14:00:54 <mvollmer> #startmeeting weekly meeting
14:00:54 <zodbot> Meeting started Mon Feb  1 14:00:54 2016 UTC.  The chair is mvollmer. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:54 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
14:00:54 <zodbot> The meeting name has been set to 'weekly_meeting'
14:01:01 <mvollmer> .hello mvo
14:01:02 <zodbot> mvollmer: mvo 'Marius Vollmer' <marius.vollmer@gmail.com>
14:01:37 <dperpeet> .hello dperpeet
14:01:38 <zodbot> dperpeet: dperpeet 'None' <dperpeet@redhat.com>
14:01:54 <stefw> .hello stefw
14:01:55 <zodbot> stefw: stefw 'Stef Walter' <stefw@redhat.com>
14:02:39 <mvollmer> #topic Agenda
14:03:05 <mvollmer> * new sink (low prio)
14:03:09 <stefw> * Devconf presentation
14:03:44 <dperpeet> * test image size (virtual machines)
14:03:58 <mvollmer> * iscsi update
14:05:32 <mvollmer> * weekly images
14:05:37 <mvollmer> let's go
14:05:46 <mvollmer> #topic Devconf presentation
14:06:06 <stefw> i've added a folder here with some work on the Devconf presentation
14:06:06 <stefw> https://github.com/stefwalter/cockpit-slides
14:06:17 <stefw> and am waiting for contributions from Peter and Dominik for their parts of the slides
14:06:23 <stefw> actually here: https://github.com/stefwalter/cockpit-slides/tree/master/devconf
14:06:50 <stefw> since the tuned stuff landed, and now with mvollmer's work on weekly image updates
14:07:04 <petervo> I'm working on writing and recording video, should have something to add soon
14:07:05 <stefw> we're in a pretty good place to be talking about things that people would like to contribute to or interact with
14:07:18 <stefw> I've updated the Hackfest tutorials so they work with the new CSP changes
14:07:30 <dperpeet> I think I will add something on tuned later tonight or tomorrow morning
14:07:31 <stefw> and i think we'll want to do a 0.94 release tomorrow to have things in good shape
14:07:58 <stefw> dperpeet, are you thinking someone else will write the selinux related stuff?
14:08:13 <dperpeet> stefw, no, that's coming too
14:08:27 <stefw> oh good, i guess we should talk about whether we want to practice tomorrow
14:08:57 <dperpeet> talk tomorrow or talk about practicing tomorrow?
14:09:12 <stefw> actually practice
14:09:40 <dperpeet> we can do that early tomorrow, if you wish
14:09:47 <stefw> ok cool
14:10:15 <stefw> the other thing remaining is that we need to give karma to https://bodhi.fedoraproject.org/updates/cockpit-0.93-1.fc23
14:10:51 <stefw> i can post the commands used
14:10:52 <stefw> if that helps
14:11:29 <dperpeet> what is the rpmlint failure?'
14:11:32 <dperpeet> I can't access the log
14:11:46 <stefw> not sure
14:11:52 <stefw> rpmlint always has broken failures
14:13:06 <dperpeet> ok
14:13:19 <stefw> that's probably it on that topic
14:14:08 <mvollmer> nice multihost demo
14:14:20 <mvollmer> #topic  test image size (virtual machines)
14:14:34 <dperpeet> I talked about this with mvollmer on irc
14:14:59 <dperpeet> and we can reduce the sparse image size a bit, about 700MiB for f23
14:15:28 <dperpeet> at the cost of increasing the non-sparse image size (reported size)
14:15:42 <mvollmer> how does that look after compressing?
14:15:45 <dperpeet> is this something we want to do? I believe so
14:15:54 <dperpeet> I didn't check that
14:15:59 <dperpeet> it shouild also help
14:16:06 <dperpeet> since compression looks at everything
14:16:25 <dperpeet> https://github.com/dperpeet/cockpit/commit/f40817fdd09a2aafb35510dc8c299a5be3a14bcd
14:16:34 <mvollmer> it should be faster, at the least
14:17:02 <dperpeet> mvollmer and I believe this is a good reason to port our setup scripts to be more modular
14:17:14 <dperpeet> some already support that, they're in test/lib
14:17:16 <mvollmer> yes
14:17:34 <dperpeet> I wouldn't want to add libguestfs to the vm creation anymore
14:17:40 <dperpeet> this should be inside the setup script, in my opinion
14:17:47 <dperpeet> and a per configuration choice
14:18:35 <mvollmer> no objections here.
14:18:47 <dperpeet> ok
14:19:13 <dperpeet> this is rather low priority, let me know if it becomes more important suddenly
14:19:16 <dperpeet> end of topic
14:19:34 <mvollmer> #topic  iscsi update
14:19:47 <mvollmer> so I have been poking at #2807 a bit
14:19:56 <mvollmer> rebased, etc
14:20:12 <mvollmer> the necessary versions of everything are in F23 updates-testing now
14:20:21 <mvollmer> and the test passes
14:20:31 <dperpeet> nice
14:20:42 <mvollmer> storaged should go to fedora-23 stable any day now, I think.
14:21:07 <mvollmer> https://bodhi.fedoraproject.org/updates/FEDORA-2016-6dcf004b0e
14:21:11 <mvollmer> let's see what happens
14:21:21 <mvollmer> the iscsi patch needs to go, too, I guess.
14:21:56 <mvollmer> so I will be monitoring this
14:23:19 <mvollmer> eot
14:23:30 <mvollmer> #topic weekly images
14:23:52 <mvollmer> we are pretty close to automatically making weekly images
14:24:00 <dperpeet> yay on that
14:24:12 <dperpeet> and it was nice to see the bots opening so many branches and pull requests last week
14:24:21 <stefw> yup
14:24:23 <mvollmer> i have been running this on my laptop, but the real test comes when we merge to master
14:25:19 <mvollmer> the new sink has been deployed, so yeah, looks like we finally get this! :-)
14:25:35 <mvollmer> let's see how this behaves in practice
14:25:51 <mvollmer> how we deal with changes that need to go along with the new images, for example
14:26:17 <dperpeet> well, I think we can probably cherry pick the commit and roll with it
14:26:30 <mvollmer> and how fast we overflow our quota
14:26:41 <dperpeet> quota can be added :)
14:26:51 <mvollmer> well...
14:27:03 <mvollmer> okay, a quick summary how this goes in practice?
14:27:18 <mvollmer> when github-task decides to refresh a a image:
14:27:24 <mvollmer> - opens a issue with "bot" label
14:27:30 <mvollmer> - description has link to log
14:27:47 <mvollmer> - when done with image or when it fails, adds a comment
14:27:55 <mvollmer> - turns issue into pull request
14:28:04 <mvollmer> - explicitly triggers some tests that are relevant
14:28:36 <mvollmer> when image creation fails, the pull request is still created, with a deliberately broken image link
14:28:43 <mvollmer> so all tests should fail quickly
14:28:44 <dperpeet> the trigger is notable: e.g. a fedora 23 image will also trigger a fedora-atomic check
14:29:37 <mvollmer> yeah, because fedora-atomic builds on fedora-23
14:29:56 <mvollmer> i wasn't sure when to trigger the "image/..." tests
14:30:05 <mvollmer> do they use any of the regular images?
14:30:42 <stefw> i think so
14:30:45 <stefw> we need to rename those
14:30:55 <stefw> otherwise they get confusing
14:31:04 <stefw> even though it was nice to call them container "images"
14:31:10 <stefw> they're really better called "containers"
14:31:36 <petervo> yeah, that's what the test dir is called anyways
14:32:03 <petervo> they use fedora-23 built rpms
14:32:10 <mvollmer> should we move the old tests into a directory as well?
14:32:33 <mvollmer> mv check-verify verify/run-tests
14:33:19 <dperpeet> I think this would make sense
14:33:49 <dperpeet> better grouping than filename magic
14:34:06 <mvollmer> yeah
14:34:22 <mvollmer> okay, I can take that action point
14:35:10 <mvollmer> #action mvo to move old 'verify' tests into their own directory
14:35:55 <mvollmer> okay, next?
14:36:17 <mvollmer> hmm, nothing left, really.
14:36:21 <mvollmer> #topic new sink
14:36:29 <mvollmer> so there is a new sink version
14:36:39 <mvollmer> and master will start using the new API soon
14:36:44 <mvollmer> I hope :-)
14:37:17 <mvollmer> now the sink can make pretty arbitrary REST requests to github
14:37:21 <mvollmer> which is pretty cool
14:37:45 <mvollmer> so if you are using your own sink, you need to update it
14:38:14 <dperpeet> now all we need to do is get rid of the badge-related code
14:38:20 <mvollmer> yeah
14:38:50 <dperpeet> monitoring github would be more consistent than hijacking the logging service
14:38:58 <dperpeet> with the rest of Cockpit
14:39:11 <dperpeet> and as stefw said, it would be nice to use http://shields.io/
14:39:41 <mvollmer> so, we want a service that takes a github url to a commit and a context and returns a badge png, right?
14:39:56 <mvollmer> (it also takes some text, color etc for the badge I guess)
14:40:48 <dperpeet> yeah, and either scanning or with github hooks
14:40:57 <mvollmer> can we cobble together something on fedorapeople ourselves?
14:41:27 <mvollmer> hmm, why scanning?
14:41:40 <mvollmer> to go from "master" -> sha1?
14:41:42 <dperpeet> github hooks are preferable
14:41:51 <dperpeet> that get called when master changes
14:41:57 <dperpeet> or we have results
14:42:11 <dperpeet> I think we can push this issue back a bit
14:42:19 <mvollmer> yes
14:42:21 <dperpeet> and discuss it when it's implementation time
14:45:41 <mvollmer> here is an issue that should soon turn into a pull request: https://github.com/cockpit-project/cockpit/issues/3651
14:46:00 <mvollmer> half of the time is spent compressing and uploading the image
14:46:06 <mvollmer> with no visible progress
14:46:11 <mvollmer> we could improve that, too
14:46:34 <dperpeet> it's probably a question of flushing
14:46:38 <dperpeet> among other things
14:46:49 <mvollmer> also, let's see how the real CI machines do this
14:47:02 <mvollmer> they hopefully have a fatter pipe to our image store
14:48:45 <mvollmer> #topic open floor
14:50:18 <stefw> Be sure to give karma to: https://bodhi.fedoraproject.org/updates/cockpit-0.93-1.fc23
14:50:39 <dperpeet> already on it
14:51:52 <dperpeet> it's stable
14:52:08 <dperpeet> or rather, it has enough karma
14:58:17 <mvollmer> okay
14:58:20 <mvollmer> #endmeeting