13:02:34 <mvollmer> #startmeeting meeting 13:02:34 <zodbot> Meeting started Mon Jun 6 13:02:34 2016 UTC. The chair is mvollmer. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:02:34 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic. 13:02:34 <zodbot> The meeting name has been set to 'meeting' 13:02:43 <mvollmer> .hello mvo 13:02:44 <zodbot> mvollmer: mvo 'Marius Vollmer' <marius.vollmer@gmail.com> 13:03:03 <dperpeet> .hello dperpeet 13:03:04 <zodbot> dperpeet: dperpeet 'None' <dperpeet@redhat.com> 13:03:29 <mvollmer> #topic Agenda 13:03:41 <mvollmer> * network teaming 13:03:54 <mvollmer> * docker storage setup 13:07:54 <mvollmer> alright 13:08:01 <mvollmer> #topic network teaming 13:08:16 <larsu> .hello larsu 13:08:17 <zodbot> larsu: larsu 'Lars Uebernickel' <lars@uebernic.de> 13:08:21 <mvollmer> i did some reading/research and came up with a trello card 13:08:48 <mvollmer> https://trello.com/c/PFP9ruOv/306-spike-research-team-support-in-networking-interface 13:09:10 <mvollmer> i got confirmation from the NM and GCC guys that the plan is basically sound 13:09:22 <mvollmer> let's see if we can pull it off in a reasonable time 13:09:42 <mvollmer> I want to plan this out more for the next three weeks 13:09:42 <dperpeet> nice 13:10:02 <dperpeet> is there anything we should get started on or research for now? 13:10:36 <mvollmer> a review of the existing UI, maybe 13:11:04 <mvollmer> but I don't think we have more experience now than at the time we made this version 13:11:25 <dperpeet> ok 13:11:44 <mvollmer> i had a quick look at bond2team also, and it doesn't look daunting. 13:12:16 <mvollmer> 90% looks like code for parsing stuff and writing JSON from bash 13:12:37 <mvollmer> and quite simple actual conversion logic 13:12:55 <mvollmer> so I stand by the opinion that this isn't rocket surgery 13:12:56 <jscotka> I've discussed teaming with vbenes. 13:13:27 <mvollmer> okay, what did he say? :-) 13:14:00 <jscotka> And he said that logic there could be same as NM has, so add interfaces and paste/edit JSON file, there should not be any predefined values 13:14:03 <cockpitbot> 2 tests failed - http://fedorapeople.org/groups/cockpit/logs/master-576199a4-verify-rhel-7/log.html 13:14:30 <cockpitbot> 1 tests failed - http://fedorapeople.org/groups/cockpit/logs/master-576199a4-verify-rhel-atomic/log.html 13:14:51 <mvollmer> jscotka, well, we said that we can't ask people to input JSON 13:15:16 <mvollmer> that's not good enough 13:15:33 <jscotka> I wanted to know some basic scenarios, but probably using teaming should not be similar as using bond devices. I also wanted to know the most common option and allow only edit these options, but he was against that 13:15:37 <dperpeet> Cockpit users won't paste JSON if I have a say 13:16:13 <mvollmer> jscotka, there are not fantastically many options, are there? 13:16:41 <jscotka> mvollmer, I'm not sure, but probably fewtimes more than bond 13:16:52 <mvollmer> in any case, there is always time for improvements later 13:17:39 <jscotka> mvollmer, exactly. at the begin, we can start with adding some interfaces to teaming, without putting json sonfig file. 13:17:55 <jscotka> mvollmer, it should also somehow work. 13:18:17 <mvollmer> of course 13:18:42 * mvollmer kinda expects to find some bugs in NM, etc... 13:19:00 <mvollmer> eot? 13:19:57 <cockpitbot> 2 tests failed - http://fedorapeople.org/groups/cockpit/logs/master-576199a4-verify-debian-unstable/log.html 13:20:21 <mvollmer> #topic docker storage setup 13:20:29 <mvollmer> i have removed the "WIP" label... 13:20:40 <dperpeet> aha! I didn't see 13:20:43 <mvollmer> fedora 24 updates-testing has new enough versions of all the dependencies 13:20:45 <dperpeet> nice job :) 13:20:56 <mvollmer> so I try to make the tests pass there 13:21:25 <mvollmer> the feature should also nicely turn itself off where it isn't yet supported 13:21:36 <mvollmer> I'll test for that, too, instead of skipping 13:21:55 <mvollmer> e.g, assert on rhel-7 that the curtain comes down etc 13:22:21 <mvollmer> i also made some trello cards for follow up work 13:22:34 <mvollmer> i stopped pushing things down into atomic for now 13:22:42 <mvollmer> but I should pick that up again 13:22:47 <mvollmer> for removing of drives 13:22:56 <mvollmer> and for getting info like usage and total 13:23:25 <dperpeet> yeah, skipping should only be for stuff we can't fix 13:23:34 <mvollmer> dan wants a d-bus api for atomic, so that might be the way forward 13:24:06 <mvollmer> and we should be seeing a lot more Atomic use cases coming our way 13:24:37 <mvollmer> maybe even replace the whole docker page with a new one that talk exclusively to the atomic d-bus api 13:24:45 <mvollmer> but that's a long way down the road 13:25:59 <dperpeet> that seems far off 13:25:59 <dperpeet> yeah 13:26:13 <mvollmer> but would make sense to me 13:27:30 <mvollmer> i have open question about docker storage... 13:27:42 <mvollmer> one open question 13:27:43 <mvollmer> so 13:28:07 <mvollmer> right now the code has its own low-level implementation of computing usage and total 13:28:21 <mvollmer> instead of using "docker info" 13:29:12 <dperpeet> that seems high maintenance 13:29:16 <mvollmer> my code is mostly a proposal, so maybe I should just switch back to "docker info" and try to sell the proposal independently 13:29:33 <mvollmer> yeah 13:30:28 <mvollmer> let's see what docker info says. 13:30:52 <mvollmer> docker info is not an API, but well... 13:31:04 <mvollmer> you get to parse thigns like "10G" 13:31:56 <mvollmer> ok, eot. 13:31:59 <dperpeet> wait 13:32:12 <dperpeet> if the low level implementation works for now, I don't see the need to replace it 13:32:13 <dperpeet> with docker info 13:32:58 <achakrab> Hi, I will be working with integrating features of Atomic onto Cockpit. 13:33:09 <mvollmer> achakrab, hi! 13:33:15 <achakrab> Would there be a good time to discuss which features of Atomic to specifically implement? 13:33:46 <mvollmer> I think now is good. 13:33:49 <achakrab> and the approach in doing so 13:33:56 <mvollmer> achakrab, we are officially in a meeting 13:34:01 <mvollmer> but the agenda is done, no? 13:34:21 <mvollmer> so we just continue with your topic. 13:34:34 <achakrab> @mvollmer Hi 13:34:43 <achakrab> So I had spoken to Dan 13:34:53 <mvollmer> achakrab, but be aware that we post the minutes on the mailing list 13:35:18 <mvollmer> #topic Atomic and Cockpit 13:35:49 <mvollmer> achakrab, go ahead. :-) 13:37:36 <cockpitbot> 2 tests failed - http://fedorapeople.org/groups/cockpit/logs/master-576199a4-verify-fedora-atomic/log.html 13:38:17 <achakrab> @mvollmer okay 13:38:25 <achakrab> so as for features to implement 13:38:41 <achakrab> will the atomic features be features on the UI? 13:38:57 <mvollmer> yes, Cockpit is pretty much all UI, no? 13:39:16 <mvollmer> you can think of this as a UI for the atomic utility 13:39:39 <mvollmer> but a better way is to start with use cases and work down from there 13:40:18 <mvollmer> i am sure the atomic utility itself has been written to concrete use cases, of course 13:42:58 <achakrab> okay so along with its current features such as system. services, containers, logs, storage, networking and tools 13:43:10 <achakrab> we will also now include features of atomic on that bar? 13:43:28 <achakrab> and there will be a call made to atomic to get these features? 13:43:32 <achakrab> a dbus call? 13:43:37 <dperpeet> is "atomic" a feature from the user's perspective? 13:43:39 <mvollmer> that's what we have to figure out 13:43:44 <dperpeet> those things are task oriented 13:44:28 <mvollmer> dperpeet, /usr/bin/atomic is a collection of high-level utilities for a Atomic Host installation 13:44:58 <dperpeet> yes, but have we done any design of where those things could fit in with cockpit? 13:45:23 <mvollmer> no 13:45:27 <mvollmer> except for storage 13:46:35 <mvollmer> the containers page could use "atomic run" and "atomic install" instead of "docker run" and "docker pull", for example. 13:47:00 <mvollmer> one aspect is that "atomic" can work with more than Docker, eventually. 13:47:11 <dperpeet> yeah, but most of that would be transparent to the user, right? 13:47:35 <mvollmer> i guess 13:48:07 <mvollmer> so, it's a big topic, I would say 13:48:17 <mvollmer> achakrab, what do you think? 13:48:35 <mvollmer> does it make sense what I say? 13:49:23 <mvollmer> dperpeet, Atomic also has the concept of a "tools container" and we could support that more directly. 13:49:30 <cockpitbot> 3 tests failed - http://fedorapeople.org/groups/cockpit/logs/master-576199a4-verify-fedora-23/log.html 13:49:31 <dperpeet> right 13:49:41 <mvollmer> the tools container contains all the things that Atomic doesn't want, like man pages, sosreport, gdb, ... 13:49:58 <mvollmer> i think it might be a spc, but I am not sure acutally.... 13:51:19 <mvollmer> one concrete idea would be to read the LABELs out of the images and use them to pre-fill the run dialog. 13:51:39 <achakrab> @mvollmer 13:51:53 * mvollmer shuts up 13:52:18 <dperpeet> well, we should take that from the top 13:52:22 <cockpitbot> 3 tests failed - http://fedorapeople.org/groups/cockpit/logs/master-576199a4-verify-fedora-24/log.html 13:52:31 <dperpeet> seems like right now we're trying to map features into cockpit 13:53:04 <achakrab> @mvollmer so in the containers section, it would have an option to call atomic run 13:53:08 <achakrab> and atomic install? 13:53:17 <mvollmer> yes 13:53:30 <achakrab> okay and the languages used here would be python? 13:53:39 <achakrab> and how would the connecting call be made? 13:53:50 <mvollmer> the cockpit code is JavaScript 13:54:17 <mvollmer> it could either spawn the "atomic" executable with some command line parameters 13:54:26 <achakrab> okay and a lot of the atomic code is in python? 13:54:31 <achakrab> okay 13:54:36 <mvollmer> or it could call a D-Bus method on the atomic service 13:54:41 <achakrab> and then makes the call to atomic run 13:54:53 <achakrab> hmm okay 13:55:12 <mvollmer> achakrab, i think it would make sense to talk you through the basic Cockpit architecture 13:55:24 <achakrab> @mvollmer absolutely 13:55:32 <mvollmer> yeah 13:55:34 <dperpeet> but this should be a follow up to the weekly meeting 13:55:44 <mvollmer> yes 13:56:19 <mvollmer> achakrab, I'll get in contact to arrange something, alright? 13:58:20 <achakrab> @mvollmer yes that works 13:58:34 <achakrab> @mvollmer should i also speak with Dan at some point as well? 13:58:35 <achakrab> about this 13:59:24 <mvollmer> achakrab, of course 13:59:50 <mvollmer> dan has a much better idea of where he wants to go, of course 13:59:52 <achakrab> @mvollmer, okay also i have been studying the code for Cockpit 14:00:07 <achakrab> to try and understand how to make the necessary changes 14:00:10 <mvollmer> okay 14:00:16 <mvollmer> I'll close the meeting real quick... 14:00:18 <mvollmer> #endmeeting