14:01:46 <mvollmer> #startmeeting meeting
14:01:46 <zodbot> Meeting started Mon Feb 15 14:01:46 2016 UTC.  The chair is mvollmer. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:01:46 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
14:01:46 <zodbot> The meeting name has been set to 'meeting'
14:01:50 <mvollmer> .hello mvo
14:01:51 <dperpeet> .hello dperpeet
14:01:52 <zodbot> mvollmer: mvo 'Marius Vollmer' <marius.vollmer@gmail.com>
14:01:55 <zodbot> dperpeet: dperpeet 'None' <dperpeet@redhat.com>
14:01:55 <andreasn> .hello andreasn
14:01:58 <zodbot> andreasn: andreasn 'Andreas Nilsson' <anilsson@redhat.com>
14:02:49 <mvollmer> #topic Agenda
14:02:53 <mvollmer> * weekly images
14:03:18 <mvollmer> * pcp/metric tests
14:03:27 <mvollmer> * storage polish
14:03:29 <andreasn> * package updates
14:03:59 <andreasn> * issue handling
14:04:02 <dperpeet> * compressed test images
14:04:02 <github> [cockpit] mvollmer opened pull request #3751: test: Use timedatectl. (master...test-timedatectl) https://git.io/vg9px
14:05:23 <mvollmer> alright!
14:05:31 <mvollmer> I start...
14:05:37 <mvollmer> #topic weekly images
14:05:52 <mvollmer> so we should see pull requests with new images now
14:05:56 <mvollmer> in fact, we saw some
14:06:08 <abhishekg5> Is the meeting started ?
14:06:14 <mvollmer> abhishekg5, yes, welcome!
14:06:33 <mvollmer> abhishekg5, do you want to put something on the agenda?
14:07:24 <mvollmer> github-scan now shows when images are due
14:07:37 <mvollmer> so if you are curious, run "./github-scan -d"
14:07:45 <dperpeet> default is every 7 days
14:07:48 <mvollmer> yes
14:08:10 <mvollmer> i have some code for private images as well, rhel-7 and rhel-atomic
14:08:17 <mvollmer> but we need to sort out the details
14:08:32 <mvollmer> right now, we have this:
14:08:38 <mvollmer> fedora-23 will be refreshed in 1.2 days.
14:08:38 <mvollmer> fedora-22 will be refreshed in 4.9 days.
14:08:38 <mvollmer> fedora-atomic will be refreshed in 1.2 days.
14:08:38 <mvollmer> fedora-testing will be refreshed in 1.2 days.
14:08:38 <mvollmer> debian-unstable will be refreshed in 3.0 days.
14:09:23 <mvollmer> the machinery seems to work, but I have seen pull requests without any verify/* statuses
14:09:31 <mvollmer> let's watch that
14:09:44 <dperpeet> nice work!
14:09:48 <mvollmer> maybe the sink now runs into github api rate limiting?
14:10:01 <dperpeet> does the sink have credentials?
14:10:08 <mvollmer> it gets a token
14:10:11 <abhishekg5> mvollmer> I am still solving certain easy fixes. Facing some little issues in them
14:10:52 <mvollmer> abhishekg5, okay, let's talk about that now. :-=
14:10:54 <mvollmer> :-)
14:10:54 <dperpeet> maybe we can add some debug output regarding the rate limits to the end of test logs
14:10:55 <mvollmer> sorry
14:11:10 <mvollmer> dperpeet, yes
14:11:25 <mvollmer> the sink might swallow some errors
14:11:34 <mvollmer> otherwise they should appear
14:11:49 <mvollmer> next?
14:12:04 <dperpeet> yes
14:12:18 <mvollmer> abhishekg5, do you want to report a bit?
14:12:53 <mvollmer> or after the meeting?
14:13:19 <abhishekg5> mvollmer> I have already posted the query here https://github.com/cockpit-project/cockpit/issues/3489
14:13:26 <mvollmer> okay
14:14:04 <andreasn> oh, I saw this one and forgot about it again with the travelling.
14:14:55 <andreasn> so you have the code checked out and can do edits to it, but it won't show up?
14:15:02 <dperpeet> abhishekg5, when you change code, you need to make install and refresh the cockpit session (log out, browser refresh, login)
14:15:06 <abhishekg5> andreasn> yes
14:15:15 <mvollmer> andreasn, do you think you can help abhishekg5 get started?
14:15:30 <mvollmer> maybe not right now during the meeting. :-)
14:15:31 <andreasn> yeah, lets figure it out
14:15:36 <mvollmer> okay, cool.
14:15:45 <andreasn> abhishekg5: lets try and set it up after the meeting
14:15:45 <mvollmer> sorry to be so german about this
14:16:02 <mvollmer> #topic  pcp/metric tests
14:16:06 <abhishekg5> mvollmer, andreasn> Cool
14:16:13 <mvollmer> so that's what I want to do nextish.
14:16:27 <mvollmer> we have some issues open in this area, and no real tests.
14:16:51 <mvollmer> i'll try to make some tests to catch completely break down
14:17:13 <mvollmer> but I also try actually measure things to see whether the numbers make sense
14:17:14 <dperpeet> will that include the numbers or graphs, too?
14:17:32 <mvollmer> not sure how to approach this
14:17:52 <andreasn> what kind of bugs for the graphs do we get the most?
14:17:59 <mvollmer> i might use a plain metric channel for the calibration tests
14:18:16 <mvollmer> andreasn, our different graphs are not consistent
14:18:20 <dperpeet> we had a few issues with scale, labels et cetera
14:18:27 <mvollmer> numbers are twice what people expect them to be
14:18:33 <dperpeet> I wanted to say that graph tests should wait until the graphs are consistent
14:18:49 <dperpeet> note changed patternfly graphs
14:19:02 <mvollmer> yeah, I wasn't thinking of testing the actual drawing, just the number gathering
14:19:10 <dperpeet> great, then I agree
14:19:25 <andreasn> I haven't heard back from Patternfly folks with regards to the graphs, I'll check what's up there
14:19:36 <mvollmer> so, maybe via playground/metrics or something better
14:20:09 <mvollmer> (I might have to learn what all the memory numbers mean... again.)
14:21:34 <mvollmer> okay, next?
14:22:08 <mvollmer> #topic storage polish
14:22:19 <mvollmer> that's what I want to do after the metric checks
14:23:02 <mvollmer> things like catching more errors and giving a good response instead of shoveling ten lines of D-Bus code in the face of the user
14:23:22 <mvollmer> sliders for sizing partitions
14:23:28 <andreasn> and maybe some relayout for the storage details
14:23:36 <andreasn> storage details pages
14:23:39 <mvollmer> maybe some better layout together with andreasn
14:23:45 <mvollmer> exactly!
14:24:03 <mvollmer> maybe I get to this towards the end of the week
14:24:19 <dperpeet> I think we should look at use cases and the layout first
14:24:27 <mvollmer> then I can sleep better. :-)
14:24:41 <mvollmer> it really bothers me that storage is so unfinished
14:25:47 <mvollmer> everyone still here?
14:25:58 <petervo> yep
14:26:09 <mvollmer> yeah
14:26:31 <dperpeet> yes
14:26:38 <mvollmer> okay
14:26:42 <mvollmer> #topic package updates
14:26:55 <andreasn> so for the user cases, I've been trying to get some, but no luck so far :) I'm happy to poke again though
14:27:08 <abhishekg5> yes
14:27:45 <andreasn> and fixing up some details to make mvollmer sleep better at night we can do regardless
14:27:55 <mvollmer> :-)
14:28:19 <mvollmer> andreasn, do you want to start right away on the storage polish?
14:29:25 <andreasn> I want to finish the writeups of issue/notifications/fires handling and OS software stuff first
14:29:37 <andreasn> but after that I'm happy to get to storage
14:29:46 <mvollmer> ok, nice.
14:30:26 <mvollmer> doen?
14:30:38 <mvollmer> *done?
14:30:39 <andreasn> sure
14:30:51 <mvollmer> #topic issue handling
14:31:28 <andreasn> just a note that I'm starting to write this up https://github.com/cockpit-project/cockpit/wiki/Feature:-Issues
14:31:34 <andreasn> but it's pretty bare right now
14:31:42 <andreasn> we discussed this a ton in Brno last week
14:32:28 <andreasn> is there a trello card for it? I can't find it
14:33:26 <dperpeet> I don't think there is
14:33:29 <andreasn> I'll create one then
14:33:35 <dperpeet> but shouldn't it be troubleshooting?
14:33:46 <dperpeet> or do we want to focus on specific issues?
14:34:09 <andreasn> no, we can call it Troubleshooting
14:34:37 <andreasn> https://github.com/cockpit-project/cockpit/wiki/Feature:-Troubleshooting
14:35:27 <andreasn> that was it on that
14:35:37 <mvollmer> thanks
14:35:51 <mvollmer> #topic compressed test images
14:36:22 <dperpeet> so, we decided that our testing images were getting too large
14:36:28 <dperpeet> on the scale of almost 4 GiB
14:36:51 <dperpeet> using the internal qcow2 compression, we've been able to reduce that to roughly 1.5
14:37:03 <dperpeet> or at least <2
14:37:21 <dperpeet> testing didn't show significant changes regarding testing runtime
14:37:42 <andreasn> nice
14:37:43 <dperpeet> empirical tests showed a fluctuation of +/- 10%
14:38:18 <mvollmer> now we can also get rid of the xz compression, right?
14:38:23 <dperpeet> we could
14:38:23 <mvollmer> during up/download
14:38:46 <dperpeet> but we don't really lose much by doing xz compression
14:38:52 <dperpeet> it costs a bit more during creation
14:39:16 <dperpeet> but since the download ist streamed anyway and the compression is asymmetric (faster decompression), even a slight compression gain can save time
14:39:23 <github> [cockpit] stefwalter opened pull request #3752: ws: Remove cockpit-testing.service and cockpit-testing.socket (master...cockpit-testing) https://git.io/vgHU6
14:39:28 <mvollmer> okay
14:39:40 <dperpeet> .xz seems to gain 3-5% compression
14:39:54 <dperpeet> so ~80MiB for some of the images I tested
14:40:23 <dperpeet> at this point, I think it isn't worth touching our xz related code and risk that we break something
14:40:59 <mvollmer> yeah, and it allows us to switch the internal compression off again if something goes wrong
14:41:38 <dperpeet> exactly
14:41:49 <dperpeet> my disk seems happy
14:42:15 <dperpeet> the overlay when building/installing is non-compressed
14:42:16 <mvollmer> dperpeet, the debian image is still making trouble, right?
14:42:21 <mvollmer> all others have been updated?
14:42:24 <dperpeet> yes, but that's not compression-related
14:42:28 <mvollmer> no
14:42:32 <dperpeet> I believe so, maybe not openshift
14:42:34 <mvollmer> this is the first time we update Debian
14:42:43 <mvollmer> and it is "interesting"
14:43:09 <mvollmer> but I think it is by and large our workarounds that break, no?
14:43:26 <mvollmer> so let's try to make them less tricky and less specific.
14:43:39 <mvollmer> and fewer of course
14:43:51 <dperpeet> sounds like a plan
14:44:13 <dperpeet> we also made some assumptions in our tests
14:44:23 <dperpeet> like the ssh availability
14:44:29 <dperpeet> that we have to carefully evaluate
14:46:25 <mvollmer> yep
14:46:39 <mvollmer> done?
14:47:14 <dperpeet> yes
14:49:18 <mvollmer> #topic open floor
14:49:44 <andreasn> I've done some work on the software updates https://github.com/cockpit-project/cockpit/wiki/Feature:-OS-Package-Installation-and-Updates
14:49:57 <andreasn> please look through and see if I missed anything
14:52:17 <andreasn> that's it from me
14:52:33 <mvollmer> looks good
14:52:36 <mvollmer> nice read
14:52:43 <mvollmer> I should read more of them :-)
14:52:53 <mvollmer> alright
14:53:01 <mvollmer> thanks everyone!
14:53:05 <mvollmer> #endmeeting