15:02:31 <mboddu> #startmeeting RELENG (2020-09-08)
15:02:31 <zodbot> Meeting started Tue Sep  8 15:02:31 2020 UTC.
15:02:31 <zodbot> This meeting is logged and archived in a public location.
15:02:31 <zodbot> The chair is mboddu. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:02:31 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:02:31 <zodbot> The meeting name has been set to 'releng_(2020-09-08)'
15:02:31 <mboddu> #meetingname releng
15:02:31 <mboddu> #chair nirik sharkcz pbrobinson pingou mboddu dustymabe ksinny jednorozec
15:02:31 <mboddu> #topic init process
15:02:31 <zodbot> The meeting name has been set to 'releng'
15:02:31 <zodbot> Current chairs: dustymabe jednorozec ksinny mboddu nirik pbrobinson pingou sharkcz
15:02:39 <nirik> morning
15:02:43 <sharkcz> hi all
15:02:56 <mboddu> Hello nirik and sharkcz
15:03:35 <jednorozec> hello all
15:03:43 <mboddu> Hey jednorozec
15:04:22 <mboddu> Okay, I dont specific things to talk, but couple of things for open floor, so, lets see
15:04:38 <mboddu> #topic Alternate Arch Updates
15:04:42 <mboddu> sharkcz: Anything to report?
15:04:59 <sharkcz> not this time, I think
15:05:08 <mboddu> I wish we can find a fix for imagefactory and kernel 5.9 issue - https://bugzilla.redhat.com/show_bug.cgi?id=1871958
15:05:21 <nirik> yeah, thats an anoying one
15:05:24 <sharkcz> let me see ...
15:06:15 <sharkcz> ah, aarch64 one
15:06:46 <mboddu> Yup
15:07:01 <mboddu> We untagged at least 4 kernel builds so far
15:07:09 <mboddu> Including one from today
15:07:38 <mboddu> pwhalen: Any update? (sorry to bother you randomly here :D)
15:08:02 <nirik> he's just now mentioning it in the arm/aarch64 status in meeting-2 ;)
15:08:14 <pwhalen> mboddu: none :(
15:08:22 <mboddu> lol :D
15:08:57 * mboddu jumps there for a min
15:09:12 <pwhalen> Sorry. I know pbrobinson is also going to look but I dont see any errors, kickstart works outside of imagefactory
15:10:09 <sharkcz> as I just commented in the bz - it might oz/imagefactory passing some wrong options now, or not passing the right ones to qemu/libvirt
15:10:26 <sharkcz> I had something similar on s390x recently too
15:11:05 <mboddu> sharkcz: Interesting...
15:11:18 <mboddu> pwhalen: Yeah, its kinda weird
15:11:51 <sharkcz> the result was https://github.com/clalancette/oz/pull/282
15:12:11 <sharkcz> and IIRC there are bits specific to aarch64 too
15:13:14 <sharkcz> unless there is some virt bug breaking things with 5.9 kernels
15:14:13 <nirik> I think normal virt instances are fine...
15:14:29 <nirik> so yeah, it's likely something being passed on oz/IF only
15:16:08 <sharkcz> I would look at https://github.com/clalancette/oz/blob/master/oz/Guest.py#L448
15:16:51 <mboddu> Thanks sharkcz
15:17:22 <sharkcz> np
15:18:06 <mboddu> #info BZ #1871958 is still annoying, we untagged all kernel 5.9 builds to get the composes out. We need a fix for it asap.
15:18:21 <sharkcz> I think the xml file is somewhere in the imagefactory log, so it could be tested outside IF/oz
15:18:35 <mboddu> #info A place to look at is https://github.com/clalancette/oz/blob/master/oz/Guest.py#L448
15:19:00 <pwhalen> thanks sharkcz, I'll see if I can find it and try
15:19:15 <sharkcz> pwhalen: you are welcome
15:19:26 <mboddu> Okay, moving on
15:21:37 <mboddu> #topic #9737 Please delete various accidental branches in dist-git
15:21:43 <mboddu> #link https://pagure.io/releng/issue/9737
15:22:15 <mboddu> We now have a way to remove branches in dist-git - https://pagure.io/fedora-infra/howtos/blob/master/f/remove_branch_distgit.md
15:22:27 <mboddu> pingou and I cleared some branches last week
15:22:52 <nirik> great
15:23:22 <jednorozec> nice
15:23:59 <mboddu> On that topic, if you are around pingou - https://pagure.io/releng/issue/9737#comment-676053, any idea how thats possible?
15:25:50 <jednorozec> mboddu, https://pagure.io/releng/pull-request/9739#request_diff
15:26:40 <mboddu> jednorozec: Nope, that was from chruchyard to fix why some the script thinks some branches are not removable
15:27:03 <jednorozec> mboddu, isnt that to resolve package names as-well?
15:27:34 <mboddu> My question was, the refs are missing in the git repo but we can be able to git checkout to those branches
15:28:07 <jednorozec> oh
15:28:39 <mboddu> jednorozec: Nope, its about crazy rpm version comparison
15:33:50 <mboddu> Anyway, I will catch up with pingou later
15:34:29 <mboddu> #topic #9630 openh264 debuginfo repo is broken and #9468 Please send openh264 2.1.1 builds to Cisco
15:35:19 <mboddu> So, this is FINALLY fixed
15:36:35 <mboddu> I need to update a small piece of the document https://pagure.io/releng/pull-request/9708
15:37:16 <nirik> awesome.
15:37:18 <mboddu> I need to talk to jkaluza about it, once thats done, we have a new process (kinda convoluted) to get the openh264 repos distributed
15:38:03 <mboddu> convoluted because, there is no easy way to rsync content from one host to other, so your local machine becomes a middle man
15:39:09 <nirik> we can (and likely should) setup a rsync or something.
15:40:40 <mboddu> That would be great
15:41:47 <mboddu> #info We now have a way to compose openh264 repos using pungi/odcs
15:41:51 <mboddu> #topic Open Floor
15:42:10 <mboddu> Well, the two things I had was the two tickets I discussed before
15:42:14 <mboddu> Anybody has anything else?
15:42:28 <mboddu> We have a go/no-go meeting this Thu without a RC request so far
15:42:57 <mboddu> There are bunch of blockers - https://qa.fedoraproject.org/blockerbugs/milestone/33/beta/buglist
15:43:07 <sharkcz> then it should default to no-go ...
15:43:47 <mboddu> Yeah
15:43:58 <mboddu> So far we got only 1 blocker fix
15:44:08 <mboddu> That too we need to push them
15:46:31 <mboddu> Anybody got anything else?
15:47:10 <nirik> nothing off hand
15:47:29 <sharkcz> nope
15:47:48 <nirik> oh, the drpm thing came up again...
15:48:07 <jednorozec> I have found email from nirik regarding FMW signing keys. Going to sign the build first thing tomorrow
15:48:29 <nirik> yeah, I was going to update that ticket as well.
15:48:46 <jednorozec> any luck with it?
15:48:47 <nirik> I got my mac connected up yesterday, but likely it needs to upgrade to sierria
15:48:51 <mboddu> Oh right, thanks nirik for your help on getting the keys
15:49:12 <mboddu> And thanks jednorozec on the win signing part
15:49:14 <nirik> was going to try and poke at it today, if I get time
15:49:28 <mboddu> nirik: So, drpm thing...
15:51:20 <mboddu> #link #7215 older drpms not synced
15:51:23 <mboddu> #undo
15:51:23 <zodbot> Removing item from minutes: <MeetBot.items.Link object at 0x7f83c9ad2a10>
15:51:29 <mboddu> #link https://pagure.io/releng/issue/7215
15:52:26 <nirik> so, probibly the least messy thing would be to modify new-updates-sync to collect the drpms from older updates composes put them in some place, make repodata and sync that as drpms... but not going to be nice...
15:55:06 <nirik> or... decouple drpms from pungi/updates pushes and make it a seperate script that generates the ones we want...
15:55:16 <nirik> no quick fix. ;)
15:55:29 <mboddu> drpms are messy, I understand the importance of them, but if foobar 1.0 is installed and got updated to foobar 1.1 and then updated to foobar 1.2, delta rpms of no use then...
15:55:47 <mboddu> If trying to update from 1.0 to 1.2
15:56:03 <mboddu> As the drpm was generated for 1.1 to 1.2 at that point of time
15:56:32 <jednorozec> drpm make sense for centos like distros
15:56:49 <mboddu> Right
15:56:58 <mboddu> As the updates are not so frequent as fedora
15:58:24 <jednorozec> I never considered in the energy needed to generate them.
15:59:29 <mboddu> And to recreate them on the client
15:59:41 <mboddu> them = the whole newer rpm
15:59:46 <nirik> you can generate a foo 1.0 to 1.2 drpm
15:59:56 <nirik> it just depends on what ones you want to generate.
16:00:14 <mboddu> nirik: Right, but the default option is to generate from the last version to the latest
16:00:22 <nirik> right now we are generating drpms in updates only when: a) There is a previous update and b) there is an update for that in the current push
16:00:39 <mboddu> Or else, we will end up with a bunch of drpms
16:00:54 <nirik> the problem is... we don't keep them around.
16:00:57 <mboddu> default = current
16:00:59 <mboddu> Yes
16:01:16 <mboddu> Anyway, its something to discuss
16:01:20 <nirik> so foo goes from 1.0 to 1.2 we make a drpm... but then the next day it's gone.
16:02:20 <nirik> another big use would be from GA to current updates... but we don't do that
16:03:10 <mboddu> The right thing would be, keep the drpms around for every update push we run
16:03:31 <mboddu> But thats a lot of churning....
16:04:32 <mboddu> Anyway, its past the scheduled time
16:04:56 <mboddu> Lets catch up in the next meeting and we can keep updating the ticket
16:04:58 <nirik> some of those would also be useless. but yeah, that would be a first cut
16:05:40 <mboddu> Thanks guys for joining
16:05:43 <mboddu> #endmeeting