16:01:00 <mboddu> #startmeeting RELENG (2019-06-27)
16:01:00 <zodbot> Meeting started Wed Jun 26 16:01:00 2019 UTC.
16:01:00 <zodbot> This meeting is logged and archived in a public location.
16:01:00 <zodbot> The chair is mboddu. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:01:00 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:01:00 <zodbot> The meeting name has been set to 'releng_(2019-06-27)'
16:01:00 <mboddu> #meetingname releng
16:01:00 <zodbot> The meeting name has been set to 'releng'
16:01:00 <mboddu> #chair nirik tyll sharkcz masta pbrobinson pingou puiterwijk maxamillion mboddu dustymabe ksinny jednorozec
16:01:00 <zodbot> Current chairs: dustymabe jednorozec ksinny masta maxamillion mboddu nirik pbrobinson pingou puiterwijk sharkcz tyll
16:01:13 <nirik> morning
16:01:28 <mboddu> nirik: Hello, how goes?
16:01:49 <nirik> mboddu: we should probibly adjust that chair line... I think a lot of those people aren''t around/don't care about releng meetings anymore. ;)
16:01:59 <nirik> not too bad, how about with you?
16:02:24 <mboddu> nirik: Yes, I will adjust the people and include only that showed up in the last few weeks
16:02:34 <mboddu> nirik: Doing okay
16:03:49 <mboddu> I have to cut this meeting short due to some work
16:04:10 <mboddu> But I was thinking and wanted to talk about few stuff before we end the meeting
16:04:18 <mboddu> #topic EPEL8
16:04:50 * dustymabe waves
16:04:52 <mboddu> #info We are at a position to build the packages in epel8
16:05:06 <dustymabe> yay
16:05:18 <nirik> (in staging)
16:05:31 <mboddu> #undo
16:05:31 <zodbot> Removing item from minutes: INFO by mboddu at 16:04:52 : We are at a position to build the packages in epel8
16:05:38 <mboddu> #info We are at a position to build the packages in epel8 in stg koji
16:05:54 <mboddu> #info mboddu is testing stg bodhi, but the composes seems to get stuck in 'Requested' state, cverna is looking into it
16:06:08 <mboddu> I couldn't figure it out as there wasn't any processes running on the box
16:06:25 <mboddu> I tried resuming, kill it and push again, but nothing helped
16:06:35 <nirik> mboddu: if we are going to do a beta period with rawhidelike composes, we also will need a pungi config for it...
16:06:44 <nirik> odd. ;(
16:06:58 <mboddu> FWIW, it worked in my previous tests (which failed due to other reasons), but now they seems stuck
16:07:46 <mboddu> nirik: I thought about rawhide like composes, but then while its being tested, we could try to fix bodhi and if it works we will just keep using it for epel8 ga
16:08:32 <mboddu> nirik: On that note, does epel8 ga will overwrite epel8 beta content? Or it will synced to a new location and changing the repo location?
16:08:59 <nirik> we could do either...
16:09:39 <mboddu> I prefer the first one if its not going to break anything and people can just start getting ga content if they were on beta
16:09:45 <nirik> the reason for wanting a rawhide like beta (nightly compose, all builds just go in) is to get the repo populated... if everything has to spend 2 weeks it will be hard to get things done
16:09:49 <mboddu> Rather than asking them to change again
16:10:23 <nirik> and it would cause a ton of overrides (basically everything)
16:10:28 <mboddu> nirik: Ahhh, okay, got it, then we could use that and for ga we will enable bodhi
16:10:31 <nirik> sure, we could do it that way, thats fine.
16:10:39 <nirik> yeah, that was my thought.
16:11:11 <mboddu> #info mboddu will work on pungi config for epel8 beta and will create the nightly compose scripts for testing
16:11:28 <mboddu> While testing, they wont be synced to mirrors until beta goes live
16:12:26 <mboddu> Anything else before I jump to next topic which is related to epel8
16:12:32 <mboddu> ?
16:15:12 * nirik is fine moving on
16:15:19 <mboddu> Okay, next thing I need to talk about epel8 is branching
16:15:44 <mboddu> Currently I am doing it which can be scripted easily, provided there is a list of packages
16:16:01 <nirik> well, one thing that someone noted this last week...
16:16:03 <mboddu> Its kinda scripted, but not great
16:16:27 <nirik> our usual process for branches/etc... already deals with pretty vast numbers of things...
16:16:44 <nirik> like there was several hundred go packages the other day
16:17:17 <nirik> so, perhaps we just tell people to use the normal process? but then they may not have it scripted to request on there end
16:17:58 <mboddu> nirik: Yes, my suggestion is updating 'fedpkg request-branch' and 'fedscm-admin' to handle epel8 branch requests
16:18:25 <mboddu> Then give them a shell script, which will help them to request epel8 branches for all the packages they maintain
16:18:35 <nirik> the other question is: from what branch does it make the new one from? master?
16:19:10 <mboddu> nirik: Thats what I am thinking, since it might still support epel8 and carry the latest patches that they might require
16:19:23 <mboddu> I will definitely wouldn't suggest epel7
16:19:41 <dustymabe> mind pinging me when we get to open floor?
16:20:00 <mboddu> dustymabe: Sure :)
16:20:32 <mboddu> By support I mean, build-able on epel8
16:20:57 * tdawson gives a +1 to branching from master.
16:21:02 <nirik> well, it's hard to say... but I guess master is the best if we don't give them the choice.
16:21:15 <nirik> with epel7 we allowed them to choose, but thats a extra pain
16:21:50 <mboddu> nirik: Well, we could give that option but as you said, might be an extra pain
16:22:05 * mboddu will see what he can do
16:24:39 <cyberpear> today, I use the F28 repo as my "EPEL 8" until the real thing is available
16:24:42 <nirik> master is likely fine, they can merge other or do whatever after.
16:26:24 <mboddu> #info We will update fedpkg and fedscm-admin to support requesting epel8 branches
16:27:37 <tdawson> cyberpear: The problem with that, is that most developer "fixes" (things that allow the pacakges to build on RHEL8) were put in F29 and F30.
16:28:10 <tdawson> Up to F28, all the fixes were mostly done by me, and there were plenty that I didn't get to.
16:28:11 * nirik looks at empty coffee cup.
16:28:12 <mboddu> cyberpear: Yeah, I would suggest f29+
16:28:29 <mboddu> nirik: Few more min and I got to run as well
16:28:33 <mboddu> #topic Open Floor
16:28:38 <mboddu> dustymabe: You are up
16:29:00 <mboddu> nirik: In the mean time quick question, does the infra openshift instance has access to /mnt/koji?
16:29:22 <nirik> mboddu: it could, I don't think we have anything that has it currently...
16:29:23 <dustymabe> mboddu: thanks
16:29:24 <mboddu> Basically thinking of creating a service to run minimal composes
16:29:44 <nirik> mboddu: odcs won't work?
16:29:49 <mboddu> nirik: +1, then I will work on it after epel8
16:29:59 <dustymabe> mboddu: just wanted to draw you and kevin to give expert advice on https://pagure.io/fesco/issue/2152#comment-579046
16:30:02 <mboddu> nirik: It might, now that they support pungi configs
16:30:23 <dustymabe> we don't have to discuss it here, but it would be nice to get an answer in the ticket (or tag in other people who would know)
16:31:38 <nirik> dustymabe: yeah, have been following that, just haven't had time to chime in. Basically I think the process should work fine, and we can always adjust it...
16:32:53 <mboddu> dustymabe: I guess you can be able to specify a commit from which you can submit a build, but I dont think that will solve your issue, right?
16:34:22 <dustymabe> mboddu: basically can we push a git commit into a repo (not on a branch) and tag it
16:34:23 * nirik thinks the branches are fine personally.
16:34:24 <dustymabe> and
16:34:37 <dustymabe> do a fedpkg build from a commit/tag
16:35:09 <mboddu> dustymabe: Where do you push that commit? If not on a branch?
16:35:17 <dustymabe> mboddu: directly to the repo
16:35:36 <dustymabe> alternatively can we push a tag+commit at the same time
16:36:49 <mboddu> dustymabe: That I am not sure
16:37:03 <nirik> what good is the tag? just to make it more clear what you are doing?
16:37:34 <dustymabe> nirik: correct - people like the idea of using a tag for each backport, rather than a branch
16:38:06 <nirik> sure, that works too I guess...
16:38:13 <cyberpear> just create a branch as called `f30-hotfix` as needed for these items? (then only even merge f30 into this one, never the other way?)
16:38:37 <cyberpear> (and once that branch is created, fcos would follow that one instead of f30?)
16:38:55 * nirik had one last item for open floor too if we have time
16:39:36 <mboddu> cyberpear: Thats what the initial request is
16:39:58 <nirik> it wouldn't follow it forever, these are all short term backports.
16:40:13 <mboddu> dustymabe: I guess a separate dist-git branch should work, but I still dont see any use case where you would two branches to carry two different patches
16:40:15 <nirik> after a few weeks it would go back to following the fedora branch
16:41:29 <mboddu> dustymabe: But I will take a look at the ticket and will comment on it
16:41:37 <mboddu> nirik: Sure, go ahead
16:42:13 <nirik> just a quick thing:
16:42:13 <dustymabe> sorry nirik mboddu - i got pulled away
16:42:16 <dustymabe> reading scrollback
16:42:31 <nirik> I spent some time yesterday on container/cloud composes.
16:42:47 <nirik> containers are back to all working except armv7, and pbrobinson is going to look into that.
16:43:05 <nirik> cloud I think is all back to working
16:43:31 <mboddu> nirik++
16:44:02 <nirik> we set armv7 to non blocking on f30, but not f29, so those container runs are failing.
16:44:04 <mboddu> #info Cloud composes are back to working and except for armv7 container composes are also working, pbrobinson is looking into the armv7 issue
16:44:11 <nirik> the 30 ones are finishing incomplete
16:44:35 <mboddu> nirik: Yeah, aarch64 became blocking for f30
16:44:40 <nirik> so hopefully builders are back to normal/stable finally after the great python3 upgrade
16:44:59 <nirik> the aarch64 ones are all finishing fine now. ;)
16:45:06 <mboddu> Yay
16:48:05 * nirik has to get coffee... back in a bit if anyone is still here.
16:48:05 <mboddu> Anything else?
16:48:14 <mboddu> nirik: I guess thats it
16:48:28 <mboddu> Thanks everyone for joining
16:48:34 <mboddu> #endmeeting