15:37:05 <dgilmore> #startmeeting RELENG (2016-03-14)
15:37:05 <zodbot> Meeting started Mon Mar 14 15:37:05 2016 UTC.  The chair is dgilmore. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:37:05 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:37:05 <zodbot> The meeting name has been set to 'releng_(2016-03-14)'
15:37:05 <dgilmore> #meetingname releng
15:37:05 <dgilmore> #chair dgilmore nirik tyll sharkcz bochecha masta pbrobinson pingou maxamillion
15:37:05 <zodbot> The meeting name has been set to 'releng'
15:37:05 <zodbot> Current chairs: bochecha dgilmore masta maxamillion nirik pbrobinson pingou sharkcz tyll
15:37:08 <dgilmore> #topic init process
15:37:30 <maxamillion> .hello maxamillion
15:37:30 <zodbot> maxamillion: maxamillion 'Adam Miller' <maxamillion@gmail.com>
15:38:03 <nirik> morning
15:38:20 * pbrobinson waves
15:38:32 * masta is here
15:38:47 <dgilmore> sorry for those who this is an hour early
15:40:08 <maxamillion> oh ritght
15:40:50 <maxamillion> dgilmore: is it technically meeting time or did we jump the gun? (not sure if the meeting is held in UTC or if it travels around with DST)
15:41:19 <pbrobinson> maxamillion: it travels around with dgilmore ;-)
15:41:23 <dgilmore> maxamillion: it si
15:41:26 <dgilmore> it is
15:41:30 <maxamillion> rgr
15:41:37 <dgilmore> maxamillion: the meeting is always 10:30am US Central
15:42:31 <dgilmore> lets get started
15:42:39 <dgilmore> #topic Secondary Architectures updates
15:42:57 <dgilmore> lets start with generic secondary arch discussions
15:43:17 <dgilmore> I know pbrobinson got pungi composes working for ppc and aarch64
15:43:21 <maxamillion> dgilmore: rgr that
15:43:33 <pbrobinson> well I got them working on aarch64
15:43:40 <pbrobinson> ppc needs a patch to lorax
15:43:47 <pbrobinson> I've sent a pull request upstream for that
15:44:02 <pbrobinson> but over all they're mostly working there as well
15:44:30 <pbrobinson> I'm just cleaning them all up today and getting them to a point they can run from a cron job
15:44:37 <pbrobinson> should have that crossed off today
15:44:45 <dgilmore> okay
15:44:58 <dgilmore> there was a complaint that aarch64 sent out broken dep emails
15:45:01 <pbrobinson> at the moment they're just a Everything and Server compose
15:45:13 <pbrobinson> dgilmore: already replied to that email, it was a bug in my script
15:45:53 <nirik> cool.
15:46:13 <dgilmore> cool
15:46:14 <pbrobinson> once I have the above bits cleaned up I'll be added extra bits like qemu and docker images
15:46:18 <dgilmore> its all on the mirrors also
15:46:27 <dgilmore> will have to tweak mirrormanager
15:46:30 <pbrobinson> and hacking in stuff like arm images
15:46:42 <pbrobinson> dgilmore: cool, the secondary mirror sync was on my todo list
15:46:44 <dgilmore> it requires going in and manually fixing up urls to the base
15:47:17 <dgilmore> pbrobinson: http://dl.fedoraproject.org/pub/fedora-secondary/development/24/
15:47:45 <pbrobinson> dgilmore: I think I have a means of moving s390 over to the process just for an Everything repo compose, I'll be looking at that later in the week
15:47:50 <dgilmore> pbrobinson: what was the issue with lorax on ppc?
15:47:57 <nirik> dgilmore: hang on...
15:47:58 <dgilmore> pbrobinson: okay
15:48:06 <nirik> can we not go in and fix the db in mm for a bit?
15:48:08 <pbrobinson> dgilmore: it wanted docker on ppc64 which we don't currently have
15:48:15 <nirik> so that mm folks can see how it's broken so we can actually fix it?
15:48:24 <dgilmore> nirik: was not going to look at it until later
15:48:28 <pbrobinson> dgilmore: it'll be the same issue for s390, so I fixed that at the same time
15:48:40 <dgilmore> nirik: I used the web interface for primary
15:48:42 <nirik> sure, but I'd like to make sure we have a chance to record how it's broken at least for them...
15:48:57 <nirik> yes, but thats not a real fix. ;) it should auto detect it right... we need to make it so
15:48:59 <dgilmore> pbrobinson: lorax wants docker? that seems wrong
15:49:24 <dgilmore> nirik: it just does not update the pointer to the new location
15:49:24 <pbrobinson> dgilmore: don't disagree, see the latest, or one of the latest changelog
15:49:40 <dgilmore> nirik: https://mirrors.fedoraproject.org/mirrorlist?repo=rawhide&arch=aarch64
15:49:48 <pbrobinson> dgilmore: as in it pulls in docker as part of building sutff
15:50:06 <pbrobinson> dgilmore: not as the lorax rpm actually depends on it
15:50:07 <dgilmore> nirik: https://mirrors.fedoraproject.org/mirrorlist?repo=fedora-24&arch=aarch64 is correct
15:50:07 <nirik> right, will ask mm folks to look.
15:50:51 <pbrobinson> dgilmore: rawhide looks a bit messed up too on the primary mirror http://dl.fedoraproject.org/pub/fedora-secondary/development/rawhide/
15:51:29 <dgilmore> pbrobinson: how?
15:51:48 <pbrobinson> dgilmore: it has Everything, Server plus all the old arch directories
15:52:02 <dgilmore> pbrobinson: s390 and ppc have not moved yet afaik
15:52:06 <dgilmore> you said ppc failed
15:52:16 <dgilmore> so I expect that to be the case until they move
15:52:27 <dgilmore> aarch64 is gone
15:52:36 <dgilmore> and it is the only one under Server and Everything
15:52:47 <pbrobinson> dgilmore: ah, OK, I didn't realise it would deal with that, I thought it was one or the other
15:53:01 <dgilmore> it just gets what is on the arches
15:53:10 <dgilmore> it does not know anything about the change
15:53:19 <dgilmore> and for the most part does not need to
15:53:32 <pbrobinson> dgilmore: and ppc has a Everything directory there already, just not Server
15:53:44 <pbrobinson> dgilmore: so it should just be s390
15:54:16 <dgilmore> pbrobinson: probably landed after the last sync
15:54:26 <dgilmore> and will get fixed next time round
15:54:30 <pbrobinson> dgilmore: that landed Friday night
15:54:39 <pbrobinson> dgilmore: actually before aarch64 ;-)
15:55:12 <dgilmore> pbrobinson: perhaps rsync to the hub is broken
15:55:37 <pbrobinson> dgilmore: possibly, lets sort it OOB and move on
15:56:17 <pbrobinson> I have nothing really else for secondary, basically we're almost ready for alpha, finally!
15:56:37 <dgilmore> rsync: failed to connect to ppc.koji.fedoraproject.org (10.5.124.182): No route to host (113)
15:56:40 <dgilmore> rsync error: error in socket IO (code 10) at clientserver.c(122) [Receiver=3.0.9]
15:56:43 <dgilmore> Command '['rsync', '-avhHp', '--exclude=.snapshot', '--exclude=archive', '--exclude=SRPMS', '--exclude=source', '--exclude=arm', '--exclude=armhfp', '--exclude=aarch64', '--exclude=s390', '--exclude=s390x', '--exclude=repodata', 'rsync://ppc.koji.fedoraproject.org/fedora-ppc', '/srv/pub/fedora-secondary/']' returned non-zero exit status 10
15:56:48 <dgilmore> rsync: failed to connect to ppc.koji.fedoraproject.org (10.5.124.182): No route to host (113)
15:56:51 <dgilmore> rsync error: error in socket IO (code 10) at clientserver.c(122) [Receiver=3.0.9]
15:56:54 <dgilmore> Command '['rsync', '-avhHp', '--exclude=.snapshot', '--exclude=archive', '--exclude=SRPMS', '--exclude=source', '--exclude=arm', '--exclude=armhfp', '--exclude=aarch64', '--exclude=s390', '--exclude=s390x', '--delete-after', 'rsync://ppc.koji.fedoraproject.org/fedora-ppc', '/srv/pub/fedora-secondary/']' returned non-zero exit status 10
15:56:59 <dgilmore> it is broken
15:57:02 <dgilmore> pbrobinson: okay cool
15:57:47 <dgilmore> I need to sit down and write up a script to do TC's and RC's
15:58:06 <dgilmore> I will do a better job of accounting for secondary arches with it
15:59:35 <pbrobinson> dgilmore: I think in most cases it should be OK, I think for f24 we're going to need to run on a branch anyway
16:00:01 <pbrobinson> I was also wondering how we're planning on doing bleed repo with pungi
16:00:17 <pbrobinson> I was thinking maybe a locked down tag like f24-bleed or similar
16:00:47 <dgilmore> pbrobinson: tags
16:00:57 <dgilmore> probably f24-compose
16:01:09 <dgilmore> and we will do TC and RC composes against it
16:01:22 <dgilmore> they will have to have inheritance on
16:01:39 <dgilmore> which may make things tricky if packages get pulled in from f23
16:03:53 <dgilmore> pbrobinson: anything you want to cover as part of the different arches?
16:04:02 <dgilmore> or just skip that and move on to open floor?
16:04:23 <pbrobinson> dgilmore: don't think so
16:04:31 <dgilmore> #topic open floor
16:04:38 <dgilmore> I have one quick thing
16:04:59 <dgilmore> there has been a raft of people doing silly things with packages causing nvr's to go backwards
16:05:12 <dgilmore> only one of which was an intentional go backwards
16:06:38 <masta> yeah... that is silly stuff
16:06:47 <pbrobinson> dgilmore: it's not unusual it's just that the report is now highlighting them
16:06:52 <nirik> perhaps we could draft a standard email and mail them about it?
16:07:09 <nirik> or can pungi stop allowing it? ;)
16:07:10 <dgilmore> pbrobinson: sure
16:07:14 <pbrobinson> dgilmore: I've often seen them and just fixed them because shadow/compare bits for secondary pick them up
16:07:22 <dgilmore> but there has to be things that can be done
16:07:30 <dgilmore> like a pre fedpkg build test
16:07:52 <maxamillion> I have something real quick also
16:07:57 <pbrobinson> yes, I think a check in fedpkg would be the best spot to handle that
16:08:04 <masta> I's say fedpkg is a good place to catch this... easy NVR comparison
16:08:21 <pbrobinson> it means the packager can fix it before it lands anywhere
16:08:57 <dgilmore> Package:      eclipse-launchbar-1.0.1-1.gitedd5f69.fc25
16:08:57 <dgilmore> Old package:  eclipse-launchbar-1.0.2-0.1.git93cdb07.fc25
16:08:57 <dgilmore> Summary:      Eclipse Launchbar plug-in
16:08:57 <dgilmore> RPMs:         eclipse-launchbar
16:08:57 <dgilmore> Size:         181382 bytes
16:08:59 <dgilmore> Size change:  -57708 bytes
16:09:02 <dgilmore> Changelog: * Thu Mar 10 2016 Mat Booth <mat.booth@redhat.com> - 1:1.0.1-1.gitedd5f69 - Take a post-release snapshot of 1.0.1 due to API breakage in newer versions
16:09:10 <dgilmore> this one they went backwards and did not add an epoch
16:09:17 <maxamillion> fun
16:09:32 <mbooth> dgilmore: It should have an epoch
16:09:54 <mbooth> http://pkgs.fedoraproject.org/cgit/rpms/eclipse-launchbar.git/tree/eclipse-launchbar.spec
16:10:10 <masta> there is a handy compare_tags_by_evr() routine in the rpmUtils python...
16:10:45 <masta> err... compareEVR()
16:11:01 <dgilmore> mbooth: so it does. must be a bug in the report tool
16:12:28 <pbrobinson> might not actually take epoch into account
16:13:08 <dgilmore> likely
16:13:15 <dgilmore> just filed https://pagure.io/compose-utils/issue/4
16:14:13 <dgilmore> one other thing I wanted to mention is that pungi is now running its test suite whenever a pull request comes in
16:14:30 <dgilmore> we also run the full test suite when doing builds
16:14:50 <dgilmore> while there is not full coverage, its getting better all the time
16:14:54 <pbrobinson> nice!
16:15:05 <dgilmore> which gives us more confidence of changes coming in
16:15:29 <maxamillion> is there a test case/scenario that does an actual compose on a small set of packages to prove that it works end to end?
16:15:35 <dgilmore> I think we finally have everything we need for Alpha
16:15:44 <dgilmore> maxamillion: there is
16:15:54 <maxamillion> dgilmore: rocking
16:16:18 <dgilmore> maxamillion: I just enabled that running in the rpm builds last week
16:16:27 <dgilmore> because it started working
16:16:37 <maxamillion> +1
16:16:46 <dgilmore> all the tests run in Fedora's Jenkins
16:17:14 <dgilmore> lsedlar is looking at getting fedpkg in there
16:17:20 <dgilmore> rpkg already is setup
16:17:25 <dgilmore> but not tied into pagure
16:18:23 <dgilmore> I would like to get tests for other things we depend on running also
16:18:56 <dgilmore> one other piece of info from me
16:19:30 <maxamillion> I have one quick one when ever you're done
16:19:36 <dgilmore> last week we had to coordinate changes to productmd, pungi, and pdc to support a request from QA
16:20:19 <dgilmore> which reminded me, we probably need to document the tooling and things we use along with the ways they are tied together
16:20:31 <dgilmore> thats it, maxamillion you're up
16:20:52 <maxamillion> just a quick note: we're trying to diagnose a pycurl issue that is causing the koji-containerbuild plugin to fail in stage koji, so stage may or may not be in various states of functional for today (and maybe tomorrow) during the debug efforts .... apologies in advance
16:21:21 <maxamillion> if this is problematic with something that others are doing, please let me know
16:21:26 <dgilmore> maxamillion: kinda what it is for since we do not have a dev koji
16:21:27 <pbrobinson> good for me
16:21:38 <maxamillion> dgilmore: +1
16:21:56 <dgilmore> I did ask for a dev koji, but I doubt it will get funded
16:22:07 <maxamillion> rgr
16:22:54 <dgilmore> anyone have anything else?
16:24:12 <pbrobinson> not from me
16:24:35 <maxamillion> no I
16:24:38 <maxamillion> not I*
16:24:40 <maxamillion> bleh, I can't type
16:25:10 <dgilmore> #endmeeting