cockpit
MINUTES
13:00:20 <mvollmer> #startmeeting
13:00:20 <zodbot> Meeting started Mon Jun  1 13:00:20 2015 UTC.  The chair is mvollmer. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:00:20 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
13:00:23 <mvollmer> :-)
13:00:26 <mvollmer> .hello mvo
13:00:27 <dperpeet> .hello dperpeet
13:00:29 <zodbot> mvollmer: mvo 'Marius Vollmer' <marius.vollmer@gmail.com>
13:00:29 <dperpeet> on the dot!
13:00:32 <zodbot> dperpeet: dperpeet 'Dominik Perpeet' <dperpeet@redhat.com>
13:00:35 <andreasn> .hello andreasn
13:00:36 <zodbot> andreasn: andreasn 'Andreas Nilsson' <anilsson@redhat.com>
13:00:41 <Shad0w_Crux> .hello tengland
13:00:42 <zodbot> Shad0w_Crux: tengland 'Turner England Jr' <turner.england@gmail.com>
13:00:51 <mvollmer> wow, everyone is ready....
13:01:15 <mvollmer> welcome, Shad0w_Crux
13:01:25 <Shad0w_Crux> Hello
13:01:42 <mvollmer> ok, let's collect the agenda
13:01:46 <mvollmer> #topic Agenda
13:01:53 <mvollmer> * rawhide
13:02:03 <mvollmer> * storage rewrite
13:02:40 <dperpeet> * fedora atomic host
13:03:27 <mvollmer> * little kubernetes update
13:03:54 <mvollmer> pstree maybe?
13:04:00 <mvollmer> *ostree
13:04:02 <mvollmer> sorry
13:04:05 <mvollmer> muscle memory
13:04:12 <andreasn> :)
13:04:53 <mvollmer> Shad0w_Crux, do you have a topic you would like to discuss?
13:05:32 <mvollmer> okay, let's start
13:05:36 <dperpeet> I think it would be nice to have a quick intro from Shad0w_Crux and links to the blog et cetera in the meeting notes
13:05:36 <mvollmer> #topic rawhide
13:05:45 <dperpeet> at the end
13:05:55 <mvollmer> ok.
13:06:21 <mvollmer> ahh, now I get the picture.
13:06:28 <mvollmer> ok, so rawhide.
13:06:45 <mvollmer> I have started testing the new storaged code on rawhide, because that is where storaged is,
13:06:57 <mvollmer> i found two bad bugs
13:07:22 <mvollmer> GFileMonitor is broken (I think) and udev
13:07:28 <mvollmer> i dig out the links...
13:07:48 <mvollmer> Please stand by .
13:08:06 <mvollmer> #info https://bugzilla.redhat.com/show_bug.cgi?id=1225905
13:08:27 <mvollmer> #info https://bugzilla.redhat.com/show_bug.cgi?id=1225957
13:08:31 <mvollmer> sorry for not being prepared
13:08:47 <mvollmer> there are workarounds, so I can continue testing
13:09:15 <mvollmer> looks good so far, right now I am doing the lvm2 tests, about 70% of everything, maybe
13:09:37 <mvollmer> i haven't looked at other tests so far
13:09:52 <dperpeet> is it too early to tell whether someone will try to fix those bugs or if we have to target them?
13:10:06 <mvollmer> they must be fixed, I'd say
13:10:48 <dperpeet> do you need verification on the GFileMonitor issue or should we wait?
13:11:06 <mvollmer> verification would be nice
13:11:23 <mvollmer> i was going to make a small reproducer
13:11:26 <dperpeet> ok, if I have some spare time I'll try to reproduce
13:11:32 <mvollmer> nice, thanks.
13:12:29 <mvollmer> everyone can now run the tests on rawhide by setting TEST_OS=fedora-rawhide
13:12:45 <mvollmer> hubbot doesn't do it yet, because we know what'll happen.
13:12:56 <mvollmer> it runs them on master, though
13:13:09 <mvollmer> another interesting point:
13:13:25 <mvollmer> rawhide and f22 images are now created from scratch, without the magic base tarballs.
13:13:31 <mvollmer> finally!
13:13:34 <dperpeet> good news!
13:14:08 <mvollmer> next would be to set up hubbot to create new ones once per week or so.
13:14:10 <mvollmer> or daily.
13:14:20 <dperpeet> but with manual merge?
13:14:23 <mvollmer> yes
13:14:46 <mvollmer> we should get to the point where merging a PR will automatically take the new images into use
13:14:49 <dperpeet> I think this should be a hubbot issue
13:15:09 <dperpeet> can you file this in https://github.com/cockpit-project/hubbot?
13:15:09 <mvollmer> right now, one has to run vm-publish-image separately
13:15:17 <mvollmer> ok
13:15:27 <dperpeet> I can tackle that soonish, as it'll save us manual work
13:15:49 <mvollmer> #action mvo to record automatic image installation
13:16:33 <dperpeet> #action dperpeet to verify bug https://bugzilla.redhat.com/show_bug.cgi?id=1225957
13:16:41 <mvollmer> ok, that's the state of rawhide
13:16:45 <mvollmer> testable but failing
13:17:07 <mvollmer> I am really curious how the udev bug will be resolved
13:17:19 <dperpeet> it sounds pretty serious
13:17:30 <mvollmer> yeah
13:17:33 <dperpeet> in the sense that it should have been noticed
13:17:42 <mvollmer> I think systemd had a biggish cleanup in udev in 220
13:17:53 <mvollmer> but don't quote me
13:18:04 <mvollmer> i wasn't really looking carefully
13:18:21 <mvollmer> well, we noticed it.
13:18:23 <mvollmer> :-)
13:18:47 <mvollmer> okay, next?
13:18:58 <andreasn> yep
13:18:58 <dperpeet> I'm glad we're testing rawhide now, easier to catch things like that in time
13:19:23 <mvollmer> yep, let's get it green.
13:19:37 <mvollmer> #topic storage rewrite
13:19:46 <mvollmer> i think that was mostly covered above
13:19:58 <mvollmer> i am now porting over the tests
13:20:01 <mvollmer> looks good so far
13:20:09 <dperpeet> and hopefully splitting them up? =)
13:20:14 <mvollmer> the same old irritating race conditions
13:20:24 <mvollmer> yes, there will be ten files, or so.
13:20:33 <mvollmer> and storagelib.py
13:21:00 <dperpeet> if you're rewriting any tests, feel free to squelch expected output
13:21:09 <dperpeet> unless it's undesired of course
13:21:19 <mvollmer> okay
13:21:24 <mvollmer> good point
13:21:25 <dperpeet> our tests have become more quiet, but they aren't quite there yet
13:21:55 <mvollmer> yep, but it's riksy to just throw away all output
13:22:01 <dperpeet> yes
13:22:06 <mvollmer> i think I'll grep out expected things
13:22:27 <dperpeet> I think it's only worth doing if you're looking at a command anyway
13:22:30 <mvollmer> one new invention that is worth discussing:
13:22:46 <mvollmer> I decided to do "deep inspection" in the tests,
13:23:01 <mvollmer> which means that the page javascript has a hook for evaluating arbitrary code in its scope
13:23:15 <dperpeet> that is nice
13:23:26 <mvollmer> oh, good that you say this. :-)
13:23:37 <dperpeet> it might save us some filesystem magic
13:23:43 <dperpeet> in other tests
13:23:58 <mvollmer> it's one of these typical backdoors "for testing" that might come back later to bite us
13:24:19 <dperpeet> we have sizzle anyway, so it's not like we changed the concept
13:24:24 <mvollmer> it might make it possible to inject code into our origin etc
13:24:38 <dperpeet> as long as the hook isn't invasive by default (i.e. changes behavior), I think it's good
13:24:45 <mvollmer> so we have to harden the hook, I guess
13:25:06 <mvollmer> let's check it in review, but I think it is a bit nasty right now.
13:25:44 <mvollmer> https://github.com/mvollmer/cockpit/blob/direct-storaged/pkg/storaged/devices.js#L15
13:26:48 <dperpeet> mvollmer, I would have expected this to be in the testing code
13:26:51 <mvollmer> It probably shouldn't be on window, but on something that only PhantomJS can access
13:27:44 <mvollmer> anyway, needs more thought.
13:27:59 <mvollmer> but the feature is really nice
13:28:25 <mvollmer> okay, I am taking too much time.
13:28:48 <mvollmer> #topic fedora atomic host
13:29:08 <dperpeet> thanks for you review on https://github.com/cockpit-project/cockpit/pull/2231 mvollmer
13:29:33 <mvollmer> is there "Fedora Atomic Server" and "Fedora Atomic Workstation"?
13:29:41 <dperpeet> good suggestions, and I'll work on the bootstrap script first - my pull request originated before we had that in our tests
13:29:44 <andreasn> no, not yet at least
13:29:51 <dperpeet> I think atomic is a server
13:30:07 <dperpeet> that's how the environment is currently understood
13:30:48 <dperpeet> we even define the users at bootup using a separate image
13:30:57 <dperpeet> cloud-init
13:31:04 <dperpeet> anyway, the tests are looking good
13:31:27 <dperpeet> a few tests we have decided to skip
13:31:30 <andreasn> the workstation stuff needs to be able to install things that are not in the core os image https://lists.fedoraproject.org/pipermail/desktop/2015-May/012362.html
13:32:10 <dperpeet> either because they don't make sense (e.g. check-loopback, which tests ssh loopback connects that all other tests use anyway on atomic)
13:32:25 <dperpeet> or because of dependencies we have in our testing daemon
13:32:31 <dperpeet> testing service, I mean
13:32:38 <dperpeet> that is resolvable, but future work
13:32:59 <dperpeet> in the meantime, check-pages and check-roles are the only two tests that fail
13:33:13 <dperpeet> and I believe a second set of eyes would help on that
13:33:22 <mvollmer> ok, I have a look.
13:33:32 <dperpeet> I propose that I integrate your comments first
13:33:35 <mvollmer> I need help to get started, though.
13:33:38 <dperpeet> should be done by end of today
13:33:39 <mvollmer> okay
13:33:47 <dperpeet> and I'll get it running on the CI server
13:33:52 <mvollmer> ok, good.
13:34:14 <dperpeet> #action dperpeet to run atomic test via hubbot
13:34:42 <dperpeet> end of topic
13:34:56 <dperpeet> (should be ready for merge this week)
13:35:12 <mvollmer> I'll be away starting on Friday
13:35:20 <dperpeet> let's get it done before then
13:35:30 <mvollmer> okay
13:35:54 <mvollmer> #topic little kubernetes update
13:36:11 <mvollmer> I merged some of stefs kubernetes work, and it's really nice
13:36:41 <mvollmer> now there are shells in the containers
13:36:44 <mvollmer> and graphs
13:36:50 <mvollmer> and plots
13:37:03 <mvollmer> and a whole lot of stuff going on
13:37:08 <mvollmer> and it all looks very nice
13:37:11 <andreasn> looks really cool
13:37:13 <mvollmer> IMO
13:38:09 <mvollmer> some filtering is coming, I think, which should help keep this all useful for non-toy deployments
13:38:20 <andreasn> yup
13:39:51 <mvollmer> (heh, lvm2 tests have just passed here :-)
13:40:19 <mvollmer> okay, next?
13:40:28 <andreasn> sure
13:40:29 <mvollmer> GSoC?
13:40:55 <Shad0w_Crux> So for the summer I'll be working on adding a UI for Rolekit. I spent last week reading documentation and source code. I'm really new to working on open source stuff and so I don't have a lot of experience working with someone else's code (or the workflow in general). I also spent some time with a few VMs doing various things with both Cockpit and Rolekit.
13:41:09 <mvollmer> #topic GSoC
13:41:09 <Shad0w_Crux> My updates are plugged into Planet Fedora and can be seen here:
13:41:09 <Shad0w_Crux> http://planet.fedoraproject.org/summer-coding/ ..and here: http://turnerengland.com/
13:41:29 <Shad0w_Crux> (oops sorry, a little early)
13:41:39 <mvollmer> no problem :-)
13:41:59 <andreasn> cool
13:42:06 <andreasn> nice to meet you Shad0w_Crux!
13:42:10 <dperpeet> yes, welcome
13:42:18 <Shad0w_Crux> Thanks everyone.
13:42:42 <dperpeet> I like the level of detail in what you already posted
13:43:00 <Shad0w_Crux> Thanks. I hope to be able to keep it at that level.
13:43:26 <sgallagh> dperpeet: Have you had much interaction with jskorepa?
13:43:27 <dperpeet> if you need help with the dev environment or workflow, don't hesitate to ask
13:43:46 <dperpeet> sgallagh, sadly not, he hasn't been online that much here
13:43:55 <Shad0w_Crux> Ok. Will probably have multiple questions as the week progresses. :)
13:44:14 <dperpeet> sgallagh, I think we will have to set some fixed meeting times
13:44:14 <sgallagh> Shad0w_Crux: Any rolekit questions can be directed at me. I'm usually around.
13:44:30 <Shad0w_Crux> sgallagh: Then we'll be in touch. :)
13:44:34 <dperpeet> Shad0w_Crux, if you have nested vm questions, let me know
13:44:43 <dperpeet> as far as I know few people do that
13:44:50 <dperpeet> in case you want to play it safe :)
13:44:56 <Shad0w_Crux> dperpeet: Got it. :)
13:45:18 <andreasn> Shad0w_Crux: and I'm very happy to help with the design side of things. Just ping me at any time
13:45:31 <Shad0w_Crux> andreasn: Will do.
13:47:19 <mvollmer> I'll answer about the internal APIs then. :-)
13:47:30 <sgallagh> Ignore that nick change; I meant to type /whois :-p
13:47:53 <andreasn> Shad0w_Crux: where are you based?
13:48:08 <Shad0w_Crux> andreasn: USA (Indiana)
13:49:25 <andreasn> me, mvollner and dperpeet are all based in Europe
13:49:38 <andreasn> and so is stefw
13:49:46 <andreasn> but sgallagh and petervo are both based in the US
13:50:01 <dperpeet> but we're all online at odd times, so someone should always be here :)
13:50:51 <mvollmer> ok, any other topic?
13:52:22 <andreasn> nope
13:52:30 <mvollmer> yep, nope
13:52:39 <mvollmer> thanks everyone!
13:52:44 <mvollmer> #endmeeting