically just 3 people... and we were all traveling and doing the mass rebuild...
18:48:05 <skywalker> the goal is just to have those packages on EPEL 8. i am willing to have their ownership for EPEL or at least have commit/push permissions
18:48:23 <nirik> the current maintainers haven't replied to bugs?
18:48:31 <skywalker> oh, i see. i didn't know that, nirik. it's alright then
18:48:45 <skywalker> they have not. i filed most of those bugs back in 2019. in sep/oct
18:48:46 <nirik> sorry it's slow, but.... sometimes it is. ;(
18:48:55 <nirik> wow... thats a while. ;)
18:49:40 <skywalker> yes, right? but that's ok. i can wait :) i see that the releng team have ~130 tickets right now
18:50:32 <nirik> I thought it was 170? oh... 165 looks like
18:50:43 <nirik> anyhow, I can try and get those this week for you....
18:51:14 <skywalker> thanks, if i can help somehow, please let me know
18:52:18 <smooge> ok thanks skamath
18:52:25 <smooge> ok thanks skywalker <tab complete>
18:52:33 <skywalker> :) thank you guys
18:52:39 <smooge> I have another meeting in 5 so I am going to put htis in into end of meeting
18:52:43 <smooge> Any other replies?
18:52:48 <smooge> requests
18:52:52 * nirik too
18:52:53 <smooge> returns?
18:53:01 <smooge> #endmeeting