15:00:14 <jednorozec> #startmeeting RELENG (2022-06-21)
15:00:14 <zodbot> Meeting started Tue Jun 21 15:00:14 2022 UTC.
15:00:14 <zodbot> This meeting is logged and archived in a public location.
15:00:14 <zodbot> The chair is jednorozec. Information about MeetBot at https://fedoraproject.org/wiki/Zodbot#Meeting_Functions.
15:00:14 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:00:14 <zodbot> The meeting name has been set to 'releng_(2022-06-21)'
15:00:14 <jednorozec> #meetingname releng
15:00:14 <zodbot> The meeting name has been set to 'releng'
15:00:14 <jednorozec> #chair nirik sharkcz pbrobinson phsmoura dustymabe jednorozec
15:00:14 <jednorozec> #topic init process
15:00:14 <zodbot> Current chairs: dustymabe jednorozec nirik pbrobinson phsmoura sharkcz
15:00:28 <nirik> morning
15:01:25 <phsmoura> morning
15:02:25 <jednorozec> lets start
15:02:44 <jednorozec> #topic compose-tracker
15:03:18 <jednorozec> so we are running f36 in staging
15:03:52 <jednorozec> nirik, you renewed pagure token for the prod deployment right? Is there some how to how to do it?
15:04:17 <nirik> I did, but that still didn't get it fully working... it needed a new perm...
15:04:36 <jednorozec> because the stg is using one that is missing modifi project
15:04:37 <nirik> so I just went and added that perm in the database, since I couldn't figure out any better way to do it. ;(
15:04:45 <nirik> I can do the same in stg...
15:05:03 <jednorozec> so you did psql query and updated that?
15:05:39 <jednorozec> I think we are supposed to generate new one in pagure...
15:06:13 <jednorozec> anyhow, it is working in prod with f34 and if the token get right perms it should work on f36 on stg
15:06:29 <jednorozec> I am thinking about migrating prod to f36
15:08:00 <nirik> well, I couldn't not find a way to do that... the pagure-admin doesn't allow making an admin token with that perm
15:08:05 <nirik> and I can't login as releng
15:08:11 <jednorozec> huh
15:08:44 <jednorozec> I tried to login as releng but was unable to do so today
15:09:07 <nirik> so I think perhaps it's a bug in pagure-admin and it should be able to set those acls... but doesn't?
15:11:20 <nirik> anyhow, inserted that acl in stg... should work now.
15:12:13 <jednorozec> awesome we will see how it goes
15:12:27 <jednorozec> if everything works are you ok with updating prod tomorrow?
15:12:59 <nirik> sure.
15:13:01 <nirik> please do
15:13:03 <jednorozec> I will do it in local morning so if something odesnot work I can fix/revert it
15:13:04 <nirik> or I can
15:13:08 <nirik> ok
15:13:26 <jednorozec> ack
15:13:28 <jednorozec> moving on
15:13:48 <jednorozec> .releng 10849
15:13:49 <zodbot> jednorozec: Issue #10849: Untag trashcan from python-bottle - releng - Pagure.io - https://pagure.io/releng/issue/10849
15:13:54 <jednorozec> so we dont do this right?
15:14:17 <jednorozec> there already is +1 NVR in koji
15:14:18 <nirik> no, we can... as long as the builds aren't already deleted
15:14:34 <nirik> but yeah, often it's easier for them to just rebuild if the build is that old
15:15:30 <nirik> basically what happens is:
15:16:12 <nirik> if there's a build and it's not in a tag we care about (or any tag) after a month I think it is, it gets marked with trashcan... then after another month or so, it gets deleted.
15:16:29 <nirik> it can be untagged from trash but if it's deleted it's gone forever.
15:17:16 <nirik> so, I guess I'd say close this saying there's a newer build and ask them to reopen if its somehow still needed?
15:17:45 <nirik> but why did they request it... huh, dunno
15:17:51 <jednorozec> yeah I am asking what it is they expect to happen when there already is new build in koji
15:18:14 * nirik nods.
15:18:39 <jednorozec> moving on
15:18:42 <jednorozec> this one
15:18:45 <jednorozec> .releng 10839
15:18:46 <zodbot> jednorozec: Issue #10839: Rawhide .composeinfo missing - releng - Pagure.io - https://pagure.io/releng/issue/10839
15:19:15 <jednorozec> so it seems we did not have a FINISHED compose since cca the date they refer in the compose name
15:20:30 <nirik> I'm not sure what makes that .composeinfo...
15:20:30 <nirik> it should be on incomplete ones too no?
15:20:46 <jednorozec> I didnt see it there
15:22:34 <nirik> 2022-06-19 11:05:01 [INFO    ] [BEGIN] Writing composeinfo
15:22:35 <nirik> 2022-06-19 11:05:01 [INFO    ] [DONE ] Writing composeinfo
15:22:43 <nirik> in the logs for the last incomplete compose...
15:22:47 <jednorozec> hm
15:24:32 <nirik> nightly.sh has:
15:24:33 <nirik> nightly.sh:$RSYNCPREFIX ./releng/scripts/build_composeinfo "$RSYNCTARGET/" --name "$NEWCOMPOSE_ID"
15:25:04 <nirik> oh it's there.
15:25:07 <jednorozec> where did you get tle log from?
15:25:11 <jednorozec> I am looking at https://kojipkgs.fedoraproject.org/compose/rawhide/Fedora-Rawhide-20220621.n.1/logs/global/pungi.global.log
15:25:33 <nirik> https://dl.fedoraproject.org/pub/fedora/linux/development/rawhide/.composeinfo
15:25:55 <nirik> I was looking at the 20220619 one as thats the last one that finished incomplete.
15:26:45 <jednorozec> hm
15:27:45 <nirik> oh, they are looking at the composes directly, not the synced one... so I bet it's not in place there.
15:29:04 <nirik> so, perhaps we need to adjust nightly.sh to also put the .composeinfo there?
15:30:20 <nirik> but... why did it stop on the 24th?
15:30:39 <jednorozec> I am looking into ansible history
15:30:45 <jednorozec> if we changed something
15:31:49 <nirik> So, on 2022-05-24 I reinstalled compose-rawhide01 with f36... but I wouldn't think that would cause a missing file
15:33:03 <jednorozec> that is weird
15:33:12 <jednorozec> the script runs according to the pungi log
15:34:29 <nirik> so yeah... dunno. I guess I'd say more investigation needed.
15:35:48 <jednorozec> yeah
15:36:05 <jednorozec> I can look into it
15:36:07 <nirik> might be some more info in the releng-cron output...
15:36:48 <jednorozec> thanks
15:36:54 <jednorozec> .releng 10679
15:36:55 <zodbot> jednorozec: Issue #10679: Cannot install python38-pytest into EPEL8 buildroot - releng - Pagure.io - https://pagure.io/releng/issue/10679
15:36:58 <jednorozec> this can be closed
15:37:06 <nirik> yep. +1 close.
15:38:01 <jednorozec> .releng 10847
15:38:02 <zodbot> jednorozec: Issue #10847: Clean up airspyhf-1.6.8-1 repo and make sure fedscmadmin doesn't allow that moving forward - releng - Pagure.io - https://pagure.io/releng/issue/10847
15:38:05 <jednorozec> that is by you nirik
15:38:24 <nirik> ah yeah, it came up on the devel list.
15:38:37 <nirik> I'm not sure there's any easy check we can add, but open to ideas...
15:39:11 <jednorozec> well, if one request xyz-123 its hard to check it
15:39:38 <nirik> I guess reading from the right you could make sure there's not a number a dash and another number (because that shouldn't be legal in a compat package either)
15:40:26 <nirik> or we could make people add a --compat if it's a compat package, otherwise reject? that might get pushback tho
15:40:38 <jednorozec> yeah
15:40:50 <jednorozec> this needs fixing on our side
15:41:13 <jednorozec> regexp wold work but also not all the time.
15:41:18 <nirik> perhaps we could just highlight it / warn the admin?
15:41:29 <nirik> warning: package name has numbers in it, is this a compat package?
15:41:36 <jednorozec> hmm
15:41:58 <jednorozec> I dont think people read messages from the tool
15:42:02 <nirik> ideally reviewers should catch this kind of thing. ;(
15:42:06 <jednorozec> it has auto--approve param...
15:42:12 <jednorozec> -auto-approve
15:42:23 <nirik> also, we are going to automate it...
15:42:41 <jednorozec> yup
15:42:47 <jednorozec> but the solution will be the same
15:42:54 <jednorozec> its python to python
15:43:20 <nirik> yeah, dunno... but worth thinking on...
15:43:21 <jednorozec> so if I add one if into fedscm-admin I can add the same one into the toddlers
15:44:00 <nirik> right, we will need to keep fedscmadm working for exceptions...
15:44:02 <jednorozec> anyhow, what do you mean by cleanup in this case?
15:44:21 <jednorozec> should i retire it or compeltly remove the repo
15:44:27 <nirik> retire https://src.fedoraproject.org/rpms/airspyhf-1.6.8-1
15:44:32 <jednorozec> ack
15:44:40 <f_crudge> Hi all :)
15:44:46 <nirik> I suppose the repo could be nuked, but then you have to do all the other steps manually...
15:44:53 <nirik> so retirement is likely better.
15:44:58 <nirik> morning f_crudge
15:45:11 <jednorozec> hello f_crudge
15:45:38 <LunaJernberg[m]> Hello
15:46:04 <f_crudge> Can anyone attend the meeting? I'm interested in the infrastructure team
15:46:34 <nirik> morning LunaJernberg[m]
15:46:45 <jednorozec> package is retired
15:46:46 <nirik> f_crudge: sure, but this is the release engineering meeting. ;)
15:47:19 <jednorozec> moving on the next ticket
15:47:22 <jednorozec> .releng 10852
15:47:23 <zodbot> jednorozec: Issue #10852: %dist is not set correctly on `epel9-next-build` - releng - Pagure.io - https://pagure.io/releng/issue/10852
15:47:37 <nirik> yeah, this does seem a problem, but I am not sure how to fix it...
15:48:08 <nirik> it's set in koji.
15:48:19 <nirik> and I don't think it can do anything with macros.
15:48:57 <nirik> do a 'koji taginfo epel9-next-build'
15:49:00 <nirik> Tag options:
15:49:00 <nirik> rpm.macro.dist : '.el9.next'
15:49:12 <jednorozec> I see
15:49:40 <nirik> so, no idea how to fix this.
15:49:49 <jednorozec> what would be the consequences to setting it just to .next
15:50:00 <nirik> I mean, we could make a special epel9-next-release with macros I suppose.
15:50:29 <nirik> it doesn't append or anything, it overrides I am pretty sure.
15:50:36 <jednorozec> oh
15:50:37 <jednorozec> ok
15:50:48 <nirik> ie, it sets dist to exactly that for mock
15:52:14 <jednorozec> hmm
15:52:38 <jednorozec> so the only real option is to create -release for el9.next that would expand macros correctly
15:52:54 <nirik> as far as I can think of...
15:53:06 <nirik> we might ask others in case I am missing something.
15:53:46 <jednorozec> I highlighted carl in the ticket
15:53:54 <jednorozec> Writing comment right now
15:54:43 <nirik> ok, sounds good
15:55:43 <jednorozec> oh
15:55:53 <jednorozec> 5 mins left
15:55:59 <jednorozec> #topic open Floor
15:56:27 <nirik> python 3.11 tag was merged back in... rawhide has been failing composes since then... investigation still underway.
15:59:15 <mhroncok> nirik: let me know if I can help
15:59:46 <nirik> mhroncok: will do. The current failure doesn't make much sense to be python related...
16:00:43 <jednorozec> ok
16:00:47 <jednorozec> time is up!
16:00:52 <jednorozec> thank you all
16:00:56 <jednorozec> #endmeeting