15:02:56 <mboddu> #startmeeting RELENG (2020-04-21)
15:02:56 <zodbot> Meeting started Tue Apr 21 15:02:56 2020 UTC.
15:02:56 <zodbot> This meeting is logged and archived in a public location.
15:02:56 <zodbot> The chair is mboddu. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:02:56 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:02:56 <zodbot> The meeting name has been set to 'releng_(2020-04-21)'
15:02:56 <mboddu> #meetingname releng
15:02:56 <zodbot> The meeting name has been set to 'releng'
15:02:56 <mboddu> #chair nirik sharkcz pbrobinson pingou mboddu dustymabe ksinny jednorozec
15:02:56 <mboddu> #topic init process
15:02:56 <zodbot> Current chairs: dustymabe jednorozec ksinny mboddu nirik pbrobinson pingou sharkcz
15:04:07 <nirik> morning
15:04:30 * mboddu waves at nirik
15:05:07 <mboddu> nirik: So, I have two tickets to discuss, 1 is normal and able to fix it easily and the other is interesting
15:05:24 <nirik> oh boy
15:05:28 <mboddu> Which one do you think we should start with? :P
15:05:38 <nirik> doesn't matter to me
15:05:47 <mboddu> Lets start with the interesting one
15:06:00 <mboddu> #topic #9408 Tag ghc into EPEL8 buildroot for ghc bootstrap
15:06:07 <mboddu> #link https://pagure.io/releng/issue/9408
15:06:18 <mboddu> So, apparently it was not part of module or crb or anything
15:06:22 <mboddu> Its a buildroot only package
15:06:39 <mboddu> I have no frigging idea how to get it in in epel
15:07:12 <nirik> yeah, not sure either.
15:08:14 <nirik> can you even download those from the buildsystem? dunno...
15:08:30 <nirik> I guess if they can figure out how to get a directory of them, we can make a external repo out of it...
15:08:34 <mboddu> No idea, nothing as I can think of
15:09:01 <nirik> is it just one package? or a pile?
15:09:19 <mboddu> I am guessing a pile which are needed by ghc
15:09:30 <mboddu> Maybe some of them are already in normal repos
15:09:35 <nirik> the ticket makes it sound like just ghc
15:10:22 <mboddu> Is it going to make a difference if its one or more? We still need to pull them some how
15:10:48 <mboddu> Also, can rhel publish buildroot only builds?
15:11:01 <nirik> well, just how easy or hard it might be. ;)
15:11:16 <nirik> if we can get the rpms we could also koji import them into epel...
15:11:57 <nirik> so I think we should ask internally... I can do that if you like...
15:12:07 <mboddu> I was under the impression that rhel doesn't publish buildroot builds
15:12:18 <mboddu> nirik: Sure, go for it and add me to the ticket or email
15:12:38 <mboddu> I dont know where to start as well, other than asking internal rcm and going from there
15:14:04 <nirik> well, not publically no.
15:15:29 <mboddu> Hmmm, then I will see what options do we have
15:16:16 <mboddu> #info nirik will check with rhel folks to figure out how we can get the ghc builds that are used in rhel buildroot
15:16:44 <mboddu> Okay, moving on
15:16:47 <mboddu> #topic #7265 blender git repo contains large files causing git checkout to timeout during builds
15:16:53 <mboddu> #link https://pagure.io/releng/issue/7265
15:17:21 <mboddu> I guess we increased the timeout for texlive, can we do that same here?
15:17:42 <mboddu> And did we increase checkout time for texlive or something else?
15:19:01 <nirik> texlive is just large naturally... this is not the same issue.
15:19:16 <nirik> this is due to someone checking in the tar.bz into git... so checkout and commits are really slow
15:20:04 <nirik> The thing fesco approved was around branches that have never been used for a build
15:20:33 <nirik> This is more removing something in an existing branch and I am not sure how it would work... so someone who followed that would have to do this work
15:22:02 <nirik> I think someone... churchyard? said they would be happy to look at these cases for us
15:23:14 <mboddu> nirik: So, basically going into every branch and see if there is source tarball commited in the branch and remove it?
15:24:02 <nirik> I think it's looking for the binary stuff and pointing those commits to a different branch (archive), so they are still there, but not in master or whatever
15:25:13 <mboddu> I am not seeing an archive branch, unless I am missing something
15:25:41 <mboddu> Or you are saying a branch that is archived?
15:26:09 <nirik> it would be a new branch. let me find the link to the meeting
15:27:05 <nirik> https://meetbot-raw.fedoraproject.org/teams/fesco/fesco.2018-03-02-15.00.log.html#l-393
15:28:49 <mboddu> I commented on the ticket with the meeting minutes from ^ and asking chruchyard if he is still interested in helping us
15:29:01 <nirik> +1 sounds great to me
15:30:31 <mboddu> #info mboddu commented on the ticket with the meeting minutes from https://meetbot-raw.fedoraproject.org/teams/fesco/fesco.2018-03-02-15.00.log.html#l-393 and asked chruchyard if he is still interested in helping us
15:31:42 <mboddu> #topic #8948 Update list of critpath packages
15:31:48 <mboddu> #link https://pagure.io/releng/issue/8948
15:32:44 <mboddu> nirik: I am thinking of running it on pdc-backend01, wdyt?
15:32:46 <nirik> so we need just to try and automate this?
15:33:00 <mboddu> We tried already, but we need to automate it
15:33:02 <nirik> well, I would probibly hold off until 32 is done
15:33:06 <mboddu> Sure
15:33:09 <nirik> it might affect the compose somehow
15:33:23 <mboddu> pdc-backend01 has the pdc token, which is required by this script
15:34:50 <mboddu> #info we will deploy this on pdc-backend01 and run it as part of cron everyday
15:34:59 <nirik> could we add it to loopabull somehow?
15:35:18 <nirik> but I guess it doesn't trigger, it just should run from time to time
15:35:56 <mboddu> Yes
15:37:12 <mboddu> #topic #8879 Create pungi config for Rawhide/F32 ursa prime buildroot
15:37:18 <mboddu> #link https://pagure.io/releng/issue/8879
15:37:42 <nirik> I thought this got put on hold... but I can't really recall.
15:37:56 <mboddu> sgallagh: ^ We blocked it on due to an issue, but I am not able to recollect why?
15:38:03 <mboddu> nirik: Haha, same here
15:38:37 <mboddu> I want to comment on it for ktdreyer
15:39:50 <nirik> could odcs do this compose now?
15:40:18 <mboddu> It could, but it was blocked on some technical issue
15:40:38 <nirik> ok
15:42:18 <mboddu> Well, I guess we could move on and ask sgallagh to comment on the ticket if he can remember what it was
15:43:07 <mboddu> Oh I think I sorta remember now
15:43:12 <mboddu> It was only approved on EPEL8
15:43:30 <mboddu> But EPEL8 buildroot comes from external repo and ursa prime cannot handle that?
15:43:46 <mboddu> I think
15:45:17 <nirik> yeah, it makes no sense in epel8...
15:46:33 <mboddu> Anyway, lets move on
15:46:37 <mboddu> #topic Open Floor
15:47:03 <mboddu> #info RC 1.4 is done and we found couple more issues, maybe we need to run another RC
15:47:25 * nirik nods
15:49:00 <mboddu> I was soo happy that we got a request in less than a week from the freeze started
15:49:10 <mboddu> AFAIK, it never happened before
15:49:58 <mboddu> But now we delayed the release and we are getting close to that "hey, we dont have a RC yet state"
15:49:58 <nirik> well, we did start late. (the freeze)
15:50:33 <mboddu> Sure, which was why I was soo happy, it was quick turn around
15:50:40 <mboddu> quickest*
15:51:09 <jkaluza> mboddu: nirik: which compose?
15:51:17 * jkaluza has highlight on odcs
15:51:23 <nirik> ha. ;)
15:51:37 <mboddu> jkaluza: Haha, its ursa prime compose
15:51:37 <nirik> the default and buildroot modules... for the non modular buildroot
15:51:41 <jkaluza> odcs prod in fedora can do any compose which does not require Koji task
15:51:47 <jkaluza> so, no runroot
15:52:07 <jkaluza> and the resulting compose is stored in /mnt/odcs NFS storage (no /mnt/redhat)
15:52:09 <mboddu> jkaluza: Well, it can also do runroot once I fix up the koji runroot policy
15:52:13 <jkaluza> yes
15:52:26 <jkaluza> once we change koji configuration, it can also do runroot, but *right now* it cannot :)
15:52:56 <jkaluza> there also is a way how to move compose from /mnt/odcs to /mnt/redhat (odcs-promote-compose), but it has not been used in production yet.
15:53:17 <nirik> yeah, this would need that I guess... but just something to think about.
15:53:18 <mboddu> I think you mean /mnt/koji/
15:53:26 <jkaluza> yes, sorry
15:53:35 <jkaluza> it's hard to switch between these two worlds constantly :(
15:53:42 <mboddu> jkaluza: Remember upstream first :P
15:54:20 <jkaluza> but it would be great to use ODCS for that compose and test promotion to /mnt/koji
15:54:24 <mboddu> Thanks jkaluza for stepping in, you made us end the meeting on funny note :)
15:54:33 <jkaluza> especially if it can wait after the freeze :)
15:54:37 <jkaluza> mboddu: :)
15:54:45 <nirik> and congrats on getting the first eln composes working. ;)
15:54:48 * jkaluza needs to leave for the dinner.
15:54:50 <jkaluza> yay :)
15:54:51 <mboddu> Sure, but we will try it even before we get to ursa prime
15:54:52 <jkaluza> thanks
15:55:00 <mboddu> Yes, jkaluza++
15:55:11 <jkaluza> it's actually empty compose, but that's going to change soon I hope
15:55:14 <jkaluza> ;)
15:55:27 <mboddu> As long as it didn't fail, its a success to me :)
15:55:43 <mboddu> Anyway, anything else nirik
15:55:52 <mboddu> ?
15:55:59 <sgallagh> mboddu, nirik: I think it got blocked because module defaults themselves were getting challenged
15:56:08 <sgallagh> And ultimately got reverted.
15:56:21 <nirik> ah, that could be...
15:56:36 <mboddu> Yeah, seems like it
15:56:53 <mboddu> Well, I am going to close those tickets for now and we will reopen them if needed
15:57:01 <sgallagh> ok
15:58:17 <nirik> +1
15:58:54 <mboddu> And that makes it 99 tickets on releng repo
15:58:59 <mboddu> Good work everyone :)
15:59:05 <mboddu> 99 open tickets*
15:59:45 <mboddu> #info Now releng repo has *only* 99 tickets
16:00:05 <mboddu> Thanks everyone for joining
16:00:09 <mboddu> #endmeeting