16:00:26 <mboddu> #startmeeting RELENG (2019-10-10)
16:00:26 <zodbot> Meeting started Wed Oct  9 16:00:26 2019 UTC.
16:00:26 <zodbot> This meeting is logged and archived in a public location.
16:00:26 <zodbot> The chair is mboddu. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:26 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:00:26 <zodbot> The meeting name has been set to 'releng_(2019-10-10)'
16:00:27 <mboddu> #meetingname releng
16:00:27 <zodbot> The meeting name has been set to 'releng'
16:00:27 <mboddu> #chair nirik sharkcz pbrobinson pingou mboddu dustymabe ksinny jednorozec
16:00:27 <mboddu> #topic init process
16:00:27 <zodbot> Current chairs: dustymabe jednorozec ksinny mboddu nirik pbrobinson pingou sharkcz
16:01:34 <smooge> .here
16:03:21 * mboddu waves at smooge
16:07:08 <mboddu> Okay, lets get started
16:07:23 <mboddu> #topic Ursa Prime
16:07:43 <mboddu> # link https://pagure.io/releng/issue/8879
16:08:08 <mboddu> #link https://pagure.io/releng/issue/8879
16:08:12 <mboddu> #link https://pagure.io/releng/issue/8880
16:08:15 <mboddu> #link https://pagure.io/releng/issue/888
16:08:17 <mboddu> #undo
16:08:17 <zodbot> Removing item from minutes: <MeetBot.items.Link object at 0x7f938c64b350>
16:08:21 <mboddu> #link https://pagure.io/releng/issue/8881
16:08:40 <mboddu> smooge: So, seems like we have to enable it for only epel8 for now?
16:08:50 <mboddu> https://pagure.io/releng/issue/8879#comment-603556
16:08:57 * nirik is sort of here, but needs coffee
16:09:40 <nirik> I'm really not clear on what if anything needs doing for epel8... we should talk more with sgallagh and/or contky
16:10:16 * nirik goes to get coffee, back in a min
16:12:56 <smooge> mboddu, I am in the same boat as nirik on this.. I think it is very handwavey what is needed
16:13:04 <mboddu> As I understand it, it will help us move away from grobisplitter and achieve rpm building using modules
16:14:19 <smooge> i guess so. I haven't done much with it
16:14:21 <mboddu> smooge: Okay, because of my understanding ^ (related to grobisplitter), I pinged you as well
16:14:28 <mboddu> Okay
16:14:34 <smooge> this was my first hearing of it
16:15:28 <nirik> we do not use ursa major in epel8 either... but yeah, I am not sure what needs to be done, we need to talk more about it
16:18:27 <mboddu> Right
16:20:32 <mboddu> Now I think about it, how is it different from ursa major?
16:21:36 <nirik> ursa major is a script that only acts on koji... it finds the modules that should be in the buildroot and tags them into it... also injecting a package of conflicts/adjustments so everything is working right
16:22:13 <nirik> epel8 takes the rhel repos, runs gobisplitter on them, puts them back in one non modular flat repo and makes that an external repo for koji to use as the buildroot.
16:22:15 <sgallagh> Sorry, missed the ping above
16:22:55 <sgallagh> nirik: If that's true, it's going to break eventually. They need to have the modular metadata in the repo.
16:23:06 * sgallagh thought he added that feature to grobisplitter... was that a fever-dream?
16:23:14 <nirik> we tried that and it failed I thought?
16:23:49 <sgallagh> I don't remember now.
16:25:10 <sgallagh> https://github.com/fedora-modularity/GrobiSplitter/pull/4
16:25:18 <sgallagh> Yeah, that has been there for a while.
16:25:45 <nirik> but are we using it? ;)
16:25:47 <nirik> smooge: ^
16:26:27 <smooge> no. i have not added the flag
16:27:01 <sgallagh> What flag?
16:27:14 <sgallagh> The code isn’t condition so far as I’m aware.
16:27:34 <smooge> sorry I need to context switch here from anotehr fire
16:27:42 <smooge> give me a minute to look at what I have
16:27:58 <sgallagh> Or do you mean that recreating the combined repo drops the metadata too?
16:28:30 <smooge> well it does but I am not seeing it being generated
16:28:49 <sgallagh> Too many pronouns, please rephrase.
16:29:12 <smooge> if I understood what I was writing in the first place I wouldn't have used pronouns
16:29:33 <sgallagh> ...
16:29:50 <smooge> grobisplitter was updated to the latest version. It runs and splits out packages to different repositories in one tree
16:30:15 <smooge> looking in those what I am seeing is just the package there.. no metadata. So I assumed I needed to add a flag
16:30:19 <sgallagh> Each of those individual repos should be carrying the associated metadata.
16:30:35 <sgallagh> Argh
16:31:26 * sgallagh rechecks the patch
16:31:28 <smooge> after each of those small repos are done, we have to do a reposync --latest from that data to a final repo because koji gets confused at times
16:32:00 <smooge> and only wants 1 version of glibc etc while the grobisplitter would have older versions
16:32:17 <sgallagh> No, the patch definitely calls the right modifyrepo command.
16:32:21 <smooge> so let us deal with this outside of the meeting
16:32:25 <sgallagh> Okay
16:33:02 <smooge> because I am not sure what this has to do with ursa-prime which I thought replaced grobisplitter
16:33:40 <mboddu> sgallagh: Doesn't ursa-prime replace grobisplitter?
16:34:21 <sgallagh> It should, yes
16:34:33 <nirik> I'm not sure how it will.
16:35:01 <nirik> fedora: a cron job runs that does a pungi compose using koji/modules to make a modular repo
16:35:21 <nirik> epel8: we have no access to brew, so we... just use appstream and crb directly?
16:35:29 <sgallagh> oh right. We still have the split koji problem
16:35:50 <sgallagh> Sorry, I’m a little slow today
16:36:08 <nirik> naw, this stuff is all confusing... at least I have to try and remember things every time. ;(
16:36:13 <mboddu> Ah right, I forgot about that as well, and I am also slow today :(
16:36:33 <nirik> I think we did try and use the repos directly at first but that fails for some reason which I cannot now recall.
16:36:59 <smooge> it pulled in the wrong things
16:37:21 <sgallagh> Define “wrong things”?
16:37:22 <smooge> things like perl-generators from the 5.24 would get pulled in for some of the 5.26 items
16:37:41 <sgallagh> Was it a matter of the module metadata not being respected?
16:38:10 <smooge> sgallagh, other times it was ursine packages not getting pulled in even though they were there
16:38:56 <sgallagh> That doesn’t seem explicable
16:40:04 <smooge> that was the point where i was seeing the packages listed in filters in modulemd and thought that was causing it.. and we did a 'no that is used for something else.. does this other thing make it work.. ok use that'
16:41:10 <smooge> i can go through 3 month old irc logs to see what it was.. but at the time we had a 'get it out by friday' going
16:42:22 <smooge> anyway.. where are we in this meeting since I think I derailed it
16:45:02 <mboddu> Well, you didn't derail, I wanted to bring it up
16:45:32 <mboddu> So, my next question is, do we still need to add ursa-prime to epel8?
16:45:52 <sgallagh> mboddu: I think it may require some re-investigation.
16:46:11 <sgallagh> I need to understand why just using RHEL 8 BaseOS/AppStream/CRB as external repos doesn't work
16:46:27 <mboddu> Sure sgallagh
16:46:45 <mboddu> We can setup sometime and figure it out in stg
16:46:51 <mboddu> smooge: ^ wdyt?
16:47:12 <sgallagh> mboddu: Is Friday possible? I'm out tomorrow
16:47:25 <smooge> sgallagh, I was asked to set it up as one repository like it was done for all the other releases.
16:47:25 <mboddu> sgallagh: Sure, that works for me
16:48:03 <sgallagh> smooge: Who set that constraint?
16:48:09 <sgallagh> Because that... seems problematical.
16:48:58 <sgallagh> At the least, it wouldn't have worked with mergerepo_c prior to 0.13.0 IIRC
16:49:34 <smooge> It was over 4 months ago.. I just remember it was what was asked to make it as easy to debug as possible
16:50:16 <sgallagh> smooge: Before or after May 3rd?
16:50:36 <smooge> since I didn't get to actually being able to touch koji until late July
16:50:41 <smooge> I would say afterword
16:50:46 <sgallagh> hmm
16:51:30 <sgallagh> Oh, crud. RHEL 8 still has 0.11.x
16:51:32 * sgallagh facepalms
16:51:43 <smooge> I remember we were waiting for mergerepo_c updates in June and then some other items
16:52:05 <smooge> we also wanted it work on RHEL-7
16:53:58 <sgallagh> OK, looks like we need to re-examine this outside the meeting.
16:54:10 <sgallagh> I think there are things we can do to make this markedly less awful.
16:54:55 <nirik> I cannot seem to find where I tested that. ;( I know I did tho.
16:55:58 <mboddu> #info We need to re-investigate why we setup epel8 in the current way and will look into this later
16:56:08 <mboddu> With 5 min left
16:56:12 <mboddu> #topic Open Floor
16:56:27 <mboddu> #info F31 Final Freeze started yesterday, Oct 8th 2019
16:56:43 <mboddu> I am doing the manual pushes while we are in freeze
16:56:46 <mboddu> Anything else?
16:57:00 <nirik> mboddu: you want me to do the f30updates-testing? or leave them all to you/
16:57:01 <nirik> ?
16:57:10 <smooge> nirik, you tested it in early august before FLOCK. You had wanted to see if various flag settings in koji alone would build things. People started having problems with glibc and some other packages not installable in the buildroot
16:57:24 <mboddu> nirik: You mean f31-u-t? I can take care of it as well
16:57:56 <nirik> mboddu: cool.
16:58:04 <nirik> smooge: sounds likely.
16:58:28 <nirik> mboddu: one final thing... we were going to try and work on one backlog item last week? I think we didn;t? can we try again this coming week?
16:59:16 <mboddu> nirik: Yes, we could
16:59:34 <nirik> perhaps schedule a window in advance to work on it, so we don't forget. :)
16:59:39 <mboddu> nirik: Also, do you why this happened - https://pagure.io/releng/issue/8878 ?
16:59:41 <mboddu> nirik: Sure
17:00:47 <nirik> not sure. I can look.
17:01:22 <nirik> it looks latest to me...
17:01:32 <mboddu> nirik: Yes, but:
17:01:48 <mboddu> koji list-tagged f32-build glibc32
17:01:48 <mboddu> Build                                     Tag                   Built by
17:01:48 <mboddu> ----------------------------------------  --------------------  ----------------
17:01:48 <mboddu> glibc32-2.20-7.3.fc28.7                   f32-build             releng
17:02:08 <nirik> that should get untagged...
17:02:24 <nirik> I thought we moved away from special tagging for it
17:02:33 <mboddu> Even though glibc32-2.30-6.1.fc32 is tagged into f32
17:02:46 <mboddu> Ahh, might be a left over?
17:02:57 <mboddu> I will untag it from f32-build
17:03:20 <nirik> and any other glibc32s in there
17:03:28 <mboddu> Nope, thats the only one
17:03:55 <nirik> cool, that should fix it I would think
17:04:21 <mboddu> nirik: I untagged it and rerunning the repo-regen
17:04:32 <mboddu> Anyway, thats all I would think
17:04:36 <mboddu> Thanks everyone for joining
17:04:40 <mboddu> #endmeeting