meeting
LOGS
13:06:22 <mvollmer> #startmeeting meeting
13:06:22 <zodbot> Meeting started Mon Aug  8 13:06:22 2016 UTC.  The chair is mvollmer. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:06:22 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
13:06:22 <zodbot> The meeting name has been set to 'meeting'
13:06:27 <mvollmer> #tpic Agenda
13:06:32 <mvollmer> #topic Agenda
13:06:38 * mvollmer is multitasking too much
13:06:44 <harish> .hello harishanand
13:06:44 <zodbot> harish: harishanand 'Harish Anand' <harishanand95@gmail.com>
13:06:48 <andreasn1> .hello andreasn
13:06:49 <zodbot> andreasn1: andreasn 'Andreas Nilsson' <anilsson@redhat.com>
13:06:50 <fabiand> stefw, thanks
13:06:55 <mvollmer> .hello mvo
13:06:55 <harish> * timers
13:06:55 <zodbot> mvollmer: mvo 'Marius Vollmer' <marius.vollmer@gmail.com>
13:07:15 <dperpeet> .hello dperpeet
13:07:15 <zodbot> dperpeet: dperpeet 'None' <dperpeet@redhat.com>
13:08:09 <andreasn1> * storage volumes
13:10:11 <mvollmer> * showing all networking interfaces
13:10:18 <mvollmer> let's start
13:10:23 <mvollmer> #topic timers
13:10:44 <harish> https://github.com/cockpit-project/cockpit/pull/4645
13:11:29 <andreasn1> I haven't read through it all (yet). What's the general status?
13:11:30 <harish> we have completed all the works pending, with tests to be corrected for its failures
13:11:40 <dperpeet> there are a few races in the tests
13:11:54 <harish> tests were a failures due to race conditions
13:11:56 <dperpeet> but I think it will be ready to get merged soon
13:12:16 <harish> I will make sure that we get this merged this week itself
13:12:27 <harish> https://medium.com/@harishanand95/gsoc-week-9-integration-test-cases-ca53eb3c3739#.8q3xg23pl
13:12:51 <harish> in this blog post I have talked about all the test cases given for each timers
13:14:33 <harish> dperpeet here are the listing of all blogs i have did as part of GsoC https://medium.com/@harishanand95/
13:14:50 <dperpeet> harish, would you like to comment on why you test via cli and not wait for new timer info in the cockpit ui?
13:14:54 <dperpeet> after triggering a timer in the tests
13:15:17 <harish> yes i will comment it in #4645
13:15:23 <dperpeet> I mean here :)
13:15:26 <harish> or do you want me to explain it here?
13:15:29 <harish> oh yes
13:15:29 <dperpeet> briefly
13:15:54 <dperpeet> usually one would expect to see the tests cover that, so it's unusual to do it this way
13:17:29 <harish> we use CLI because the timer interface next run and last run gives different results even after we set the test machine to 2020-01-01 UTC
13:18:11 <harish> this happens because the PhantomJS has the time of the server it is running not that of the test machine
13:18:45 <dperpeet> it's possible to get this time difference in the tests, but I think it would be good to push this to a follow-up
13:18:50 <dperpeet> and it's not essential
13:19:41 <harish> yes there is also the differences in timezones to look forward to
13:19:53 <harish> https://medium.com/@harishanand95/gsoc-week-8-different-dates-issue-testing-41a582ce2aa6#.dn8dfq3x6
13:20:21 <harish> I have detailed out the issue here. We could calculate the difference and do the testing
13:20:40 <harish> which we could do it as a follow up.
13:20:41 <dperpeet> great
13:21:12 <harish> I think rest of the things are good.
13:21:46 <harish> andreasn1 dperpeet there is also a follow up we discussed on setting timers for existing services as well
13:21:59 <andreasn1> yeah, sounds good
13:22:15 <harish> here we create new services, we have to do that case as well.
13:22:34 <dperpeet> good work
13:22:48 <dperpeet> let me know when I should review again
13:22:55 <harish> thank you ! :)
13:23:29 <mvollmer> next?
13:23:36 <harish> eot
13:23:40 <mvollmer> #topic storage volumes
13:23:58 <andreasn1> I thought I had some questions, but I actually didn't I realized
13:24:05 <andreasn1> so we can continue this on the PR
13:24:17 <mvollmer> alright! :-)
13:24:31 <mvollmer> #topic showing all networking interfaces
13:24:50 <mvollmer> so, I have done this: https://github.com/cockpit-project/cockpit/pull/4825
13:25:04 <mvollmer> we used to only white-list a few interface types
13:25:11 <mvollmer> like ethernet, bond, bridge, ...
13:25:29 <mvollmer> this meant that wifi, inifiniband, tun/tap, etc would not be shown at all
13:25:38 <mvollmer> not even in the traffic graphs
13:25:54 <mvollmer> the now shows everything
13:26:00 <andreasn1> was there an issue that triggered this change?
13:26:12 <andreasn1> not against this change per se, mostly curious
13:26:15 <mvollmer> people were requesting infiniband support
13:26:33 <andreasn1> what's that? infiniband
13:26:38 <mvollmer> and it always confused me on my laptop that the main wifi interface didn't show up
13:26:46 <mvollmer> mostly its traffic
13:27:00 <mvollmer> i think infiniband is ethernet from IBM
13:27:08 <mvollmer> more better
13:27:10 <andreasn1> right, and on most servers, the wifi would be non-existant anyway, so it won't hurt
13:27:14 <mvollmer> some other cables etc
13:27:25 <andreasn1> yeah, makes sense
13:27:52 <mvollmer> i had some fear that showing wifi will just mess it up
13:27:57 <mvollmer> but it works well
13:28:04 <mvollmer> i can even bring it down and up again etc
13:28:14 <andreasn1> nice
13:28:31 <mvollmer> just not do anything wifi specific like select a hotspot and put in a password etc
13:28:46 <mvollmer> but as long as it is configured already, Cockpit will do the right thing and not destroy anything
13:28:52 <dperpeet> that's good
13:29:16 <mvollmer> stef pointed out that tests are missing...
13:29:23 <mvollmer> which is true
13:29:31 <mvollmer> but itmight be hard to test this in a VM
13:29:43 <mvollmer> i don't hink we can add infiniband to a vm, no?
13:30:02 <mvollmer> hmm, but tun/tap should work
13:30:51 <dperpeet> maybe not test it with infiniband
13:30:55 <dperpeet> but with something else
13:32:38 <mvollmer> yeah, something that doesn't need VM support, like tun
13:32:51 <andreasn1> http://opennebula.org/community-extension-enables-infiniband-in-kvm-virtual-machines/ ?
13:33:53 <mvollmer> andreasn1, is this passthrough, or emulated?
13:34:04 <andreasn1> I couldn't figure out
13:34:18 <andreasn1> I just googled for "kbm infiniband"
13:34:25 <andreasn1> kvm infiniband even
13:34:38 <andreasn1> here is someone asking about emulation http://serverfault.com/questions/252407/virtual-infiniband-on-kvm-qemu-or-another-open-source-platform
13:35:43 <mvollmer> thanks
13:35:54 <mvollmer> i go with tun anyway I think.
13:36:08 <dperpeet> mvollmer, if we need to play around with NIC models, let me know
13:36:11 <dperpeet> http://libvirt.org/formatdomain.html#elementsNICSModel
13:36:59 <mvollmer> dperpeet, we would need to make a interface show up in NetworkManager that is not "ethernet"
13:37:08 <mvollmer> like wifi or infiniband
13:37:27 <mvollmer> or tun, which I think is easiest and doesn't involve qemu at all
13:37:34 <mvollmer> i think
13:37:37 <dperpeet> right
13:37:52 <dperpeet> well, we can discuss this outside of the meeting
13:37:59 <dperpeet> let me know if I should dig into libvirt
13:38:25 <mvollmer> okay, let me first try tun
13:38:32 <mvollmer> thanks!
13:40:37 <mvollmer> done?
13:40:56 <mvollmer> #topic any other business
13:41:10 <dperpeet> I'm (finally) writing some code to prune logs
13:41:18 <dperpeet> that goes into the sink
13:41:58 <dperpeet> so for example pull request logs will get removed after 30 days
13:42:04 <dperpeet> is that too soon for known issues?
13:42:39 <mvollmer> sounds okay
13:43:01 <dperpeet> and I'm setting it up in a way that the script runs daily
13:43:41 <dperpeet> or rather, each time the sink is called it will check if prune was run in the last 24 hours
13:43:46 <dperpeet> and if not, it will prune
13:44:35 <dperpeet> or should it be only a timer event?
13:44:43 <dperpeet> i.e. if you don't want pruning, you don't set that up?
13:45:03 <mvollmer> whatever is easier...
13:45:19 <dperpeet> easiest would be to make it timer based
13:45:28 <dperpeet> that way it doesn't interfere with the regular sink use
13:45:31 <mvollmer> a cron entry in $HOME?
13:45:35 <mvollmer> yeah
13:45:42 <dperpeet> or systemd timer :)
13:45:59 <mvollmer> ahh, can we have them as non-root?
13:46:13 <mvollmer> probably
13:46:20 <mvollmer> anyway, sounds all good to me
13:46:52 <dperpeet> great, thanks
13:46:57 <dperpeet> I will exempt release logs
13:47:24 <dperpeet> should we keep the verify log output?
13:47:27 <dperpeet> for statistics
13:47:50 <dperpeet> or rather, the "log" file
13:48:57 <mvollmer> hmm, if we keep the verify logs, what is there to prune?
13:49:08 <dperpeet> system logs, screenshots
13:49:10 <mvollmer> am i misunderstanding?
13:49:11 <dperpeet> lots of stuff
13:49:15 <mvollmer> ahh, I see
13:49:20 <dperpeet> I propose keeping the summary
13:49:22 <dperpeet> so to speak
13:49:28 <mvollmer> alright, yeah
13:49:33 <dperpeet> I can also zip those
13:49:38 <dperpeet> and put them in an archive folder
13:50:07 <dperpeet> ok, I'll tinker with that
13:51:45 <mvollmer> cool
13:52:38 <dperpeet> eot
13:53:01 <mvollmer> #endmeeting