13:01:45 <andreasn> #startmeeting
13:01:45 <zodbot> Meeting started Mon Mar 30 13:01:45 2015 UTC.  The chair is andreasn. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:01:45 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
13:02:02 <andreasn> .hello andreasn
13:02:04 <zodbot> andreasn: andreasn 'Andreas Nilsson' <anilsson@redhat.com>
13:02:29 <dperpeet> .hello dperpeet
13:02:30 <zodbot> dperpeet: dperpeet 'Dominik Perpeet' <dperpeet@redhat.com>
13:02:43 <sgallagh> .hello sgallagh
13:02:44 <zodbot> sgallagh: sgallagh 'Stephen Gallagher' <sgallagh@redhat.com>
13:03:17 <andreasn> petervo, stefw, you here too?
13:03:22 <stefw> .hello stefw
13:03:24 <zodbot> stefw: stefw 'Stef Walter' <stefw@redhat.com>
13:03:35 <andreasn> #topic Agenda
13:03:44 <andreasn> what do we have today?
13:03:47 <andreasn> * Fedora 22
13:04:06 <petervo> yep
13:04:30 <sgallagh> * GSoC
13:04:56 <andreasn> * Patternfly
13:04:57 <stefw> * User syncing without accountsservice
13:05:10 <petervo> * storage daemon
13:06:07 <github> [cockpit] dperpeet opened pull request #2059: Test: add support for RHEL guest (master...rhel_ci) http://git.io/jKFW
13:06:18 <andreasn> all right
13:06:26 <andreasn> #topic Fedora 22
13:06:59 <stefw> If we still want to be able to add Fedora 22 servers to future dashboards ... (ie: backwards compatibility) then we are not ready yet
13:07:02 <sgallagh> I noticed that the branding suddenly appeared in a recent update, so thanks for that
13:07:22 <stefw> We need to migrate away from the cockpitd Accounts code in order to have that backwards compatibility
13:07:43 <stefw> WIP branch: https://github.com/cockpit-project/cockpit/pull/2057
13:07:58 <mvollmer> (in theory the setup code could fall back to using cockpitd when it detects it.)
13:08:13 <stefw> true, except for even now that code errors out
13:08:59 <stefw> so i'd like to see if we can do our Fedora 22 release later tonight?
13:09:07 <stefw> and try and include the code that migrates away from cockpitd Accounts
13:09:37 <sgallagh> stefw: That's going to be a bit tight for getting it into stable, but we can manage it as long as we get the necessary karma
13:09:54 <sgallagh> (Otherwise we have to aim for a blocker or freeze exception bug)
13:10:13 <stefw> between petervo, sgallagh, and me, we can find 2 karma
13:10:26 <stefw> which is usually what we require
13:11:11 <sgallagh> /me nods
13:11:38 <stefw> anything else on Fedora 22?
13:11:44 <andreasn> don't think so
13:11:52 <andreasn> #topic GSoC
13:13:01 <sgallagh> So, we have three candidates for Cockpit in GSoC
13:13:03 <dperpeet> I haven't heard from any candidates personally
13:13:23 <sgallagh> One of them may not actually be eligible to participate, we're sorting that out.
13:13:29 <andreasn> do you have a URL to the tasks?
13:14:00 <sgallagh> https://fedoraproject.org/wiki/Summer_coding_ideas_for_2015#Cockpit_UI_for_Rolekit
13:14:03 <andreasn> thanks
13:14:10 <andreasn> #info https://fedoraproject.org/wiki/Summer_coding_ideas_for_2015#Cockpit_UI_for_Rolekit
13:14:31 <sgallagh> #info Tasks are "Cockpit UI for Rolekit", "Docker Volume Support" and "Cockpit support for systemd timers"
13:15:03 <dperpeet> primary topic was rolekit, we added the others last minute
13:15:15 <andreasn> so three candidates as people who reached out and said they are interested in these tasks, or 3 candidates as in 3 tasks?
13:16:06 <dperpeet> they all signed up for rolekit, but after we posted the other two as alternatives, two of the candidates also posted their interested in the two new topics
13:16:17 <sgallagh> andreasn: Three candidates (all with well thought-out proposals) applied for the rolekit one
13:16:27 <andreasn> ah, I see
13:16:52 <sgallagh> We added others because it looked like we might end up with more good candidates than tasks :)
13:17:11 <andreasn> the other two tasks looks excellent as well
13:17:39 <sgallagh> Unfortunately, as I mentioned above, one of the applicants may be ineligible (I'm not going to name the student publicly) due to a possible conflicting relationship with the mentoring organization.
13:17:47 <andreasn> right
13:18:54 <sgallagh> Anyway, the main point here is that dperpeet and I need to get in touch with the applicants and decide how to proceed.
13:19:03 <sgallagh> #action sgallagh and dperpeet to review GSoC applications
13:19:21 <dperpeet> sgallagh, I sent you an e-mail with preliminary thoughts on this, but we can discuss further
13:19:41 <sgallagh> Yeah, I saw the email but haven't gotten that far through my weekend backlog :)
13:19:58 <sgallagh> We can take that outside of the meeting.
13:21:02 <dperpeet> ok
13:21:15 <andreasn> all right, anything else on that topic?
13:21:22 <sgallagh> Not for now. More as it develops :)
13:21:35 <andreasn> excellent, thanks for driving this!
13:21:44 <sgallagh> Happy to :)
13:21:45 <andreasn> #topic Patternfly migration
13:22:24 <andreasn> pull request is here https://github.com/cockpit-project/cockpit/pull/2058
13:22:37 <andreasn> will need to test a bit more and see I didn't forget any old spinners in there
13:23:02 <andreasn> nothing more to report
13:23:36 <dperpeet> andreasn, thanks for that - looking forward to seeing the current style used consistently
13:23:58 <andreasn> yeah, would be great to leave the old version behind us
13:24:07 <andreasn> ok, next topic I guess
13:24:10 <andreasn> #topic User syncing without accountsservice
13:24:49 <stefw> i've been working on this today
13:24:54 <stefw> i went down the wrong road at first
13:25:19 <stefw> because of the (somewhat unfortunate) way we handle running the remote cockpit-bridge during the 'add to dashboard' setup process
13:25:33 <stefw> i wasn't able to implement the logic for syncing users in javascript
13:25:52 <stefw> this is because the javascript code requires multiple channels, and we can only open one channel to the remote cockpit-bridge during setup
13:26:09 <stefw> so this means essentially either spawning a remote process, or using a remote dbus API
13:26:24 <stefw> i've opted for the latter, which is to implement an internal DBus API in cockpit-bridge, by which the users are synced
13:26:37 <stefw> it's a cockpit.Setup() API, that has various phases, and is extensible for domain support later
13:26:48 <stefw> this is the pull request: https://github.com/cockpit-project/cockpit/pull/2057
13:26:52 <stefw> i'm writing detailed tests right now
13:27:11 <stefw> that's it
13:27:44 <mvollmer> cool
13:28:04 <mvollmer> what's the reason for having only one channel?
13:28:12 <mvollmer> the different credentials?
13:28:16 <stefw> and host keys, yes
13:28:20 <mvollmer> right
13:28:28 <stefw> we sorta cringed when we were implementing that in cockpitwebservice, i remember
13:28:35 <mvollmer> yes
13:28:36 <stefw> but it's too late to change it now :S
13:29:06 <stefw> there is one thing we could do to make it "less magical"
13:29:13 <stefw> and that is to make opening a private channel explicit
13:29:20 <stefw> rather than implicit
13:29:40 <stefw> right now it's implicitly private if there is a 'user' option and/or a 'host-key' option
13:29:42 <stefw> (maybe others?)
13:30:03 <mvollmer> why can't the js open multiple private channels?
13:30:12 <stefw> N ssh connections?
13:30:15 <mvollmer> if it can open one, it can open many, no?
13:30:21 <stefw> sure, but it would be pretty nasty
13:30:25 <mvollmer> i see
13:30:30 <mvollmer> true
13:30:31 <stefw> lets say you want to read /etc/passwd /etc/group and /etc/shadow
13:30:34 <stefw> that's 3 channels
13:30:38 <stefw> then you want to check if users exist
13:30:40 <stefw> then you want to run useradd
13:30:43 <stefw> then you want to change passwords
13:31:00 <stefw> and you've just created N + 5 channels
13:31:02 <mvollmer> right, so we are taken dozens of channels.
13:31:03 <stefw> where N is the number of users
13:31:04 <stefw> right
13:31:08 <stefw> so anyway
13:31:16 <stefw> lets just stick with what we have
13:31:23 <mvollmer> yep
13:31:24 <stefw> and live within that
13:33:27 <andreasn> all right, next topic?
13:33:30 <stefw> yup
13:33:37 <andreasn> #topic storage daemon
13:33:45 <petervo> so PR is in
13:33:55 <stefw> i was wondering about the name
13:34:13 <petervo> github is still having issues
13:34:16 <stefw> can we keep cockpit-storaged for the storaged we will likely have to self-host for RHEL?
13:34:23 <petervo> what were you thinking?
13:34:26 <stefw> and maybe use cockpit-udisks as the client-side binary name?
13:34:36 <petervo> ok
13:34:43 <stefw> or something like that?
13:34:52 <petervo> https://github.com/cockpit-project/cockpit/pull/2050
13:34:55 <stefw> otherwise the name 'storaged' is used in both things
13:35:39 <mvollmer> i was thinking cockpitd-storage
13:36:11 <stefw> it's sorta like cockpit-pcp
13:36:11 <mvollmer> petervo, pr looks good to me.
13:36:19 <mvollmer> petervo, what are the next steps?
13:36:19 <stefw> but it's for udisks
13:36:35 <mvollmer> import storaged?
13:36:41 <mvollmer> and rename it a bit?
13:37:30 <petervo> i wasn't sure if wanted to move storaged inside on fedora as well
13:37:47 <mvollmer> right
13:37:49 <stefw> it would allow the new storaged to start replacing us on rawhide
13:37:54 <stefw> even if we don't yet have time to migrate to the new storaged
13:37:58 <petervo> or if that was something we were going to do just as part of the build process for OSest that need it
13:38:26 <mvollmer> i agree with stefw, it would be nice to make space for the new one.
13:38:32 <mvollmer> in fedora.
13:38:46 <petervo> ok
13:39:10 <petervo> so cockpit-storaged and cockpit-udisks
13:39:24 <petervo> everyone good with those 2?
13:39:30 <stefw> ++
13:39:41 <stefw> or cockpit-storaged and cockpit-storage
13:39:47 <stefw> if people have problems with 'udisks'
13:39:54 <stefw> but i really think this is udisks specific implementation
13:40:11 <mvollmer> yes
13:40:11 <stefw> even the fact that the old storaged exists
13:41:54 <petervo> i think that's it for that topic
13:42:06 <andreasn> all right
13:42:10 <andreasn> #topic Open Floor
13:42:52 <dperpeet> pr for RHEL support is in https://github.com/cockpit-project/cockpit/pull/2059
13:42:53 <stefw> can anyone else think of anything else for Fedora 22
13:42:54 <dperpeet> finally
13:43:25 <andreasn> I e-mailed Ryan Lerch twice and asked if he had any feedback on the current branding
13:43:34 <andreasn> but I never got a reply back, so I guess it's ok
13:43:46 <stefw> cool
13:43:49 <sgallagh> andreasn: How recently?
13:44:16 <sgallagh> /me notes that Ryan has been on his way to Australia for about half a week and will be out of contact until next week.
13:44:18 <andreasn> on the 24th
13:44:38 <sgallagh> He left on the 25th I think, so he may not actually have had a chance to examine it.
13:44:50 <sgallagh> Might be worth pinging mizmo or edirsh
13:45:00 <andreasn> all right
13:46:53 <andreasn> anything else for the open floor?
13:48:47 <andreasn> sounds not
13:48:51 <andreasn> #endmeeting