14:02:23 <mvollmer> #startmeeting Meeting
14:02:23 <zodbot> Meeting started Mon Jan 23 14:02:23 2017 UTC.  The chair is mvollmer. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:02:23 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
14:02:23 <zodbot> The meeting name has been set to 'meeting'
14:02:33 <mvollmer> .hello mvo
14:02:34 <zodbot> mvollmer: mvo 'Marius Vollmer' <marius.vollmer@gmail.com>
14:02:37 <dperpeet> .hello dperpeet
14:02:38 <zodbot> dperpeet: dperpeet 'None' <dperpeet@redhat.com>
14:03:43 <mvollmer> #topic Agenda
14:04:04 <dperpeet> * the GitHub BUG label
14:04:09 <mvollmer> * Network integration tests
14:07:45 <mvollmer> anything else?
14:08:25 <mvollmer> #topic the GitHub BUG label
14:08:51 <dperpeet> we were discussing this last week on IRC
14:09:00 <pitti> background: last week I tagged a PR with "bug" (old habit from systemd where we categorize PRs and issues) and was told not to
14:09:00 <dperpeet> and the question is: do we still need the BUG label?
14:09:18 <dperpeet> originally we had this for bugs in the issues
14:09:37 <pitti> indeed, if we don't do that let's just remove them -- it's dead cheap to put back if we ever want it again, after all?
14:09:39 <dperpeet> and we still quite a few https://github.com/cockpit-project/cockpit/issues?q=is%3Aopen+is%3Aissue+label%3Abug
14:10:05 <dperpeet> but theoretically our issues should all be bugs
14:10:07 <mvollmer> yeah, I don't pay attention to that label
14:10:17 <pitti> dperpeet: you don't use issues for RFEs?
14:10:18 <larsu> yeah, I agree that this label is a bit useless
14:10:39 <dperpeet> we found that wishes don't age well
14:10:47 <dperpeet> either it's on our trello roadmap
14:10:53 <dperpeet> or someone wants to implement something
14:10:57 <pitti> in systemd we use it for "bug" vs "RFE", and have additional label for the component (network, pid1, nspawn, etc) -- in cockpit this could be "storage", "kubernetes", etc.
14:11:12 <larsu> component ones I would understand
14:11:14 <dperpeet> just putting a wish out there with nobody to work on it usually doesn't lead to any progress
14:12:01 <dperpeet> I mean, ideas are awesome
14:12:10 <dperpeet> but we have more of those than we can implement
14:12:30 <dperpeet> so either you start implementing something
14:12:35 <dperpeet> or convince others to implement something
14:12:41 <pitti> so, if RFE is frowned upon and we don't use it, then "bug" is useless too indeed
14:13:03 <dperpeet> pitti, RFEs are very valuable
14:13:10 <dperpeet> but they need to be in the right form
14:13:23 <pitti> https://github.com/cockpit-project/cockpit/labels/enhancement
14:13:29 <pitti> 72 of them
14:13:36 <dperpeet> yeah, this is for small stuff usually
14:13:41 <dperpeet> that's a different discussion I think
14:14:21 <dperpeet> ok, anybody against removing the "bug" label from github?
14:14:45 <dperpeet> we have "knownissue" also
14:14:50 <dperpeet> that's good to keep
14:14:52 <pitti> let's rename it to "improvement-potential" :)
14:14:56 <dperpeet> heh
14:15:07 <pitti> . o { Lernchance } *cough*
14:15:14 <dperpeet> ok, I'll delete it
14:15:16 <dperpeet> end of topic :)
14:15:58 <mvollmer> roger
14:16:08 <mvollmer> #topic Network integration tests
14:16:39 <mvollmer> so I refactored the tests and changed them at the same time, and also we have some changes pending to the actual code
14:16:55 <mvollmer> and the tests are failing in new and exciting ways now
14:17:02 <mvollmer> so I am looking at that
14:17:18 <dperpeet> thanks for the refactor
14:17:24 <dperpeet> but you're right, lots of changes at once :)
14:17:25 <mvollmer> i might find some real bugs, hopefully
14:17:33 <mvollmer> not soooo much
14:17:37 <mvollmer> but maybe too much
14:18:02 <mvollmer> so now it looks like we are getting wrong property values on d-bus
14:18:23 <mvollmer> this might be NM specific, but in any case the events in JavaScript don't match the events on D-Bus
14:19:03 <mvollmer> anyway, a typical failure is that the health check fails, or is ignored
14:19:18 <dperpeet> that sounds bad
14:19:21 <mvollmer> so if you see that, you might want to ignore it
14:19:57 <mvollmer> I also see some spurious disconnections during manual testing, sometimes
14:20:34 <dperpeet> I tried checking, but didn't go into depth: did you make sure that nothing runs in parallel now that requires the same fixed ip?
14:20:45 <mvollmer> I am pretty sure
14:20:58 <mvollmer> there isn't any fixed mac anymore in the tests
14:21:36 <mvollmer> but some devices used to be on dead vlans, and now they are all on cockpit1
14:21:43 <dperpeet> right
14:22:04 <mvollmer> we can put the "switch of rp filter" hack back
14:22:11 <mvollmer> I make a PR for that
14:22:22 <mvollmer> and debug into the D-Bus property mystery
14:22:50 <mvollmer> the symptom of that is that a new bond shows up as unmanaged in the cockpit ui
14:23:01 <mvollmer> because we miss the change that sets Managed = true
14:23:21 <mvollmer> but that happens only on rhel-7 in one of the PRs
14:23:25 <mvollmer> interesting
14:23:33 <mvollmer> so, this is just a heads up
14:24:01 <mvollmer> let's hope I find some real bug... :-)
14:24:21 <dperpeet> ok, I'll be looking forward to seeing all those tests reliably green :)
14:25:25 <mvollmer> yeah
14:27:00 <stefw> hey i'm around ... on the train ... working on my talk for devconf
14:27:07 <mvollmer> eot.
14:30:03 <mvollmer> #topic Other business
14:31:58 <mvollmer> looks like we are done?
14:32:27 <dperpeet> it appears so
14:32:53 <mvollmer> alright
14:32:58 <mvollmer> #endmeeting