15:01:01 #startmeeting Server Working Group Weekly Meeting (2016-07-26) 15:01:02 Meeting started Tue Jul 26 15:01:01 2016 UTC. The chair is mjwolf. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:01:02 Useful Commands: #action #agreed #halp #info #idea #link #topic. 15:01:02 The meeting name has been set to 'server_working_group_weekly_meeting_(2016-07-26)' 15:01:13 #chair danofsatx nirik stefw adamw simi mhayden jds2001 mjwolf hanthana 15:01:13 Current chairs: adamw danofsatx hanthana jds2001 mhayden mjwolf nirik simi stefw 15:01:20 .hello sgallagh 15:01:21 sgallagh: sgallagh 'Stephen Gallagher' 15:01:29 .hello blc 15:01:29 .hello kevin 15:01:29 bconoboy: blc 'Brendan Conoboy' 15:01:30 mjwolf: (it's simo, not simi, FYI) 15:01:32 nirik: kevin 'Kevin Fenzi' 15:01:54 ah crud I cant cut and paste then :( thanks I will update my notes 15:02:26 thanks 15:02:32 .hello mhayden 15:02:34 mhayden: mhayden 'Major Hayden' 15:02:59 .hello langdon 15:03:00 langdon: langdon 'Langdon White' 15:05:06 /me is concerned by the lack of attendees. 15:05:11 This is an important meeting 15:05:53 Lots of people getting ready for flock? I know I need a new little toothpaste tube. 15:06:12 * mhayden is in a videoconf meeting along with this meeting 15:06:14 adamw is on vacation. 15:06:34 mjwolf: Well, even lacking quorum we should probably discuss the Flock sessions 15:07:02 agreed. it is a bit past the hour I think we have all the people that are going to show 15:07:34 So I will be at flock and can attend the PRD session. I believe sgallagh you mentioned another working session 15:07:45 my main concern is the server talk itself 15:08:15 Right, there is a "Server SIG Pow Wow" scheduled at 13:30 on Friday. 15:08:41 Unfortunately, my flight got moved up and I won't be able to attend it, but it was intended to be a block of time we could use to kickstart our year's efforts. 15:08:48 With whatever we decide we're going to focus on 15:08:54 ok and what is that? Is that a session to help determine where the server should go 15:09:04 ok I can attend that as well 15:09:08 I dont leave until Sat 15:09:21 and take notes 15:09:28 * nirik should be at both barring being sucked into something else. :) 15:09:36 It's meant to be a working session, which may mean active planning, but I think we want to have the broad strokes worked out at the State of the Union session if possible 15:09:42 So people know what they're coming to work on 15:10:18 I'll be there on Friday though I'm not an official serverwg member 15:10:56 bconoboy: WG is only for voting. Anyone who wants to contribute is always welcome. 15:11:05 (And may eventually be asked to become a voting member) 15:11:20 and the State of the Union is the part I'm most uncomfortable with. Not really sure what should be covered 15:12:11 well, I'd say a chunk could talk about f24 stuff... rolekit and partitioning changes and whatever else we did in f23/f24? 15:12:20 mjwolf: I generally ran it as a two-part session: the first part was a recap of the changes made since the previous Flock. 15:12:36 The rest of the session was interactive with the attendees about where we should go next 15:12:50 ah interactive.... 15:13:08 although it's sometimes hard to get people to chime in... 15:13:25 Right, although I think bconoboy and langdon may have something to add there... 15:13:35 * bconoboy nudges langdon 15:13:49 ok. so I know one of the things mentioned in this meeting was replacing rolekit with openshift and there other ideas to stimulate the conversation 15:13:49 ha 15:13:55 * langdon in two meetings 15:14:04 langdon: I can take it if you prefer 15:14:11 * langdon typing 15:14:12 so really it's: a) go on as we have and try and find people to work on rolekit, etc... b) switch to our roles being containers and have a openshift install to run them, c) something else? 15:14:30 we have a c) 15:14:35 well.. coming from the modularity-wg .. we have "doing a server edition in modules" by f26 in our plan.. 15:14:49 probably a "built in fedora-stg" version 15:15:01 ah, that could indeed be interesting. 15:15:06 langdon: Maybe you could talk a little bit about what that might look like? 15:15:42 well.. modularity needs a "release channel".. and we thought server edition was the "simplest" given how we are proceeding with modules.. 15:16:09 the basic idea being this thing we call base-runtime module stack to light up a machine.. and then "things on top" with independent lifecycles 15:16:27 we also have the concept of "install profiles" which are largely similar to roles.. but less interactive 15:16:36 so modules wouldn't be a 1 for 1 replacement for roles right? ie, if we made a freeipa module it would more be the packages and such needed, but wouldn't handle config or the like? or would it? 15:16:45 ah ha. 15:17:02 right now.. modules don't have the "interactive part".. but i think it may be something to explore.. 15:17:27 our idea was more like "i need httpd to develop a website" vs "i need httpd to production serve a website" 15:17:37 * nirik nods. 15:17:46 or w/ base-runtime "i need the bits for a container image" vs "i need the bits to build a vm" 15:17:54 *container base image 15:18:11 so.. we have been planning to get more tightly integrated with the server-wg post flock.. 15:18:26 langdon: nirik asks a good question though: are modules just bits or are they bits+configuration/deployment? 15:18:33 so we could start working on how it would / could be modularized.. and what problems server has that we haven't though of 15:18:47 as far as some of this perhaps cockpit could gain module handling abilities, since it already does a lot of config? 15:18:51 langdon: I think "post-flock" should be changed to "at Flock", don't you? 15:19:14 I like "at flock" idea. 15:19:27 sgallagh, arggh.. i think that is a tough q.. the UX is still the most up in the air.. but I think they could be bits+config+deployment.. but we have mostly explored bits+config 15:19:35 definitely at flock 15:19:58 * nirik thinks this at least gives us several things to float in the talk and discuss at the workshop. ;) 15:20:01 sgallagh, re:flock) perhaps.. "post my preso and have stopped losing my mind at flock" would be more accurate ;) 15:20:17 langdon: Tall order ;-) 15:20:20 especially if we're planning for a full 12 months 15:20:44 because the infrastructure isn't completely ready to go today, but it will be in short order 15:20:50 nirik, re: cockpit) we have actually been building the "ux" as a library to be included in dnf, or cockpit or gnome-software or whatever 15:21:02 so I'm confused about what is the difference between the 2nd half of the state of the union and the friday session 15:21:06 so waiting until the next flock, or even devconf, is more of a wait than needed 15:21:14 cool. ;) 15:21:20 Yeah, I'd love to see us have a prototype or "beta" around the F26 timeframe with something production-ready by F27. 15:21:29 So bring up these ideas in the state of the union and flush out the detail in the working session? 15:21:37 mjwolf: Exactly. 15:21:44 what bconoboy said.. we are trying to have "most of module building" in fedora-stg by flock.. build server for f26 in that stg env by devconf.. and then, hopefully, build for prod in f27 15:21:47 mjwolf: yeah, discussion first, then working/planning on it in the workshop 15:21:51 Or rather "bring up and gather ideas at the state of the union". 15:21:58 There may be audience members with good questions too 15:22:21 right I understand now. 15:22:45 langdon: "library" in what sense? A literal C/Python/Golang module or a language-agnostic API? 15:22:59 sgallagh, python 15:23:00 Maybe that's too low-level for this discussion. I withdraw it. 15:23:14 but the server has an api.. restful.. so agnostic too 15:24:12 * langdon hopes the state of the union is not at the same time as modularity talk 15:24:34 it's not 15:24:38 Man, I wish the schedule wasn't set in stone already. With all this to work on, I'd have asked to swap it to the morning so I wouldn't miss it. Maybe I can BlueJeans into it from the airport... 15:25:11 sgallagh: perhaps we can trade with some other workshop eariler? 15:25:25 nirik: Well, the conbooks are printed, so it would likely be a pain 15:25:33 ah, ok. 15:26:15 so.. can i ask a more meta question? 15:26:25 That said, there's nothing stopping us from hashing some things out on the boat cruise or something :) 15:26:29 langdon: Of course 15:26:51 what do y'all think of the modules plan? do you think it would make sense for server to be the flagship? do you know much about the modularity work? 15:27:13 I do not do you have a url? 15:27:29 I think I know more about it than most (and still much less than I'd like to), but I think Server is an obvious choice. 15:27:38 s/but/and/ 15:27:47 I've followed along as best I can with the time I have... and yeah, I think it might be a good flagship case. 15:27:56 I've always assumed Server would be the first user of modularity. 15:28:09 One of the biggest reasons people hesitate to use Fedora on a server is the Too Fast/Too Slow problem. 15:28:10 mjwolf, https://fedoraproject.org/wiki/Modularity 15:28:15 thanks 15:28:16 This is a chance to directly address that 15:28:27 sgallagh: exactly 15:28:28 yeah. 15:28:47 too fast/ too slow in a release sense? 15:28:59 fundamentally.. modularity is about allowing for lifecycle independence of "modules" 15:29:01 ... 15:29:28 modules are "pieces of software" .. be it the bits that light up the hardware or a web server or a set of things like a LAMP stack 15:29:49 does that help? 15:30:17 mjwolf: It's more: "I want this system to remain exactly the same for five years, except this 1% case that I always want to be tracking upstream" 15:30:27 ha 15:30:36 ah ok. 15:30:47 and.. we would like you to be able to "codify" that in the system itself.. so "yum update" just works 15:31:43 yeah, the old... "I want super stable, only security updates, don't bother me, except for php, which I need the very newest version" 15:31:54 (and do that without experiencing the dangers of getting stuff uncurated from upstream) 15:32:01 So, going back to flock... where are we? 15:32:36 I think we are in good shape for more discussion there. ;) 15:32:37 ugh it would seem that if you pick the wrong thing the dependancy list could be larger than you like 15:32:43 Does anyone want to speak *against* making modular delivery a key part of the F26/F27 strategy? 15:32:48 and yes lots to dicuss 15:32:52 mjwolf, magic happens there :) 15:33:18 Right, don't look too hard at the sausage. It's from Ankh-Morpork. 15:33:56 mjwolf: do you have enough for the talk? 15:34:13 (I'll also point out that, unlike some of our other efforts in the past, modularity has people actively working on it who just volunteered to work with the Server SIG) 15:34:29 yes I think there are plenty of seeds to get folks thinking. and like sgallagh points out I would hope for more suggestions from the audience 15:34:29 nirik: We might want to walk through a list of the last year with him 15:34:37 Unless you think we should take that to the mailing list instead 15:34:50 "had planned to work with the server sig" ... just bringing it up now because we had some concerns about making sure we fit in to your plans 15:34:51 either way. I'd have to look up f23 changes... it's been so long 15:35:59 langdon: Well "had planned to" < "is now doing" :) 15:36:30 sgallagh, true.. i am just pointing out .. we had this in our plan.. so we have staffing for it.. not just a "shot in the dark and a prayer" 15:36:48 mjwolf: I might also mention at the session that there is at least one open seat in the WG. simo reminded me that he resigned several weeks ago 15:37:07 yes I had that on my mental list already 15:37:36 didnt make sense to bring it up for the meeting. hoping for candidates and either settle at flock or at the next meeting after 15:38:11 so for f25 we are kinda in a holding pattern with the current setup right? 15:38:13 I mean get candidate names at flock or deal with it immediately after flock 15:38:14 If this modularity thing gets traction, maybe someone from that side will volunteer for a seat on the Server WG to maintain communication lines? 15:38:26 canddiates should probably go hand in hand with planning at flock, ne? 15:38:31 +1 15:38:35 nirik: Yeah, F25 is not really going to see any changes beyond package rebases. 15:38:48 since you want people on the wg that represent the direction the wg chooses 15:38:50 right. but we need to try and keep everything working. ;) 15:38:57 i do have someone taking on leadership of base-runtime module which i think maps most directly in the near term.. 15:40:22 * langdon notes there isn't a great reference to point at for what that is yet aside from a recent demo : https://www.youtube.com/watch?v=IKUioQTl-Bo&list=PLcwHJG45BmAN0XNiNsLbF7203HCNBPZWG&index=2 15:41:05 * nirik saves link for later 15:41:34 mjwolf: So what else do you need to prepare? 15:41:50 * nirik gets more coffee, back in a min. 15:42:04 Do you want to send a mail to the list asking for a list of stuff for the first part of the session? 15:42:07 * langdon glares at empty coffee pot 15:42:14 we covered everthing here. I would still like an conversation on the list about f23/f24 changes 15:42:27 OK, kick it off and I'll add my notes after lunch 15:42:33 ok 15:43:06 so are there any other topics people want to cover today? 15:44:28 please join langdon's modularity presentation :-) 15:44:40 there will be live demos! 15:44:47 so you know that will be fun 15:44:50 bconoboy: This is his Flock session? 15:44:58 sgallagh, yeah 15:45:00 https://flock2016.sched.org/event/6yoy/modularity-why-where-we-are-and-how-to-get-involved 15:45:01 https://flock2016.sched.org/event/6yoy/modularity-why-where-we-are-and-how-to-get-involved 15:45:07 ahh there its 15:45:10 *it is 15:45:17 Someone want to #link that? 15:45:19 11am wednesday 15:45:23 yes I'm marking that on my schedule 15:45:31 #link https://flock2016.sched.org/event/6yoy/modularity-why-where-we-are-and-how-to-get-involved 15:45:32 we also were trying to get a workshop on how to build a module .. but we kinda dropped the ball on getting it scheduled :( 15:46:08 langdon: If things continue in this vein, maybe that happens at the Pow Wow 15:46:29 * nirik nods 15:46:37 sgallagh, sure.. (/me admits he is not entirely sure what all the events you were discussing were) 15:46:58 https://flock2016.sched.org/event/76ol/server-sig-pow-wow 15:47:12 friday from 1:30pm to 3:30pm 15:47:19 langdon: There are two main Server sessions: the State of the Fedora Server Union, which is a recap of this year and a 50,000ft planning session for the next year 15:47:24 oh.. that might be super tasty 15:47:39 And the Pow Wow, which was meant to be a working session to kickstart the new year ideas. 15:47:41 could we do a 20m session on "how to build a module"? 15:47:42 it's good timing and a nice thick 2 hour session 15:47:58 wait is it 3 or 2 slots? 15:48:39 langdon: Two slots: https://flock2016.sched.org/event/76mM/state-of-the-fedora-server-union and https://flock2016.sched.org/event/76ol/server-sig-pow-wow 15:48:44 Friday, August 5 • 13:30 - 15:30 15:48:54 The first one is 50 minutes 15:48:56 On Tuesday 15:49:13 The second one is a workshop/hackfest on Friday 15:49:27 ok.. but where is the 50k view for planning bit? 15:49:46 the state of the union 15:49:58 ahh gotcha.. 15:50:02 langdon: It starts with a recap and then goes into the 50k planning 15:50:23 langdon: can you send me a few bullet points or a slide about it and I will include it in the state of the union address 15:50:28 well.. if y'all like.. we could do a bit on how to build a module in the pow-wow.. we should be able to do it in 30m or less 15:50:39 As for "How to build a module", mjwolf do you want to give the first 20 minutes of the Pow Wow to Langdon? 15:50:43 otherwise I will go off of the notes from this meeting 15:50:53 sure 15:51:02 that sounds nice to me 15:51:12 mjwolf, slide for what exactly? how i want modules for server or whatnot? 15:51:12 if there werent other ideas we can let it keep going 15:51:29 if there were other ideas at some point we can cut it off and discuss the other ideas 15:52:04 langdon: how would you like me to introduce the modules idea at the 50k foot level to the server people 15:52:25 mjwolf, ack.. ill see if i can build you something 15:52:34 mjwolf@us.ibm.com 15:54:19 ok any other subjects or do we call it a day? 15:55:04 /me has nothing else 15:55:12 * nirik either 15:55:24 ok. thanks everyone. It was a great discussion today 15:55:39 mjwolf++ 15:55:39 sgallagh: Karma for mjwolf changed to 1 (for the f24 release cycle): https://badges.fedoraproject.org/tags/cookie/any 15:55:40 #endmeeting