21:03:08 <sgallagh> #startmeeting Server Working Group Weekly Meeting (2016-11-29)
21:03:08 <zodbot> Meeting started Tue Nov 29 21:03:08 2016 UTC.  The chair is sgallagh. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:03:08 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
21:03:08 <zodbot> The meeting name has been set to 'server_working_group_weekly_meeting_(2016-11-29)'
21:03:09 <sgallagh> #chair nirik sgallagh mhayden dperpeet smooge jds2001 vvaldez adamw mjwolf
21:03:09 <zodbot> Current chairs: adamw dperpeet jds2001 mhayden mjwolf nirik sgallagh smooge vvaldez
21:03:10 <sgallagh> #topic roll call
21:03:11 <sgallagh> .hello sgallagh
21:03:12 <zodbot> sgallagh: sgallagh 'Stephen Gallagher' <sgallagh@redhat.com>
21:03:12 <sgallagh> Does anyone see my messages?
21:03:15 <dustymabe> .hello dustymabe
21:03:16 <zodbot> dustymabe: dustymabe 'Dusty Mabe' <dustymabe@redhat.com>
21:03:18 <vvaldez> .hello vvaldez
21:03:19 <zodbot> vvaldez: vvaldez 'Vinny Valdez' <vvaldez@redhat.com>
21:03:21 <mjwolf> .hello mjwolf
21:03:22 <smooge> yes
21:03:25 <zodbot> mjwolf: mjwolf 'Michael Wolf' <mjwolf@us.ibm.com>
21:03:26 <smooge> .hello smooge
21:03:28 <zodbot> smooge: smooge 'Stephen J Smoogen' <smooge@gmail.com>
21:03:29 <sgallagh> IRC is being finnicky for me today, sorry.
21:03:29 <vvaldez> sgallagh: ack
21:03:31 <mhayden> .hello mhayden
21:03:32 <zodbot> mhayden: mhayden 'Major Hayden' <major@mhtx.net>
21:03:42 <dustymabe> hey sgallagh, i'm here but in another video meeting right now
21:03:47 <sgallagh> I started the meeting several minutes ago, then discovered I had dropped
21:03:50 <dustymabe> hopefully will be able to contribute
21:04:07 <adamw> .hello adamwill
21:04:09 <zodbot> adamw: adamwill 'Adam Williamson' <awilliam@redhat.com>
21:04:16 <adamw> i'm seeing ya, sgallagh
21:04:41 <sgallagh> Thanks, disconnecting and reconnecting to my bouncer seems to have addressed it
21:05:44 <sgallagh> #topic Agenda
21:06:12 <sgallagh> I don't have an agenda today. Still digesting Thanksgiving turkey.
21:06:47 <sgallagh> Thanks to everyone who participated in getting Fedora Server 25 out the door.
21:06:50 <sgallagh> /me applauds
21:07:31 * smooge gives virtual virtual cookies to everyone for their work
21:08:05 * adamw has updated all his personal servers and the openQA staging instance to f25
21:08:06 <sgallagh> So that's it, we're done, right? Wait, what do you mean we have to work on Fedora Server 26 now?
21:08:23 <mhayden> haha, wooo
21:08:34 <sgallagh> adamw++
21:08:35 <zodbot> sgallagh: Karma for adamwill changed to 5 (for the f25 release cycle):  https://badges.fedoraproject.org/tags/cookie/any
21:08:47 <adamw> we're fixing some cases where selinux is denying postfix and spamd operations, that's likely to affect quite a few people
21:08:52 <smooge> no we are going to skip Fedora 26.. let Fedora 25 age like a fine wine
21:09:01 <jds2001> .hello jstanley
21:09:02 <zodbot> jds2001: jstanley 'Jon Stanley' <jonstanley@gmail.com>
21:09:05 <mhayden> smooge++
21:09:05 <zodbot> mhayden: Karma for smooge changed to 3 (for the f25 release cycle):  https://badges.fedoraproject.org/tags/cookie/any
21:09:11 <sgallagh> I'm pretty sure we can't get away with thta.
21:09:11 <adamw> there is also an issue where the default PHP config now generates a ton of AVCs due to PCRE JIT compilation doing execmem operations
21:09:24 <sgallagh> Also, every time I've tried to age wine, I got expensive vinegar for my troubles
21:09:38 <adamw> that one kinda needs help from upstream pcre to really be fixed. details on commonbugs page
21:10:11 <sgallagh> #info Fedora Server 25 is released! Gratitude all around
21:10:28 <sgallagh> #info Some late-breaking SELinux issues have cropped up and need to be looked into
21:10:48 <sgallagh> #undo
21:10:48 <zodbot> Removing item from minutes: INFO by sgallagh at 21:10:28 : Some late-breaking SELinux issues have cropped up and need to be looked into
21:10:49 <sgallagh> #undo
21:10:49 <zodbot> Removing item from minutes: INFO by sgallagh at 21:10:11 : Fedora Server 25 is released! Gratitude all around
21:10:52 <adamw> well, they're pretty much looked into, updates are submitted.
21:11:00 <sgallagh> #topic Fedora 25 Release
21:11:02 <sgallagh> #info Fedora Server 25 is released! Gratitude all around
21:11:04 <sgallagh> #info Some late-breaking SELinux issues have cropped up and need to be looked into
21:11:08 * nirik is sort of here now
21:11:11 <sgallagh> adamw: OK, good to know
21:11:34 * dperpeet is here now
21:11:43 <sgallagh> OK, let's talk a little about F26.
21:11:51 <sgallagh> #topic Fedora 26 Planning
21:12:23 <sgallagh> We've made some good progress on organizing our thoughts here, but we (read: I) stalled out on the requirements we were trying to write up.
21:12:40 <sgallagh> We also need to finish the discussion on role design.
21:12:58 <sgallagh> (Neither of these things should happen in this meeting today, as no one is likely prepared)
21:13:06 <vvaldez> ah yes sgallagh, I owe respnses to your comments
21:13:40 <dperpeet> I have a comment on the roles
21:13:49 <sgallagh> #topic Role Design
21:13:53 <sgallagh> dperpeet: You have the floor
21:14:19 <dperpeet> I have a feeling that part of these "roles" is still a bit nebulous
21:14:28 <dperpeet> i.e. how will this all work in the end
21:14:58 <sgallagh> dperpeet: That's kind of why I asked the coordinators to write up a set of use-cases and high-level requirements for discussion
21:14:59 <dperpeet> looking at what we want to achieve, I feel like we should pick one and try to implement it as far as possible
21:15:12 <dperpeet> the Cockpit team feels like NFS is a good candidate for this
21:15:41 <sgallagh> dperpeet: Please be more specific. NFS Server?
21:15:46 <dperpeet> right
21:15:52 <sgallagh> Sounds good to me.
21:15:56 <dperpeet> the document has received some attention
21:15:57 * jds2001 too
21:16:02 <dperpeet> and we feel that we can move forward with that
21:16:16 <dperpeet> and treat this as a case study of how to make it work in Cockpit
21:16:48 <dperpeet> so we'll design a bit
21:16:53 <dperpeet> or rather, andreasn will :)
21:17:19 <dperpeet> and I'll post an update on the mailing list once there is something tangible
21:17:22 <dperpeet> and we can iterate from there
21:17:32 <sgallagh> That would be really excellent, indeed.
21:17:33 <dperpeet> I'm sure we'll identify issues we haven't thought of yet
21:17:54 <dperpeet> especially how to choose a role
21:17:56 <dperpeet> et cetera
21:18:17 <sgallagh> dperpeet: That's great from a user perspective, and we should absolutely do that as soon as we can.
21:18:27 <sgallagh> Some of the open questions we still have are backend implementation questions.
21:18:27 <dperpeet> agreed, so effort there has begun
21:18:56 <dperpeet> I'm hopeful to have some design to discuss in 2016 still
21:19:00 <sgallagh> We've basically settled on using ansible as a means to fire off the deployments, but there's plenty that's ambiguous about how we're doing it under the hood.
21:19:12 <dperpeet> or the end of the next (not this) sprint, if you're following the Cockpit schedule
21:19:16 <sgallagh> Will all deployments be in containers? All on bare metal? A mixture?
21:19:38 <dperpeet> right, but I think for Cockpit we want to look at the other side: UX
21:19:57 <dperpeet> how we drive that under the hood is almost secondary, I think at this point
21:20:13 <sgallagh> dperpeet: Not secondary, but parallel
21:20:21 <dperpeet> fair enough
21:20:34 <sgallagh> But it has to be considered in terms of how we're going to handle scaling our deployments.
21:20:47 <dperpeet> right, but for a user it should "just work"
21:20:51 <sgallagh> NFS is kind of an edge case in this situation.
21:20:55 <dperpeet> or at least guide the user through the necessary steps
21:21:12 <sgallagh> Because you *probably* aren't going to be doing NFS as a microservice.
21:21:19 <dperpeet> might be, but I feel it's fairly easy to get a working prototype
21:21:38 <dperpeet> and once we have user interaction, we can refine the actual behavior/content
21:21:43 <sgallagh> dperpeet: Sure, but I don't want us to build a prototype we can't use as a framework to build further roles either
21:22:06 <sgallagh> Right, I'm not disagreeing with you.
21:22:28 <dperpeet> ok, I'll make sure we pause after the first design stage
21:22:50 <dperpeet> then we'll want all role coordinators to think if what is presented can work for their roles
21:23:07 <sgallagh> dperpeet: And the UI *may* be limited in what we can do with the tech under the hood too.
21:23:25 <sgallagh> For example, iterative interaction may be impossible if our selected tool is declarative.
21:23:29 <dperpeet> I'm fairly certain it will be
21:23:39 <dperpeet> we don't want to project all capabilities into the GUI
21:23:56 <sgallagh> dperpeet: I think you read that backwards.
21:24:17 <sgallagh> I'm saying that things like "we will use ansible as our deployment tool" implies certain limitations on how one can design the UX.
21:24:26 <sgallagh> Like it means that all information has to be provided up-front.
21:24:26 <dperpeet> hm
21:24:31 <dperpeet> true
21:24:56 <sgallagh> That doesn't mean that the UI cannot be iterative, but it can't effect any changes in the middle.
21:25:01 <dperpeet> I'll make sure this is considered
21:25:56 <sgallagh> On the other hand, maybe it's best if you have andreasn just do an idealized design and then we can do a second iteration taking into account tech limitations
21:26:03 <sgallagh> (sorry, thinking as I type)
21:26:31 <dperpeet> in the first stage we usually take tech limits into account, but not necessarily as hard constraints
21:26:32 <sgallagh> If the UI tradeoff is harsh enough, it might drive a change in implementation
21:26:36 <dperpeet> exactly
21:27:00 <dperpeet> that said, once we have something, we'll need feedback to keep momentum
21:27:07 <sgallagh> dperpeet: Agreed
21:27:34 <sgallagh> jds2001: Could you pretty-up and provide your set of use-cases and requirements to the cockpit team?
21:28:00 <jds2001> sgallagh: yeah, how should I do that?
21:28:03 <sgallagh> And since you're acting as coordinator here, I'd like to nominate you for carrying the ball here and making sure that any questions are asked broadly enough?
21:28:05 <dperpeet> what's there looks good, but I think it's important to make sure that the use cases are well-rounded
21:28:16 <jds2001> sgallagh: do i need to fill out a template in triplicate :)
21:28:23 <dperpeet> jds2001, make sure no important use cases are missing :)
21:28:36 <sgallagh> dperpeet: Right, perhaps that would be worth doing iteratively between you, andreasn and jds2001
21:28:48 <dperpeet> certainly
21:28:57 <sgallagh> Perhaps with adamw's expertise in catching cases we might not consider.
21:29:05 <sgallagh> (if he's willing)
21:29:13 <dperpeet> jds2001, what you can do right away is make sure the use cases cover what we need
21:29:41 <dperpeet> jds2001, then once we have a design, you can look over that and see if we interpreted anything wrong :)
21:29:44 <jds2001> dperpeet: certainly
21:30:06 <dperpeet> jds2001, thanks!
21:30:52 <sgallagh> OK, so in order to keep the momentum on this, shall we set a check-in point?
21:31:25 <dperpeet> sgallagh, end of next Cockpit sprint: Dec 13
21:31:38 <dperpeet> we have a meeting on that day
21:31:42 <sgallagh> "at" or "by"?
21:31:49 <sgallagh> Or "on"?
21:31:50 <sgallagh> :)
21:31:52 <dperpeet> on
21:32:05 <dperpeet> let's add it to the agenda that day
21:32:09 <sgallagh> I'd prefer to see some discussion before that, at least a use-case review
21:32:21 <sgallagh> Could we aim to have that done sooner?
21:32:48 <dperpeet> sounds good, but I wouldn't want to promise a certain date
21:32:48 <sgallagh> That way your next sprint has that as input data?
21:33:43 <dperpeet> yeah
21:33:56 <sgallagh> jds2001: I'm going to leave it to your best judgement how to handle that.
21:34:12 <dperpeet> sgallagh, andreas has work on this scheduled for this sprint
21:34:22 <sgallagh> ok
21:34:30 <dperpeet> jds2001, https://trello.com/c/00TuMHlI/414-design-nfs-server-configuration
21:34:51 <dperpeet> I haven't added the doc link there yet
21:35:07 <dperpeet> but as stuff happens, it'll show up there
21:35:31 <sgallagh> dperpeet: OK, I interpreted your earlier statements as being "we plan to start looking at the design on Dec. 13"
21:35:40 <sgallagh> If that's already on the queue, then great!
21:36:01 <dperpeet> I'm hoping we have some design ideas by dec 13th
21:36:55 <sgallagh> #action dperpeet and jds2001 will discuss UX design for NFS role with Cockpit designers
21:37:24 <sgallagh> #info Server SIG will have a check-in on NFS server on Dec. 13th after the Cockpit sprint planning
21:38:41 <sgallagh> Anything else on the roles topic?
21:38:57 <dperpeet> not from me, thanks
21:39:15 <vvaldez> not today
21:39:50 <sgallagh> vvaldez: Sometime before the winter holidays, we should sit down and discuss the domain controller.
21:40:06 <sgallagh> #topic Open Floor
21:40:11 <vvaldez> sgallagh: agree
21:40:52 <sgallagh> Anything for Open Floor today?
21:40:58 <sgallagh> Otherwise, we'll make it a short meeting.
21:43:36 <sgallagh> OK, a short meeting it is, then.
21:43:40 <sgallagh> Thanks for coming, folks.
21:43:43 <sgallagh> #endmeeting