13:03:13 <andreasn> #startmeeting
13:03:13 <zodbot> Meeting started Mon Oct 12 13:03:13 2015 UTC.  The chair is andreasn. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:03:13 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
13:03:18 <andreasn> .hello andreasn
13:03:19 <zodbot> andreasn: andreasn 'Andreas Nilsson' <anilsson@redhat.com>
13:03:25 <dperpeet> .hello dperpeet
13:03:26 <zodbot> dperpeet: dperpeet 'Dominik Perpeet' <dperpeet@redhat.com>
13:04:07 <dperpeet> * tests: shutdown-restart
13:04:38 <andreasn> * SOS-reports
13:05:35 <mvollmer> upps
13:05:38 <mvollmer> .hello mvo
13:05:39 <zodbot> mvollmer: mvo 'Marius Vollmer' <marius.vollmer@gmail.com>
13:05:56 <mvollmer> * plot polish
13:07:43 <stefw> .hello stefw
13:07:44 <zodbot> stefw: stefw 'Stef Walter' <stefw@redhat.com>
13:08:01 <stefw> * Git commit messages
13:08:22 <stefw> * Redesigning the add machine UI
13:08:42 <github> [cockpit] mvollmer opened pull request #2947: Prettier graphs (master...prettier-graphs) http://git.io/vC8Hy
13:09:07 <andreasn> ok, lets start
13:09:16 <andreasn> #topic tests: shutdown-restart
13:09:33 <dperpeet> ok, we have an issue that system reboots aren't properly registering within our tests
13:09:35 <dperpeet> #link https://github.com/cockpit-project/cockpit/issues/2915
13:09:43 <stefw> do we know that the system is rebooting?
13:09:49 <dperpeet> getting to that
13:10:03 <dperpeet> this cropped up shortly after switching to libvirt, so we first thought the issue must lie there
13:10:28 <dperpeet> but since even alternate detection isn't working, stefw added some more logging
13:10:36 <dperpeet> #link https://github.com/cockpit-project/cockpit/commit/86f1120a0522b87bbc627aea8addf3cf651cf4e4
13:10:55 <dperpeet> since we don't know if it's a failure in the test or if the system really isn't restarting
13:11:18 <dperpeet> it happened again with the extended logs, so we can look into those and get a bit more clarity
13:11:31 <dperpeet> and hopefully resolve the issue soon
13:12:13 <dperpeet> end of topic
13:12:26 <petervo> what i did in the ostree tests is wait for the disconnected dialog
13:12:33 <dperpeet> hm
13:12:39 <petervo> and then call the wait reboot
13:12:43 <dperpeet> did that work?
13:13:08 <petervo> yes, but obviously those tests aren't in master so haven't been run as much
13:13:17 <dperpeet> is the system just taking longer to reboot than it did?
13:13:26 <dperpeet> I can't reproduce the issue on my system
13:13:28 <stefw> if you look at the logs
13:13:31 <stefw> at the journal
13:13:37 <stefw> i don't think the system is rebooting at all
13:13:43 <stefw> there is no journal logs past the reboot
13:13:48 <stefw> there are no messages about shutdown ofthe system either
13:13:55 <stefw> and the PNG indicates that the shutdown command is hanging
13:14:15 <stefw> i am of the opinion that there is a bug that prevents the 'shutdown' command from working at a particular point during boot
13:14:26 <stefw> we invoke shutdown with try: superuser
13:14:32 <stefw> so you can see in the logs where the privilege escalation takes place
13:14:37 <stefw> and that takes place before the system has booted
13:14:42 <stefw> even before systemd reports the stats about the boot
13:15:17 <dperpeet> hm, now that you say it, we don't do anything in that cockpit session before testing the restart
13:15:36 <stefw> right, so we're really really fast
13:16:01 <dperpeet> this might be something we should catch in cockpit though
13:16:10 <dperpeet> if the shutdown command isn't being accepted properly
13:16:22 <stefw> it seems that the command is hanging
13:16:34 <stefw> because the promise returned from cockpit.spawn() never completes
13:16:37 <dperpeet> the question is, whether this could happen in real life
13:16:40 <stefw> obviously all of this assumes there are no other bugs
13:17:26 <andreasn> next topic?
13:17:35 <dperpeet> hm
13:17:39 <dperpeet> one second andreasn
13:17:48 <andreasn> sure, sorry :)
13:17:52 <dperpeet> we should decide on the action here
13:17:54 <stefw> well this is pretty much the most important thing that's breaking every single pull requests tests
13:18:13 <stefw> a work around to test thish theory would be adding a 30 second wait
13:18:17 <stefw> before logging into cockpit
13:18:18 <dperpeet> yeah, I was thinking that
13:18:27 <stefw> and i don't mind digging into it after thath point
13:18:28 <dperpeet> just to get it working again
13:18:30 <stefw> and trying to make a reproducer
13:18:39 <stefw> with a oneshot service that calls shutdown at the same point during boot that we did
13:18:41 <dperpeet> I'll add the wait, stefw
13:18:47 <dperpeet> and make a pr
13:18:50 <stefw> ok, once we've confirmed the theory
13:18:52 <stefw> then i can do the upstreaming
13:18:56 <stefw> not that it'll get fixed
13:19:02 <dperpeet> :)
13:19:03 <stefw> I don't think our SIGKILL bug ever got fixed
13:19:09 <stefw> but at least we've reported it properly
13:19:14 <dperpeet> right
13:19:16 <mvollmer> the distributions are patching it one by one, no?
13:19:20 <stefw> yup
13:19:39 <dperpeet> ok andreasn, next topic!
13:19:44 <andreasn> sure
13:19:48 <andreasn> #topic SOS-reports
13:20:03 <andreasn> there is some cases here https://github.com/cockpit-project/cockpit/wiki/Feature:SOS-reporting
13:20:07 <andreasn> #info https://github.com/cockpit-project/cockpit/wiki/Feature:SOS-reporting
13:20:29 <andreasn> and I attached some initial designs https://github.com/cockpit-project/cockpit/wiki/Feature:SOS-reporting
13:20:34 <andreasn> https://raw.githubusercontent.com/cockpit-project/cockpit-design/master/sos-reports/create-sos-report.png
13:20:45 <stefw> I believe this is server specific, and shouldn't be associated with the global user session
13:21:01 <stefw> doesn't it go under Tools?
13:21:52 <andreasn> yeah, could be
13:22:00 <dperpeet> yes, I think at this stage it'd only be useful as a server specific tool
13:23:05 <andreasn> all right, makes sense
13:23:16 <andreasn> I think that was it on that
13:23:57 <andreasn> if anyone can think of more feedback, please add it to the trello card or the wiki page
13:24:02 <andreasn> all right, next topic
13:24:14 <andreasn> #topic plot polish
13:24:22 <mvollmer> i have started on that
13:24:33 <mvollmer> based on andreas' mockup
13:24:35 <andreasn> what's the pull req now again?
13:24:47 <mvollmer> just opened it: https://github.com/cockpit-project/cockpit/pull/2947
13:25:07 <mvollmer> the server page with the four plots looks much better now
13:25:23 <mvollmer> the stacked plots look funny
13:25:31 <andreasn> I'll check it out
13:25:35 <mvollmer> like memory for multiple containers
13:26:07 <mvollmer> we might want to use stronger colors than shades of blue
13:26:15 <andreasn> I based the mockups on patternfly widgets https://www.patternfly.org/wp-content/uploads/2015/07/dashboard_callout.png
13:26:20 <stefw> thish is temporary right, until we can sync up with patternfly on this no?
13:26:30 <mvollmer> and put those colors into the row of the 'object', like on the dashboard
13:26:31 <andreasn> yeah, this should be closer to patternfly
13:27:23 <andreasn> and will need a bit more work, as followup to match better with those
13:28:15 <mvollmer> oky, I'll look into the stacked plots some more
13:28:21 <andreasn> I'll take a look too
13:28:36 <andreasn> next topic?
13:28:37 <mvollmer> maybe we just need to replace zeros with nulls so that only things are drawn that are interesting.
13:28:41 <mvollmer> yep
13:28:59 <andreasn> lets talk more on the pull or outside the meeting about the details
13:29:01 <andreasn> all right
13:29:06 <andreasn> #topic Git commit messages
13:29:36 <stefw> I'd just like to highlight how to right a good git commit message
13:29:38 <andreasn> protip: probably not how I do them ;)
13:29:46 <stefw> there's a nice summary there
13:29:53 <stefw> "The seven rules of a great git commit message"
13:30:02 <andreasn> do you have a url?
13:30:07 <stefw> http://chris.beams.io/posts/git-commit/
13:30:15 <stefw> http://chris.beams.io/posts/git-commit/#seven-rules
13:30:17 <andreasn> #info http://chris.beams.io/posts/git-commit/
13:30:27 <stefw> when they're not in that format, then someone else is going to have to rewrite them
13:30:45 <stefw> In addition 'Fixes #NNN' or 'Closes #NNN' go on separate lines after the main body of the message
13:31:05 <stefw> obviously people who do review still fix up commit messages
13:31:08 <stefw> and add Closes etc.
13:31:19 <stefw> but i just figured it would be worth mentioning, since this doesn't happen by itself :)
13:31:25 <andreasn> I usually use -m "haaaands", should I rather skip that flag and edit in the editor?
13:31:39 <stefw> naughty naughty
13:31:46 <stefw> yeah
13:31:50 <stefw> otherwise someone else has to do that part
13:32:04 <andreasn> yup, ackowledged
13:32:22 <andreasn> and thanks for the link. I'll check it out!
13:32:30 <stefw> np
13:33:01 <andreasn> next topic?
13:33:06 <andreasn> #topic Redesigning the add machine UI
13:33:34 <stefw> I'm working on some new dialogs for the add machine UI
13:33:38 <stefw> which both allow adding a machine
13:33:50 <stefw> and then changing stuff, like address, or authentication info
13:33:53 <stefw> after it's been added
13:34:11 <stefw> and Peter is working on helping with some of the implementation
13:34:24 <andreasn> is there a issue or a pull request url?
13:34:26 <stefw> https://trello.com/c/0B6y8SXP/216-better-workflow-for-adding-machines
13:34:31 <stefw> not yet, just the card
13:34:38 <andreasn> #info https://trello.com/c/0B6y8SXP/216-better-workflow-for-adding-machines
13:35:27 <stefw> that's it
13:35:29 <stefw> just a heads up
13:36:11 <andreasn> looks good
13:36:19 <dperpeet> yeah, that'll help a lot of people
13:37:29 <andreasn> all right, anything else on that?
13:37:40 <stefw> onpe
13:37:50 <andreasn> we're moving along quickly today it seems
13:37:56 <andreasn> #topic open floor
13:38:42 <mvollmer> the server page still looks unhappy to me
13:38:46 <mvollmer> unbalanced
13:39:00 <mvollmer> i should describe this better
13:39:03 <andreasn> even with the new graphs?
13:39:06 <mvollmer> yes
13:39:22 <stefw> i think it really could be redesigned to use the new Patternfly dashboard concept
13:39:36 <stefw> it is a server specific dashboard, after all
13:39:42 <mvollmer> e.g., what about changing the 'info column' with all the text and buttons to be a row on top of the graphs?
13:39:42 <stefw> https://www.patternfly.org/patterns/dashboard-layout/
13:40:23 <andreasn> is there any new information we'll need to fit there soon?
13:40:36 <andreasn> so we can plan ahead a bit with a relayout?
13:40:50 <stefw> subscription status was supposed to go there
13:40:52 <mvollmer> the info column is very cramped, and the whitespace below it is wasted
13:41:23 <stefw> an ostree summary should eventually go there
13:41:25 <dperpeet> if we have a new design with a place for subscription status, that'd be a good time to add it :)
13:41:25 <andreasn> yeah, that should be an easy fix to just spread it out a bit more. And I agree about the misbalance
13:41:28 <mvollmer> i have no good constructive ideas, unfortunately
13:41:48 <andreasn> it's good you bring it up though
13:41:57 <mvollmer> it's the default page that people see, so I think it should be our best one.
13:42:10 <dperpeet> do we have enough content to make it a bit dynamic?
13:42:19 <andreasn> it also ties in slightly to the alerts I think
13:42:34 <dperpeet> like a basic grid where you can drag stuff around
13:42:40 <dperpeet> but that might be out of scope
13:42:46 <dperpeet> customizable
13:43:05 <mvollmer> oho
13:43:08 <andreasn> might be good, but if it's unbalanced and bad by default, customizing it only goes so far
13:43:11 <mvollmer> overlapping windows...
13:43:18 <andreasn> tsssss
13:43:23 <dperpeet> nothing overlapping mvollmer
13:43:37 <mvollmer> :)
13:43:44 <dperpeet> kind of like a grid of tiles where consumers take mxn tiles
13:44:15 <stefw> lest not design this here
13:44:21 <stefw> but i think mvollmer's point is a good one
13:44:33 <stefw> i agree we should relayout this page
13:44:51 <andreasn> I'll create a trello card and add it to the backlog
13:45:01 <andreasn> and link to the patternfly dashboard mockups
13:46:07 <andreasn> maybe even create a wiki page
13:46:11 <mvollmer> nice, thanks!
13:46:17 <andreasn> and then we can add requirements and ideas there
13:47:11 <andreasn> #action andreasn to create trello card and wiki page for dashboard redesign
13:47:43 <andreasn> anything else before I end the meeting?
13:48:33 <github> [cockpit] petervo opened pull request #2948: test: Testing wait_reboot issues (master...test-reboot) http://git.io/vC8h2
13:48:48 <andreasn> just a quick note that I started a wiki page about container scanning, haven't gotten it running yet though
13:48:51 <andreasn> https://github.com/cockpit-project/cockpit/wiki/Feature:-container-security-scanning
13:50:05 <andreasn> all right, thanks for the meeting everyone!
13:50:09 <andreasn> #endmeeting