15:00:00 <bowlofeggs> #startmeeting Bodhi stakeholders (2017-03-27)
15:00:00 <zodbot> Meeting started Tue Mar 27 15:00:00 2018 UTC.  The chair is bowlofeggs. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:00 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:00:00 <zodbot> The meeting name has been set to 'bodhi_stakeholders_(2017-03-27)'
15:00:00 <bowlofeggs> #meetingname bodhi_stakeholders
15:00:00 <bowlofeggs> #topic salutations
15:00:00 <bowlofeggs> #chair bowlofeggs adamw kparal masta mboddu nirik puiterwijk Kellin
15:00:00 <zodbot> The meeting name has been set to 'bodhi_stakeholders'
15:00:00 <zodbot> Current chairs: Kellin adamw bowlofeggs kparal masta mboddu nirik puiterwijk
15:00:53 <puiterwijk> hello
15:01:26 * jsmith lurks
15:01:51 * mboddu kinda here
15:03:14 <bowlofeggs> cool, well it'll be a small meeting but i'll get started
15:03:26 <bowlofeggs> #topic announcements and information
15:03:26 <bowlofeggs> #info 3.5.2 deployed yesterday
15:03:26 <bowlofeggs> https://bodhi.fedoraproject.org/docs/user/release_notes.html
15:03:26 <bowlofeggs> #info FESCo has decided to suspend batching while they decide what the future of batching will be. in the meantime, bodhi now dequeues batched updates daily
15:03:30 * nirik is around somewhat
15:03:41 <jsmith> Thanks for doing that, bowlofeggs
15:03:57 <bowlofeggs> jsmith: sure thing :)
15:03:57 <jsmith> I know it was extra work, but will help avoid confusion until we have a better plan of action for batched updates
15:04:03 <jsmith> bowlofeggs++
15:04:14 <bowlofeggs> jsmith: it actually wasn't that much work. it was a s/2/*/ in a cron job :)
15:04:25 <jsmith> bowlofeggs: Even better :-)
15:04:32 <bowlofeggs> well, and seeking a freeze break request from infra, but that was also not hard
15:05:01 * puiterwijk wonders if we should make the FBR process harder for bowlofeggs if it's that easy :)
15:05:07 <bowlofeggs> haha
15:05:39 <bowlofeggs> i would like a FBR to re-write bodhi in a mixture of Rust, Erlang, and lolcode
15:06:03 <puiterwijk> -50
15:06:05 <bowlofeggs> ok, if there are no other thoughts on the announcements, i'll move on to the next topic, 3.6.0!
15:06:20 <bowlofeggs> #topic bodhi-3.6.0
15:06:20 <bowlofeggs> There are a number of exiting patches that are merged and waiting for Bodhi 3.6.0:
15:06:20 <bowlofeggs> * Bodhi can now "compose" containers (no more manual container publishing for releng!), though it does not yet support BROs on containers.
15:06:20 <bowlofeggs> * Bodhi is raised from 95% line test coverage to 98%, on track to reach 100% during the summer.
15:06:20 <bowlofeggs> * There is now a REST API to see the status of running or failed composes.
15:06:22 <bowlofeggs> * The Bodhi CLI now supports Python3 with 100% test coverage.
15:06:24 <bowlofeggs> * The Bodhi server theoretically also supports Python 3, but only has 78% coverage (due to skipping tests that fail in Py3) so we won't switch to Python 3 there for a few more releases.
15:06:26 <puiterwijk> bowlofeggs: hah, that's a good way to silence votes, just change topic :)
15:06:29 <bowlofeggs> * Documentation is expanded.
15:06:31 <bowlofeggs> * A variety of bug fixes.
15:06:33 <bowlofeggs> #info Bodhi 3.6.0 release notes available at https://github.com/fedora-infra/bodhi/pull/2249
15:06:35 <bowlofeggs> haha
15:07:04 <mboddu> bowlofeggs: +50 for container support
15:07:16 <bowlofeggs> nice, now i'm back to 0!
15:07:17 <mboddu> negating puiterwijk -50 :)
15:07:19 * puiterwijk notes that does not offset the -50
15:07:22 <puiterwijk> Nope
15:07:26 <mboddu> puiterwijk: :)
15:07:26 <bowlofeggs> haha
15:07:34 <puiterwijk> -50 -> immediate binning of proposal
15:07:45 <puiterwijk> But yeah, nice one on the container work :)
15:08:06 <bowlofeggs> i was hoping to have a 3.6.0 beta deployed to stg before this meeting, but i've been working on converting the spec file to do python 3 only for /usr/bin/bodhi and it was slightly tricky
15:08:23 <bowlofeggs> so i don't quite have it deployed, but i think i might be able to get it deployed this week
15:08:38 <bowlofeggs> there's also a snag with python-pyramid-fas-openid (i need to update the version in f27)
15:08:58 <bowlofeggs> so look out for an announcement about that
15:09:12 <puiterwijk> Oh? I guess I should go look at that...
15:09:25 <bowlofeggs> puiterwijk: i can do it - i have ACLs
15:09:30 <bowlofeggs> it just needs a version bump
15:09:39 <puiterwijk> Right. I'm just curious what changed :)
15:09:40 <bowlofeggs> i looked at the diff and i dont' think there are backwards compat concerns
15:09:48 <bowlofeggs> mostly it grew python 3 support
15:10:07 <bowlofeggs> the spec file still runs the python 3 tests even though it doesn't ship server in python 3
15:10:16 <nirik> python3 is really starting to arrive finally. ;)
15:10:17 <bowlofeggs> i guess i could stop it from doing that, but extra testing is a good thing :)
15:10:29 <puiterwijk> bowlofeggs: yeah, just keep doing that.
15:10:32 <puiterwijk> (I'd say)
15:10:47 <bowlofeggs> yeah that's my thought about it - we can just bump that package in f27 to get going
15:10:50 <bowlofeggs> nirik: yeah it's exciting!
15:11:49 <bowlofeggs> oh i forgot to mention above (but it is in the release notes) - the bodhi CLI can now display the status of runnign composes too
15:12:02 <bowlofeggs> so bodhi-monitor-composes is deprecated (but not removed until bodhi 4)
15:12:12 <bowlofeggs> this is nice because you dont' need to ssh to backend01 to run it :)
15:12:22 <mboddu> bowlofeggs: When do you think BRO for containers will be done?
15:12:37 <bowlofeggs> mboddu: i'm hoping "soon", hopefully for bodhi 3.7.0
15:12:38 <mboddu> Seems like thats one important requirement for me
15:12:39 <bowlofeggs> it should not be hard
15:12:43 <bowlofeggs> yeah it is important
15:12:53 <bowlofeggs> however, bodhi supporting BROs is only half the equation
15:13:06 <bowlofeggs> we also need OSBS to *use* the BROs
15:13:24 <bowlofeggs> the bodhi side of it is easy, but i'm not sure how easy the OSBS side of things will be
15:13:44 <puiterwijk> Shouldn't be too hard. But I'm not volunteering anymore.
15:13:49 <puiterwijk> (I've learned from the last week)
15:13:52 <mboddu> bowlofeggs: Right
15:13:55 <bowlofeggs> haha
15:14:07 <bowlofeggs> i should probably coordinate with cverna about this
15:15:06 <bowlofeggs> mboddu, puiterwijk, nirik: one question i have is whether we want to start using bodhi to ship containers even if it doesn't support BROs. imo, it is still useful without that functionality
15:15:18 <bowlofeggs> because it can save releng from running skopeo
15:15:26 <puiterwijk> bowlofeggs: +1
15:15:28 <puiterwijk> ship it
15:15:41 <nirik> yeah, would be nice to have things more in line with how we do packages, etc. c
15:15:43 <puiterwijk> And primarily, it makes people able to do things themselves
15:15:46 <bowlofeggs> the way we'll "ship it" is bascialyl just to create a container release once we deploy
15:15:57 <bowlofeggs> or container release(s), (f26, f27, f28?)
15:16:15 <puiterwijk> bowlofeggs: hah.... We should probably update the regex then to f\d[MC]
15:16:24 * puiterwijk submits PR
15:16:38 <bowlofeggs> yeah that's a good point
15:16:49 <puiterwijk> Luckily, that's trivial
15:17:11 <bowlofeggs> though i was able to run this stuff locally and didn't hit that problem, but perhaps i also didn't name my release object in such a way that would reveal the problem
15:17:21 <mboddu> bowlofeggs: I am totally +1 using it but it should not create more problems though :)
15:17:23 <bowlofeggs> oh and i think i actually did my testing before you merged that regex patch actually
15:17:26 <puiterwijk> Right. I don't think version_int is used in the container mashing
15:17:41 <bowlofeggs> yeah that's probably true
15:17:48 <bowlofeggs> the container "masher" is very small actually
15:17:50 <bowlofeggs> which is nice
15:17:59 <puiterwijk> So it should probably not matter, but I still think we should fix it
15:18:25 <nirik> pungier?
15:18:37 <mboddu> pungify
15:18:38 <mboddu> ?
15:18:59 <bowlofeggs> this is tne entirety of the container "masher" (I've been calling it a composer): https://github.com/fedora-infra/bodhi/blob/eea4039/bodhi/server/consumers/masher.py#L863-L902
15:19:07 <bowlofeggs> puiterwijk: agred
15:19:37 <mboddu> bowlofeggs: quick question, does container releases depend on fedora release state in bodhi?
15:20:30 <bowlofeggs> mboddu: i'm not sure i fully grasp the question - do you mean the settings in production.ini that configure whether f28 is pre-beta, post-beta, etc.?
15:21:09 <mboddu> bowlofeggs: Sorta, for example bodhi is not enabled in rawhide, but we build containers for rawhide and release them
15:21:31 <nirik> not yet anyhow. ;)
15:21:33 <mboddu> Also, bodhi is not enabled for two weeks after we branch, but we will build containers in the mean time as well
15:21:52 <puiterwijk> mboddu: no, layered containers should always go through Bodhi with this system I'd say
15:22:02 <puiterwijk> Worst case we set minimum_days_testing=0
15:22:06 <bowlofeggs> mboddu: ah. well the container release object in bodhi is separate of the fedora rpm release object, so we could use bodhi to do that in theory, though if we made it inconsistent with rpms that might confuse packagers
15:22:30 <mboddu> puiterwijk: Okay
15:22:39 <bowlofeggs> mboddu: also there has been some discussion about the possibility of using bodhi on rawhide, though that's still being debated
15:23:01 <puiterwijk> bowlofeggs: it'll be inconsistent (for now?), but it'd make sure releng never has to do that
15:23:21 <bowlofeggs> yeah
15:23:29 <mboddu> bowlofeggs: So, we should talk to WG and let them know about how we plan to release them using bodhi and the time constraints
15:23:30 <bowlofeggs> i'm certainly not opposed to doing it - it should work in theory
15:24:01 <bowlofeggs> basically we'd create a rawhide container release object in bodhi and go to town
15:24:16 <mboddu> If we decide to follow rpm schedule for consistency
15:25:04 <puiterwijk> mboddu: if we decide to do the same as for RPMs, we'd either 1. need to re-implement it again, or 2. have releng still do manual pushes for rawhide.
15:25:07 <nirik> we could also just do like we do for rpms... and have the nightly compose deal with them... but that would take some work
15:25:11 <puiterwijk> Neither of which I think is a good idea
15:25:21 <puiterwijk> nirik: yeah, and need that code written
15:25:29 <nirik> yeah.
15:25:39 <mboddu> nirik: Yes
15:25:59 <mboddu> puiterwijk: Yes, I agree
15:26:53 <bowlofeggs> i do think bodhi would get the job done with its existing code, though i haven't tested it for rawhide - i don't know of anythign special about rawhide that would be different other than the naming conventions
15:27:29 <bowlofeggs> the container release's branch attribute is used to determine which prefix to put in front of the container in the registry
15:27:44 <bowlofeggs> for example, f27/httpd - the "f27" comes from the release's branch
15:28:05 <bowlofeggs> we might want to lie to bodhi and not say "master" for --branch on rawhide, which might be a little weird
15:28:29 <mboddu> Rawhide is always special
15:28:36 <bowlofeggs> i actualyl don't see rawhide layered images in our prod registry right now, fwiw
15:28:47 <bowlofeggs> so it looks like we just don't ship them today?
15:29:00 * mboddu remembers there is a ticket about it
15:29:04 <mboddu> I will go and find it
15:29:04 <bowlofeggs> ah ok
15:29:23 <bowlofeggs> well if we want them to look like rawhide/httpd, we can just set --branch to "rawhide" and bodhi won't know we're lying
15:29:34 <bowlofeggs> it's a little hacky but it shoudl work
15:30:00 <bowlofeggs> that's the only special thing i can think of
15:30:11 <mboddu> https://pagure.io/atomic-wg/issue/439
15:30:24 <mboddu> Okay, they plan on doing it for latest stable and branched releases
15:31:42 <bowlofeggs> fwiw, i also wonder whether it is useful for us to prepend the branch name onto the container name
15:32:11 <bowlofeggs> do users care which base os the container is when they just want httpd?
15:32:24 <bowlofeggs> or maybe it's a way to handle backwards incompatible changes?
15:32:30 <bowlofeggs> maybe it is useful
15:32:49 <bowlofeggs> in any case, yeah, we can start using bodhi to ship containers with 3.6.0 :)
15:32:54 <bowlofeggs> any other thoughts?
15:33:03 <mboddu> I think it is useful, but not sure if they care about non-stable releases(rawhide)
15:33:07 <mboddu> bowlofeggs: +1
15:35:46 <bowlofeggs> alright, let's move on
15:36:00 <bowlofeggs> #topic Looking forward
15:36:00 <bowlofeggs> I'm still working on giving Bodhi a REST API for managing composes. The read only support is there, we just need to make it have control support now.
15:36:26 <bowlofeggs> 3.6.0 has a read-only REST API that is the beginnings of this, and it has CLI support to use it too
15:37:00 <bowlofeggs> the idea is that some not-too-distance bodhi release will allow releng to use the regular bodhi CLI from the comfort of their own laptop to drive and monitor composes
15:37:17 <bowlofeggs> i.e., no need to ssh to bodhi-backend01 to start or monitor composes
15:37:37 <bowlofeggs> admittedly, read support is easier than write support, but it's a step :)
15:38:04 * nirik looks forward to it.
15:38:27 <bowlofeggs> whenever that is supported, bodhi-push will become deprecated like bodhi-monitor-composes
15:38:41 <bowlofeggs> it'll probably be pretty similar to bodhi-push though
15:40:05 <bowlofeggs> alright, next topic
15:40:07 <mboddu> bowlofeggs: How many cookies you want?
15:40:17 <bowlofeggs> i love cookies
15:40:23 <bowlofeggs> #topic triage
15:40:23 <bowlofeggs> #info Bodhi's high priority issues: https://github.com/fedora-infra/bodhi/issues?q=is%3Aopen+is%3Aissue+label%3A%22High+priority%22
15:40:23 <bowlofeggs> Are there any items that are not on that list that should be? Anything on that list that should not be?
15:40:42 * mboddu is ready to give all the cookies to bowlofeggs for making bodhi pushes run on personal laptops
15:41:12 <bowlofeggs> there are also 2 critical issues open: https://github.com/fedora-infra/bodhi/issues?q=is%3Aopen+is%3Aissue+label%3ACritical
15:41:30 <bowlofeggs> i think puiterwijk actually already fixed this one though: https://github.com/fedora-infra/bodhi/issues/2211
15:41:41 <bowlofeggs> the other one is critical because it sucks and i need to make sure it is fixed before f29
15:41:56 <nirik> there may also be some work/issues from rawhide gating once we dive into that...
15:43:02 <bowlofeggs> true
15:43:08 <mboddu> bowlofeggs: Please make sure we get https://github.com/fedora-infra/bodhi/issues/1337 fixed before F29
15:43:36 <bowlofeggs> mboddu: yeah that's why i marked it critical (and it's now the only critical ticket)
15:43:37 <mboddu> Huh, you already messaged about it
15:43:43 <bowlofeggs> mboddu: also that ticket is so 1337
15:43:45 <nirik> that issue is so 1337
15:43:48 <bowlofeggs> hahaha
15:43:48 <nirik> ha.
15:44:03 <bowlofeggs> mboddu: that ticket will also bite us when we make container releases
15:44:32 <bowlofeggs> i probably won't be able to fix it in time for that, but i can manually fix them again like last time
15:44:50 <bowlofeggs> i do think i can do this for f29 though
15:45:01 <mboddu> bowlofeggs, nirik : haha :)
15:45:27 <mboddu> bowlofeggs: Okay, thanks
15:45:54 * masta looks in
15:46:01 <masta> hey sry late
15:51:30 <bowlofeggs> ok, let's move on
15:51:38 <bowlofeggs> #topic Open floor
15:54:44 * nirik has nothing aside from the rawhide gating work which might heat up bodhi work...
15:58:01 <bowlofeggs> yeah
15:58:16 <bowlofeggs> at this point it seems too fuzzy to discuss in the last minutes of an IRC meeting
15:58:23 <bowlofeggs> but agreed that might be a lot of work :)
15:58:30 <bowlofeggs> alright, thanks for coming everyone!
15:58:31 <nirik> yep. just mentioning it... nothing to really discuss here.
15:58:36 <bowlofeggs> #endmeeting