15:04:17 #startmeeting Server Working Group Weekly Meeting (2016-08-09) 15:04:17 Meeting started Tue Aug 9 15:04:17 2016 UTC. The chair is nirik. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:04:17 Useful Commands: #action #agreed #halp #info #idea #link #topic. 15:04:17 The meeting name has been set to 'server_working_group_weekly_meeting_(2016-08-09)' 15:04:23 #chair danofsatx nirik stefw adamw simi mhayden jds2001 mjwolf hanthana 15:04:23 Current chairs: adamw danofsatx hanthana jds2001 mhayden mjwolf nirik simi stefw 15:04:28 #topic Agenda 15:04:39 nirik: "simi" is still misspelled. 15:04:45 sgallagh: oops. 15:04:51 I just grabbed this from the last meeting 15:05:02 Hi, here is Robert Richter from Cavium 15:05:21 morning 15:05:25 Christopher Covington from Qualcomm 15:05:29 hi all 15:05:39 .hello sgallagh 15:05:40 sgallagh: sgallagh 'Stephen Gallagher' 15:06:14 .hello blc 15:06:15 bconoboy: blc 'Brendan Conoboy' 15:06:37 hello everyone. 15:07:00 I thought we might use todays meeting to go over discussions at flock and roadmap and such 15:07:49 cov, rric: Nice to meet you! 15:07:51 nirik: sounds good. Jon Stanley can't be the only one looking for info 15:07:59 hi, all 15:08:13 my first time on the meeting her 15:08:16 here 15:09:47 #topic Flock discussions / Roadmap for f26 and beyond 15:10:10 So, there's been some email on list this morning going over things. 15:10:14 welcome mjwolf 15:10:26 hello nirik 15:10:51 We had some great discussions and input from folks at flock... 15:12:10 My take away from that was that we would look at seeing if we could use rolekit's interface/frontend, but replace it's backend with ansible and make roles ansible playbooks. No one from rolekit was there, so we don't know how feasable this might be. 15:12:25 and then also work with modularity folks to move to modules. 15:12:44 this would all be f26 and beyond, as f25 is already in alpha freeze 15:13:26 there was also discussion on whether an "undo" would be needed 15:13:35 rolekit currently supports that 15:13:40 nirik: I assume you're talking about the Pow Wow? 15:13:41 .hello langdon 15:13:42 langdon: langdon 'Langdon White' 15:13:47 sgallagh: yeah. 15:13:52 * langdon sorta lurks, also running modularity wg meeting 15:14:10 sgallagh: perhaps you could chime in from a rolekit perspective. is that at all feasable to replace backend? 15:14:22 there was also discussion about name-spacing galaxy playbooks for ansible 15:14:52 nirik: It *might* be, but given that the usage of it is minimal today, I'd think it would be nice-to-have rather than need-to-have. 15:14:53 I think thats further down the road, but yeah, something to look into 15:15:22 To the best of my knowledge, nothing is consuming the rolekit API outside of the rolectl CLI to date. 15:15:28 sgallagh: well, the thought is that if we can reuse it's interface, we don't have to change cockpit or break existing server users, or define a new interface. 15:15:39 isn't cockpit? 15:15:42 There's nothing stopping us from hacking up the CLI to wrap a new solution, of course. 15:16:02 nirik: No, Cockpit never got around to it. Resource contention. 15:16:08 ahh... 15:16:27 And when we set rolekit aside, I told them to take it off their priority list until and unless we restarted it 15:16:36 openqa is using it for tests... but adamw loves re-writing those. ;) 15:16:54 adamw: Are you using the D-BUS API or just the CLI? 15:17:06 Wrapping the CLI wouldn't be particularly hard, I suppose. 15:17:24 Particularly if we constrained it to just the stuff OpenQA is using for the time being. 15:17:57 * nirik is happy to do whatever works best, we just thought it might be good to re-use the existing interface if it was easy 15:18:52 .hello snavin 15:18:53 hanthana_: snavin 'Danishka Navin' 15:19:16 Unless something compelling is using it, I'd scrap it and do it better this time around 15:19:22 of course the question here is always who is going to do the work. ;) I can commit to working on the ansible playbook side of things, but beyond that I am pretty swamped. 15:20:02 I think we're getting a little too far into implementation before we decide if that's the route we want to go :) 15:20:50 sure. 15:20:54 probably more productive thinking about where you want to go first... 15:21:20 beyond ansible integration there was a lot of interest in modularity 15:21:46 There's an unfortunate scheduling problem in that the serverwg meeting happens at the same time as the modularitywg 15:22:31 that'll need to be sorted out if there is going to be cooperation between the wgs 15:23:47 * bconoboy hoping somebody else will run with this topic 15:23:56 we can adjust our meeting time, sure... 15:24:36 I think modularity will get us a lot of things people have been asking for: different lifecycles, being able to switch versions of stacks easily, less dependencies/footprint, etc 15:24:57 anyway, the modularity wg would like to have the server edition be the first instance of a modularized OS. maybe a PoC version in f25, delivering an actual release with f26 15:25:18 bconoboy: Well, F25 is three months away. 15:25:21 That seems... ambitious. 15:25:28 sgallagh: openqa test uses the cli 15:25:29 and we just froze for alpha today 15:25:31 I think a realistic goal would be F26 15:25:36 they are nothing if not ambitious ;-) a PoC wouldn't be an official build. 15:25:37 sorry, got a few different things happening. 15:26:19 Anyway, sorting out what a PoC even means will be interesting... and the entire enterprise is a cool direction to take the serverwg 15:26:25 sgallagh: https://bitbucket.org/rajcze/openqa_fedora/src/6fdc9934d41ff4b00cdf165f2502ea3c82c67e88/tests/role_deploy_domain_controller.pm?at=develop&fileviewer=file-view-default shows you everything it does. 15:26:31 adamw: RE: openqa, I'd either hand you a wrapper or modify the test for you. Whichever ends up being less work. 15:26:39 fine by me. 15:26:57 well, we'd rather edit the test, i think, because the point of the test is to make sure the intended interface for rolekit works... 15:27:10 anyhow. 15:27:18 true enough 15:28:00 bconoboy: No, but I think we should probably look at working in Rawhide for now. 15:28:10 That way we don't cause any side-effects that would impact F25 15:29:15 sgallagh: wfm 15:30:05 I still think we're jumping ahead a bit though. 15:30:33 I'm going to try in the next few days to work on putting the Logic Model we worked out at the PRD session into a pretty graphical display so we can discuss and adjust it. 15:30:48 Then we can use that to direct our decision-making. 15:30:51 yeah, we need to go over the PRD.... 15:30:59 I haven't had time to really look at that. 15:32:21 I don't think much of the content will ultimately change, but I think we'll have a better description of how we got to each of the goals. 15:33:07 For those who weren't there for the PRD session and logic model: 15:33:32 mattdm presented the kellogg logic model at his star trek talk. Pointer to the model: https://www.wkkf.org/resource-directory/resource/2006/02/wk-kellogg-foundation-logic-model-development-guide 15:34:00 A very succinct image showing the model: 15:34:04 http://toolkit.pellinstitute.org/wp-content/uploads/2010/06/Reading-a-Logic-Model.png 15:34:13 You plan from right to left, execute from left to right 15:34:17 The short version is that it planning should happen from long-term vision down to implementation details, while work should go the other way. 15:34:30 bconoboy put it more elegantly 15:34:41 "You plan from right to left, execute from left to right" - mattdm 15:34:50 I totally stole that from mattdm :-) 15:35:11 Anyway, at the PRD session on Friday there was an effort to use this model for the server wg 15:35:38 We got through impact, outcomes, and some of the outputs- but did not cover activities or resources/inputs 15:36:13 There is a link to this work here: https://fedorahosted.org/council/ticket/63 15:36:49 Right, and not covering activities and resources was a deliberate choice (left for the WG) 15:37:07 The version in that ticket is in text and not very easy to follow. 15:37:14 As I said, I'll try to figure out a better way to present it 15:37:43 cool 15:38:10 it really is helpful to have it in the diagram form with the arrows 15:39:09 * nirik pulls the pdf to read when time permits 15:39:10 bconoboy: I think I could make something functional without arrows just by using Trello/Taiga 15:39:30 I may do that today as a short-term fix 15:39:49 that'd be great 15:41:46 OK, I'll see what I can do and I'll post it to the list when it's ready 15:42:11 cool. Thanks sgallagh 15:42:25 so I guess thats our next step? PRD rework/discussion and then from that a roadmap ? 15:43:08 The thing that wasn't clear to me is if the PRD as discussed there is in some way "final" or if it was an exercise for the serverwg as a whole, some of who were not present 15:43:37 We should probably also do a check-in to see who is still interested in being on the WG 15:43:47 I know we have at least one open seat: Simo's 15:43:56 #info sgallagh to work on a PRD draft/kellog model for server this week 15:44:07 #info should check who is still interested in being a part of the group 15:44:10 bconoboy: That was meant to seed the discussion and focus it, but ultimately the WG has to approve it. 15:44:21 Particularly since I don't think a single sitting WG member was present at the time... 15:44:37 sgallagh: OK, so it sounds like it was a trial run. that's great- it would be nice to have the full WG contribute 15:44:37 yeah, it needs approval/input from the working group 15:44:42 I mostly ask about membership because if we move into a closer alignment with modularity, we may want to cross-pollinate... 15:45:27 * nirik nods. 15:47:30 i don't think i want to be cross-pollinated. it sounds painful. 15:47:42 heh 15:47:55 ok, anything further today? Or shall we call it a meeting and continue on the list? 15:48:26 what was the plan for the meeting time? 15:48:49 I can send out a whenisgood, but then we need to know who all is involved first? 15:49:22 ah, so maybe wg membership, then meeting time? 15:49:25 guys, I was suppose to write a blogpost but could not find a time as I was really busy. I will be working on it. 15:50:00 bconoboy: yeah, thats my thought... 15:51:06 sounds good- some modularity team members might want to join in 15:52:38 so... what was the membership plan? :-) 15:53:17 I can send an email and ask folks to express their continued interest or not? we see who we have left and how many open seats to fill... 15:53:31 nirik: That would probably be best. 15:53:47 Send to the list and CC all the sitting members? 15:54:03 say hold it open for a week then check next meeting? or more time better? 15:55:34 Let's give two weeks for people to respond with yay or nay. If only because it's vacation season 15:55:47 yeah. 15:56:07 #info check existing members for continued interest on list over next 2 weeks. 15:57:12 sounds good. Anything else? 15:57:14 #topic Open floor 15:59:03 ok then, will close out the meeting in a minute if nothing more. 15:59:51 Thanks for coming everyone! 15:59:54 #endmeeting