=== bfiller is now known as bfiller_afk | ||
=== vrruiz_ is now known as rvr | ||
=== Ursinha-afk is now known as Ursinha | ||
rsalveti | didrocks: sergiusens: so, I just noticed a qtubuntu mr got merged automatically after approved | 09:25 |
---|---|---|
rsalveti | thought it'd be on hold for the landing slot stuff | 09:25 |
rsalveti | https://code.launchpad.net/~ricmm/qtubuntu/build-multiple-binary-fix/+merge/203975 | 09:25 |
rsalveti | should we release it via the old method? | 09:25 |
didrocks | rsalveti: let me check if it's in CI Train | 09:26 |
didrocks | (it's still possible to release trunk with an empty MP at worse) | 09:26 |
didrocks | rsalveti: there is no lander for qtubuntu | 09:27 |
didrocks | I wonder what sergiusens is doing! | 09:27 |
didrocks | https://docs.google.com/a/canonical.com/spreadsheet/ccc?key=0Au6idq7TkpUUdC05a2ZQSmgwU2NFYnJQOE9qMDRYa3c#gid=1 | 09:27 |
didrocks | sergiusens: putting that on your shoulders for citrain? | 09:28 |
didrocks | plars: psivaa: joining us? | 09:32 |
plars | didrocks: we're in a meeting at the moment | 09:32 |
plars | didrocks: sorry | 09:32 |
didrocks | ah | 09:32 |
popey | outrage! | 09:32 |
didrocks | plars: joining afterwards? | 09:32 |
popey | I was early and everything | 09:32 |
didrocks | :p | 09:32 |
didrocks | poor popey and Mirv ;) | 09:32 |
plars | didrocks: the results I looked at on mako seemed pretty random over the past 2 builds - same number of failures but different tests that failed, so nothing that's a clear regression | 09:33 |
Mirv | :( | 09:33 |
plars | didrocks: I believe psivaa said maguro was the same | 09:33 |
Mirv | now I have to spend quality time with popey! | 09:33 |
ogra_ | the whoopsie-upload crash is new though | 09:33 |
didrocks | plars: hum, ok, we'll do without you | 09:33 |
ogra_ | (in install-and-boot) | 09:33 |
didrocks | thanks | 09:33 |
sergiusens | didrocks, I didn't do anything | 09:33 |
didrocks | popey: Mirv: joining your party | 09:33 |
didrocks | sergiusens: that's the issue! :) | 09:33 |
didrocks | you slacked over and didn't assign to you qtubuntu :p | 09:34 |
popey | typical fin, mirv is on mute | 09:34 |
didrocks | sergiusens: more seriously, are you going to take care of qtubuntu as a lander? | 09:34 |
Mirv | silence is golden | 09:34 |
sergiusens | didrocks, yes | 09:34 |
sergiusens | didrocks, yeah, I had to discuss that one; it's an edge case | 09:35 |
sergiusens | didrocks, but seems it falls into our team | 09:35 |
rsalveti | yeah, it was in a gray area | 09:35 |
ogra_ | http://ci.ubuntu.com/smokeng/trusty/touch/mako/156:20140131:20140115.1/6364/security/ | 09:41 |
ogra_ | Mirv, ^^ | 09:41 |
Mirv | hmm, I wonder why my screen stays black after flashing even after two reboots. unity8 seems to be running. and yes libunity-mir is installed | 10:08 |
Mirv | um, I guess I need to total-destruction-flash | 10:14 |
sil2100 | hm, here it works fine, just flashed it clean | 10:16 |
Mirv | sil2100: ok I've working device, but I still don't have explanation on where were the test tools that contain eg. that run_test.sh? | 10:52 |
cjwatson | any reason I can't upload livecd-rootfs? we have an urgent ppc64el deadline which is going to require a change there | 10:53 |
cjwatson | [ Chris Wayne ] | 10:53 |
cjwatson | * Add hostname and machine-info to the writable paths, so that hostnamectl | 10:53 |
Mirv | not eg at https://wiki.ubuntu.com/Touch/Testing | 10:53 |
cjwatson | can change them to create a sane default bluetooth device name. | 10:53 |
cjwatson | is the other change | 10:53 |
ogra_ | popey, could you reproduce the UI hang you had during the meeting ? | 10:54 |
cyphermox | eh, why is that in livecd-rootfs? | 10:54 |
cjwatson | dunno | 10:55 |
cjwatson | don't have time to look | 10:55 |
cyphermox | I would have thought lxc-android-config :/ | 10:55 |
cyphermox | cjwatson: if it's the change I think, it's pretty harmless | 10:55 |
ogra_ | cyphermox, thats the weird way ... | 10:55 |
cjwatson | http://paste.ubuntu.com/6848788/ is the full patch | 10:55 |
cjwatson | which makes sense to me I think | 10:55 |
cyphermox | right | 10:55 |
cyphermox | safe | 10:55 |
cjwatson | good | 10:55 |
ogra_ | we have two variants of writable files ... one requires the file to be touched and to pre-exist in the image | 10:55 |
ogra_ | cyphermox, there is a second half that i'm not actually happy with that does the boot time bits for setting the hostname | 10:57 |
ogra_ | (for lxc-android-config | 10:57 |
ogra_ | ) | 10:57 |
cyphermox | ogra_: why are you not happy with it? | 10:57 |
ogra_ | cyphermox, because it runs at every boot | 10:57 |
cyphermox | only if the value isn't set, otherwise it's a no-op | 10:57 |
ogra_ | while it should be a one time/ first boot thing | 10:57 |
cyphermox | (not that it's much better to even start it at all) | 10:58 |
Mirv | sil2100: ok found btw finally | 10:58 |
ogra_ | i would pretty much like us to have some way of running things at first boot once | 10:58 |
Mirv | didrocks: sil2100: on another note, qmlscene maguro crash might be fixed by https://code.launchpad.net/~ricmm/unity-mir/catch-starting-failure/+merge/203920 | 10:58 |
didrocks | Mirv: yeah, we still need to have more infos from the toolkit/qt side on why the app failed to start though :) | 10:59 |
sil2100 | Mirv: yes, we know ;) | 10:59 |
ogra_ | while it might not add much, the processing happens in a "script" block in the upstart job ... that means a shell is fired up just for checking the file and killing the job again | 10:59 |
cyphermox | ogra_: well, right now it's half-way there, it actually does anything only on first boot | 10:59 |
Mirv | aha :) | 11:00 |
cyphermox | right | 11:00 |
sil2100 | Mirv: oh, sorry, missed your prev message! | 11:00 |
cyphermox | I know it's not ideal, but there doesn't seem to be an alternative yet | 11:00 |
cjwatson | oh, that livecd-rootfs change was already uploaded but not committed | 11:00 |
cjwatson | god's sake people | 11:00 |
cyphermox | ugh | 11:00 |
* ogra_ checks | 11:01 | |
ogra_ | phew, wasnt me | 11:01 |
cjwatson | so never mind, not for this channel | 11:01 |
ogra_ | cyphermox, i'll land the lxc-android-config bit today ... but we really need to find some better way for that first-boot-run-once stuff | 11:02 |
popey | ogra_: no | 11:03 |
popey | ogra_: i think it was some odd quirk that happend when i took a screenshot | 11:03 |
ogra_ | ah, k | 11:04 |
ogra_ | we just sasw a hang on the new N7 with the latest zip and were wondering if the two were related | 11:04 |
popey | ogra_: I've now taken 80 screenshots | 11:04 |
popey | and it's not hung yet | 11:04 |
ogra_ | (but we use cdimage images so its not very likely) | 11:04 |
Mirv | sil2100: ok still not everything, missing "qrt_tests"? | 11:06 |
Mirv | aha, setup.sh.. | 11:09 |
Mirv | didrocks: sil2100: I'm getting a pass on 156 of click-apparmor | 11:20 |
=== alan_g is now known as alan_g|dentist | ||
sil2100 | Mirv: did you try many retries? | 11:22 |
Mirv | sil2100: 5 so far | 11:22 |
sil2100 | Mirv: I'm trying to retrace the unity8 crash but I seem to be getting a strange python error while trying to fetch more information from the coredump and crash file | 11:22 |
Mirv | sil2100: didrocks: ok on 7th try I got one upstart failure, but under different subtest than in the logs. so seems flaky. | 11:25 |
didrocks | Mirv: hum, interesting | 11:25 |
sil2100 | Did anyone get an error like this before when trying to apport-cli a crash: struct.error: unpack requires a bytes object of length 4 | 11:28 |
sil2100 | (python error) | 11:28 |
sil2100 | Mirv, didrocks: so, the unity8 crash on the install-and-boot has a damaged (not completely downloaded) CoreDump, so we can't really retrace it properly | 11:34 |
=== MacSlow is now known as MacSlow|lunch | ||
sil2100 | Mirv: did you try pinging someone that could help us out with these? | 12:25 |
sil2100 | Mirv: with the failing security tests | 12:26 |
Mirv | sil2100: no I didn't. I'm not sure if it should upstart devs or apparmor? | 12:29 |
sil2100 | Mirv: that's a valid question... | 12:31 |
sil2100 | Mirv: you could start off with Jamie I guess? | 12:31 |
Mirv | jdstrand: ^ can you think of something why click-apparmor test has become flaky? https://jenkins.qa.ubuntu.com/job/trusty-touch-mako-smoke-security/130/console | 12:51 |
Mirv | jdstrand: I could reproduce it on my mako ca. 1/5 or 1/10, although the exact test case might vary | 12:51 |
Mirv | sil2100: argh, sorry, I'm running out of time now on all fronts and I'm into hangout now | 13:01 |
=== alan_g|dentist is now known as alan_g|lunch | ||
=== josepht changed the topic of #ubuntu-ci-eng to: Ubuntu CI Engineering Team | Vanguard: josepht | Landing instructions: http://paste.ubuntu.com/6292280/ | Known issues: - | ||
sil2100 | Mirv: ok, I'll try to poke this as well | 13:03 |
jdstrand | Mirv: re click-apparmor> I don't know why it has become flaky, no. there is a sleep in there though to wait on upstart starting an app-- it is 15 seconds though. if it is failing, I would think it has something to do with app startup | 13:18 |
jdstrand | Mirv: did you say it always failed or intermittently failed? | 13:20 |
* jdstrand assumes 1/5 and 1/10 means intermittently | 13:20 | |
sil2100 | jdstrand: yes, it seemingly failed every 5 or 10 times | 13:21 |
jdstrand | https://jenkins.qa.ubuntu.com/job/trusty-touch-mako-smoke-security/130/artifact/clientlogs/application-click-com.example.lp1245677_lp1245677_0.1.log/*view*/ shows: ** (process:14151): WARNING **: Unable to exec: Permission denied | 13:22 |
Mirv | jdstrand: the delays seem quite near 15 seconds though every time | 13:22 |
Mirv | sil2100: thanks | 13:22 |
Mirv | (the delay that I see the line about upstart launch in the test) | 13:23 |
sil2100 | Mirv: did you try downgrading any of the listed risky packages? | 13:23 |
Mirv | ah, except that if there's no output whatsoever and it's always the same sleep | 13:23 |
jdstrand | there is a TODO to remove the sleep because I don't know how else to know when upstart has *actually* started the thing. it will return 'started' immediately iirc, but then may not immediately do its thing | 13:23 |
jdstrand | Mirv: you said different tests within click-apparmor would fail? | 13:25 |
Mirv | (sorry still in hangout) | 13:28 |
Mirv | jdstrand: yeah for example I had it in com.example.am-i-confined_am-i-confined_0.1.desktop, during the checking application upstart job, while that ^ log shows it under different one. | 13:29 |
=== alan_g|lunch is now known as alan_g | ||
jdstrand | Mirv: but is it always com.example.am-i-confined_am-i-confined_0.1.desktop? | 13:30 |
Mirv | I mean here it was https://jenkins.qa.ubuntu.com/job/trusty-touch-mako-smoke-security/130/console com.example.lp1245677_lp1245677_0.1.desktop | 13:30 |
Mirv | while I had it during checking application upstart job of another one | 13:31 |
jdstrand | https://jenkins.qa.ubuntu.com/job/trusty-touch-mako-smoke-security/130/console and http://ci.ubuntu.com/smokeng/trusty/touch/mako/156:20140131:20140115.1/6364/security/719045/ failed in the same place on the same test | 13:32 |
jdstrand | (the 2nd was the original you pointed me to | 13:33 |
jdstrand | ) | 13:33 |
sil2100 | I think it's the same test-run even | 13:33 |
jdstrand | can you point me to different runs with the failure? | 13:34 |
Mirv | sil2100: I downgraded upstart packages to the previous one. no failure yet, but it's hard to be conclusive | 13:34 |
Mirv | jdstrand: http://pastebin.ubuntu.com/6849408/ | 13:34 |
jdstrand | is http://ci.ubuntu.com/smokeng/trusty/touch/mako/156:20140131:20140115.1/6364/ the first time is started failing? | 13:36 |
Mirv | jdstrand: looks like, on http://ci.ubuntu.com/smokeng/trusty/touch/mako/156:20140131:20140115.1/6364/security/ you can see "Result History" for previous 5 | 13:36 |
Mirv | and click for more. so yes. | 13:36 |
jdstrand | ok | 13:37 |
sil2100 | jdstrand: yes, we were suspecting that some of the last uploads we made to the image caused it to start happening | 13:37 |
jdstrand | fwiw, I hate this test because of the sleep, but it could be showing a problem somewhere else-- 15 seconds is a *long* time to wait to see if an app started | 13:38 |
tedg | Wow, I'm getting even faster at scaring didrocks away. | 13:40 |
jdstrand | upstart-app-launch has quite a few changes that could conceivably affect this test... | 13:40 |
sil2100 | Mirv: ok, so it seems no failures with upstart bits reverted, right? | 13:58 |
=== MacSlow|lunch is now known as MacSlow | ||
jdstrand | by upstart bits, do you mean upstart-app-launch? | 14:03 |
sil2100 | jdstrand: yes, I guess that's what you upgraded Mirv ? | 14:04 |
sil2100 | s/upgraded/downgraded | 14:04 |
ogra_ | sidegraded :P | 14:04 |
jdstrand | hmm, but the error is when using aa-exec click | 14:15 |
jdstrand | right, so upstart-app-launch isn't being used by click-apparmor | 14:19 |
jdstrand | it is calling the upstart job directly with start, so upstart-app-launch should be off the table | 14:19 |
* tedg should have a stable connection now | 14:20 | |
jdstrand | oh, what a second | 14:21 |
jdstrand | wait* | 14:21 |
jdstrand | durr, start application | 14:21 |
jdstrand | so upstart-app-launch could be on the table, but it shouldn't be for that one test cause it is happening when using aa-exec | 14:22 |
jdstrand | meh, let me actually read this thing (apparently I was too clever when I wrote this test script :) | 14:23 |
jdstrand | sorry-- I confused myself-- ignore everything I said in the last 5 minutes | 14:24 |
sil2100 | hmm | 14:25 |
jdstrand | start application APP_ID="..." >/dev/null && sleep 15 && [ -f "$test_path" ] *is* what's failing | 14:25 |
sil2100 | Ok ;) So indeed upstart-app-launch might be related | 14:25 |
tedg | jdstrand, Could that be because we change it to a task so that'd get delayed? | 14:26 |
tedg | jdstrand, What is "test_path" ? | 14:26 |
jdstrand | that was what I was thinking | 14:26 |
jdstrand | tedg: $HOME/.local/share/$pkgname/pass | 14:27 |
jdstrand | tedg: https://jenkins.qa.ubuntu.com/job/trusty-touch-mako-smoke-security/130/artifact/clientlogs/application-click-com.example.lp1245677_lp1245677_0.1.log/*view*/ | 14:27 |
tedg | Hmm, seems that shouldn't be an issue. Really, it should be more reliable now with that task marking. | 14:27 |
jdstrand | tedg: that ^ is what's weird. it should look like: https://jenkins.qa.ubuntu.com/job/trusty-touch-mako-smoke-security/130/artifact/clientlogs/application-click-com.example.am-i-confined_am-i-confined_0.1.log/*view*/ | 14:28 |
jdstrand | tedg: so the binary isn't even getting run for some reason-- but it will most of the time | 14:28 |
tedg | jdstrand, Well the call to exec is failing. | 14:28 |
jdstrand | yes | 14:29 |
jdstrand | but why only occasionally? | 14:29 |
tedg | I'm pretty sure that calling a syscall isn't my fault :-) | 14:29 |
=== bfiller_afk is now known as bfiller | ||
tedg | Can we turn on debugging to see what teh exec line is? | 14:29 |
tedg | Perhaps it's not getting set correctly? | 14:29 |
jdstrand | how do you do that? | 14:30 |
jdstrand | note, I cannot reproduce locally | 14:30 |
jdstrand | Mirv seems to be able to | 14:30 |
tedg | You need to set G_MESSAGE_DEBUG=all | 14:30 |
tedg | I put it in an Upstart job that is start on starting dbus | 14:30 |
tedg | So it's always on for my system. | 14:31 |
tedg | http://paste.ubuntu.com/6849653/ | 14:31 |
jdstrand | tedg: so what does kenvandine's patch do? | 14:33 |
tedg | jdstrand, The triplet API one? | 14:33 |
jdstrand | "Added upstart_app_launch_triplet_to_app_id to construct an appid from pkg, app, version triplet. If a specific version isn't provided, click is used to determine the version from the manifest." | 14:33 |
tedg | jdstrand, It's just putting some common code that used to be in URL dispatcher in libUAL so he can use it in content hub. | 14:34 |
tedg | jdstrand, It's not used in the upstart jobs directly. | 14:34 |
jdstrand | my thinking was if Mirv is mostly hitting one path, but occasionally hitting the other | 14:34 |
jdstrand | hmm | 14:34 |
tedg | There was an issue where exec wasn't getting set correctly, but that was an issue with not flushing the dbus connection. | 14:34 |
tedg | It looks like that patch is in trunk. | 14:35 |
jdstrand | tedg: so, I guess you'd want to see the exec line for both the failing and passing case, correct? | 14:35 |
tedg | jdstrand, Well, more the failing :-) | 14:35 |
tedg | I think it'll be obvious if it's broken. | 14:35 |
sil2100 | I'm not sure if Mirv is around anymore... | 14:35 |
jdstrand | right, so, I can't seem to help with that | 14:35 |
jdstrand | Mirv: can you get what tedg needs? | 14:35 |
sil2100 | jdstrand: did you try re-running it multiple times on the latest image on a mako? | 14:36 |
sil2100 | jdstrand: I might try achieving the same here, but I'll have to prepare my device | 14:36 |
jdstrand | I can't on mako (it is my dogfood device, I'm an avenger, etc, etc). I am running it in a loop on grouper. it has gone 30 times or more | 14:36 |
sil2100 | jdstrand: let me try giving it a shot then | 14:37 |
sil2100 | Might take some moments | 14:38 |
jdstrand | sil2100: thanks-- might set up the upstart job like tedg suggested | 14:38 |
tedg | It makes you appreciate logrotate ;-) | 14:38 |
bfiller | sil2100: how do I know when it's ok to press the "Merge and Clean" button on Landing-002 silo? | 14:56 |
bfiller | sil2100: I think the packages are published | 14:57 |
seb128 | bfiller, when the packages are in the release pocket, if you press too soon the jenkins job is going to fail and you are good to retry later | 14:58 |
seb128 | e.g no issue trying | 14:58 |
bfiller | seb128: how do I know when they hit the release pocket? | 14:58 |
seb128 | bfiller, you look on https://launchpad.net/distros/ubuntu/+source/<source> | 14:59 |
bfiller | seb128: ok thanks | 14:59 |
seb128 | yw | 14:59 |
ogra_ | or use rmadison :) | 14:59 |
seb128 | or if you have bugs linked in the changelog, that's when the bugs get autoclosed by launchpad | 14:59 |
seb128 | that generates emails | 14:59 |
bfiller | I've seen those, cool | 15:00 |
sil2100 | jdstrand: ok, checking first if I can reproduce the failure at al | 15:03 |
sil2100 | bfiller: as seb128 said ;) | 15:03 |
jdstrand | cool | 15:18 |
cjwatson | seb128: you must use rmadison not lp | 15:27 |
cjwatson | lp will tell you it's in release when it's finished processing accepted, which is before it hits disk | 15:27 |
cjwatson | rmadison updates almost as soon as possible after it hits disk in a way that you can see from jenkins builds etc. | 15:27 |
cjwatson | so it's far more accurate for this and doesn't have false positives | 15:28 |
sil2100 | jdstrand: damn, I'm running and running and I cannot get things failing ;/ | 15:44 |
=== josepht changed the topic of #ubuntu-ci-eng to: Ubuntu CI Engineering Team | Vanguard: cihelp | Landing instructions: http://paste.ubuntu.com/6292280/ | Known issues: - | ||
jdstrand | sil2100: sounds like we need Mirv :\ | 15:49 |
sil2100 | uh! | 15:56 |
sil2100 | Wait! Finally! | 15:56 |
sil2100 | Ok, but without the debugging - but since I made it once, I'll repro it again | 15:57 |
balloons | sergiusens, a few questions about click-buddy if you have a moment | 16:04 |
sil2100 | jdstrand, tedg: when G_MESSAGE_DEBUG=all is set, does it output the logs somewhere? | 16:08 |
tedg | sil2100, In the application log. | 16:09 |
tedg | sil2100, They should be in ~/.cache/upstart/application* | 16:10 |
sil2100 | tedg: let me pastebin this | 16:12 |
sil2100 | tedg, jdstrand: not sure if this helps, but here are the logs: http://paste.ubuntu.com/6850111/ | 16:14 |
sil2100 | tedg, jdstrand: this time it failed on com.example.am-i-confined_am-i-confined_0.1 | 16:14 |
tedg | Hmm, doesn't seem to have the debug messages. | 16:17 |
=== gatox is now known as gatox_lunch | ||
tedg | sil2100, Can you pastebin ~/.cache/upstart/application.log as well? | 16:19 |
tedg | sil2100, Perhaps you could just run "APP_ID=onethatsfailing /usr/lib/*/upstart-app-launch/click-exec" in a loop. | 16:20 |
sil2100 | tedg: one moment | 16:20 |
tedg | See if it ever prints a different APP_EXEC line | 16:21 |
sil2100 | tedg: this one is poorer than I thought: paste.ubuntu.com/6850144/ | 16:22 |
sil2100 | http://application log. | 16:22 |
sil2100 | geh | 16:22 |
sil2100 | Wrong paste | 16:22 |
sil2100 | http://paste.ubuntu.com/6850144/ | 16:22 |
tedg | Hmm, don't see any debug messages there either. | 16:24 |
tedg | sil2100, Did you add the upstart job or just set it on the command line? | 16:25 |
tedg | initctl get-env G_MESSAGES_DEBUG | 16:25 |
=== josepht changed the topic of #ubuntu-ci-eng to: Ubuntu CI Engineering Team | Vanguard: josepht | Landing instructions: http://paste.ubuntu.com/6292280/ | Known issues: - | ||
sil2100 | Ah, I see I made a typo | 16:27 |
tedg | Oh, good. I was getting confused :-) | 16:28 |
sil2100 | G_MESSAGES_DEBUG was missing a 'S' | 16:28 |
sil2100 | FInally it crashed | 16:36 |
seb128 | cjwatson, thanks, the launchpad summary/bug emails are usually good enough and less slow to load (though that might not be true anymore since it changed server) | 16:37 |
sil2100 | tedg: log coming right up | 16:37 |
sil2100 | tedg, jdstrand: http://paste.ubuntu.com/6850227/ | 16:38 |
=== greyback_ is now known as greyback|food | ||
tedg | jdstrand, That looks right to me... | 16:42 |
tedg | jdstrand, Not sure that you need the "./" but, that should work. | 16:42 |
tedg | sil2100, Is there a DENY in the syslog? | 16:43 |
sil2100 | tedg: hm, doesn't seem like it | 16:45 |
cjwatson | seb128: rmadison is way faster now than it used to be, indeed - it doesn't usually annoy me any more | 16:45 |
seb128 | cjwatson, good to know, I should change my habits then ;-) | 16:45 |
cjwatson | occasionally takes a little while to recache but nothing like the minutes that it sometimes was | 16:46 |
* cjwatson nukes raring from its config which can't hurt | 16:46 | |
bfiller | sil2100: can you publish lines 34 and 35 in CI Train. testing complete | 16:50 |
sil2100 | bfiller: will do publishing bits in a moment, still counting on finding the badguy for the security test failures | 16:51 |
sil2100 | tedg and jdstrand are helping me | 16:51 |
tedg | sil2100, I don't know what else to suggest. Not sure how we could have a permission error there. | 16:51 |
tedg | We should probably print the executable name on error. | 16:52 |
tedg | But I don't think that's this problem. | 16:52 |
jdstrand | that is totally odd | 16:56 |
jdstrand | the previous run used the same APP_EXEC | 16:56 |
jdstrand | and it succeeded | 16:56 |
jdstrand | tedg: what is (process:3330) in '** (process:3330): WARNING **: Unable to exec: Permission denied'? | 16:57 |
jdstrand | up above, I see 'libust[3330/3332]: Error:...' | 16:58 |
tedg | jdstrand, That's the return from the call to exec | 16:58 |
=== greyback|food is now known as greyback | ||
jdstrand | no, I mean, what process is that? | 16:58 |
jdstrand | 3330 | 16:58 |
jdstrand | process 3330 is the one that isn't able to do the exec, aiui | 16:59 |
jdstrand | so that is what we should be looking at | 16:59 |
tedg | jdstrand, It's the one in the exec line of the job. exec-line-exec | 16:59 |
jdstrand | perhaps it is spitting out the wrong error. perhaps it is doing the right thing, but DAC is preventing it | 17:00 |
sil2100 | ogra_: could you ACK this packaging change? http://162.213.34.102/job/landing-009-2-publish/lastSuccessfulBuild/artifact/packaging_changes_unity-scopes-shell_0.2.8+14.04.20140131.2-0ubuntu1.diff | 17:00 |
sil2100 | popey: I guess no meeting today! robru is away anyway | 17:01 |
popey | ok | 17:01 |
popey | sil2100: i tried flashing an old image on my phone | 17:01 |
popey | didnt work well | 17:02 |
popey | will try harder and let you know what i find | 17:02 |
sil2100 | popey: ok, thanks! I'll try that as well, maybe during the weekend even | 17:02 |
tedg | jdstrand, It's just calling strerror on the return code from execvp | 17:02 |
sil2100 | Today sadly I was super busy with *everything* | 17:03 |
popey | sil2100: http://popey.mooo.com/mirror is my archive | 17:03 |
tedg | jdstrand, Not much processing in between. | 17:03 |
jdstrand | ok, then that would suggest either MAC or DAC. MAC (apparmor) would show in the log, but it doesn't, so we should be looking at DAC | 17:03 |
tedg | jdstrand, I need to run for a bit (reading at my son's school). Drop me a mail if you need something. | 17:04 |
* jdstrand notes that this is not a click-apparmor thing, but will try to help | 17:04 | |
sil2100 | jdstrand, tedg: give me a sign if you find anything as well - I'll fill in a bug regarding this | 17:05 |
jdstrand | sil2100: how are you running the click-apparmor test | 17:05 |
sil2100 | So, I'l logged into the device, I copied the qrt_tests directory with the tests to my main directory, as well as the run_test.sh script (from the branch) and run ../run_test.sh click-apparmor in the qrt_tests directory | 17:06 |
jdstrand | sil2100: ok, so if I give you a patch to the click-apparmor test, you'd be able to apply it and we could get more debugging info? | 17:07 |
sil2100 | jdstrand: I think I could do that :) | 17:10 |
sil2100 | slangasek: http://162.213.34.102/job/landing-009-2-publish/lastSuccessfulBuild/artifact/packaging_changes_unity-scopes-shell_0.2.8+14.04.20140131.2-0ubuntu1.diff <- pkg changes (only debian/ modifications and makefiles shown) | 17:12 |
ogra_ | sil2100, ack | 17:13 |
slangasek | sil2100: so, where is libunity-scopes 0.2.8? The archive has 0.2.6 | 17:13 |
slangasek | sil2100: also, why is there a hard-coded dep on libunity-scopes0, instead of this being handled via shlibdeps? | 17:14 |
sil2100 | slangasek: libunity-scopes is part of the landing, so it's being published along with this package | 17:15 |
ogra_ | right | 17:15 |
sil2100 | slangasek: as for the second question - not sure, it seems to have been like that for a while, would have to check and ask for the rationale | 17:16 |
=== gatox_lunch is now known as gatox | ||
ogra_ | and we had bad issues with shlibdeps recently (because the code simply didnt do the right thing) in unity8 | 17:16 |
slangasek | ok, that's fine then; but the hardcoded lib dep looks wrong, I'm not ok to sign off on a version bump of it without knowing why it's there in the first place | 17:16 |
slangasek | there needs to be an explicit rationale for such things | 17:16 |
ogra_ | slangasek, all these unity packages have the libs hardcoded that way, upstream is slowly moving over to shlibdeps (but the unity8 disaster this week showed that this doesnt always work OOTB) | 17:20 |
slangasek | sil2100, ogra_: so no, it looks to me like the shlibs in libunity-scopes itself are wrong and need to be fixed, and the hard-coded dep in unity-scopes-shell dropped | 17:20 |
slangasek | ogra_: "this is how it's done" is not an excuse for any core-dev signing off on such things that are explicitly contrary to Ubuntu best practices | 17:20 |
ogra_ | ok | 17:21 |
sil2100 | hmm... | 17:21 |
* ogra_ would agree if just using shlibdeps would work, apparently it doesnt always though | 17:21 | |
sil2100 | Can I ask for the same for http://162.213.34.102/job/landing-005-2-publish/1/artifact/packaging_changes_gallery-app_0.0.67+14.04.20140131-0ubuntu1.diff then? | 17:22 |
* ogra_ doesnt like that second one | 17:22 | |
ogra_ | that mmeans it will always pull autopilot in if you install gallery-app | 17:23 |
sil2100 | But that's for the autopilot package of gallery-app | 17:24 |
sil2100 | So gallery-app-autopilot | 17:24 |
ogra_ | oh, ok, just for the autopilot package of it ... then i'm fine | 17:24 |
ogra_ | (that info gets kind of swallowed in the diff) | 17:25 |
ogra_ | (it woud be nice if there waas a changelog entry for it the next time, but i wont block on it ...) | 17:25 |
sil2100 | ogra_: true... | 17:26 |
=== josepht changed the topic of #ubuntu-ci-eng to: Ubuntu CI Engineering Team | Vanguard: cihelp | Landing instructions: http://paste.ubuntu.com/6292280/ | Known issues: - | ||
ogra_ | sil2100, since you published after my ACK (before slangasek NACKED), can you contact upstream to change to shlibdeps with the next upload ? | 17:26 |
slangasek | ogra_: shlibdeps works *fine* if people aren't *failing to use them* | 17:27 |
* sil2100 sighs | 17:27 | |
ogra_ | slangasek, as long as the linking of the code is right, yes | 17:27 |
sil2100 | Maybe we can somehow back if off? | 17:27 |
ogra_ | slangasek, it obviously didnt work with unity8 on wed. so that libunity-mir1 vanished off the image | 17:28 |
slangasek | ogra_: so, er, fix the code linkage, don't pile on kludges | 17:28 |
ogra_ | (which broke the world then) | 17:28 |
sil2100 | They were dlopen'ing libunity-mir1 | 17:28 |
slangasek | ok, that's a reason for not relying on shlibs | 17:29 |
slangasek | but that's not the case here | 17:29 |
sil2100 | Let me check | 17:29 |
sil2100 | bfiller: I published 2 of your packages btw. - you can stay on a lookout for those being ready to clean up | 17:33 |
bfiller | sil2100: thanks, will do | 17:33 |
jdstrand | sil2100: ok, can you pull r2110 of QRT | 17:37 |
jdstrand | sil2100: then instead of running '../run_test.sh click-apparmor', just run 'click-apparmor -d' directly | 17:37 |
jdstrand | sil2100: do that in a loop until it fails. that should run 'click pkgdir' to get the symlink, run ls on it, then resolve it and run ls -ld on the dir, and ls -lR on the dir | 17:39 |
jdstrand | sil2100: perhaps that will give us a clue as to what is happening | 17:39 |
sil2100 | jdstrand: I'll try, but I'll be sadly EOD'ing soon | 17:42 |
jdstrand | ack | 17:43 |
sil2100 | jdstrand: runnin | 17:49 |
sil2100 | jdstrand: oh! Failures, a lot of them! On the first run | 17:49 |
sil2100 | Oh, wait, no | 17:50 |
cjwatson | [6~ | 17:52 |
cjwatson | [6~ | 17:52 |
cjwatson | [6~ | 17:52 |
cjwatson | argh | 17:52 |
cjwatson | need beer | 17:52 |
ogra_ | ++ | 17:53 |
sil2100 | jdstrand: ok, I couldn't reproduce it yet | 18:07 |
jdstrand | ok | 18:07 |
sil2100 | jdstrand: I'll keep trying when I'm at the hotel, but for now I guess I'll EOD to get something to eat | 18:08 |
sil2100 | jdstrand: thanks for all the help! :) | 18:08 |
sil2100 | jdstrand: it's a strange issue, but let's work on that again on Monday | 18:08 |
jdstrand | sil2100: I feel like I didn't do much-- but hopefully the debugging will be enlightening | 18:08 |
sil2100 | Let's hope so | 18:08 |
jdstrand | sil2100: have a nice weekend :) | 18:09 |
=== alan_g is now known as alan_g|EOW | ||
sil2100 | You too! | 18:09 |
tedg | jdstrand, Anything exciting? | 18:31 |
jdstrand | tedg: no. I updated the click-apparmor test to add a debug flag that would show the permissions of various files and directories. sil2100 tried it, but couldn't reproduce before he went eod | 18:32 |
tedg | jdstrand, Interestingly the Jenkins run, in the app log, it says "pass" http://ci.ubuntu.com/smokeng/trusty/touch/mako/156:20140131:20140115.1/6364/security/719045/ | 18:33 |
jdstrand | where? | 18:34 |
jdstrand | I see: | 18:34 |
jdstrand | Click apparmor tests | 18:34 |
jdstrand | Passed: 9/10 | 18:34 |
jdstrand | Failed: 1/10 | 18:34 |
jdstrand | ADB_RC=1 | 18:34 |
jdstrand | oh, in the app log | 18:35 |
jdstrand | let me look at that | 18:35 |
jdstrand | tedg: right, https://jenkins.qa.ubuntu.com/job/trusty-touch-mako-smoke-security/130/artifact/clientlogs/application-click-com.example.am-i-confined_am-i-confined_0.1.log/*view*/ did pass, but https://jenkins.qa.ubuntu.com/job/trusty-touch-mako-smoke-security/130/artifact/clientlogs/application-click-com.example.lp1245677_lp1245677_0.1.log/*view*/ did not | 18:35 |
jdstrand | tedg: application-click-com.example.am-i-confined_am-i-confined_0.1.log vs application-click-com.example.lp1245677_lp1245677_0.1.log | 18:36 |
tedg | Ah, I see | 18:36 |
tedg | I proposed this merge, perhaps to get easier debug output next time. | 18:38 |
tedg | https://code.launchpad.net/~ted/upstart-app-launch/print-exec-on-error/+merge/204285 | 18:38 |
tedg | Perhaps sil2100 can use that. | 18:38 |
jdstrand | yeah | 18:41 |
jdstrand | it sure is a weird issue | 18:42 |
jdstrand | I let it run for an hour in a loop on grouper without any issues | 18:42 |
ChrisTownsend | Hi, is anyone available to help with an issue I'm seeing on the Intel CI Autopilot machine? | 19:01 |
=== bfiller is now known as bfiller_afk |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!