cockpit_weekly_meeting
LOGS
14:02:50 <andreasn_> #startmeeting
14:02:50 <zodbot> Meeting started Mon Mar  9 14:02:50 2015 UTC.  The chair is andreasn_. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:02:50 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
14:02:58 <andreasn_> #meetingname Cockpit Weekly Meeting
14:02:58 <zodbot> The meeting name has been set to 'cockpit_weekly_meeting'
14:03:12 <mvollmer> .hello mvo
14:03:13 <zodbot> mvollmer: mvo 'Marius Vollmer' <marius.vollmer@gmail.com>
14:03:28 <dperpeet> .hello dperpeet
14:03:29 <zodbot> dperpeet: dperpeet 'Dominik Perpeet' <dperpeet@redhat.com>
14:03:42 <sgallagh> .hello sgallagh
14:03:43 <zodbot> sgallagh: sgallagh 'Stephen Gallagher' <sgallagh@redhat.com>
14:03:44 <andreasn_> #chair andreasn mvollmer stefw sgallagh petervo dperpeet
14:03:44 <zodbot> Current chairs: andreasn andreasn_ dperpeet mvollmer petervo sgallagh stefw
14:03:48 <andreasn_> .hello andreasn
14:03:49 <zodbot> andreasn_: andreasn 'Andreas Nilsson' <anilsson@redhat.com>
14:04:35 <andreasn_> #topic Agenda
14:04:47 <andreasn_> what do we have this week?
14:05:07 <mvollmer> * Optional PCP update
14:05:13 <mvollmer> * Avcocado updates
14:05:22 <mvollmer> "Avocado", sorry.
14:05:26 <dperpeet> * Subscriptions
14:06:20 <sgallagh> Timeline for Fedora 22 stabilization?
14:06:53 <andreasn_> * iSCSI initiator support
14:07:39 <andreasn_> I guess that's it? we can also have stuff on open floor
14:07:47 <andreasn_> #topic Optional PCP update
14:07:58 <andreasn_> so is the update optional, or PCP? :)
14:08:55 <mvollmer> PCP :-)
14:08:59 <dperpeet> hehe, leading is so stressful
14:09:20 <mvollmer> ok, stefw started a branch to put the old samplers behind the new metrics1 channel.
14:09:44 <mvollmer> we have old samplers for cgroups, net devices, etc.
14:09:47 <andreasn__> eeek
14:09:50 <andreasn__> network error
14:09:55 <andreasn__> but back now
14:10:10 <mvollmer> ohh, didn't even notice.
14:10:16 <mvollmer> ok
14:10:31 <andreasn> please continue
14:10:37 <mvollmer> i have that almost finished
14:10:47 <mvollmer> tests are missing
14:11:03 <mvollmer> and the webapp needs to fall back to them when PCP isn't running.
14:11:27 <mvollmer> there wont be any archives without PCP, so no zooming, scrolling, etc, either.
14:11:41 <mvollmer> this is on hold until the avocado stuff is sorted out.
14:11:58 <mvollmer> done.
14:12:01 <dperpeet> sounds good
14:12:02 <andreasn> thanks!
14:12:35 <andreasn> sounds like Avocado is next up then
14:12:37 <dperpeet> how long can we allow this to be blocked by avocado? or should we implement "old-style" tests sometime?
14:13:13 <dperpeet> maybe we can table that question until we know the state of avocado tests
14:13:41 <andreasn> #topic Avocado updates
14:13:52 <andreasn> then :)
14:14:17 <mvollmer> ok, jscotka and me are getting the test infrastructure ready to be merged.
14:14:52 <mvollmer> example: http://files.cockpit-project.org/hubbot/66fcbc53a39075ae196b7edc13bb914046a0722f.2/hubbot.html
14:15:03 <dperpeet> do you have an ETA on that? I need to implement tests for subscriptions soon - which framework should I use?
14:15:07 <mvollmer> we are getting close
14:15:38 <mvollmer> there are a lot of review comments so still work to do
14:15:46 <dperpeet> days or weeks?
14:15:53 <mvollmer> a week.
14:16:02 <dperpeet> thanks!
14:16:05 <mvollmer> after that, we still need to port all the tests.
14:16:27 <mvollmer> until we can retire the old stuff.
14:16:40 <petervo> are we going to have the ability to target tests at specific os'?
14:17:01 <mvollmer> petervo, yes.
14:17:09 <petervo> or would that be a comes later enhancement
14:17:15 <mvollmer> we have that already now, someone just has to do it. :-)
14:17:17 <petervo> cool
14:17:40 <mvollmer> but with virt-deploy (new style) it should be much easier.
14:18:14 <mvollmer> right now I am struggling with avocado a bit, trying to understand it better, etc, and find a place where to make screenshots when a test fails.
14:18:30 <mvollmer> it's interesting, so I'll explain a bit
14:18:38 <dperpeet> will the infrastructure of how we provide services for tests change? e.g. setting up a candlepin instance
14:18:44 <mvollmer> Avocado is 'smart' and fishes out the class to test from your sources
14:19:04 <andreasn> oOoh, screenshots
14:19:05 <andreasn> nice
14:19:28 <mvollmer> it searches for classes that are derived from avocado.test.Test
14:19:52 <mvollmer> I tried to make a convenience class that derives from test.Test, and then we derive our actual tests from that.
14:20:17 <mvollmer> Of course, Avocado picks the convenience class as the class to test, and crashes.
14:20:31 <mvollmer> so, hrrr.
14:20:37 <mvollmer> stuff like that
14:21:00 <mvollmer> I am not saying our old code is better, but it is ours and easier to change. :-)
14:21:34 <mvollmer> ok, I'll worh through this with jscotka and the avocado guys and we'll figure something out.
14:22:09 <mvollmer> for users, the most interesting bit is probably the different 'philosophy'.
14:23:04 <mvollmer> that is, when to create and snapshot and update a test machine.
14:23:45 <mvollmer> the new stuff is much more incremental, and we might be getting new packages and new breakage more spontaneously and less controlled than before.
14:23:56 <mvollmer> that will be interesting
14:24:23 <mvollmer> the old stuff was much more careful, but we also were lazy with updates, so...
14:24:31 <dperpeet> how does the speed compare for a development iteration? (changed source, rerun a test)
14:25:05 <mvollmer> i think the new stuff is more flexible, and can be made faster for individual cases.
14:25:14 <dperpeet> good news :)
14:25:27 <mvollmer> you can run a test locally in a VM, for example.
14:25:47 <mvollmer> no need to boot anything or create an image, just install cockpit and point the test at localhost.
14:26:18 <mvollmer> it also means the test environment is less well defined
14:27:01 <mvollmer> also, VMs are reused until forever, so they might acquire cruft over time.
14:27:17 <mvollmer> my ipa machine started losing its default route, for example.
14:27:37 <mvollmer> quite annyoing when that happens in the CI machinery.
14:27:47 <mvollmer> but yeah, let's see.
14:28:15 <dperpeet> so a bad pull request can break hubbot?
14:28:35 <mvollmer> so, summary: we pretty close to putting it "in production", but there is a huge amount of work still waiting to be done after that
14:28:51 <mvollmer> dperpeet, yes.
14:28:57 <mvollmer> dperpeet, that's a good way to put it.
14:29:19 <mvollmer> dperpeet, but we'll put in some more snapshots to help a bit.
14:29:20 <dperpeet> :o then we should migrate the tests carefully and slowly :)
14:29:51 <mvollmer> dperpeet, it's not as bad, actually.
14:30:03 <mvollmer> now that I think about it.
14:30:14 <mvollmer> i have to admit that not all consequences are clear in my mind
14:30:43 <mvollmer> the VMs are cleaned up after a test, but not from one "make install" to the next.
14:30:54 <mvollmer> so we should just do that.
14:31:29 <mvollmer> ok, let's stop. :-)
14:32:05 <mvollmer> let's say the future is a bit experimental but I am confident. :-)
14:32:19 <andreasn> all right, next topic then
14:32:22 <mvollmer> and it's not going to be a huge blocker
14:32:28 <andreasn> #topic Subscriptions
14:32:32 <mvollmer> if it's all crap, we keep doing the old stuff.
14:32:36 <mvollmer> (sorry)
14:32:40 <dperpeet> :)
14:33:01 <dperpeet> subscriptions: first pull request is waiting for andreas-approval and code review
14:33:21 <dperpeet> once that is merged (blocking on tests-change) I can start integrating the tests
14:33:31 <dperpeet> I've set up a working candlepin locally
14:33:36 <dperpeet> based on fedora server
14:33:38 <andreasn> https://github.com/cockpit-project/cockpit/pull/1857/
14:33:49 <petervo> does this 1st PR include the json only status feature?
14:33:56 <dperpeet> and talked to the candlepin people - they have scripts to populate with dummy data
14:33:59 <petervo> or is that a seperate PR?
14:34:11 <dperpeet> petervo, what do you mean?
14:34:19 <dperpeet> the status summary?
14:34:35 <dperpeet> stefw wanted to finish the cache for that first
14:34:46 <petervo> ok
14:34:48 <dperpeet> so we decided to refactor separately
14:34:53 <petervo> make sense
14:34:58 <petervo> makes*
14:35:09 <dperpeet> the proposed order right now is: basic functionality (current pr)
14:35:11 <dperpeet> then tests
14:35:18 <dperpeet> then refactor and link to system page
14:35:20 <andreasn> the ui is in good shape, and I'm going to test it out on a RHEL system today
14:35:28 <dperpeet> sounds good
14:35:48 <dperpeet> once design is worked out maybe mvollmer or petervo can review the code, since stefw's away this week
14:36:33 <petervo> sure
14:36:56 <dperpeet> I expect to have the dummy server running this week
14:37:08 <dperpeet> but I won't merge that until we have the new test structure
14:37:08 <andreasn> \o/
14:37:20 <dperpeet> that's it on subscriptions
14:37:34 <andreasn> thanks
14:37:46 <andreasn> #topic iSCSI initiator
14:38:13 <andreasn> so this is a feature Fabian Deutsch is working on
14:38:20 <andreasn> https://github.com/cockpit-project/cockpit/wiki/Feature:-iSCSI-Initiator
14:39:02 <andreasn> stories and workflows are in good shape, and I'll do designs this week
14:39:32 <andreasn> should hopefully fit in quite well on the storage page
14:39:48 <andreasn> in the end
14:39:57 <andreasn> but I think he'll do it as a plugin first
14:40:12 <andreasn> that's it about that I think
14:41:27 <andreasn> anything for the open floor?
14:41:35 <andreasn> #topic Open Floor
14:41:50 <sgallagh> Yeah, I have one item
14:42:22 <andreasn> go ahead
14:42:26 <dperpeet> we forgot the timeline
14:42:26 <sgallagh> Fedora 22 plans; specifically that we need to account for the Beta Freeze (read, essentially complete except for release-blocking bugs)
14:42:40 <sgallagh> #link https://fedoraproject.org/wiki/Releases/22/Schedule
14:43:03 <sgallagh> Beta Freeze is currently scheduled for March 31
14:43:25 <mvollmer> right
14:43:35 <sgallagh> At which point we want to have something essentially stabilized (and branding re-applied)
14:43:40 <dperpeet> #link https://trello.com/c/duRQw1rU/113-milestone-fedora-22
14:44:06 <andreasn> Patternfly is updated, but not all pages are migrated
14:44:14 <andreasn> will get to some of those this week
14:44:26 <andreasn> and PCP is getting there
14:44:36 <mvollmer> but not fully optional.
14:44:40 <andreasn> so I think we're on good track
14:44:41 <dperpeet> stefw is on removing the daemon
14:44:46 <sgallagh> /me nods
14:45:06 <mvollmer> dperpeet, what daemon?
14:45:12 <dperpeet> the old code
14:45:15 <dperpeet> I meant
14:45:22 <mvollmer> cockpitd?  yeah.
14:45:44 <mvollmer> that's not a blocker for Fedora 22, I'd say.
14:45:59 <dperpeet> will we dedicate some time before march 31st to closing issues/bugs?
14:46:08 <mvollmer> we should.
14:46:20 <dperpeet> the list has been growing nicely
14:46:25 <mvollmer> maybe we need to put the avocado stuff on hold a bit.
14:46:30 <mvollmer> wdyt?
14:46:39 <dperpeet> I don't think so, a hold on that would cost us a lot
14:46:46 <mvollmer> would it?
14:46:52 <dperpeet> you're right in the middle
14:46:56 <mvollmer> yeah.
14:47:13 <dperpeet> since you have a lot of experience with cockpit, you could help triage the issues
14:47:32 <mvollmer> and we can merge it even without screenshots.
14:47:46 <dperpeet> true, those could be separate
14:47:54 <dperpeet> I can work on issues after subscriptions are tested
14:48:12 <dperpeet> the new journal look isn't that important for fedora, I think
14:48:30 <andreasn> no, it would be nice, but not a blocker in any way
14:48:43 <dperpeet> how about we decide next Monday?
14:48:46 <andreasn> we have a lot of new stuff compared to f21 already
14:48:47 <dperpeet> then we still have enough time
14:48:57 <mvollmer> sgallagh, have you tried recent cockpit releases?
14:49:08 <andreasn> does the branding still work?
14:49:29 <mvollmer> i have the feeling that right now I am too close to see the big picture.
14:49:36 <sgallagh> andreasn: Branding isn't working on my system at present, but I think I'm pointed at the cockpit-preview repo
14:49:39 <sgallagh> Let me double check
14:49:53 <mvollmer> but I'd say it's pretty okay regression wise.
14:49:55 <andreasn> sgallagh: do you have a url to a nightly build of server?
14:50:30 <andreasn> I always have a hard time navigating to the right images
14:50:36 <sgallagh> I'm using the official builds: no branding
14:50:59 <sgallagh> andreasn: We don't exactly have nightly builds, but since we just approved Alpha on Friday, you can use those
14:51:15 <andreasn> where can I find that?
14:51:18 <sgallagh> #link http://dl.fedoraproject.org/pub/alt/stage/22_Alpha_RC3/Server/
14:51:32 <andreasn> oh, ok. Let me take a look at the branding then
14:51:40 <sgallagh> ok
14:52:27 <andreasn> dperpeet: monday next week sounds good btw
14:52:31 <dperpeet> ok
14:52:41 <andreasn> mvollmer: would you agree?
14:52:50 <sgallagh> (RE: "nightly" builds. adamw has a tool that automatically sanity checks the nightly composes and declares some of them testable)
14:53:09 <sgallagh> If you subscribe to the test@lists.fp.o mailing list, you'll see those announcements
14:53:21 <andreasn> cool
14:53:33 <mvollmer> andreasn, ok.
14:54:04 <andreasn> greaty
14:54:08 <andreasn> great even
14:54:14 <andreasn> so I guess that
14:54:21 <andreasn> 's it for this weeks meeting?
14:54:40 <mvollmer> I'd say so.
14:54:43 <sgallagh> /me nods
14:54:53 <andreasn> #endmeeting