IRC logs for #baserock for Monday, 2014-10-13

radiofreejjardon: libtasn1 branch baserock/gnome00:27
radiofreemorph is
radiofreeno idea if the morph file is necessary anymore, but it built for me00:27
jjardonradiofree: new jjardon/gnome_and_systemd pushed06:12
* pedroalvarez waves from dusseldorf06:47
* pedroalvarez also had a distbuild error:
pedroalvarezurgh... twice06:54
pedroalvarezi guess one osf the workers is misbehaving 07:03
radiofreeif you have a repo with submodules, do you have to manually commit the changes to the submodule repos (i.e to use gbo) or will morph do that for you?07:08
paulsherwoodpedroalvarez: ^^07:09
pedroalvarezradiofree: afaik you have to change the url of the submodule to use the upstream: alias07:10
radiofreethat's a bit annoying :\07:11
paulsherwoodit's a minor annoyance :)07:12
*** dutch [] has joined #baserock07:12
pedroalvarezlast week was mentioned an idea of a new feature, to be able to specify in a chunk morphology that you have to fetch a repository inside the source of the chunk before building07:13
pedroalvarezbut indeed is a bit annoyin07:13
* pedroalvarez restarts the workers, the controller, and prays07:27
jjardonNice, that would be much better than having to commit in the repo07:29
straycatradiofree, I don't understand, what's causing it to take 6 minutes?07:46
*** franred [] has joined #baserock07:47
straycatI'm sure we could look at trying to optimise the parts of the process paulsherwood mentioned on the mailing list.07:50
* straycat notes that the speedups jjardon noticed are due to richard_maw's fix that removes queueing rather than anything he's done07:56
*** mSher [] has joined #baserock08:05
*** franred [] has quit [Remote host closed the connection]08:19
*** jonathanmaw [] has joined #baserock08:19
*** tiagogomes [] has joined #baserock08:24
*** franred [] has joined #baserock08:30
*** ssam2 [] has joined #baserock09:04
pedroalvarezradiofree: weston-simple-egl crashes09:30
paulsherwoodjoin #datacentred10:26
*** mlilenium_ [~mlilenium@] has joined #baserock10:51
*** mlilenium_ [~mlilenium@] has quit []10:51
*** thecorconian [~thecorcon@] has joined #baserock11:20
*** thecorconian [~thecorcon@] has quit []11:20
pedroalvarezcontroller-helper says: and controller says:
pedroalvarezsomething is going wrong :(11:39
ssam2connection refused11:41
ssam2I've seen this before, but I forget exactly the fix11:41
ssam2it maybe that the service is listening on the external, but something else is trying to talk to
ssam2*external IP11:42
ssam2could you paste  /etc/morph-*.conf somewhere ?11:43
ssam2or I could give you my public key and poke around the machine myself, if you trust me not to break it further :)11:44
pedroalvarezssam2: I trust you always you don't reboot it11:48
pedroalvarezssam2: these are the conf files:
pedroalvarezsometimes it works, and sometimes it doesn't :/11:56
wikicatWiki change: Link to all Trove documentation in trove page;a=commitdiff;h=45ff9c911:58
ssam2pedroalvarez: oh, that's not good at all11:59
ssam2when I saw 'connection refused', it never worked12:00
pedroalvarezssam2:  maybe is only one worker?12:01
ssam2yeah .. but those errors are from the controller12:02
ssam2I see quite a lot of 'connection refused' errors in the logs, actually12:02
ssam2to all of the boards12:03
ssam2it maybe that there's still problems with the underlying networking12:03
ssam2and the distbuild code isn't being fully robust against that12:03
pedroalvarezso, internal connection problems12:06
pedroalvarezi think I'm going to use internal IPs to configure the workers instead of using public Ips12:07
pedroalvarezmaybe that improves the connection12:07
ssam2ok, yeah12:07
ssam2if it's intermittent and it's an error that doesn't normally happen it must be the networking, really12:07
pedroalvarezthey don't have internal ip addresses12:11
* pedroalvarez sighs12:11
*** Kinnison [~dsilvers@gateway/shell/pepperfish/x-rwcvpoastxeguvkx] has quit [Ping timeout: 245 seconds]12:22
*** Kinnison [~dsilvers@gateway/shell/pepperfish/x-foiupxvydubaayvx] has joined #baserock12:23
radiofreepedroalvarez: yep, because of --disable-egl-gallium13:18
radiofreeit's worth nothing that that argument is only mentioned in the wayland docs because they use the intel driver13:20
radiofreeprobably need the state trackers too13:20
radiofreecan someone else please review "Fix gstreamer 0.10 build with newer glibs"13:21
radiofreeit has a +1 from richard13:21
radiofreethis e-mail was sent 23/06/14 btw13:21
richard_mawradiofree: which richard? If it wasn't me, you can have another +113:22
radiofreerichard_maw: it was you :)13:22
richard_mawhave a +2 then13:22
*** vmeson [~quassel@] has quit [Ping timeout: 260 seconds]15:04
*** vmeson [~quassel@] has joined #baserock15:05
pedroalvarezssam2: I've disables the trove -gc unit in the baserock-clone trove. I think it's causing problems (see mason failures) 15:10
ssam2oh dear, yes15:11
* paulsherwood continues to see silent stop after 'computing build graph'15:11
* pedroalvarez continues to see distbuild failing to open a socket to, but only sometimes 15:13
pedroalvarezI'm sad today 15:14
paulsherwoodpedroalvarez: go to the yocto talk, that will cheer you up :)15:21
ssam2paulsherwood: 'deciding task order' is instant for me now15:27
ssam2oh, that wasn't the one you said, though15:27
radiofreego to the drone talk!15:28
ssam2pedroalvarez: I've changed the trove-gc hack to not set 'cachedir-artifact-keep-younger-than=0'15:43
ssam2I wonder if we were seeing a race condition where it deleted stuff after a remote Morph had checked that it existed15:43
ssam2I'll reenable it, if things break again then it'll need more thought15:44
pedroalvarezssam2: ok15:46
pedroalvarezThe yocto talk is being a lot of people complaining about problems... 16:02
rjekIs anybody sat looking smug because they don't have those problems?16:03
pedroalvarez"clients using different build machines could be an issue" 16:04
pedroalvarezrjek: me16:04
rjekSmugness feels nice, unless it's somebody else being smug.16:05
ssam2what !16:10
ssam2something is changing permissions on the files in /home/cache/artifacts on the datacentred Trove16:11
ssam2I guess it must be the trove-gc hack, as that's the only new thing16:11
Kinnisonis the trove-gc hack running as the cache user?16:11
ssam2that might help16:11
ssam2but I don't understand why it'd change permissions on files that it doesn't delete16:11
*** dutch [] has quit [Quit: Quit]16:11
Kinnisonmorph's probably normalising the cache16:12
Kinnisonto what *it* thinks it should be16:12
ssam2possibly, but I don't know of or see any code to do that16:12
ssam2unless its hidden in PyFilesystem16:13
pedroalvarezssam2: I copied the artifacts from one folder to another using the root user 16:21
ssam2oh, maybe it was you that broke it :)16:21
pedroalvarezBut I checked them and I think everything was alright 16:21
pedroalvarezSo leave it running but fix the owners 16:22
ssam2I've fixed the owners16:22
ssam2I added some code to print out the invalid JSON message that's causing the public Jetson distbuild to crash16:30
* ssam2 struggles to copy it due to vim's mouse integration being enabled by default in Baserock16:31
ssam22014-10-13 16:30:47 DEBUG JsonMachine: about to crash due to the following invalid message: GET HTTP/1.116:31
ssam2i don't know enough about HTTP to know what is going on there16:32
ssam2but it seems like a proxy is getting in the way somehow :)16:32
*** jonathanmaw [] has quit [Quit: Leaving]16:33
radiofreessam2: set mouse-=a16:36
ssam2radiofree: i know (thanks to you telling me about it in the past :)16:36
radiofreeit's enabled by default in the /etc/vimrc file that's installed16:36
radiofreeneed to send a patch to get rid of that16:36
ssam2yeah, that'll be the default vimrc that comes with vim16:36
ssam2vim is not exactly known for sensible defaults :)16:37
ssam2pedroalvarez: I wonder if it'd be enough to make the Json state machine ignore invalid Json and continue, instead of crashing16:37
ssam2i've hacked up the jetson controller to do so and am trying a build16:37
radiofree"" In many terminal emulators the mouse works just fine, thus enable it."16:37
radiofreeNO NO NO NO NO NO16:37
ssam2it makes sense if you're in a fullscreen terminal16:39
ssam2much less if you're in a desktop environment16:39
*** tiagogomes [] has quit [Ping timeout: 272 seconds]16:45
ssam2my conclusion on the jetson distbuild is: I have no idea what's going on16:49
ssam2but I think it'd be useful to fix distbuild so that it reports internal errors rather than silently exiting16:50
jmacsThat's roughly my opinion of chef16:50
ssam2jmacs: chef seems to largely be a way to install packages16:50
ssam2in a platform-agnostic way16:51
jmacsThat bit of it seems fine; the bit that installs packages on remote machines and phones back to the server is more tricky16:55
*** mSher [] has quit [Quit: Leaving]17:09
*** ssam2 [] has quit [Quit: Leaving]17:23
*** franred [] has quit [Read error: Connection reset by peer]23:52
*** fay_ [] has quit [Read error: Connection reset by peer]23:52
*** franred [] has joined #baserock23:52
*** fay_ [] has joined #baserock23:52

Generated by 2.15.3 by Marius Gedminas - find it at!