16:01:37 <stefw> #startmeeting Cockpit
16:01:37 <zodbot> Meeting started Mon Jan 19 16:01:37 2015 UTC.  The chair is stefw. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:01:37 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:01:49 <stefw> #chair stefw mvollmer andreasn dperpeet petervo jvance subin
16:01:49 <zodbot> Current chairs: andreasn dperpeet jvance mvollmer petervo stefw subin
16:01:56 <stefw> #topic Agenda
16:02:00 <stefw> Hello everyone :)
16:02:00 <mvollmer> .hello mvo
16:02:01 <zodbot> mvollmer: mvo 'Marius Vollmer' <marius.vollmer@gmail.com>
16:02:05 <mvollmer> Hi!
16:02:08 <subin> hi
16:02:13 <sgallagh> .hello sgallagh
16:02:14 <zodbot> sgallagh: sgallagh 'Stephen Gallagher' <sgallagh@redhat.com>
16:02:25 <sgallagh> I'm attending another meeting concurrently
16:02:30 <subin> .hello subin
16:02:31 <zodbot> subin: subin 'Subin Francis' <subin.francis@gmail.com>
16:02:48 <petervo> .hello petervo
16:02:49 <zodbot> petervo: Sorry, but you don't exist
16:02:53 <andreasn> .hello andreasn
16:02:54 <zodbot> andreasn: andreasn 'Andreas Nilsson' <anilsson@redhat.com>
16:03:05 <stefw> Anyone can submit agenda points prefixed with a *
16:03:12 <stefw> * Better IRC meeting time
16:03:17 <stefw> * Status: Kubernetes work
16:03:24 <stefw> * Status: Integration test refactoring
16:03:35 <andreasn> * Status: ongoing Docker work
16:03:40 <stefw> * Status: Metrics channels
16:03:41 <aayush-k> hi
16:03:43 <aayush-k> .hello
16:03:43 <zodbot> aayush-k: (hello <an alias, 1 argument>) -- Alias for "hellomynameis $1".
16:04:10 <aayush-k> .hello aayush-k
16:04:13 <zodbot> aayush-k: Sorry, but you don't exist
16:04:23 <andreasn> how does the hellomynameis thing work now again, is it matching it with fedora id's somehow?
16:04:31 <stefw> i have no idea
16:04:35 <stefw> i just cargo cult what others do
16:04:38 <andreasn> haah
16:04:41 <andreasn> haha
16:05:22 <stefw> * Possible next release blockers?
16:05:22 <sgallagh> andreasn: Yes, it's your FAS ID
16:05:33 <mvollmer> * Fedora 22
16:06:01 <stefw> okay, is that it for agenda? shout now if you're still typing ...
16:06:31 <stefw> #topic Better IRC meeting time
16:06:52 <stefw> so mvollmer mentioned that the current IRC meeting time doesn't work super well
16:07:09 <stefw> mvollmer, do you have any proposal for a better time?
16:07:15 <mvollmer> I like to keep my evenings free...
16:07:34 <andreasn> what's the earliest in the day that would work for you petervo?
16:07:41 <mvollmer> two hours earlier, or four hours later.
16:07:47 <petervo> 2
16:07:57 <jscotka> stefw, yes, we can,
16:08:00 <petervo> 2 hours earlier, is the earliest for me
16:08:22 <andreasn> I think I would prefer 2 hours earlier over 4 hours later
16:08:26 <mvollmer> i would prefer 2 hours earlier, obviously...
16:08:35 <sgallagh> Two hours earlier works better for me as well
16:08:43 <stefw> alright, although i may have to skip every other week
16:08:52 <stefw> or be distracted
16:08:54 <mvollmer> we can move the day, too.
16:08:58 <stefw> it'd be fine if someone else takes the lead
16:09:03 <stefw> it's a good day
16:09:07 <mvollmer> ok, sure.
16:09:09 <stefw> i can just do two things at once
16:09:24 <mvollmer> :-)
16:09:32 <stefw> so as long as someone else chairs it i'm fine with 14:00 UTC Mondays
16:09:36 <mvollmer> i have trouble with one.
16:09:45 <stefw> well i'll do one at a time
16:09:59 <mvollmer> I'll chair.
16:10:01 <stefw> but lots of meetings only need attention part time ... you can multi-task Win 3.1 style
16:10:04 <stefw> cooperatively
16:10:10 <andreasn> we also have the option to change day of the week
16:10:19 <stefw> well, lets try this
16:10:20 <andreasn> (to complicate things further)
16:10:24 <andreasn> all right
16:10:33 <stefw> #info Proposed IRC meeting time is 14:00 UTC on Monday
16:10:33 <mvollmer> yep
16:10:48 <stefw> any objections?
16:10:50 <mvollmer> excellent, thanks everyone!
16:10:57 <andreasn> great!
16:10:59 <petervo> jvance might not be able to make it
16:11:10 * mvollmer feels a bit selfish
16:11:16 <stefw> yes, that's true, but (for better or worse) not everyone needs/can make it
16:11:17 <petervo> but i guess we can address that if he brings it up
16:11:18 <subin> i am ok with 14:00 UTC on Monday
16:11:19 <mvollmer> but only a bit. :-)
16:11:47 <stefw> mvollmer, will you announce?
16:11:50 <stefw> on cockpit-devel?
16:12:37 <stefw> #action IRC meeting moved to 14:00 UTC on Monday
16:12:45 <mvollmer> yes
16:12:56 <stefw> #action mvollmer will announce on cockpit-devel
16:13:01 <dperpeet> hi and sorry I'm late - 2 hours earlier is a lot better :)
16:13:10 <stefw> great
16:13:33 <stefw> #topic Status: Kubernetes work
16:13:52 <stefw> So one big theme that a few people are working on is getting Cockpit to work well with Kubernetes orchestration of containers
16:14:01 <stefw> some proof of concept code was merged into master, than subin and I worked on
16:14:16 <stefw> it needs to be manually enabled, it's not installed or part of the packages by default
16:14:28 <stefw> subin, do you have anything to add?
16:14:55 <subin> nothing much , just basic GET queries from k8
16:15:18 <andreasn> somewhat related but not quite, does anyone have a good and simple virsh guide? I wasted way too much time trying to get this running today
16:15:20 <subin> no PUT/POST request are available on the UI. Very basic info of kubernetes
16:16:46 <stefw> In addition andreasn has been doing some design work so we can have multiple dashboards ... one of which will optionally be Kubernetes
16:16:54 <stefw> andreasn, anything to show there?
16:17:18 <andreasn> not yet, but working on it and hope to have something to show tomorrow
16:17:27 <stefw> great
16:17:35 <subin> i am trying out the latest k8 code to see if anything is new
16:17:37 <stefw> #info Proof of concept commited to git master, designs coming shortly
16:17:55 <stefw> subin, do we know yet if we have to rewrite anything for the Kubernetes v3 API?
16:18:08 <stefw> er, i mean "everything"
16:18:38 <subin> i just started exploring need a day or 2 on it. But so far old api seem to work
16:18:56 <stefw> cool
16:19:05 <stefw> alright, should we move on?
16:19:13 <subin> yeah thats all from me
16:19:43 <stefw> #topic Integration Test Refactoring
16:20:06 <stefw> So jscotka has been refactoring our integration tests, and has made a proof of concept pull request
16:20:18 <jscotka> Hi, yes
16:20:41 <jscotka> So I've created prototype based on avocado framework
16:20:52 <jscotka> Everything seems well.
16:21:07 <stefw> jscotka, what can we do as our next step to move this along?
16:21:27 <jscotka> I'm now waiting for avocado developers to make virt machine scheduler working
16:22:11 <jscotka> Next step is to make it fully automated, installation of virt image & schedule test there
16:22:31 <stefw> fsimonce pointed out that virt-deploy does this
16:22:59 <stefw> a standard set of images, support for custom repos, uses virt-builder etc.
16:23:10 <jscotka> yes, I'll use it, in case everything will go well.
16:23:37 <stefw> does avocado run inside the virtual machine?
16:24:17 <jscotka> avocado is able to run tests inside virt machine or use also remote plugin, so that it is able to schedule test everywhere
16:25:29 <stefw> can we move forward without the avocado virtual machine support, if we only (for now) use it inside the virtual machine, and not to start the virtual machine?
16:26:04 <jscotka> http://avocado-framework.readthedocs.org/en/latest/VirtualMachinePlugin.html
16:26:20 <stefw> yes, that's exactly what virt-deploy does too
16:26:51 <stefw> so my question is, if one is not working, and the other is, could we use virt-deploy instead of the avocado code that isn't ready?
16:27:02 <jscotka> I thought that wirt deploy is installation part of process (it installs disto based on template
16:27:22 <jscotka> I hope that it will be ready this week.
16:27:44 <stefw> ok
16:28:08 <stefw> #info Test refactoring waiting on incomplete Avocado virtual machine plugin
16:28:11 <jscotka> as I've discussed with Lukas, he said, that it was working well, but somebody broke it accidently
16:28:48 <stefw> alright. anything else on the integration test refactoring?
16:29:12 <mvollmer> I could briefly mention hubbot.
16:29:16 <stefw> ah yes
16:29:20 <mvollmer> ok
16:29:28 <jscotka> I would be glad if somebody is writing new test, could use avocado.
16:29:57 <jscotka> Is anybody now writing some new test?
16:30:08 <mvollmer> We used to have a trivial bot called "hubbot" that could run VERIFY for a given pull request.
16:30:36 <mvollmer> It broken when github changed something in the API re auth.
16:30:58 <mvollmer> I have resurrected it now and changed it to add statuses to github (instead of comments).
16:31:12 <mvollmer> So it's just like Travis now.
16:31:31 <mvollmer> You can run it via ssh to files.cockpit-project.org
16:31:47 <mvollmer> Right, most important bit: You need to run it manually.
16:32:00 <mvollmer> If someone wants to automate it....
16:32:01 <mvollmer> :-)
16:32:21 <mvollmer> wait, I'll start a run...
16:32:55 <mvollmer> https://github.com/cockpit-project/cockpit/pull/1677
16:33:03 <mvollmer> it's busy with that pull request now.
16:33:58 <mvollmer> I plan to use this instead of running VERIFY on my laptop.
16:34:06 <mvollmer> if you want to do that, too, let's talk.
16:34:10 <mvollmer> eof. :-)
16:34:18 <stefw> do you know enough about the github api to automate it?
16:34:43 <mvollmer> no, but it appears to be friendly.
16:34:44 <stefw> or do we have to figure out how to give people access to that server?
16:35:35 <mvollmer> giving access is just copying ssh keys into ~hubbot/ssh/
16:35:38 <petervo> what is travis running now?
16:35:48 <petervo> make check?
16:35:51 <mvollmer> yes
16:36:23 <stefw> so the goal is to have both the unit tests and the integration tests run for each pull request
16:36:36 <stefw> the integration tests can only be run once you're sure that the pull request isn't nefarious
16:36:44 <stefw> since we run it on our own server
16:37:27 <mvollmer> and we run them on our own server since they spawn virtual machines.
16:37:35 <stefw> right, for the time being
16:37:49 <stefw> soon they'll run in virtual machines, so we'll have to see how that plays out
16:37:55 <stefw> once jscotka's work lands
16:38:01 <stefw> so jscotka will you be reworking further tests?
16:38:04 <mvollmer> yep, and if that goes away, hubbot is obsolete, probably.
16:38:07 <jscotka> yes
16:38:14 <jscotka> I'll do that.
16:38:27 <mvollmer> so, I don't want to 'invest' much here.
16:38:36 <stefw> mvollmer, all the github interaction is not obselete
16:38:53 <stefw> and basically a script can still spawn the vm and run the tests in it, if we want to keep that going
16:39:03 <petervo> wouldn't we still want the github integration?
16:39:29 <jscotka> probably we could have also some openstack instances aviable somewhere (via some fedora upstream testing)
16:40:04 <mvollmer> stefw, petervo, true.
16:40:13 <mvollmer> but we might use Travis, no?
16:40:55 <petervo> is travis debian only?
16:40:56 <mvollmer> anyway, should be fun to figure out github events.
16:41:07 <petervo> or is that just how i always used it?
16:41:21 <stefw> yes travis is pretty much Debian only
16:41:25 <stefw> alright, next topic?
16:41:29 <mvollmer> (let's move on... :-)
16:41:37 <stefw> #topic Status: Ongoing Docker work
16:41:43 <stefw> andreasn, you're on
16:42:18 <andreasn> yeah, this was mostly that I wanted to highlight some of the work dperpeet is doing with docker right now
16:42:52 <andreasn> most of them are here https://github.com/cockpit-project/cockpit/labels/starter
16:43:05 <andreasn> https://github.com/cockpit-project/cockpit/issues/1587
16:43:25 <dperpeet> I implemented your design
16:43:39 <dperpeet> but I'm currently in the testing phase
16:43:43 <andreasn> cool
16:43:49 <dperpeet> and I noticed that the expose in master doesn't work for me
16:43:57 <dperpeet> so I'm looking into that
16:44:10 <andreasn> great, looking forward to see all this in master soon
16:44:15 <stefw> #info Several docker enhancements going on
16:44:15 <dperpeet> I'll add updated design screenshots tomorrow
16:44:21 <erinaboyd> stefw - sorry just saw this invite for the IRC chat!
16:44:24 <dperpeet> especially regarding disabled user input
16:44:34 <stefw> erinaboyd, no worries, welcome
16:44:41 <erinaboyd> I think I am going to have to change my working hours to match yours more
16:44:48 <dperpeet> has anyone tested the port mapping to host recently?
16:45:09 <andreasn> with the docker cli command?
16:45:20 <dperpeet> via cockpit
16:45:39 <dperpeet> I start a container, the image exposes port X and I want to map it to Y on my host
16:45:43 <dperpeet> that is current state
16:46:16 <dperpeet> the issue is to extend that to arbitrary ports when running an image
16:46:26 <stefw> is there a github issue filed for that?
16:46:48 <andreasn> ah, similar to -P?
16:46:48 <dperpeet> stefw, the extension is the issue andreasn posted
16:46:51 <dperpeet> yes
16:47:00 <stefw> cool. anything else to highlight?
16:47:17 <dperpeet> not at this point
16:47:39 <stefw> #topic Status: Metrics channels
16:47:53 <stefw> Last week the work mvollmer did on the metrics channels was merged. yay
16:48:05 <mvollmer> Yay!
16:48:07 <stefw> and it uses PCP to gather it's information
16:48:16 <mvollmer> I have archives working here, pretty much
16:48:22 <stefw> so i've been also working on porting the old monitors over no the new metrics channels
16:48:33 <stefw> because there's lots of places where PCP doesn't provide all the info we need, or it doesn't work at all
16:48:51 <stefw> A bit of cleanup here: https://github.com/cockpit-project/cockpit/pull/1677
16:48:56 <stefw> i hope that doesn't conflict with your code
16:48:58 <stefw> mvollmer,
16:49:20 <mvollmer> no, it shouldn't.
16:49:21 <stefw> and hopefully i'll have a pull request for the remaining work in the next few days
16:49:40 <mvollmer> good stuff, moving thing compression all the way into cockpitmetrics.
16:49:42 <stefw> once i've ported over the old monitors, it brings us closer to ditching the legacy cockpitd process
16:50:24 <mvollmer> I keep thinking that we might want to make custom pmda for the things that are missing from the official pcp release.
16:50:40 <mvollmer> magnitude more work, probably, so yeah...
16:50:49 <stefw> yes, except it seems that PCP wants to run processes and do lots of stuff behind our backs
16:50:56 <stefw> and as such makes cockpit-bridge not be relocatable
16:51:14 <stefw> we rely on the relocatable nature of cockpit-bridge for use with privileged containers
16:51:38 <stefw> I think the metrics channels were designed to be abstracted at that point, and we don't necessarily need to bring everything under PCP
16:52:12 <mvollmer> I agree
16:52:33 <stefw> cool to hear the archives work is nearly done?
16:52:46 <stefw> andreasn, i don't remember, was there design work done for scrolling graphs backwards and looking at old data?
16:53:12 <andreasn> I can't remember to be honest where things are with that
16:53:15 <stefw> ok
16:53:22 <stefw> alright, next topic?
16:53:22 <mvollmer> stefw, well, tests, docs, ...
16:53:32 <mvollmer> stefw, but it looks good, no blockers.
16:53:35 <stefw> #topic Possible next point release blockers?
16:53:36 <andreasn> but I'll investigate and make sure we have something if we don't already
16:54:17 <stefw> We have a regression with the dashboard now using PCP ... it no longer works with Atomic
16:54:26 <stefw> Do we think we need to sort that out before the 0.37 release?
16:54:50 <mvollmer> stefw, I don't think so.
16:55:04 <stefw> i tend to agree although we need to fix this up quickly
16:55:21 <mvollmer> What breaks it?  Super prov container?  PCp just isn't there?
16:55:41 <stefw> well first of all the library dependency isn't available
16:55:49 <stefw> but once you force that into the system, then it fails to initialize
16:56:04 <stefw> PCP hard codes its paths, and expects lots of infrastructure outside of pcp-libs to be installed before it'll even initialize
16:56:15 <stefw> some precompiler that it launches for instance
16:56:21 <mvollmer> but if we force that into the system, too?
16:56:22 <stefw> as well as the pmda modules
16:56:26 <stefw> we cannot write to /usr\
16:56:37 <stefw> at it runs it as a hard coded path
16:56:41 <stefw> so that's a dead-end
16:57:17 <stefw> this is the binary it wants to run: /usr/libexec/pcp/bin/pmcpp
16:57:26 <stefw> at least the first one, haven't gotten past that
16:57:44 <mvollmer> so pcp is broken with a ro /usr?
16:57:54 <stefw> it expects to be installed to /usr
16:58:00 <stefw> and Atomic doesn't allow anything to be installed to /usr
16:58:10 <stefw> well, at least not after the fact
16:58:12 <mvollmer> but i could be part of Atomic?
16:58:17 <stefw> it allows stuff in /usr/local
16:58:23 <stefw> just not /usr/*everything-else*
16:58:24 <mvollmer> *it
16:58:41 <stefw> in theory yes, anything could be part of Atomic
16:59:05 <mvollmer> right, so maybe that's acceptable. :-)
16:59:16 <mvollmer> after trimming down the dependencies of "pcp".
16:59:58 <mvollmer> of course, I agree that running the C preprocessor during startup is, err,  anachronistic.
17:00:05 <stefw> yup
17:00:13 <stefw> so but we agree this isn't a blocker for 0.37?
17:00:21 <mvollmer> i agree
17:00:32 <stefw> ok
17:00:39 <stefw> #topic Fedora 22
17:00:53 <mvollmer> Fedora 22 is starting to close the doors.
17:00:56 * stefw tries to move things along ... last agenda point ...
17:01:08 <andreasn> sgallagh: around?
17:01:09 <stefw> what does "close the doors" mean?
17:01:12 <mvollmer> sgallagh, still here?
17:01:19 <sgallagh> sgallagh: I am now
17:01:36 <mvollmer> great, so re Cockpit and F22.
17:01:48 <stefw> I think cockpit in Fedora could be updated until the code freeze there?
17:01:55 <sgallagh> stefw: The deadline for submitting System-Wide Change pages for Fedora 22 is less than 24 hours away
17:02:19 <stefw> does anyone have plans to make a cockpit related system wide change?
17:02:43 <sgallagh> The major effect of these Change pages is to coordinate landing things and ensure that they are put into the release notes and marketing materials
17:02:51 <stefw> there are a few i could think of (such as showing the cockpit url in the Fedora Server VT) ... but no plans here
17:03:34 <mvollmer> Should we stabilize master for F22, as we did with F21?
17:03:41 <mvollmer> or is that hopeless?
17:03:48 <andreasn> and if there was any UI for Server Roles in Cockpit, that would be a System Wide Change?
17:04:05 <stefw> andreasn, no, i don't think so
17:04:22 <sgallagh> mvollmer: If we don't stabilize master for F22, what would we do?
17:04:22 <stefw> mvollmer, that's a good question, if we want to update it at all, we probably would need to do another stable release, right?
17:04:35 <stefw> the alternative is to leave at the F21 version :(
17:04:39 <sgallagh> Release a dev pre-release as the public face of Fedora Server?
17:04:50 <stefw> no, that's not an option
17:04:51 <andreasn> ugh, the F21 version is soo old
17:04:53 <stefw> yup
17:05:36 <sgallagh> It would need to be "testable" (it already is) by Feb 24 and complete by March 31
17:05:40 <sgallagh> (As the current schedule stands)
17:06:06 <sgallagh> I'd think March 31 is a reasonable target for stabilizing a release.
17:06:09 <stefw> that seems doable (end of march)
17:06:14 <mvollmer> yep
17:06:14 <andreasn> yeah
17:06:21 <stefw> but then we can do some bug fixing
17:06:30 <mvollmer> we don't need to complete the arch rewrite, even.
17:06:35 <sgallagh> Right, after March 31 we go into bugfix-only mode
17:06:53 <stefw> People have used the Fedora Server release of Cockpit to get involved with Cockpit a lot. So delivering something there is quite beneficial to the project as a whole.
17:07:13 <stefw> we may have enough resources to have two branches for a while
17:07:23 <stefw> well enough contributors
17:08:14 <mvollmer> ok, sounds good to me!
17:08:18 <sgallagh> it's also beneficial to Fedora Server: Cockpit continues to be the most-reviewed (and most positively-reviewed) part of Fedora Server
17:08:19 <stefw> can we agree that we're not delivering any system wide changes?
17:08:23 <andreasn> so March 31 only bugfixes... for how long is it in that mode before release?
17:08:47 <sgallagh> andreasn: May 5th is Final Freeze
17:08:49 <andreasn> ok
17:08:55 <mvollmer> stefw, yes.
17:08:56 <sgallagh> So about one month
17:09:06 <andreasn> stefw: I think so, yes
17:09:20 <andreasn> although it's blurry to me what is a system wide change exactly
17:09:25 <sgallagh> Yeah, as I think on it, if we do the Domain Controller component, it's still not likely system-wide.
17:09:34 <sgallagh> andreasn: Yeah, it's intentionally left a little subjective
17:09:54 <mvollmer> it's a local change in cockpit/rolekit/freeipa.
17:09:59 <sgallagh> The idea is that "Self-Contained" Changes are "Things we want to advertise, but don't need to coordinate throughout the project"
17:10:11 <andreasn> is it so you have your back free so people don't shout at you on fedora-devel?
17:10:18 <sgallagh> System-Wide Changes are things that can pull the rug out of other groups
17:10:19 * mvollmer hasn't even read the instructions....
17:10:29 <andreasn> right
17:10:40 <andreasn> but yeah, I don't think we have any of those
17:10:52 <sgallagh> Note that: "Blocking the release of Fedora Server" makes something system-wide.
17:11:02 <sgallagh> Which is why Cockpit was treated that way in F21
17:11:15 <sgallagh> (I.e. having it at all)
17:11:55 <sgallagh> But adding a feature to it is not System-Wide unless it would require retooling other pieces.
17:12:21 <mvollmer> unless the release blocks on that feature. :-)
17:12:41 <sgallagh> The only blocking feature for Fedora Server so far this release is the creation of the Database Server Role
17:13:16 <stefw> #action Cockpit is not proposing any system wide changes for Fedora 22
17:13:17 <sgallagh> (Which I'm not going to push for inclusion into Cockpit, for the record)
17:13:39 <mvollmer> ok, I'll have to leave nowish...
17:13:41 <sgallagh> #action sgallagh to file Self-Contained Change for Domain Controller setup support in Cockpit.
17:13:41 <stefw> #action We'll try to plan out further what it means to stabilize a release/branch of Cockpit for F22
17:13:58 <stefw> #topic Open Floor
17:14:01 <stefw> anything else?
17:14:08 <andreasn> nothing from me
17:14:14 <sgallagh> ah, I'm not chaired. Oh well
17:14:16 <mvollmer> https://github.com/cockpit-project/cockpit/pull/1677
17:14:28 <mvollmer> hubbot is happy.
17:14:48 <mvollmer> ok, thanks!
17:14:58 <stefw> sgallagh, sorry
17:15:10 <stefw> #action sgallagh to file Self-Contained Change for Domain Controller setup support in Cockpit.
17:15:14 <sgallagh> Thanks
17:15:17 <dperpeet> mvollmer, always nice to have happy bots, good job
17:15:22 <stefw> #endmeeting