fedora-qadevel
LOGS
14:01:29 <tflink> #startmeeting fedora-qadevel
14:01:29 <zodbot> Meeting started Mon Sep 18 14:01:29 2017 UTC.  The chair is tflink. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:01:29 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
14:01:29 <zodbot> The meeting name has been set to 'fedora-qadevel'
14:01:29 <tflink> #topic Roll Call
14:01:33 <jskladan1> yay!
14:01:38 * lbrabec is here
14:01:42 <jskladan1> could not have been more excited!
14:01:45 <tflink> #chair kparal jskladan1 lbrabec
14:01:45 <zodbot> Current chairs: jskladan1 kparal lbrabec tflink
14:02:09 <tflink> jskladan1: well, you just got chaired so I hope you didn't explode from excitement just then :-P
14:03:05 <tflink> let's get this party started
14:03:13 * kparal is here
14:03:15 <jskladan1> https://pics.me.me/this-is-my-favourite-guitar-i-call-it-old-smokey-10620789.png
14:03:27 <tflink> in theory, the more quickly we start, the more quickly we'll be able to end the meeting :)
14:03:57 <tflink> jskladan1: RIP?
14:04:10 <tflink> #topic Announcements and Information
14:04:33 <kparal> #info rpmlint errors in debugsource rpms fixed
14:04:35 <kparal> #link https://pagure.io/taskotron/task-rpmlint/issue/14
14:05:04 <tflink> there's nothing listed in the wiki page, anything to add here?
14:05:14 <kparal> #info ansiblized libtaskotron documentation added (in a basic form) into readme
14:05:17 <kparal> #link https://pagure.io/taskotron/libtaskotron/branch/feature/ansiblize
14:05:55 <kparal> #info ansiblized libtaskotron builds available
14:05:57 <kparal> #link https://copr.fedorainfracloud.org/coprs/kparal/taskotron-ansiblize/
14:06:43 <kparal> that's all from me
14:07:36 <tflink> well, it looks like you've done more announcement worthy stuff than I did last week :-/
14:08:44 <tflink> anything else from other folks?
14:08:57 <jskladan_rip> Yeah, I really enjoyed my vacation :)
14:09:27 <tflink> :-P
14:09:35 <tflink> ok, moving on
14:09:54 <tflink> #topic standard interface tests on koji/mbs builds
14:10:28 <tflink> my idea here was mostly to sync up on what the plan is, what's been done and what's left
14:11:10 <tflink> there have been requests for taskotron to start running standard interface tests stored both in dist-git and in the upstreamfirst forge
14:11:35 <kparal> rpmlint has been converted and partially rpmgrill
14:11:44 <kparal> lbrabec knows more about rpmgrill
14:12:03 <kparal> basic (rpm-based) SI tests should work, most probably
14:12:04 <tflink> kparal: partially rpmgrill?
14:12:25 <kparal> tflink: task-rpmgrill. not sure if it's fully functional yet
14:12:37 <tflink> I don't understand what rpmgrill/rpmlint has to do with standard interface stuff
14:12:52 <kparal> they use SI+our extension to it
14:13:09 <tflink> why are we using SI for rpmgrill?
14:13:44 <kparal> because we're getting rid for runtask.yml and we need to run it somehow
14:14:13 <tflink> sure but that doesn't need to mean SI
14:14:32 <tflink> my question about whether all that was clear or not has been answered, I think :)
14:14:37 <kparal> yes, it's not 100% SI. the differences are described in the readme as lnked
14:14:39 <kparal> *linked
14:15:11 <tflink> for
14:15:12 <kparal> but we're trying to stay close to SI, so that our code base doesn't have two different code paths
14:15:47 <tflink> for the moment, I'd like to ignore that effort and talk about just getting SI based stuff that other folks are writing to run on build events
14:16:23 <kparal> ok. we should deploy the stuff to one of our servers and try how it works. it works with those extra simple examples like gzip
14:16:45 <tflink> so, there is an old-style task that can run general SI based tests: https://pagure.io/taskotron/task-standard-interface-shim
14:17:27 <tflink> eventually, that'll need to be updated to work with the new ansibleized runner but it's a place to start
14:17:28 <kparal> does it exist in case feature/ansiblized doesn't move forward fast enough?
14:17:59 <tflink> it's somewhat orthogonal, in my mind
14:18:16 <kparal> because as it is currently written, I don't think we need a shim task. the runner will support SI tasks natively
14:18:33 <kparal> we just clone the git repo and fire a task
14:18:53 <tflink> which farther answers my question about whether or not we're all on the same page there :)
14:19:29 <tflink> I don't think there's a reason to tie ourselves to SI like that - assuming I understand you correctly
14:20:21 <tflink> the ansibleize branch wasn't meant to use SI and SI only. the fact that SI is based on ansible and we're moving more toward ansible is a co-incidence
14:20:43 <kparal> sorry, I need to step away now
14:20:50 * kparal afk
14:20:54 <tflink> ok
14:21:23 <jskladan_rip> well that seems to end that discussion :)  /me is pretty much in Jon Snow mode on this matter...
14:21:33 <tflink> lbrabec: is what I'm saying consistent with what we were talking about at Flock?
14:21:39 <tflink> seem consistent, rather
14:22:08 <lbrabec> yes, pretty much
14:23:19 <tflink> anyways, the thing that I was hoping for was to be able to run SI based tests stored in dist-git or the upstreamfirst forge on koji build to start with and on mbs build shortly thereafter
14:23:53 <tflink> but it sounds like this is better put on hold until kparal is around
14:24:09 <tflink> objections to moving on and coming back if/when kparal gets back?
14:24:17 <jskladan_rip> lets
14:24:53 <tflink> #topic fedmsg versions for trigger
14:25:11 <tflink> apparently, there is a problem with the fedmsg-related code we're using right now.
14:25:30 <tflink> I don't have the links but it sounds like there is a memory error that hit prod and dev
14:26:18 <tflink> jcline is working on backporting the fixes to fedora builds but was asking about whether it'd be possible for us to upgrade to f26
14:26:28 <lbrabec> #link https://pagure.io/taskotron/issue/235
14:26:35 <tflink> thanks
14:26:49 <tflink> other than freeze, I don't have any huge objections to upgrading the master
14:26:53 <tflink> my worry is about the client-hosts
14:27:22 <jcline> upstream issue is https://github.com/fedora-infra/fedmsg/issues/475 (which links to the pyOpenSSL issues)
14:27:38 <tflink> one of these days, I'm going to get around to getting that anaconda bug fixed/mitigated but until then, it takes a crazy amount of time to get a successful install on our client-host boxes
14:27:54 <tflink> thoughts on a path for upgrading?
14:28:29 <tflink> part of me is tempted to just upgrade the master as soon as beta freeze lifts
14:28:46 <jskladan_rip> does upgrading mean all the stack, or just trigger?
14:29:11 <tflink> that's what we're discussing
14:29:37 * jskladan_rip remembers the last time :))
14:29:38 <tflink> at a minimum it would involve upgrading the master
14:30:11 <jskladan_rip> is f26 buildmaster compatible with f25 slaves?
14:30:21 <tflink> jskladan_rip: don't worry, I'm sure we won't hit any problems with bugs introduced by the upgrade
14:30:26 <jskladan_rip> :))
14:30:40 <tflink> I think so - the buildbot protocol is pretty stable and the 0.8.x branch isn't really changing anymore
14:31:21 <jskladan_rip> lets do it then..
14:31:32 <tflink> jcline: would fixing the issue outside of official fedora builds be an option?
14:31:34 <jskladan_rip> when's the freeze's end?
14:31:50 <tflink> IIRC, part of your concern was changing the cryptography package
14:32:09 <tflink> when beta is released
14:32:20 <tflink> at least another week
14:32:43 <jcline> tflink, if you're comfortable with that. I just finished reading through the cryptography changelog and I think I can safely bump it to 1.7.2 which is good enough for us. I'll chat with some of the other maintainers to make sure they're fine with that.
14:32:59 <jcline> If they're not we could always build it into an infra repo
14:33:41 <tflink> I think that upgrading is a better longterm solution but I'm ok with using infra repo builds for the short term
14:34:08 <jskladan_rip> +1
14:35:09 <tflink> we'll need to upgrade from f25 before long, anyways
14:36:52 <tflink> #info to fix the fedmsg memory issue, the plan is to use side-builds in the short term if needed, upgrade the master to f26 in the medium term and upgrade everything to f26 in the longer term
14:37:02 <jcline> tflink, okay. I'll do my research in any case and if it's easy I'll go ahead and update so you all have more breathing room.
14:37:16 <tflink> jcline: thanks
14:37:52 <tflink> any other questions on this?
14:37:56 <jskladan_rip> nn
14:38:44 <tflink> ok, moving on
14:38:48 <tflink> #topic Tasking
14:39:04 <tflink> is anyone in need/want of stuff to do?
14:40:17 * tflink takes silence as a "no" :)
14:40:23 <jskladan_rip> I'll be mostly catching up with things I guess - lbrabec told me you are working on something for trigger now - so if I can be of any help, don't hesitate and let me know!
14:40:40 <tflink> yeah, I was planning to talk about that with the SI topic
14:41:20 <tflink> the feature is to support discovery of SI tests in dist-git and check for the existence of tests in the upstreamfirst forge
14:41:53 <tflink> I think that it's done, I just haven't sat down long enough to verify that it works for builds of packages that exist in the upstreamfirst repo
14:41:59 <jskladan_rip> OK
14:42:20 <jskladan_rip> I'll have a look at the ResultsDB logs, but I doubt I'll find much useful info :/
14:42:52 <tflink> it's a small patch. i just need to find a time where a build of something that's in the upstreamfirst forge happened and run that through trigger locally
14:43:21 <tflink> but I think that brings us to ...
14:43:26 <tflink> #topic Open Floor
14:43:35 <tflink> Anything else that folks want to bring up?
14:44:06 <tflink> it sounds like the SI/ansible discussion is going to have to wait for another day
14:44:21 * jskladan_rip has nothing
14:45:40 <tflink> cool, I'm going to light the magical non-deterministic qa meeting fuse (TM), then
14:46:45 <tflink> thanks for coming, everyone
14:46:50 * tflink will send out minutes shortly
14:46:54 <tflink> #endmeeting