fginther | kgunn, I've kicked off 3 more builds | 00:19 |
---|---|---|
ogra_ | cjwatson, hmm, click is broken on the latest image ... seems download-manager changed the dbus service name without retaining backwards compatibility | 11:00 |
ogra_ | yeah, rolling back to the former download-manager version (0.2+13.10.20130906.1-0ubuntu1) helps | 11:01 |
ogra_ | asac, lool ^^^ (i guess that is the cause for other test failures too) | 11:13 |
ricmm | ogra_: ping | 13:24 |
ogra_ | ricmm, hey | 13:52 |
cjohnston | Anyone around, who cares, I've started the smoke jobs over now that things are coming back up | 15:57 |
=== Ursinha-afk is now known as Ursinha | ||
cjwatson | ogra_: nothing in click itself touches the download manager - that's the click *scope*, ralsina's stuff | 21:35 |
ogra_ | ah | 22:16 |
lool | ogra_: so surprizing that this got uploaded; I hope we didn't push it with one of the stacks | 22:21 |
lool | ogra_: the landing was meant for monday IIRC | 22:21 |
lool | ogra_: so it got released yesterday morning as part of one of the stacks | 22:22 |
lool | ogra_: do you know where the cu2d config lives? | 22:23 |
lool | lp:cupstream2distro-config apparently | 22:25 |
ogra_ | lool, it was in the 0920 image from fri ... so it must have gone into the archive earlier | 22:28 |
ogra_ | http://people.canonical.com/~ogra/touch-image-stats/20130920.changes | 22:28 |
lool | ogra_: so this is line 43, and the landing missed the fact that unity-scope-click is affected | 22:28 |
ogra_ | right | 22:28 |
lool | ogra_: basically the spreadsheet says we wanted to land this, but we missed one rdeps in review and in testing (click scope) | 22:28 |
ogra_ | i dont think we have any tests that try to install a click package | 22:28 |
lool | (that's right; I did ask for them a couple of times in past weeks :-) | 22:29 |
ogra_ | another issue i missed is that we shipped a hack in the session manager that puts an env var into ~/.bashrc ... | 22:30 |
ogra_ | the new lightdm stuff didnt have that ... | 22:30 |
ogra_ | when testing i tested with installing the new session package that was booted with the old session first .. | 22:30 |
lool | ogra_: so we could either revert download-manager, but it's also on the critical for updated system-image using download-manager | 22:30 |
ogra_ | so i did have the hack | 22:30 |
lool | ogra_: or we could ask for updated unity-scope-click | 22:31 |
lool | actually I see it's in bzr | 22:31 |
ogra_ | utah then failed because that hack wasnt in place anymore when tesed in the image | 22:31 |
lool | r46 | 22:31 |
lool | since timestamp: Wed 2013-09-18 08:58:09 +0000 | 22:31 |
ogra_ | i wonder how we can prevent such rest issues | 22:31 |
ogra_ | *test | 22:31 |
ogra_ | my borken upload broke the whole of the unity8 tests | 22:31 |
lool | hold on, just to finish on scope-click | 22:32 |
lool | ogra_: there is some issue in the landing; it didn't pick up unity-scope-click | 22:32 |
ogra_ | yeah, if we have a fix we should definitely ship it | 22:32 |
lool | ogra_: let's raise this monday, but also get someone to release unity-scope-click ASAP | 22:32 |
ogra_ | ++ | 22:33 |
ogra_ | note that i did a 59 build to actually make sure the unity tests are fixed | 22:33 |
ogra_ | sadly the datacenter went down today | 22:33 |
ogra_ | (or parts of it) | 22:33 |
ogra_ | so the tests were a bit screwed up, but unity8 finished | 22:34 |
* ogra_ is curious what that was btw ... | 22:35 | |
ogra_ | lool, oh, btw, music-app is in 59 :) | 22:36 |
lool | ogra_: sadly music experience is utterly broken :-( | 22:36 |
ogra_ | :( | 22:37 |
ogra_ | will get better with the new gstreamer stuff, i'm sure | 22:37 |
lool | ogra_: no, it's due to another bug in upstart-app-launch sadly | 22:38 |
ogra_ | aww | 22:38 |
lool | ogra_: https://bugs.launchpad.net/upstart-app-launch/+bug/1228387 | 22:38 |
ubot5 | Ubuntu bug 1228387 in upstart-app-launch (Ubuntu) "Doesn't unquote double-quotes from desktop files" [High,New] | 22:38 |
ogra_ | oh, still that one ? | 22:38 |
lool | ogra_: that's another one | 22:45 |
lool | ogra_: the first one was with spaces :-) | 22:45 |
ogra_ | ah | 22:45 |
ogra_ | heh | 22:45 |
ogra_ | oh my | 22:45 |
lool | I've updated the spreadsheet with the failure and added some colors and image numbers for ongoing landings | 22:45 |
lool | I'm sure asac will freak out with the new colors as it visually changes a lot; asac when you read this: I just changed the colors :-) | 22:46 |
lool | ogra_: also we are landing 2 other changes in upstart-app-launch for the music-experience; it's quite a nightmare | 22:46 |
ogra_ | yesh | 22:46 |
ogra_ | did nobody use it before now ? | 22:47 |
lool | there's a chance the plan changes on this on monday, so let's see | 22:47 |
ogra_ | i mean, someone should have noticed that before | 22:47 |
lool | people used it, but not with file names or double-quotes it seems :-) also it couldn't guarantee single instance | 22:47 |
plars | ogra_: lool: haven't read the backscroll yet, but I saw there was a new image that tried to get tested earlier and something new broke | 23:05 |
ogra_ | plars, well, the datacenter was down for quite a while today | 23:05 |
plars | I'm seeing this in the logs: | 23:05 |
plars | initctl: unable to determine sessions | 23:05 |
plars | ogra_: I'm getting it at home also | 23:05 |
ogra_ | ah | 23:06 |
plars | ogra_: didn't see that error in previous runs, and it looks like all the autopilot jobs are getting killed by it | 23:06 |
ogra_ | well, at least the unity8 tests run again, maguro seems to have fallen over heavily though | 23:06 |
ogra_ | plars, not on mako | 23:06 |
plars | ogra_: ... except mako, was about to add that I had just noticed that | 23:06 |
plars | ogra_: I'm just now looking | 23:07 |
ogra_ | the outage affected LP as well, not sure if PPAs or bzr branches worked | 23:08 |
ogra_ | (i definitely couldnt open bugs) | 23:08 |
plars | ogra_: not sure if it's related, but there are some new errors in the logcat.log that I didn't see on the previous build | 23:09 |
plars | D/libEGL ( 1438): loaded /vendor/lib/egl/libEGL_POWERVR_SGX540_120.so | 23:09 |
plars | E/linker ( 1438): linker.c:1095| ERROR: Library 'libOpenVG.so' not found | 23:09 |
plars | E/linker ( 1438): linker.c:1095| ERROR: Library 'libPVROGL.so' not found | 23:09 |
plars | E/linker ( 1438): linker.c:1095| ERROR: Library 'libPVROCL.so' not found | 23:09 |
ogra_ | no | 23:09 |
ogra_ | :) | 23:09 |
plars | wouldn't think so, but strange that it's only on maguro | 23:09 |
ogra_ | they arent new | 23:09 |
ogra_ | yeah, thats fine | 23:10 |
ogra_ | PVR doesnt ship and GL support :) | 23:10 |
lool | plars: how does not reproduce the initctl issue you had? | 23:10 |
ogra_ | only GLES | 23:10 |
lool | is it only when running autopilot tests? | 23:10 |
plars | lool: seems to be | 23:11 |
=== lool changed the topic of #ubuntu-ci-eng to: Tree is "ask for landing" | image 58++ blocked by unity crash lp:1228097 and broken click updates | Ubuntu CI Engineering Team | Known issues: - | Vanguard: - (http://goo.gl/dsQJaX) | Sheriff: - (http://pad.lv/~canonical-ci-sheriffs) | ||
ogra_ | plars, i would just run maguro again completely ... the outage happened exactly when the image was done, i guess not everything was up at the same time after that | 23:11 |
plars | ogra_: I can give it a try, but mako was running also at the exact same time | 23:12 |
ogra_ | mako doesnt actually look bad on the dashboard | 23:12 |
lool | so I have only mako for testing, if it doesn't happen there I can't reproduce I guess | 23:12 |
lool | plars: would be nice to find which update breaks that | 23:12 |
ogra_ | well, there is no real reaaason why it would run on mako but not on maguro | 23:12 |
ogra_ | lool, might be similar to the bashrc hack | 23:13 |
plars | I restarted maguro, we'll see what happens | 23:13 |
lool | plars: I see an unity8 crash file in http://reports.qa.ubuntu.com/smokeng/saucy/touch_ro/4390/ubuntu-weather-app-autopilot/ | 23:14 |
lool | if unity8 isn't starting, you might end up without an user session to connect to I guess | 23:14 |
ogra_ | phablet@ubuntu-phablet:~$ initctl list|wc -l | 23:14 |
ogra_ | 28 | 23:14 |
ogra_ | works all fine for me | 23:14 |
ogra_ | yeah, what lool says | 23:15 |
ogra_ | well, unity respawns ... | 23:16 |
ogra_ | but the session startup takes about a minute | 23:16 |
ogra_ | so during that time you will definitely get session errors if you try to connect | 23:16 |
plars | ah, hang on, I think I was able to find where it happens | 23:16 |
ogra_ | hmm | 23:17 |
plars | sudo -i -u phablet bash -ic "echo test" | 23:17 |
ogra_ | and the hacks used ot export the session to adb will actually not be updated ... so it can happen that you keep the old dbus address and XDG upstart info | 23:17 |
plars | ogra_: ^ works on mako, not on maguro... well, it works, but it prints the error | 23:17 |
plars | ogra_: seems somehow related to the bashrc thing? | 23:18 |
ogra_ | do you have a session on screen ? | 23:18 |
ogra_ | not really the bashrc only holds the two vars | 23:18 |
plars | ogra_: I'm just comparing them at home | 23:18 |
ogra_ | it is rather the hacks we ship in /etc/profile.d | 23:19 |
ogra_ | either the dbus one or the upstart one | 23:19 |
plars | ogra_: so that error message, though strange, might be a red herring. It seems to do what it's supposed to do, regardless of the complaint | 23:21 |
ogra_ | but there is really no reason why it would break in any device specific way | 23:21 |
ogra_ | right | 23:21 |
plars | ogra_: but it definitely seems to be a new error we're seeing in this build | 23:21 |
ogra_ | file a bug, i'll inspect it ... i dont think thats the issue though | 23:22 |
ogra_ | (against ubuntu-touch-session ... assign to me) | 23:22 |
ogra_ | F/libc ( 664): Fatal signal 11 (SIGSEGV) at 0xdeadbaad (code=1), thread 116 (SurfaceFlinger) | 23:27 |
ogra_ | ouch | 23:27 |
ogra_ | thats the calendar-app test on mako | 23:27 |
plars | ok, well it's restarted on maguro, will be back to check on it in a bit | 23:28 |
plars | hopefully it was just the bzr thing | 23:28 |
ogra_ | the gallery app failure seems to be fallout of the download-manager breakage | 23:29 |
ogra_ | the click lens is keeping the system busy on session start with attempts to fire up download-managerr | 23:30 |
ogra_ | that makes systemsettle-before fail | 23:30 |
* ogra_ will go afk now (1:30am here) ... | 23:32 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!