14:01:11 <mvollmer> #startmeeting
14:01:11 <zodbot> Meeting started Mon Nov 16 14:01:11 2015 UTC.  The chair is mvollmer. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:01:11 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
14:01:13 <dperpeet> .hello dperpeet
14:01:14 <zodbot> dperpeet: dperpeet 'None' <dperpeet@redhat.com>
14:01:15 <mvollmer> .hello mvo
14:01:19 <zodbot> mvollmer: mvo 'Marius Vollmer' <marius.vollmer@gmail.com>
14:01:32 <mvollmer> #topic Agenda
14:01:53 <mvollmer> * Mock in VM
14:01:56 <dperpeet> * fedora atomic 22 vs 23
14:02:05 <stefw> .hello stefw
14:02:06 <zodbot> stefw: stefw 'Stef Walter' <stefw@redhat.com>
14:02:28 <mvollmer> * fedora-testing
14:02:56 <dperpeet> * cockpit on debian
14:04:52 <mvollmer> okay!
14:05:04 <mvollmer> #topic Mock in VM
14:05:13 <mvollmer> Work is progressing
14:05:29 <mvollmer> #info https://github.com/cockpit-project/cockpit/pull/3138
14:05:43 <mvollmer> new images needed to be made, which found new bugs
14:05:49 <mvollmer> many are already fixed again
14:06:03 <mvollmer> so Fedora 23 setup is pretty clean now
14:06:19 <mvollmer> i want to test atomic and fedora-testing manually
14:06:26 <mvollmer> but have trouble with atomic
14:06:35 <mvollmer> it doesn't announce its IP apparently
14:06:44 <dperpeet> I'm working on reproducing that
14:06:45 <mvollmer> so the tests hang
14:07:06 <mvollmer> the mock related changes are working, though
14:07:24 <mvollmer> but would be nice to get a green light on the PR also for aotmic
14:07:41 <mvollmer> I'll 'rebase' fedora-testing on Fedora-23 now
14:07:57 <mvollmer> "should be straightforward"
14:08:14 <dperpeet> do we want a green light on atomic to be required to get this in?
14:08:19 <petervo> atomic isn't really working now
14:08:26 <petervo> so it probably shouldn't block this
14:08:29 <mvollmer> okay
14:08:36 <dperpeet> good to know!
14:08:37 <mvollmer> but let's not let it rot...
14:08:43 <petervo> though we do need it fixed
14:09:03 <mvollmer> and we should make the "mock-in-vm" stuff work for it too
14:09:15 <mvollmer> so that we don't need mess with that when we fix it for real
14:09:16 <dperpeet> that would be best
14:09:30 <petervo> yep
14:09:42 <mvollmer> okay, so mock-in-vm is close to be done.
14:09:52 <dperpeet> good job
14:09:59 <stefw> nice
14:10:02 <mvollmer> thanks
14:10:09 <mvollmer> took longer than expected, as always
14:10:29 <mvollmer> hofstadtlers law
14:10:35 <mvollmer> sp?
14:10:52 <mvollmer> https://en.wikipedia.org/wiki/Hofstadter's_law
14:11:18 <mvollmer> next?
14:11:37 <dperpeet> let me know if there are parts to review already
14:11:48 <mvollmer> yes, it can be reviewed
14:12:00 <dperpeet> ok
14:12:25 <mvollmer> ok, next.
14:12:39 <mvollmer> #topic fedora atomic 22 vs 23
14:12:54 <dperpeet> since fedora atomic 23 has been released, the question is what to focus on
14:13:03 <dperpeet> do we still need to test 22?
14:13:15 <dperpeet> or try to get 23 up and running
14:13:47 <stefw> i would suggest getting fedora 22 going
14:13:50 <stefw> at least first
14:13:59 <stefw> typically Fedora Atomic has lagged Fedora development
14:14:03 <stefw> while they roll new releases regularly
14:14:18 <stefw> it has been the case that these releases have been rolled from Fedora 22 througout the Fedora 23 development cycle
14:14:29 <stefw> i would suggest getting it working as a separate step to upgrading it to Fedora 23
14:15:09 <dperpeet> so we get 22 working first?
14:15:38 <dperpeet> atomic host is based on fedora 23
14:15:48 <dperpeet> on the official page https://getfedora.org/cloud/download/atomic.html
14:15:56 <stefw> yup
14:16:02 <mvollmer> petervo, is it possible to quickly summarize what is broken about atomic?
14:16:17 <stefw> but i would think we should work through the issues and get back to a known good working state, and then move forward
14:16:21 <mvollmer> i mean, our fedora-atomic-22 test config
14:16:26 <petervo> haven't really looked into it
14:16:27 <dperpeet> yes, I agree stefw
14:16:35 <mvollmer> okay
14:16:51 <petervo> just haven't been able to run it since we moved away from hubbot
14:17:08 <petervo> i was going to pick it next if no one else had by then
14:17:13 <dperpeet> did you try to create a new image?
14:17:17 <dperpeet> I know that stefw rolled back a few
14:17:21 <dperpeet> not sure if atomic was among those
14:17:24 <petervo> there is also this https://github.com/cockpit-project/cockpit/pull/3081
14:17:39 <stefw> petervo, yes, i didn't have time to finish new atomic images
14:18:49 <dperpeet> stefw, should mvollmer and me pick up from #3081?
14:19:55 <stefw> sure
14:20:11 <dperpeet> mvollmer, I'll look at atomic today
14:20:16 <dperpeet> let's keep in touch
14:20:23 <mvollmer> okay
14:20:48 <mvollmer> next`
14:20:57 <mvollmer> next?
14:21:03 <dperpeet> yes
14:21:17 <mvollmer> #topic fedora-testing
14:21:33 <mvollmer> as part of mock-in-vm, I'll switch fedora-testing over to Fedora 23.
14:21:37 <mvollmer> okay? :-)
14:21:50 <dperpeet> yes
14:21:54 <dperpeet> let's see everything turn red
14:22:11 <stefw> sure
14:22:11 <mvollmer> should we run it regularly?
14:22:21 <stefw> yes, it will be helpful
14:22:27 <stefw> especially when tracking updates that fix regressions
14:22:33 <mvollmer> only master or all PRs?
14:22:33 <stefw> i guess the question is, where should we run it?
14:22:48 <stefw> probably just on master, no? we don't currently have a mode for that
14:22:52 <mvollmer> now with the naughty machinery, we could run it for each PR, actually.
14:23:17 <mvollmer> check-verify --github=master should do it, no?
14:23:27 <stefw> mvollmer, hmm, yes probably
14:23:37 <mvollmer> but the container doesn't do it.
14:23:37 <stefw> although it will just test the same thing over and over again
14:23:42 <mvollmer> yep
14:23:48 <stefw> we just need more hardware in that case ... or make the verifiers just pick jobs from a list like we talked about
14:23:57 <stefw> rather than each being limited to a single TEST_OS
14:24:00 <stefw> or N TEST_OS
14:24:38 <stefw> maybe i can work on that later this week
14:24:52 <dperpeet> I think having verifiers pick a job from a list with multiple TEST_OS values would help scalability
14:24:57 <stefw> and then it should be simple to schedule tests of all the operating systems, in whatever combination we want
14:25:38 <mvollmer> yes, pretty much in the same way that hubbot did it
14:25:50 <stefw> right except distributed
14:25:54 <mvollmer> yes
14:26:10 <mvollmer> but one machine only runs one OS at any one time, right?
14:26:24 <mvollmer> but it switches between them from one run to the next
14:27:07 <stefw> yeah
14:27:14 <stefw> the TEST_OS will be set as part of --github=next
14:27:27 <mvollmer> yep
14:27:47 <mvollmer> would it then make sense to split check-verify into two parts?
14:28:04 <mvollmer> one that figures out what to test and checkout the right sources, and the other that runs all the tests?
14:28:12 <stefw> that's the way it is already
14:28:23 <mvollmer> i mean, two source files
14:28:24 <stefw> testlib.py is the second part
14:28:34 <mvollmer> hmm
14:29:37 <mvollmer> yeah, good enough, I think
14:30:07 <stefw> we can certainly move parts of it to testlib.py for any particular part where that makes sense
14:30:22 <mvollmer> maybe invoking testsuite-prepare
14:30:30 <stefw> yes, that colud be part of testlib.py
14:30:51 <mvollmer> okay
14:32:02 <mvollmer> okay, so we wait with fedora-testing until that machinery works.
14:32:14 <stefw> makes sense
14:32:22 <mvollmer> I have one more topic
14:32:41 <mvollmer> shall I go ahead?
14:32:43 <dperpeet> so do I (listed above)
14:32:52 <mvollmer> ahh.
14:32:57 <dperpeet> but go ahead if you wish
14:33:13 <mvollmer> #topic cockpit on debian
14:33:19 <mvollmer> no, sorry, it slipped my mind
14:33:28 <dperpeet> last week we told a bit about systemd.conf
14:33:45 <dperpeet> as a follow-up I wrote down how we got Cockpit working on debian: http://dominik.perpeet.eu/cockpit-on-debian-8-2
14:33:55 <dperpeet> there were a few responses
14:34:18 <dperpeet> mbiebl had a follow-up session last weekend and hacked on it a bit more
14:34:20 <dperpeet> https://anonscm.debian.org/cgit/collab-maint/cockpit.git/
14:34:36 <dperpeet> and also mentioned the progress on the systemd mailing list http://lists.alioth.debian.org/pipermail/pkg-systemd-maintainers/2015-November/009477.html
14:34:59 <dperpeet> I'm in communication with him and will also look at the packaging
14:35:26 <dperpeet> some of the issues were filed on github, thanks stefw for working on that and merging things quickly
14:35:52 <mvollmer> really nice
14:35:53 <dperpeet> mbiebl is currently just hacking on this, but not committed to become a cockpit maintainer on debian
14:35:56 <dperpeet> so we're still looking
14:36:16 <dperpeet> but the goal is to get as many of the changes we need for debian to go upstream into cockpit
14:36:29 <dperpeet> and the idea is that once it works, it'll be easier for someone to become the maintainer
14:37:52 <dperpeet> discussions and what should go in and what shouldn't we'll just have as they arise
14:37:54 <dperpeet> end of topic
14:37:57 * mvollmer is looking forward to see Cockpit with Debian branding
14:38:11 <mvollmer> #topic pyblk
14:39:01 <mvollmer> i am discussing with the pyblk author about how to improve Cockpit storage with it
14:39:24 <mvollmer> pyblk is a 'next-gen' lsblk
14:39:35 <stefw> that's a library that one could use to draw a d3 graph of the various storage objects in the system, right?
14:39:39 <stefw> because there's another one for network
14:39:42 <stefw> called plotnetconfig
14:39:44 <mvollmer> #info https://github.com/mulkieran/pyblk/
14:40:00 <mvollmer> stefw, yes, it can draw the graph
14:40:05 <mvollmer> i see
14:40:17 <mvollmer> discussion is on cockpit-devel
14:40:50 <mvollmer> https://lists.fedorahosted.org/pipermail/cockpit-devel/2015-November/000372.html
14:41:01 <mvollmer> heh, did I not reply to the list?
14:41:05 <mvollmer> I'll fix that
14:41:23 <mvollmer> i guess I replied to the 'wrong' mail
14:41:32 <mvollmer> so, discussion will be on the list. :-)
14:42:36 <mvollmer> eot?
14:42:39 <dperpeet> wait
14:42:55 <dperpeet> mvollmer, do you know if GiB/GB et cetera is also part of that design effort?
14:43:09 <mvollmer> i donÄt think it is
14:43:09 <dperpeet> looking over the doc quickly didn't tell me anything about that
14:43:23 <dperpeet> let's poke them about it
14:43:33 <dperpeet> if there are discussions
14:43:38 <mvollmer> you do that, dperpeet, you do that. :-)
14:43:47 <dperpeet> ok, I'll poke the sleeping tiger
14:44:49 <mvollmer> okay, that's it?
14:46:34 <mvollmer> all right, thanks everybody!
14:46:38 <mvollmer> #endmeeting