cockpit
LOGS
14:04:16 <andreasn_> #startmeeting Cockpit
14:04:16 <zodbot> Meeting started Mon Feb  9 14:04:16 2015 UTC.  The chair is andreasn_. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:04:16 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
14:04:27 <sgallagh> Hello, all
14:04:32 <mvollmer> .hello mvo
14:04:33 <zodbot> mvollmer: mvo 'Marius Vollmer' <marius.vollmer@gmail.com>
14:04:33 <sgallagh> .hello sgallagh
14:04:36 <zodbot> sgallagh: sgallagh 'Stephen Gallagher' <sgallagh@redhat.com>
14:04:39 <andreasn_> #chair andreasn petervo dperpeet mvollmer sgallagh stefw
14:04:39 <zodbot> Current chairs: andreasn andreasn_ dperpeet mvollmer petervo sgallagh stefw
14:04:50 <andreasn_> .hello andreasn
14:04:52 <zodbot> andreasn_: andreasn 'Andreas Nilsson' <anilsson@redhat.com>
14:05:18 <andreasn_> #topic Agenda
14:05:38 <mvollmer> * Rewrite cockpit in something else than JavaScript
14:05:45 <mvollmer> :-)
14:05:47 <andreasn_> hehehe
14:05:55 <mvollmer> sgallagh is learning JS here...
14:05:59 <andreasn_> * Fedora 22
14:05:59 <stefw> :)
14:06:58 <andreasn_> * Docker stuff
14:07:18 <andreasn_> * Devconf recap
14:07:56 <petervo> * system updates
14:08:08 <andreasn_> yes!
14:08:47 <andreasn_> ok, lets run with that agenda
14:09:02 <andreasn_> #topic Fedora 22
14:09:50 <andreasn_> so Fedora 22 is launching in April
14:10:08 <twoerner> .hello twoerner
14:10:09 <zodbot> twoerner: twoerner 'Thomas Woerner' <twoerner@redhat.com>
14:10:09 * stefw pulls up the trello roadmap card for that
14:10:22 <stefw> Fedora 22 Milestone: https://trello.com/c/duRQw1rU/113-milestone-fedora-22
14:10:25 <andreasn_> and the beta freeze is march 21st
14:10:44 <andreasn_> so there is a couple of things in that Trello card
14:11:41 <dperpeet> should we update trello deadline to March 21st? it's currently April 1st
14:11:42 <andreasn_> the easier branding is just an optimization
14:11:54 <stefw> dperpeet, yes lets
14:12:18 <stefw> dperpeet, done
14:12:46 <stefw> although it's worth noting that these deadlines usually slip a bit
14:13:00 <dperpeet> stefw, no reason to follow a bad example :)
14:13:05 <stefw> heh
14:13:07 <sgallagh> stefw: We're trying very hard not to slip this time
14:13:12 <sgallagh> (We as in FESCo)
14:13:19 <stefw> noted
14:13:59 <dperpeet> andreasn_, I think the easier branding isn't that big of an issue for us, but makes everything look a lot more polished (i.e. a good goal to leave in there)
14:14:12 <andreasn_> yeah
14:14:12 <dperpeet> good in respect to effort/gain ratio
14:14:48 <sgallagh> I just updated the Server Roles card.
14:15:01 <sgallagh> I've made headway on that during DevConf this week, so I moved it to In Progress
14:15:12 <andreasn_> \o/
14:15:20 <andreasn_> anything else for F22?
14:15:40 <stefw> sgallagh, nice
14:15:54 <stefw> #info Fedora 22 Milestone is tracked here: https://trello.com/c/duRQw1rU/113-milestone-fedora-22
14:15:54 <mvollmer> PCP is coming along nicely
14:15:54 <sgallagh> Easier branding or not, we need to make sure to address branding issues for the Fedora version by beta
14:16:07 <sgallagh> (Such as dropping the "not suitable for production" messages like we did in F21)
14:16:13 <andreasn_> yeah
14:16:13 <stefw> sgallagh, nod
14:16:18 <mvollmer> We made some plans for the missing UI goodness
14:16:37 <twoerner> A question that came up today was, if a firewall plugin in CockPit could be a replacement of firewall-config
14:16:37 <mvollmer> decision is still open for how to actually configure/control pmlogger
14:16:57 <sgallagh> twoerner: We're on the Fedora 22 agenda item right now
14:17:05 <mvollmer> the ready-made scripts offered by pcp might not be robust enough.
14:17:21 <mvollmer> but no risk for F22 there.
14:17:39 <twoerner> sgallagh: I am heading to a meeting in 13 minutes..
14:17:58 <sgallagh> twoerner: OK, we'll arrange a separate meeting on that.
14:18:01 <sgallagh> So we don't run short on time
14:18:21 <stefw> makes sense
14:18:24 <twoerner> sgallagh: thanls
14:18:26 <twoerner> thanks
14:18:51 <andreasn_> ok, next topic
14:19:02 <andreasn_> #topic Docker stuff
14:19:13 <andreasn_> so some cool stuff by petervo and dperpeet on that
14:19:32 <dperpeet> links are in, thanks to petervo
14:19:39 <andreasn_> yay!
14:19:43 <andreasn_> and works great!
14:19:53 <andreasn_> and storage is coming in https://github.com/cockpit-project/cockpit/pull/1765
14:20:04 <dperpeet> I'm currently updating the validation with the links and based on design feedback from andreasn_
14:20:32 <andreasn_> neat
14:20:51 <dperpeet> https://github.com/cockpit-project/cockpit/pull/1772 is functional, so shouldn't be much longer
14:21:18 <andreasn_> is 1765 is blocking on docker upstream?
14:21:29 <petervo> not really
14:21:57 <petervo> it has a warning, once we get the patch in, then we can add code to remove the warning
14:22:03 <andreasn_> ok
14:22:06 <petervo> for newer docker
14:22:20 <mvollmer> i heard that device mapper as a backend is on the way out.  true?
14:22:26 <petervo> i'm working on revising the patch a little today
14:22:32 <andreasn_> great
14:22:39 <petervo> then i think it will be in soon
14:22:46 <petervo> i hadn't heard that
14:23:11 <mvollmer> petervo, ok.
14:23:20 <petervo> last i heard they just made it work better with lvm
14:23:32 <petervo> but it's not like i'm really in the loop there
14:23:39 <mvollmer> maybe I confused one way of using lvm with using lvm in general.
14:23:40 <petervo> so it's possible
14:23:48 <mvollmer> me neither
14:23:59 <mvollmer> what happens if everyone switches to overlayfs over night?
14:24:20 <petervo> then no storage graph
14:24:26 <mvollmer> right
14:24:35 <mvollmer> good enough, I guess
14:24:39 <petervo> we'll need to contribute some reporting to docker
14:24:44 <petervo> for other graph backends
14:24:51 <andreasn_> right
14:24:59 <andreasn_> next topic?
14:25:08 <stefw> petervo, i discussed this with vbatts
14:25:15 <stefw> and hopefully he'll get back to you about the docker patch soon
14:25:32 <stefw> he wants to understand the problem fully, so there's probably just a bit of confusion that needs to be cleared up
14:25:45 <stefw> remember that the default configuration of docker on Atomic doesn't use loopbacks
14:26:08 <petervo> oh didn't know that
14:26:12 <petervo> what does it use?
14:26:17 <stefw> it partitions your system
14:26:20 <stefw> so there's a big lvm pool
14:26:26 <stefw> and parts of it are allocated to container storage
14:26:43 <stefw> anyway, it's a good point to try and make sure we work well in that case too
14:26:58 <petervo> yeah we should be fine there
14:27:08 <petervo> as long as the volumes aren't thin
14:27:29 <stefw> hmmm, good question
14:27:29 <petervo> but i can do some testing
14:27:29 <stefw> yeah, that would probably help
14:27:47 <stefw> and also once vbatts has a bit of spare time, we can poke him again, and he said he'll invest time in helping get the patch in
14:28:08 <petervo> ok cool
14:28:44 <stefw> next topic?
14:28:48 <andreasn_> yep
14:28:51 <andreasn_> #topic Operating System Updates
14:29:01 <petervo> that was mine
14:29:21 <andreasn_> do we have a trello card for that?
14:29:26 <petervo> i did some research on packagekit and os updates
14:29:30 <stefw> yeah
14:29:35 <petervo> https://trello.com/c/lnGL2htr/22-os-package-updates
14:29:48 <stefw> and https://trello.com/c/5j0gCd3d/26-atomic-ostree-updates
14:29:54 <sgallagh> OS updates in what sense? Fedora 21->22 or more with atomic ostree stuff?
14:29:58 <petervo> both
14:30:04 <stefw> i think security updates
14:30:09 <stefw> i don't thinkwe should tackle upgrades at this point
14:30:21 <stefw> i would submit that OS upgrades are out of scope
14:30:26 <petervo> ok
14:30:37 <sgallagh> stefw: Out of scope forever, or out of scope for the immediate future?
14:30:45 <stefw> for this initial work
14:30:48 <andreasn_> stefw, mvollmer and myself also spoke a bit to Colin Walters about OSTree updates
14:30:49 <sgallagh> (Wrapping fedup in Cockpit would actually be really easy)
14:31:01 <stefw> andreasn_, yes
14:31:11 <stefw> and we figured out that ostree nedes some serious work before it's callable from a UI
14:31:21 <stefw> not the least of which is the fact that it is not concurrency safe
14:31:24 <petervo> yeah
14:31:27 <stefw> you can totally screw up your system by running more than one ostree
14:31:40 <stefw> so colin was talking about having libostree and friends wrapped in a dbus daemon
14:31:43 <stefw> which would be perfect
14:32:02 <andreasn_> and it currently lacks any distinction of "fun" vs. "security" updates
14:32:02 <stefw> i'll add that to the card
14:32:21 <andreasn_> fun as in bugfixes
14:32:25 <sgallagh> andreasn_: From an OSTree perspective, it shouldn't matter
14:32:26 <petervo> is that something on his list to work on, or something we'll need to contribute
14:32:27 <stefw> andreasn_, good point
14:32:40 <stefw> petervo, he has it on his list
14:32:43 <sgallagh> Because the idea of the install trees is that they're tested as a complete set
14:32:53 <sgallagh> Not as individual updates
14:33:00 <stefw> petervo, and would appreciate help with, but i'm trying to see if he has someone in mind
14:33:19 <petervo> ok, that was part of my point
14:33:29 <petervo> the other is that the dbus api for this
14:33:30 <petervo> http://www.freedesktop.org/wiki/Software/systemd/SystemUpdates/
14:33:43 <petervo> is only in the very newest version of fedora
14:33:44 <andreasn_> sgallagh: I'm thinking more from the sysadmins point of view "do I need to reboot to upgrade now, or can I wait until tonight"
14:34:14 <petervo> also packagekit isn't on fedora cloud images
14:34:26 <stefw> yeah, they use ostree right?
14:34:32 <petervo> no
14:34:54 <stefw> oh, hmmm
14:34:57 <petervo> but maybe i'm wrong
14:35:23 <petervo> i've been using yum and rpm
14:35:38 <stefw> petervo, can you put your findings in that card? becuase you've done a bunch of research or is it somewhere else?
14:35:49 <petervo> i can update the card
14:36:06 <sgallagh> I don't think PackageKit is enabled by default on Fedora Cloud, no.
14:36:27 <petervo> i guess my bigger question is are we ok only targeting the latest?
14:36:34 <sgallagh> But it would be worth discussing with them whether Cockpit support was sufficient justification for it
14:36:42 <andreasn_> I have some designs for updates on paper, but nothing proper yet, and it will depend a lot of what we get in the different situations
14:37:55 <andreasn_> anything more for OS updates?
14:38:05 <stefw> well petervo brought up a question above
14:38:19 <stefw> i think we should only target reliable update mechanisms
14:38:23 <andreasn_> sorry, missed that
14:38:32 <stefw> ones that we can guarantee work, such as offline updates, or ostree
14:38:53 <petervo> older systems still have offline updates at least a few versions back
14:39:01 <stefw> if someone else (not us here in cockpit) figures out a way to reliably make a live yum update "reliable" in the sense that it knows exactly what to restart, and does that, we could suppor tit.
14:39:02 <petervo> we just can't trigger it with dbus
14:39:08 <stefw> petervo, interesting
14:39:18 <stefw> i would suggest starting with the latest then
14:39:31 <stefw> i don't think we care about older Fedoras at this point
14:39:44 <stefw> and obviously we care about other distros that have older update systems
14:39:46 <sgallagh> It depends if we care about RHEL 7, though
14:39:47 <stefw> but we have to start somewhere
14:40:08 <stefw> yes, but realistically lets start wit hthe latest, and then figure out what we need to do to backport?
14:40:10 <stefw> does that work?
14:40:14 <sgallagh> I should note: PK in EL7 does support offline updates
14:40:41 <sgallagh> It may not have the D-BUS interface for it, but it does have a command-line tool that can do it, so worst case would be to create a minimalistic D-BUS service on those platforms.
14:40:48 <stefw> right
14:42:37 <petervo> ok i think we are good on this for now, we'll develop against the latest, and wait on os tree updates
14:43:05 <stefw> makes sense
14:43:28 <andreasn_> devconf next?
14:43:43 <stefw> yup
14:43:44 <andreasn_> #topic Devconf recap
14:43:57 <sgallagh> /me takes his cap off, then puts it back on again
14:43:58 <andreasn_> so stefw, mvollmer, sgallagh and myself attended devconf
14:44:12 <andreasn_> stefw gave a talk on Saturday
14:44:16 <andreasn_> well attended and appreciated
14:44:31 <andreasn_> I failed at finding the right youtube link
14:45:36 <stefw> youtube link https://www.youtube.com/watch?v=97l1qf2sZtk
14:46:05 <sgallagh> #link https://www.youtube.com/watch?v=97l1qf2sZtk
14:46:05 <andreasn_> #info stefw's talk from Devconf https://www.youtube.com/watch?v=97l1qf2sZtk
14:46:20 <andreasn_> #undo
14:46:21 <zodbot> Removing item from minutes: INFO by andreasn_ at 14:46:05 : stefw's talk from Devconf https://www.youtube.com/watch?v=97l1qf2sZtk
14:46:42 <stefw> The hackfest was nice too, with several people adding stuff to cockpit
14:46:46 <stefw> some people learning about the protocol
14:46:51 <stefw> we got a bug fix too
14:47:00 <andreasn_> \o/
14:47:22 <sgallagh> Huzzah!
14:47:23 <stefw> dan winship was trying to refactor our websocket code into libsoup
14:47:44 <sgallagh> Did he succeed, or "No libsoup for you!"?
14:47:46 <stefw> maybe we can use that in the future for http and websocket stuff ... ?
14:47:49 <stefw> :)
14:47:56 <stefw> but there's several blockers, won't go into detail here
14:48:01 <stefw> so yes, "no libsoup for you" for now
14:48:07 <andreasn_> hehe
14:48:15 <stefw> andreasn_, helped other projects migrate to patternfly
14:48:16 <stefw> ahem
14:48:53 <andreasn_> yeah, helped Alberto with polishing up his Fleet Commander a bit
14:49:04 <mvollmer> sgallagh got started with a rolekit tool inside cockpit
14:49:27 <andreasn_> how many people came to the hackfest? 15-20?
14:49:40 <stefw> yeah started around 25  i would say
14:50:05 <sgallagh> We lost a few when I failed to get Cockpit to build on F20 (sadface)
14:50:46 <andreasn_> but overall successful
14:51:16 <sgallagh> I'm afraid I need to disappear and begin my odyssey home to the USA
14:51:23 <sgallagh> Be well, folks!
14:51:32 <stefw> o/
14:52:15 <andreasn_> good luck!
14:52:36 <andreasn_> so open floor?
14:52:42 <andreasn_> #topic Open Floor
14:54:26 <andreasn_> if nothing we can call it a day
14:54:35 <andreasn_> or, I mean, end of the meeting
14:55:46 <andreasn_> all right
14:55:52 <andreasn_> thanks everyone
14:55:56 <andreasn_> #endmeeting