15:00:01 <bowlofeggs> #startmeeting Bodhi stakeholders (2017-05-23)
15:00:01 <zodbot> Meeting started Tue May 23 15:00:01 2017 UTC.  The chair is bowlofeggs. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:01 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:00:01 <zodbot> The meeting name has been set to 'bodhi_stakeholders_(2017-05-23)'
15:00:03 <bowlofeggs> #meetingname bodhi_stakeholders
15:00:03 <zodbot> The meeting name has been set to 'bodhi_stakeholders'
15:00:04 <bowlofeggs> #topic salutations
15:00:05 <bowlofeggs> #chair acarter bowlofeggs dgilmore masta mboddu nirik pbrobinson puiterwijk trishnag
15:00:06 <zodbot> Current chairs: acarter bowlofeggs dgilmore masta mboddu nirik pbrobinson puiterwijk trishnag
15:00:25 <bowlofeggs> #chair Kellin
15:00:25 <zodbot> Current chairs: Kellin acarter bowlofeggs dgilmore masta mboddu nirik pbrobinson puiterwijk trishnag
15:01:13 * nirik is sort of here, but in another meeting, catching up on email, etc
15:01:27 * mboddu is sort of here, working on something else
15:01:54 <sochotni> really here
15:03:08 <bowlofeggs> #topic announcements and information
15:03:09 <bowlofeggs> #info A Bodhi 2.7.0 beta is deployed to stg: https://bodhi.stg.fedoraproject.org/
15:03:11 <bowlofeggs> #info Release notes available at https://bodhi.stg.fedoraproject.org/docs/release_notes.html
15:03:59 <dustymabe> .hello dustymabe
15:04:01 <zodbot> dustymabe: dustymabe 'Dusty Mabe' <dustymabe@redhat.com>
15:04:01 <bowlofeggs> i will likely release 2.7.0 this week, but due to the freeze it will just be released to fedora/epel
15:04:08 <bowlofeggs> it's got some CLI improvements from cverna
15:04:16 * cverna is around
15:04:30 <dustymabe> bowlofeggs: any chance we can get kushal's changes in before 2.7.0 ?
15:04:52 <dustymabe> another question is - how hard is it to cut a release?
15:05:16 <bowlofeggs> dustymabe: not for 2.7, but it might be able to make it for 2.8. i don't think there are any tests yet, right?
15:05:28 <bowlofeggs> tests are going to be pretty important for that patch
15:06:03 <dustymabe> bowlofeggs: is 2.8 on a timeline?
15:08:44 <bowlofeggs> dustymabe: cutting a release often takes me more time than i'd like
15:08:53 <bowlofeggs> since bodhi doesn't have a CI suite
15:09:05 <bowlofeggs> i've been working with bstinson to get access to the centos CI
15:09:20 <bowlofeggs> that will help, but i will still need a test suite that can take advantage of it
15:09:31 <bowlofeggs> anyways, i frequently find problems when building the beta
15:09:41 <bowlofeggs> it'd be nice to find those problems sooner
15:10:13 <dustymabe> bowlofeggs: we can chat about this later I guess. sorry to derail the meeting
15:10:31 <bowlofeggs> dustymabe: i do follow a timeline - i usually make a beta ~1 week before this meeting, and this meeting is every 4 weeks
15:10:50 <bowlofeggs> so basically about once a month i make a release of whatever is on the develop branch + whatever patches i make after the beta
15:10:57 <sochotni> well setting up CI for bodhi might be an important goal mid-long term
15:11:02 <bowlofeggs> usually there's a good 4-5 patches needed due to the lack of CI ☹
15:11:13 <bowlofeggs> sochotni: indeed
15:11:19 <sochotni> given that you release ~once a month I guess it will pay for itself relatively quickly
15:11:19 <dustymabe> bowlofeggs: ok - so basically the timeline we're working on is that we need bodhi calling pungi for f26 right after release
15:11:40 <dustymabe> right after f26 release
15:11:53 <dustymabe> since that's when bodhi starts making the ostrees
15:12:01 <bowlofeggs> so that should probably get in in the next ~3 weeks
15:12:30 <dustymabe> yeah - we can chat more in #fedora-cloud
15:12:32 <dustymabe> thanks bowlofeggs
15:12:34 <bowlofeggs> cool
15:12:58 <bowlofeggs> let's move on. i've got a few feature topics to give a high level update on
15:13:13 <bowlofeggs> #topic multi-type support in Bodhi
15:13:14 <bowlofeggs> #info There is a milestone tracking adding multi-type support to Bodhi: https://github.com/fedora-infra/bodhi/milestone/4
15:13:54 <bowlofeggs> we've made a little progress towards multi-type support since the last stakeholder's meeting - the database models are now in place
15:14:09 <bowlofeggs> the next area to focus on is the API, which actually i think might not need that much changed
15:14:40 <bowlofeggs> we'll see though
15:15:00 <bowlofeggs> one thing that would really help me is access to a koji instance that has some modules in it
15:15:07 <sochotni> yeah...
15:15:09 <sochotni> about that...
15:15:40 <bowlofeggs> i think we can make bodhi auto-detect whether an "nvr" is a module/rpm/container, and then the user doesn't have to do anything but type a valid koji identifier (what is currently called an nvr)
15:15:46 <sochotni> there are different ways to go about this
15:15:48 <bowlofeggs> with that, the API gets easier
15:16:05 <sochotni> yes, it should be fairly straight forward to identify
15:16:36 <sochotni> the option is either - go ahead and make necessary config modifications in stage
15:17:06 <sochotni> (but I think we need to wait for dgilmore to review https://pagure.io/releng/issue/6799)
15:17:27 <sochotni> or I can try and stand up a separate dev instance with the modules
15:17:37 <sochotni> or at least *a* module
15:17:53 * threebean nods
15:17:59 <threebean> so, we are blocked on https://pagure.io/releng/issue/6799 ?
15:18:22 <bowlofeggs> yeah either option is fine with me. it'd be nice to have something within the next week or so
15:18:25 <sochotni> threebean: well unless we want to power through - yes
15:18:49 <sochotni> I did send email earlier today to katec to see if they could bring it up at grooming or other relevant rcm meeting
15:19:04 <bowlofeggs> cool
15:19:28 <bowlofeggs> well we follow up on that later, thanks for fighting for me ☺
15:19:38 <katec> sochotni yes, sorry i haven't responded yet but i definitely can!
15:19:38 <sochotni> bowlofeggs: I'll have a quick look at standing up some koji instance with a module ... if it won't take too much time I might just do that
15:19:45 <katec> i'm going to email you with some questions
15:19:47 <bowlofeggs> not much else to report about modules for now, but I'd say the effort is ~on schedule
15:19:51 <sochotni> katec: I don't expect replies within the hour, no worries
15:19:51 <katec> our grooming is on thursdays
15:20:10 <threebean> bowlofeggs: can I introduce you to mcurlej_?
15:20:18 <bowlofeggs> threebean: of course ☺
15:20:19 <sochotni> katec: maybe send me invite? if there are questions I'll answer immediately...
15:20:23 <bowlofeggs> hi mcurlej_!
15:20:24 <threebean> he just joined the factory2 team yesterday (\ó/)
15:20:28 <bowlofeggs> oh cool
15:20:29 <bowlofeggs> welcome!
15:20:35 <mcurlej_> bowlofeggs: hi
15:20:39 <sochotni> and we won't have to do the whole ping-pong in email
15:20:54 <mcurlej_> nice to meet you all
15:21:52 <bowlofeggs> anything else on modules/multi-type? if not, i've got two more high level feature topics (one of them is actually still modularity related actually)
15:21:54 <sochotni> threebean: I'll put up the last PR piece for MBS that enables the CG code tomorrow
15:22:08 <sochotni> OK from my end
15:22:24 <bowlofeggs> cool
15:22:31 <threebean> rad :)
15:22:33 <bowlofeggs> #topic Moving away from pkgdb
15:22:35 <bowlofeggs> #info mprahl has been working on moving Bodhi away from using pkgdb and towards PDC and pagure
15:22:41 <sochotni> I'd just mention - we'll want to figure out pungi module mashing soon :-)
15:22:53 <bowlofeggs> yeah that's def important
15:22:59 <threebean> (will get there!)
15:23:07 <bowlofeggs> so the move away from pkgdb is modularity related too
15:23:09 <sochotni> anyway - back to current topic
15:23:25 <bowlofeggs> aiui, it's needed for the new branching model
15:23:49 <bowlofeggs> and bodhi uses pkgdb to find out info about packages (like ACLs, and what is critpath, and maybe other things i'm forgetting)
15:24:08 <bowlofeggs> but mprahl has a nice PR up already that's in the review dance
15:24:40 <bowlofeggs> he's also making the switch settings driven (which is also nice) so we can get it in place and switch it on by a config change
15:25:38 <bowlofeggs> that's all i have to say about it, any questions?
15:27:07 <bowlofeggs> #topic CI Gating
15:27:08 <bowlofeggs> #info pingou has been working on adding gating to Bodhi based on CI results
15:27:39 <bowlofeggs> so this one is pretty exciting i think - bodhi will soon be able to gate updates based on test results from the CentOS CI system
15:28:00 <threebean> \ó/
15:28:10 <bowlofeggs> pierre-yves has a nice PR up to do this, and i've also got a patch i've been working on that's complimentary to it
15:29:00 <bowlofeggs> making testing more automated is really going to help increase quality of fedora updates so i'm a huge +1 to this effort (not that the quality is bad, but making it better is always a good thing to do ☺)
15:29:32 <bowlofeggs> any comments/questions on that?
15:30:02 <sochotni> bowlofeggs: besides the hoorays you mean?
15:30:07 <bowlofeggs> haha ☺
15:31:08 <bowlofeggs> cool, we can move on to the more general topics we always hit
15:31:22 <bowlofeggs> #topic Looking forward
15:31:24 <bowlofeggs> #info Bodhi's high priority issue list https://github.com/fedora-infra/bodhi/issues?q=is%3Aopen+is%3Aissue+label%3A%22High+priority%22
15:31:25 <bowlofeggs> #info High priority means it's important, but not a show stopper
15:31:27 <bowlofeggs> Any filed issues that aren't on these lists that should be?
15:32:04 <bowlofeggs> it's a pretty big list
15:32:50 <bowlofeggs> #info Bodhi's critical priority list https://github.com/fedora-infra/bodhi/issues?q=is%3Aopen+is%3Aissue+label%3ACritical
15:32:56 <bowlofeggs> one of those two has a PR up from me
15:33:12 <bowlofeggs> the other i am planning to prioritize for 2.8 or 2.9 because it makes stg not work for me
15:33:36 * dustymabe looks
15:33:37 <bowlofeggs> (critical means we really can't keep living in this squalor)
15:33:50 <bowlofeggs> critical usually means that something is super broken
15:34:28 <dustymabe> yeah - i'm hoping to test some things in stg soon
15:34:33 <dustymabe> just got access to stg
15:34:38 <bowlofeggs> awesome
15:34:52 <dustymabe> might ask you some questions about this sometime soon
15:34:59 <bowlofeggs> sure, feel free!
15:35:35 <bowlofeggs> by the way, on the topic of looking forward - i expect that bodhi will have some stability issues in the next few months
15:35:42 <bowlofeggs> there are a lot of really big changes coming in
15:36:06 <bowlofeggs> and many of them are in areas that are difficult or even impossible to test in dev or stg (which is why we really need CI)
15:36:06 <dustymabe> bowlofeggs: more fun for the guy trying to get a release out the door every two weeks :)
15:36:13 <bowlofeggs> yeah
15:36:25 <bowlofeggs> dustymabe: but one of the big changes i'm referring to is for/from you ☺
15:36:34 <dustymabe> :)
15:36:35 <sochotni> :-)
15:36:43 <sochotni> dustymabe: you break it you buy it!
15:36:52 <sochotni> good thing this is all free software...
15:36:55 <bowlofeggs> but you know, we're becoming multi-type, we're going to gate on CI, we're switching to pungi, we're moving away from pkgdb
15:37:09 <bowlofeggs> i just bet that some of it isn't going to be smooth, and it's all happening at the same time
15:37:11 <dustymabe> bowlofeggs: bodhi's growing up
15:37:15 <bowlofeggs> hahah yeah
15:37:34 <bowlofeggs> it is a crazy amount of change all at once, so i just want to set the expectation that it may have some bumps along the way
15:38:14 <bowlofeggs> we'll do our deployments on mondays/tuesdays so we have time to respond to issues ☺
15:39:31 <sochotni> sounds reasonable
15:40:25 <bowlofeggs> that said, if anyone is interested in contributing to bodhi in a testing capacity, i'd love help there. writing unit tests would be great, but even better would be starting up a CI suite for bodhi with the centos CI system
15:41:00 <bowlofeggs> i hope to work on that myself once the f27 goals are met, but if someone is here looking for a way to contribute, that would be a great help
15:41:44 <sochotni> hmmm...
15:41:47 <bowlofeggs> let's do open floor
15:41:54 <bowlofeggs> #topic Open floor
15:42:35 <sochotni> bowlofeggs: I have an existing koji instance, but it's behind a firewall/VPN - is that a problem for your dev work?
15:42:49 <sochotni> I could create a module build there pretty quickly
15:43:05 <sochotni> it's also limited as far as auth options go
15:43:08 <bowlofeggs> sochotni: i think i could access that from a vagrant box on my laptop, so it should work
15:43:31 <bowlofeggs> sochotni: does it do krb?
15:43:36 <sochotni> bowlofeggs: nope
15:44:06 <bowlofeggs> sochotni: bodhi used to have code to do tls auth, but i realized this weekend that that got removed when fedora switched to krb. i think i could resurrect that old code and make it work either way again
15:44:06 <sochotni> bowlofeggs: it's using certs
15:44:29 <bowlofeggs> technically we shouldn't have removed that for semver reasons anyway
15:44:46 <bowlofeggs> not sure why i didn't realize taht at the time
15:44:55 <sochotni> anyway, it's a thought...
15:45:03 <bowlofeggs> yeah i think that could be useful
15:45:11 <sochotni> setting up a new koji dev env + kerb would be a bit time consuming for me right now
15:45:26 <sochotni> ok, let's talk about that later
15:45:31 <bowlofeggs> sochotni: fwiw, i did make a vagrantfile for koji recently
15:45:57 <bowlofeggs> https://pagure.io/koji/pull-request/420
15:46:15 <bowlofeggs> it's not fully functional yet, but it's meant to be a start
15:46:29 <bowlofeggs> i've wanted a way to have a nice koji dev box to use with bodhi's dev box
15:46:52 <sochotni> yeah, exactly what's needed here basically...
15:47:07 <bowlofeggs> it doesn't do the tls bits yet either
15:47:11 <bowlofeggs> nor krb
15:47:16 <bowlofeggs> basically it doesn't fully work ☺
15:47:22 <sochotni> :-)
15:47:35 <sochotni> don't let the perfect be enemy of "good enough" I guess :-)
15:47:41 <sochotni> or at least useful
15:48:02 <sochotni> anyway - will let you know wrt koji/module stuff
15:49:23 <threebean> bowlofeggs: fwiw, I like the idea you mentioned earlier on the multitype API.  if users just submit NVRs, and bodhi knows how to figure out the content type from the NVR, then little has to change.
15:49:43 <threebean> perhaps only the "read" endpoints of the API need to be expanded to return additional type information.
15:49:53 <threebean> the write endpoints stay the same.
15:49:57 <bowlofeggs> sochotni: exactly
15:50:13 <bowlofeggs> threebean: yeah that's what i'm hoping
15:50:16 <threebean> validators need to be enhanced to detect multi-type conflicts if (say..) you are trying to add a container nvr to an update that already contains rpm builds.
15:50:30 <bowlofeggs> threebean: we actually already have those validators in place ☺
15:50:36 <threebean> bowlofeggs++
15:50:38 <threebean> bonzer
15:51:21 <bowlofeggs> any other thoughts before we end the meeting?
15:51:54 <sochotni> not from my end
15:53:36 <bowlofeggs> cool, thanks for attending everyone!
15:53:41 <bowlofeggs> #endmeeting