/srv/irclogs.ubuntu.com/2015/09/11/#ubuntu-ci-eng.txt

Mirvmorning05:03
robruMirv: heya. I rolled out some big train changes today, had some hiccups but it mostly seems fine. There's a major performance regression in the publish and merge jobs, it scared me at first but it seems to be working at least. I'll investigate that tomorrow. I'll be up for a couple more hours, please ping me if anything horrible explodes and I'll take a look05:10
Mirvrobru: ok! sounds manageable.05:17
Mirvrobru: ^ a publishing went just fine at least05:17
robruMirv: cool05:17
=== chihchun_afk is now known as chihchun
Mirvrobru: I'm getting https://ci-train.ubuntu.com/job/prepare-silo/6125/console after the first assigning failed06:54
Mirvit's not really assigned06:54
Mirvnot sure if this is a new bug or if it's the first time I forgot first to add the test plan before trying to assign a silo..06:54
Mirvbut it feels like some bit gets incorrectly set ie the check for test plan setting is done too late or the changes are not resetted when it's noticed that a test plan is missing and assigning should be aborted06:55
Mirvwith debug https://ci-train.ubuntu.com/job/prepare-silo/6126/console06:55
MirvI left now the https://requests.ci-train.ubuntu.com/#/ticket/332 alone for debugging and created a new request which assigned fine.06:58
Mirvrobru: interesting, the plot thickens. assigning the new silo warned that the package is already in "silo 014" - which is nowhere to be seen. so the 332 assigning-aborting-on-no-testplan really did something half the way.06:59
Mirvrobru: anyway, no worries for the moment, but it's a bit borken at the moment.06:59
robruMirv: oh hrm just saw this now07:56
robruMirv: can you file a bug? With the debug log07:57
robruMirv: there wouldn't be anything "half way" on the Jenkins side, assignment is determined by a single file, so technically the silo assigned. The bug is just that it didn't say the siloname in bileto07:58
robruMirv: interesting chicken and egg, the object enforcing those field requirements only loads after the assignment is made. Not sure if i should just suppress the checks at assignment time or perhaps catch the error and delete the assignment file...08:00
Mirvrobru: I'm filing a bug, ok08:04
robruMirv: tomorrow I'll focus on bug fixing, too scary to make big changes on a Friday before vacation ;-)08:06
Mirvrobru: ok :) bug #1494628 for your tomorrow then. have a great holiday!08:07
ubot5bug 1494628 in Bileto "Silo configured "half way" after an error about lack of test plan" [Undecided,New] https://launchpad.net/bugs/149462808:07
robruMirv: thanks08:09
Mirvjibel: joining?08:31
ogra_Mirv, i dont think your search includes the right things ... try https://bugs.launchpad.net/canonical-devices-system-image/+bugs?field.tag=hotfix608:40
Mirvogra_: oh! I thought https://bugs.launchpad.net/ubuntu/+source/unity8/+bug/1491566 would be on that list but seems not08:44
ubot5Ubuntu bug 1491566 in Canonical System Image "Shell not responsive after an incoming SMS" [Critical,Confirmed]08:44
nerochiarocihelp: hi, any idea why pretty much all the tests fail on this MR when they were successful before ? it could really be a mistake in the code, but they seem to run fine locally: https://jenkins.qa.ubuntu.com/job/generic-deb-autopilot-runner-vivid-mako/3357/08:44
psivaanerochiaro: do you have a link for the passing job for me?08:47
nerochiaropsivaa: https://jenkins.qa.ubuntu.com/job/generic-mediumtests-builder-vivid-armhf/4041/08:52
Mirvrobru: go. to. sleep. I know you're there!08:53
Mirvand thanks for the fix :)08:53
robruMirv: no I'm not here you're imaginging it08:53
robruMirv: you're welcome.08:53
robruMirv: it should appear in production in 20 minutes. I tested it in staging so I'm confident nothing will explode.08:54
psivaanerochiaro: Isn't a different job, one is running on mako and the other is on a cyclops node?08:54
nerochiaropsivaa: let me find the link to the right one, sorry08:54
nerochiaropsivaa: this should be the passing one on mako http://jenkins.qa.ubuntu.com/job/generic-deb-autopilot-runner-vivid-mako/3303 sorry for the noise08:55
psivaanerochiaro: np, thanks for the link, let me take a look08:56
morphisMirv: don't remember well, was there something more to change for a component to get it build with MPs in a silo?09:16
Mirvmorphis: more than what?09:17
morphisbluez wasn't handled with MPs yet09:18
morphisI changed that and now ready to use MPs and want to test that with citrain09:18
Mirvmorphis: so the first requirement is that there's an upstream project for it on LP, but Bluez is not developed by Ubuntu or on LP09:18
morphisso I was just wondering if there is anything more than adding a MP to the request09:18
morphisMirv: https://code.launchpad.net/~bluetooth/bluez/vivid-phone-overlay09:19
morphisthat should be the baseline09:19
Mirvmorphis: it seems https://launchpad.net/bluez is registered though09:19
Mirvmorphis: I think that since it's reserved for upstream's usage and not us, we can't take it (and we don't have permissions to do so)09:20
morphisMirv: and what is with lp:~bluetooth/bluez/vivid-phone-overlay when I created MPs against that one?09:20
morphisMirv: ofono does that similar: see https://requests.ci-train.ubuntu.com/#/ticket/33809:21
Mirvmorphis: we can check what happens if we add ci-train-bot to ~bluetooth09:21
Mirvmorphis: right, sounds good!09:21
Mirvmorphis: so, since you're an admin, please add ci-train-bot to https://launchpad.net/~bluetooth/+members09:21
morphisdlne09:22
morphiss/dlne/done/09:22
Mirvmorphis: and then you can try to add  https://code.launchpad.net/~bluetooth/bluez/mer-patches/+merge/270782 to landing09:22
morphisI need a new silo for that and can't reuse one with a manually upload package, right?09:22
Mirvmorphis: if you also added https://wiki.ubuntu.com/DailyRelease/InlinePackaging .bzr-builddb to it09:22
Mirvmorphis: you can reuse, just remove it from the manual upload packages and add the MP09:22
morphisok09:22
morphislet me try that09:23
morphisMirv: ok, looks like I need to get rid of all the quilt patches etc09:25
Mirvmorphis: hmm, right.09:28
psivaanerochiaro: so a rerun also is seeing the same failures, i see _usr_bin_webbrowser-app.32011.crash inside the device09:44
psivaaalso _usr_bin_gdb.0.crash09:44
nerochiaropsivaa: ok, let me investigate that09:46
psivaanerochiaro: ack09:46
Mirvjibel: FYI I've documented a total hang I've experience with OTA-5 and OTA-6 at bug #1494692 now that I got it again10:28
ubot5bug 1494692 in mir (Ubuntu) "Total hang of the krillin with a graphics artifact" [Undecided,New] https://launchpad.net/bugs/149469210:28
Mirvunfortunately not possible to adb in during the hang so not possible to trace any process or such10:29
jibelMirv, you cannot even ssh?11:30
Mirvjibel: so ssh would be enabled by default when developer mode is enabled? that actually counters my trust in using the dev mode, the 4-digit pin-code is not really nice enough if there's a wide open ssh running..11:31
Mirv(on a personal phone)11:31
davmor2Mirv: no device has to be unlocked again11:32
Mirvdavmor2: do you mean ssh port is open when the device has screen on / not in lock screen, otherwise closed?11:33
davmor2Mirv: only works via phablet-shell you have to enable ssh otherwise11:34
davmor2Mirv: and only over cable11:34
jibelafter I used phablet-shell once and it copied my keys to the phone, I can ssh to the phone locked or unlocked. I don't remember I did anything else.11:35
Mirvdavmor2: right, but that needs adb too?11:35
davmor2Mirv: no that uses ssh, but should be block the same as adb iirc11:35
Mirvdavmor2: uses ssh, but it requires adb to forward the ssh port in the first place etc?11:36
davmor2Mirv: so if the screen isn't unlocked there is no connection, I could be wrong on that though11:36
Mirvhmm, it starts to be a bit too open for my taste to enable that one11:37
jibeldavmor2, you can ssh even with the screen locked, don't you?11:37
jibeland dev mode on11:37
MirvI wonder how it manages it so that the ssh is not available otherwise than via the cable11:37
davmor2jibel: no11:37
jibeldavmor2, why can I do it then?11:38
davmor2jibel: did you enable it?11:38
jibeldavmor2, I enabled developer mode and used phablet-shell once.11:39
Mirvreading https://sturmflut.github.io/ubuntu/touch/2015/05/08/hacking-ubuntu-touch-part-5-adb-shell-vs-phablet-shell/ it'd sound like the ssh still usess an active adb connection for the forwarding11:40
jibelMirv, sorry but I'm currently connected to 2 phones over ssh and there is no cable between the laptop and the phones :)11:40
Mirvjibel: ok, hmm11:41
ogra_cables are so last century11:41
Mirvso I guess the adb forward is nothing something that "uses" adb constantly, just sets ports up11:41
davmor2Mirv: the ssh uses keys11:41
jibelogra_, especially on arale which is not recognized on usb11:41
ogra_yup :)11:41
Mirvanyway, I'm sure someone else should find the symptoms of my total hangs familiar11:42
Mirvif no-one else gets them, it might be even a hw fault11:42
jibelMirv, after you rebooted is there any interesting event in syslog at the same time than the hang?11:43
ogra_Mirv, android-gadget-service enable ssh11:43
ogra_that gives you the non adb ssh11:43
jibelMirv, davmor2 phablet-shell enables non adb ssh and copies the ssh key11:46
Mirvjibel: not really sure... attached syslog snippet from the time to he bug11:47
=== alan_g is now known as alan_g|lunch
Mirvthere are a few things that sound like interesting, but have happened also before the hang and nothing unique during the 12:48 when the hang happened11:59
Mirvwell actually, there is mali_timeline_sync_fence_create() fail!12:01
Mirvthat might be it12:01
Mirvjibel: thanks, I believe that might be a good indicator here12:13
jibelMirv, np, you did all the work.12:16
pmcgowanwhat do I need to do to have update to a newer package version from a silo? its not showing as a candidate12:29
pmcgowanMirv, ^^12:29
pmcgowanjibel, ^^12:30
Mirvpmcgowan: https://wiki.ubuntu.com/LandingTeam/SiloTestingGuidelines#Install_silos_with_overlay_PPA_enabled <- nowadays the instruction there is correct. add a silo.pref file12:32
jibelpmcgowan, you have to pin the silo higher than the overlay ppa12:32
Mirvjust change the silo number from the example12:32
Mirvapt upgrade behavior changes immediately after the file is created12:32
jibelpmcgowan, either use citrain tool but it'll upgrade everything in the silo, if you want to do it manually add something like12:32
jibelPackage: *12:33
jibelPin: release o=LP-PPA-ci-train-ppa-service-landing-04812:33
jibelPin-Priority: 100212:33
pmcgowanthanks guys I knew something changed a while back12:33
jibelto /etc/apt/preferences.d/extra-ppas.pref12:33
pmcgowancitrain always fails for me12:33
Mirvphablet-tools-citrain package version needs to be either from wily or the vivid-overlay (20150519-0ubuntu1)12:35
pmcgowanMirv, hmm really? why isnt that package in the sdk tools ppa12:37
pmcgowanall dev packages should be in sdk-team12:37
pmcgowanbzoltan, ^^12:38
Mirvpmcgowan: right, good point. it seems it's there for vivid, but the trusty version is too old.12:40
Mirvthe SDK PPA's vivid version is 15.04.20150722-0ubuntu112:40
pmcgowanthe vivid version is old as well12:40
Mirvlet's try to building it for trusty tooo12:40
Mirvit's from July, the pinning was added in May and is visible in its changelog so it does have it12:41
pmcgowanok12:41
pmcgowanbut lets keep it in sync, not sure what we have to do but this keeps happening12:41
Mirvright12:41
pmcgowanphablet-tools in general not up to date in the ppa12:41
Mirvwe'll update it for vivid+trusty to the latest12:42
Mirvok vivid is the latest (same contents as wily, newer date). trying to compile for trusty.12:44
Mirvbzoltan: I uploaded 1.1+15.10.20150519-0ubuntu1~trusty1 to https://launchpad.net/~ubuntu-sdk-team/+archive/ubuntu/testing/ - let's copy SDK Release PPA after testing it for a while (made an AP for myself)12:46
=== alan_g|lunch is now known as alan_g
bzoltanMirv: pmcgowan:  it is about staging those packages... I am testing those packages and usually I wait for big landings to test the whole chain before copying over. We were burned so many times by these tools.13:41
bzoltanpmcgowan :  Mirv: the phablet tools are build from their trunk by a recipe in the Dev Tools PPA13:42
pmcgowanbzoltan, what do we need to do to ensure sdk ppa always has latest phablet tools13:44
bzoltanpmcgowan: If you want to to an automatic landing of the phablet-tools on each and every developers machine whenever the lp:phablet-tools trunk is updated then it should have a test plan and it should be tested from a landing silo and/or a staging repository just as any other component what is landed.13:48
bzoltanpmcgowan: Without it there will be a sometimes not so comfortable gap between the lp:phablet-tools trunk and the SDK PPA13:48
pmcgowanbzoltan, I just want it fixed of course13:49
bzoltanpmcgowan: everybody wants quick release ... but when a non tested component burns developers then we suffer.13:50
pmcgowanbzoltan, so i agree, lets control it but lets actually get the update13:50
bzoltanpmcgowan: +1 i am checking the lp project and the staging ppa right now13:52
LaneyMirv: how do I train something to the overlay ppa that is already in wily?13:52
Laneyan empty commit or something?13:53
bzoltanMirv: where the phablet-tools is coming from actually? I see this https://code.launchpad.net/~phablet-team/phablet-tools/trunk no recent update13:54
Laneycan I use sync source for this?14:44
* Laney is bold14:46
anpokcihelp: is our mir ci job mir-clang-vivid-amd64 using the stable-phone overlay?14:46
* Laney got it wrong14:47
fgintheranpok, no, it does not14:50
Laneyok I don't know how to do this :)14:50
Laneyrobru: he;lp14:50
anpokfginther: i just realized that also mir-mediumtests-builder-vivid-armhf needs the ppa to successfully build14:50
anpokhttps://code.launchpad.net/~andreas-pokorny/mir/libinput-platform/+merge/270313 this is the MP I am trying to get built14:51
anpokwhich relies on a libinput version only in vivid+overlay14:51
anpokhm maybe I should try getting it through a SRU?14:52
Mirvbzoltan: there's no development since May, it's the correct trunk14:54
fgintheranpok, that might be best. In the past, I thought the goal was to keep mir buildable w/o the overlay PPA, but I could be wrong. This is entirely up to the mir team.14:55
MirvLaney: either a new no-op dual landing or a vivid only landing with the sync-from-wily feature (examples in wiki)14:57
Mirvno-op eg an empty MP14:57
LaneyMirv: I'm trying to sync14:58
anpokfginther: I will discuss in a few minutes.. I thought since we are currently targeting vivid+overlay and wily adding the PPA is the right way to go.14:58
Mirvbzoltan: trusty phablet-tools hadn't been updated for ages, I think that was the only omission14:58
Laneybut I got a KeyError, help meeeeeeeeeeeeeeeeeeeeeeeeeeeee :)14:58
Laneyhttps://requests.ci-train.ubuntu.com/#/ticket/34414:58
Laneythere are two normal uploads (dput, already there) and one which could be synced14:59
Laneythat's lxc-android-config14:59
MirvLaney: sorry, I'm off already and doing this on my Bq and it seems our Bileto is not very optimized yet for our browser. please check the syntax from wiki, sync:ubuntu,wily lxc-android-config or some such15:04
* Mirv goes back to eating pizza15:04
LaneyMirv: ok I'm waiting for someone else15:04
Mirvrobert should be here in 1h or so15:04
LaneyI think I used the syntax the wiki says15:05
Laneybut I don't know if it supports only syncing some things15:05
Laneyenjoy pizza15:05
Mirvrobru: bileto should work on Ubuntu phone :)15:05
balloons_ping cihelp. Can I get utopic jobs disabled on ubuntu-docviewer-app-reboot-ci and ubuntu-docviewer-app-reboot-ci-autolanding?15:16
josephtballoons_: I'll take care of that for you15:24
anpokfginther: ok discussed with mir team - we only target vivid+overlay and wily15:43
anpokso we want those jobs to include the stable phone ppa15:43
fgintheranpok, will do. Should have this finished today.15:44
fgintheranpok, do you have an MP that depends on this work?15:44
anpokyes15:45
anpokhttps://code.launchpad.net/~andreas-pokorny/mir/libinput-platform/+merge/27031315:45
robruLaney: "sync:" is spreadsheet talk, you don't need to specify "sync:" in the sync field, it's redundant.15:50
robruLaney: does the wiki still say to put in "sync:"? I thought i got rid of that15:50
Laneyrobru: https://wiki.ubuntu.com/citrain/LandingProcess#Syncing_from_and_to_the_Stable_Overlay_PPA15:50
robruLaney: fixed, try building.15:53
Laneyrobru: can I say to just sync one thing?15:53
Laneylike if I put it in the packages to rebuild field15:53
robruLaney: yes, "PACKAGES_TO_REBUILD" is a misnomer, it means "act upon only these packages"15:54
LaneyI guess it existed before syncs were a thing15:54
robruyes15:54
Laneyalright, this seems to be doing thing which is good thing15:55
Laneythanks!15:55
robruLaney: you're welcome15:55
Laneyhm15:56
Laneyit didn't use a different version15:56
Laneyoh no lies15:56
josephtballoons_: this is done!16:00
* balloons_ looks16:00
balloons_k, trying a build to check16:01
=== balloons_ is now known as balloons
robrukenvandine: hey, you around?16:53
kenvandinerobru, yeah, what's up?16:53
robrukenvandine: can you run this job for me? https://ci-train.staging.ubuntu.com/job/ubuntu-landing-004-2-publish/parambuild/?IGNORE_VERSIONDESTINATION=true&DEBUG=true&ACK_PACKAGING=true (don't worry, staging area doesn't have permission to do anything)16:54
robruI'm debugging some train stuff16:54
kenvandineok16:54
kenvandine404 error16:54
robrukenvandine: ugh, sorry, log in & then go to the link16:55
robrukenvandine: log in at https://ci-train.staging.ubuntu.com16:55
robrukenvandine: and make sure your teams are checked16:55
robruit won't 404 after you log in16:56
kenvandinehttps://ci-train.staging.ubuntu.com/job/ubuntu-landing-004-2-publish/6/console16:56
robruthanks!16:57
kenvandinenp16:57
robruok, I gotta run to the doctor, should be back in 2 hours or so!17:02
robrukenvandine: can I get a packaging ACK? https://ci-train.ubuntu.com/job/ubuntu-landing-029-2-publish/42/ (and if you do publish, please enable DEBUG)18:24
Trevinhorobru: what's up with AP legacy?18:25
robruTrevinho: I tried to publish your silo but I can't because there's packaging changes18:25
Trevinhorobru: mh, I just applied a patch that in debian/patches with no sense right now...18:25
TrevinhoI could even put it back where it was, but I don't see the reasen18:26
Trevinhoreason*18:26
robruTrevinho: do you or do you not want your silo published right now?18:26
* kenvandine is looking at it18:26
Trevinhorobru: well, if we can...18:26
Trevinhokenvandine: thanks18:27
robruTrevinho: ok but did you just say you were still working on it? Because "Publish without QA" means "publish now"18:27
Trevinhorobru: no, I wansn't working on it.. it was ready to publish to me18:27
Trevinhorobru: in case you need it to be changed, I could, but I'd prefer not18:27
robruTrevinho: well ken has the right to refuse your packaging if he finds a mistake. but I don't personally care.18:28
TrevinhoI see... I've noticed the debian changelog entry for the patch is wrong though18:28
robruTrevinho: I didn't really understand what you said about applying a patch that doesn't make sense. so the silo in it's current state has a patch that doesn't make sense, but isn't worth removing?18:29
robruthat seems like a weird thing to want to publish18:29
Trevinhorobru: no, maybe I've not been clear. there was a debian/patch that it didn't make sense to be there as downstream patch. So I've just applied it upstream.18:29
robruTrevinho: oh, well that's good then18:30
kenvandineit's fine18:30
robruTrevinho: thanks for clarifying18:30
Trevinhoyeah, sorry... I thought it was clear by diffs but maybe not :)18:30
robrukenvandine: please publish with DEBUG, I need DEBUG on to profile the performance of the job18:30
kenvandinedone with debug :)18:31
Trevinhoyay18:31
robrukenvandine: thanks18:32
kenvandine2015-09-11 18:32:36,303 ERROR https://code.launchpad.net/~3v1n0/autopilot/badwindow-errors-protect-legacy/+merge/26899518:32
kenvandine2015-09-11 18:32:36,304 ERROR Some merges have unbuilt revisions.18:32
kenvandineugh18:32
kenvandineTrevinho, silo needs a rebuild18:33
Trevinhoouch18:33
TrevinhoI thought I had18:33
kenvandinemake sure you list autopilot in the list of packages to rebuild18:33
kenvandinethat's the only one that needs a rebuild18:33
Trevinhoyeah, I've done that... I thought I had done this morning even, but maybe I added something else. :/18:34
robrubah18:34
kenvandinegive me a shout when it needs to be published18:34
Trevinhok, thanks18:35
robrukenvandine: due to the publish failure I didn't get the info I needed from that debug log, please also set DEBUG next time you publish too18:35
kenvandinerobru, will do18:36
robrukenvandine: thanks18:36
Trevinhokenvandine: ah... might that be because the approved revision is an old one?18:36
kenvandinei don't think so18:37
kenvandinelooks like the revision built in the silo doesn't match the latest in that one branch18:37
Trevinhook, ok.. that's fine.. It's rebuilding now, it should take few minutes18:38
Trevinhokenvandine: ^18:54
kenvandinepublishing with debug https://ci-train.ubuntu.com/job/ubuntu-landing-029-2-publish/44/console18:55
kenvandinerobru, ^^18:55
robrukenvandine: thanks a bunch18:55
kenvandinenp18:59
oSoMoNtrainguards: hey, I need help with silo 35, the train refuses to do a source copy from https://launchpad.net/~ubuntu-mozilla-security/+archive/ubuntu/ppa/+packages because of the custom version, can you do a manual source copy?20:06
oSoMoN(detailed failure: https://ci-train.ubuntu.com/job/ubuntu-landing-035-1-build/48/console)20:06
robruoSoMoN: sure one sec20:06
robruoSoMoN: wait, you want a source copy? not a binary copy?20:06
oSoMoNrobru, yes, because oxide-qt links against libmedia-hub, and there are ABI changes in the version in the overlay PPA20:07
robruoSoMoN: good to know20:07
robruoSoMoN: just to confirm you want 1.9.2-0ubuntu0.15.04.1 copied?20:08
robruof oxide-qt20:08
oSoMoNrobru, correct20:08
robruok20:08
oSoMoNI don’t mind if the train adds a custom suffix to the version number20:09
robruoSoMoN: manual PPA copy like this won't change the version number.20:09
oSoMoNok, even better then :)20:09
robruoSoMoN: please change your request to a manual source (drop the sync request) and do a WATCH_ONLY build.20:09
oSoMoNlaunched: https://ci-train.ubuntu.com/job/ubuntu-landing-035-1-build/49/console20:10
robruoSoMoN: great, thanks.20:11
oSoMoNrobru, thanks for your help20:14
robruoSoMoN: you're welcome20:14
jdstrandoSoMoN: I'm not sure it is even better-- having two differently built and linked packages with the same version number may be an issue long term. not saying we should block this upload, but food for thought20:15
robrujdstrand: "it's just the overlay PPA". this rebuilt thing won't be in ubuntu archive.20:41
jdstrandsure20:41
robrujdstrand: but generally I agree with the premise that two different builds with same version number is bad.20:41
jdstrandit just adds an additional qualifying question20:42
jdstrand"what version are you running? oh, and on what system?"20:42
jdstrandanyway, not trying to block, just stating a preference20:42
robrujdstrand: well, apt-cache policy will show where it's from, which will answer the question as well20:42
jdstrandmaybe-- that would depend on the ordering of sources.list if -security and the overlay were enabled in sources.list20:43
robrujdstrand: yes, overlay ppa shows up in sources.list on the phones. it's also pinned higher than everything else, so policy reports that that's where packages come from20:51
oSoMoNjdstrand, good point20:59

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!