[01:18] dobey: not a clue [01:18] Saviq: your unity-mir issue should be fixed [01:19] cjohnston, may I ask what's the resolution? [01:21] Saviq: the workaround provided by didrocks [01:21] not using the ppa [01:22] cjohnston, ok yeah, still need mir to be published, since the unity-mir built in the mbs / local upstream merger repo already has unity-mir built against mir 0.1.2 [01:22] cjohnston, that will help us long-term, though [01:23] Saviq: I don't think I have the ability to do anything with that [01:23] cjohnston, yeah, for that we'll just have to wait, thanks [01:23] and I believe it still requires code work on the Mir side [01:29] cjohnston, yeah, I'm just building / testing the whole mir stack to confirm it's working [03:25] tedg, ogra_, et al.: I'm using r34 and still experience bug 1253703 [03:25] bug 1253703 in Upstart Application Launcher "[regression] Webbrowser tries to open file:///usr/share/webbrowser-app/'URL' instead of the URL directly" [Critical,Fix committed] https://launchpad.net/bugs/1253703 [08:45] ogra_: hey, seeing you online, mind kicking a new image? [08:45] === Image r35 building === [08:46] wow [08:46] r34 on mako looks *really* bad [08:46] and only 3 packages changed ... [08:46] yeah [08:47] not sure if it's just because of "not enough relaunch" or something else [08:47] let's wait on psivaa :) [08:47] well, maguro improved ... [08:47] ogra_: I guess webbrowser-app is for a lot [08:47] (on mako) [08:48] basically half of the failures [08:48] yeah [08:48] and notes is stuck once again [08:48] I bet something horrible happened [08:48] yeah [08:49] I think psivaa needs to talk to Osomon [08:49] as Bill is on holidays [08:49] so that this is looked at [08:49] i wonder if it is related to the bug above [08:58] morning [08:59] didrocks: ogra_ : let me take a look at mako runs [08:59] hey psivaa! thanks === psivaa changed the topic of #ubuntu-ci-eng to: Ubuntu CI Engineering Team | Vanguard: psivaa | Landing instructions: http://paste.ubuntu.com/6292280/ | Known issues: http://goo.gl/DStLaS [09:24] ogra_: browser app sounds good to me on mako [09:24] just trying locally [09:26] do you mean regarding the bug or regarding the tests ? [09:28] regarding the tests [09:28] I'm trying to reproduce the same here [09:29] :q [09:29] Damn, wrong window [09:29] hey sil2100 :) [09:30] landing call? [09:30] \o/ [09:30] Hello! [09:36] bug 1253703 [09:36] bug 1253703 in upstart-app-launch (Ubuntu) "[regression] Webbrowser tries to open file:///usr/share/webbrowser-app/'URL' instead of the URL directly" [Undecided,New] https://launchpad.net/bugs/1253703 [09:40] http://people.canonical.com/~ogra/touch-image-stats/20131128.changes [09:40] changes of r35 ^^^ [09:53] === Image r35 DONE === [10:01] didrocks: ogra_: could not complete the notes app test on mako with r34 because the tests have started with r35 [10:01] psivaa: ok, let's take that one as the base [10:01] didrocks: ack [10:49] psivaa, hey, we need to enable the daily-build ppa for a merge on unity8 and unity-mir, I've added the hook when starting the job, but it doesn't seem like it's executed http://s-jenkins.ubuntu-ci:8080/job/unity-mir-trusty-armhf-ci/30/console ? [10:49] ah builder_hooks? [10:49] Saviq: let me take a look [10:50] hmm no, that's not a parameter that I have access too [10:50] -o [10:58] psivaa, cjohnston changed the default configuration of the unity-mir- jobs to not include the daily-build ppa, maybe that's of interest (bug #1255578) [10:58] bug 1255578 in Ubuntu CI Services "dependency issues on libunity-mir1 in testrunner-otto" [Undecided,Confirmed] https://launchpad.net/bugs/1255578 [11:00] Saviq: ok, this is for otto run. not sure how that impacts here. [11:01] psivaa, well, otto run is affected, the hooks were global on both -ci and -autolanding jobs [11:01] psivaa, i.e. unity-mir got built against daily-build, so then no job in the unity8 stack, that doesn't include the daily-build ppa, would fail [11:12] vila: would you mind helping me out on the Saviq's req please. fginther being on holiday i dont want to make a change that i dont fully understand :) [11:15] psivaa: not sure I know more than cjohnston there, I've been mostly involved in [re]setting otto nodes rather than the cu2d part that drive them. [11:15] so, [11:16] my understanding was that a wrong package version was put in a ppa (is that correct ? which ppa ?) [11:17] and that cjohnston's MP will do the right thing by *not* referring to that ppa [11:17] Saviq: ^ [11:17] vila, no [11:17] vila, the problem was: unity-mir-autolanding ran against ppa:ubuntu-unity/daily-build - by default [11:18] vila, the resulting package got put in the mbs repo for unity8 stack [11:18] mbs ? [11:18] vila, those are local repositories built per stack from -autolanding jobs [11:19] * vila nods [11:19] vila, so whatever was trying to use that mbs repo and install libunity-mir1 for that, would fail if it didn't also have ppa:ubuntu-unity/daily-build enabled [11:19] vila, hence the bug [11:20] Saviq: oh, the other way around then ? An additional ppa needs to be added ? [11:20] vila, the solution applied (correctly) was that unity-mir-ci nor -autolanding would run against daily-build [11:20] Saviq: lacking context, I'm trying to understand what the fix should be, cleaning a ppa ? Adding a ppa as a parameter somewhere ? Rebuild something ? [11:21] bumping a revision to override the mess ? ;) [11:21] vila, misunderstanding [11:21] vila, the above bug is fixed long-term [11:21] Saviq: how ? [11:21] vila, by removing the ppa from unity-mir-ci and -autolanding jobs [11:21] in cu2d-config for that stack ? [11:22] vila, I think so, yes [11:22] Saviq: ok, progress, that was cjohnston work and MP right ? [11:22] vila, yes [11:22] ok, at least I've followed some bits [11:22] vila, now, I triggered http://s-jenkins.ubuntu-ci:8080/job/unity-mir-ci/169/rebuild/? [11:22] vila, with *manually added* daily-build ppa [11:22] vila, i.e. D09add_ppa~ubuntu-unity~daily-build [11:23] Saviq: as a one-off or should that become the default ? [11:23] vila, one-off [11:23] because we need to transition to mir 0.1.2, which is only daily-build [11:23] vila, and here comes the issue: [11:23] vila, the hook wasn't executed it seems [11:23] vila, http://s-jenkins.ubuntu-ci:8080/job/unity-mir-trusty-armhf-ci/30/console [11:24] O_o [11:24] vila, it's there in hooks=D09add_ppa~ubuntu-unity~daily-build [11:24] vila, but not in Pbuilder_hooks any more [11:24] and there's basically no mention of it past the parameter printing [11:25] http://s-jenkins.ubuntu-ci:8080/job/unity-mir-trusty-armhf-ci/30/parameters/? mentions it right [11:25] vila, as opposed to http://s-jenkins.ubuntu-ci:8080/job/unity-mir-trusty-armhf-ci/27/console - which was yesterday, and had the same hook by default [11:25] vila, yes, it's even mentioned in the console output at the top [11:25] vila, which leads me to think the hook got removed completely, and was hence ignored? [11:25] right, I can see that (thanks) [11:25] Saviq: removed from what ? [11:26] vila, from wherever the hooks are picked up from [11:26] vila, I mean the hook as in the actual script === MacSlow is now known as MacSlow|lunch [11:26] vila, that's just a suspicion, no reason to believe me :) [11:26] hmm, deployment issue you mean ? (As in, the jobs are correct but rely on stuff installed on the slaves) [11:26] ha [11:27] psivaa: did you look at the playbook ? [11:28] vila: hmm no, sorry my bad [11:28] psivaa: oh, now, just wanted to check if you knew something I didn't ;) [11:29] vila, psivaa, going for food, biab or ping me if needed [11:29] Saviq: Disable unity-mir use_stack_ppa until Mir transitions their ABI. rings a bell ? [11:29] Saviq: that's the last config that has been deployed http://s-jenkins.ubuntu-ci:8080/job/deploy-cupstream2distro-config/113/ [11:30] Saviq: ha, hmm, well, I'd better do the same to maximize the overlapping then ;) [11:30] psivaa: when do you intend to lunch ? ;) [11:32] vila: normally in about 90 mins.. but could move it to suite :) [11:32] psivaa: up to you, if you prefer to keep digging that one, just leave notes here [11:32] vila, well, it's not "until", but it should just not be enabled by default - daily-build is unsafe, as things there might never end up in distro, if they fail QA [11:33] vila: ack, will do a bit more.. [11:33] Saviq: fginther knows about that ? I'm quite surprised this hasn't been encountered in the past and I'd feel better if mir was a special case instead :-/ [11:34] Saviq: but let's have lunch first ;) [11:48] Saviq: i added D09add_ppa~ubuntu-unity~daily-build as a builder hook and re-running the job. http://s-jenkins.ubuntu-ci:8080/job/unity-mir-trusty-armhf-ci/31/console [11:48] looks like that's progressing and i have not made any changes to the configs since this is one off [11:57] Saviq: vila: and that run has succeeded. [11:58] psivaa, what's the difference between hooks and builder_hooks? [11:58] psivaa, and how can I supply builder_hooks? [11:58] Saviq: i dont know the difference between them yet, i could findout [11:58] Saviq: builder_hooks was part of the rebuild parameters, [12:01] psivaa, do you have it here http://s-jenkins:8080/job/unity-mir-ci/169/rebuild/? ? [12:01] psivaa, I only see "hooks" [12:01] psivaa, maybe that's the problem then? [12:01] that that parameter isn't "published" [12:02] psivaa, from the names I'd assume hooks was used for otto runner, for example, where builder hooks were used by builders [12:02] psivaa, so the hook basically needs adding in both, but I don't see builder_hooks as a parameter I can modify [12:06] Saviq: yes, i dont see it in the master job. may be builder_hooks is not intended to be a published parameter. [12:06] not sure if there was a reason for that and the builder hooks should not vary between runs [12:07] psivaa, at least some time ago we were able to add hooks to "hooks" parameters and they would get picked up everywhere - for both builders and runners [12:08] psivaa, so something changed there :/ [12:08] psivaa, let me file a bug [12:11] psivaa, bug #1255948 [12:11] bug 1255948 in Ubuntu CI Services "upstream merger hooks do not propagate to child jobs' builder_hooks" [Undecided,New] https://launchpad.net/bugs/1255948 [12:11] Saviq: thanks. looks like the change has happened after 26th. [12:12] psivaa, please let me know if you don't find a reason/solution you're comfortable with, we'll have to force push the two MPs for unity8 and unity-mir [12:14] Saviq: ack [12:22] psivaa: \o/ [12:23] psivaa: all good for the short term ? [12:23] vila: not yet completely, the parameters are not propogated to the child jobs [12:23] * vila thinks a bug is the right thing to do here, this seems like a deeper issue [12:24] psivaa: yup, but you are able to trigger the required jobs to unblock Saviq right ? [12:24] vila, no, you can only trigger the downstream jobs correctly [12:24] vila: yep ^ [12:25] vila, so no way to run the whole -ci or -autolanding job, as there's no way to add the hook to builder_hooks there [12:26] Saviq: I got that, hence the bug. How many job triggers do you need ? 1, 2 ? Or is this something you need for the coming days, weeks ? [12:26] vila, 2 MPs [12:26] Saviq: fginther is in vacations until next Monday and I'm far from sure we can fix that properly until then... [12:26] vila, so we'll just merge manually [12:27] Saviq: right, so manual workaround should get you out of trouble right ? [12:27] vila, yes [12:31] * vila blinks [12:31] Saviq: hmm, how about hooks: H05set_package_version D00mbs_archive A10checklicenseheaders [12:32] vila, what about it? that's the default hooks? [12:32] Saviq: tha'ts in cu2d-config/stacks/unity8.cfg with some different ones for unity-api and unity-mir [12:32] Saviq: and that doesn't match your needs ? [12:33] vila, default is fine [12:34] vila, problem was with one-off changes [12:34] s/was/is/ [12:34] vila, when you want to override the defaults for whatever reason [12:34] vila, you can't - 'cause the "hooks" param does not propagate to "builder_hooks" [12:34] so, in effect you can't override builder_hooks [12:35] Saviq: ok, let's wait for fginther , was just having a look at the config in case it provides a different workaround but that would require two deployments, not good for 2 landings ;) [12:36] vila, yeah, doesn't make sense to deploy just for this [12:37] ogra_: that's not really accurate [12:37] " [12:37] As some might have noticed we recently had a few bad image releases into [12:37] the Trusty channel that contained regressions. " [12:37] we landed one images with some regressions [12:37] then, we only get images to fix some of them [12:37] three if i didnt miscount [12:38] they didn't add more regressions [12:38] (two of them to fix the rehgressions but still knowingly shipping others) [12:38] ogra_: shipping new regressions? [12:38] no [12:38] ogra_: that's what you are implying [12:38] in your email [12:39] we had trhee regressions and released one image that still had two and one image that still had one [12:39] which makes multiple images with regressions [12:39] (and we had regressed images before, its just one current example) [12:40] ogra_: hum, "a few images with regressions" sound like we have new images promoted regressing previous promoted image [12:40] but thats not what i said [12:40] that's how it can be read :) [12:40] didrocks, let's not focus on unprecise wording, that's not the main point of that email [12:40] seb128: still, I think that set a false perception [12:41] but anyway [12:41] yeah, it is [12:41] my main point is that i want public meetings with community testers participating ... and that i want to take load off the landing team [12:41] not worth arguing over though, it's written/send and hopefully a detail people are not going to stop on [12:41] let's see… [12:42] and if that isnt coming out clear i'll correct this in a followupü [12:42] * didrocks waits now on managers reaction "why did we land multiple images with regressions" [12:42] well, that will be be direct pings I guess [12:42] because we didnt wait with releaseing until all regressions were fixed [12:42] ogra_: which was a good thing, right? [12:42] but fixed them one by one [12:42] or we will still wait [12:43] not sure ... i would presonally have waited but i didnt want you to have even more pressure from above [12:43] (which is what a community based team should solve as well) [12:43] (as long as there are clear policies) [12:43] let's see === MacSlow|lunch is now known as MacSlow [12:46] in any case i think the release process deserves a lot more attention ... and i also think the landing team doesnt need to have 2 2h meetings just because of that ... which is why i propose a new team with wider participation and more focus [12:47] ogra_: 2h meetings? [12:48] most of our meetings are under 20 minutes, if not 15 [12:48] didrocks, going through the buglist one by one, collecting info from more testers and other teams etc ... [12:48] that would add up [12:48] ah, you mean, if we collected [12:48] right [12:49] we have a huge community and i bet we can get a ton of people to participate more in testing if we would give them a voice in a public meeting [12:49] these two points make me suggest a new team [12:50] (and the hope that it frees up landing resources) [13:05] Saviq: vila: i could not modify the config of the master job to pass the new parameter with rebuilds.. but the child jobs pass individually when the parameter is passed to them [13:06] so if you want to force merge the MP, i have no objections [13:06] psivaa, yeah, I will just trigger generic-land manually [13:06] psivaa, once mir gets published, we're back on track [13:07] psivaa, bug still needs fixing, but we're unblocked at least [13:07] but i've assigned the bug to fginther, ill subscribe to it/ [13:07] sorry could not help more [13:07] s/but// === psivaa changed the topic of #ubuntu-ci-eng to: Ubuntu CI Engineering Team | Vanguard: use 'cihelp' | Landing instructions: http://paste.ubuntu.com/6292280/ | Known issues: http://goo.gl/DStLaS === alan_g is now known as alan_g|lunch [13:15] * didrocks goes for a run === psivaa is now known as psivaa-lunch === alan_g|lunch is now known as alan_g [14:42] * ogra_ files bug 1255999 [14:42] bug 1255999 in ubuntu-keyboard (Ubuntu) "[regression] testing on mako with image r32 and maguro with r34 the keyboard does not come up when an input field is focused" [High,New] https://launchpad.net/bugs/1255999 === psivaa-lunch is now known as psivaa [14:47] ogra_: let me upgrade and reproduce [14:47] ogra_: can't reproduce on image r32 on mako FYI [14:48] how long did you run it yet [14:48] ogra_: bonus points for triple 9 ? [14:48] ogra_: since this morning [14:48] I'm still on image 32, didn't reboot [14:49] ogra_: and you don't tell it's not from the start [14:49] ogra_: can't reproduce in messaging (nor anything I tried since I installed r34) [14:49] didrocks, well, on my mako it runs since we promoted it ... [14:49] ogra_: but I've seen it here on there for ages [14:49] ogra_: so, maybe you should tell that it's after a while :) [14:49] and try to find a reproducer :p [14:49] ogra_: did you switch your language? [14:49] well, i see it on two devices [14:49] the maguro was updated this morning [14:49] ogra_: which app ? [14:50] any [14:50] settings app, searching in the shell [14:50] browser too [14:50] shell and browser ok here [14:51] is your device properly set up in french and all ? [14:51] ogra_: can you tell it's after a while and not on startup? Otherwise, people will close it with not reproduceable here [14:51] ogra_: I get that in French for a while, it's in english right now [14:51] settings ok too :-/ [14:52] yep, just rebooted the maguro ... no kbd coming up [14:53] let me reboot the mako [14:53] ogra_: I've subscribed to the bug (and commented), I won't call it a regression though [14:54] i didnt have that issue every with image 10 [14:54] *ever [14:55] ogra_: I've always seen that randomly, I can ensure you [14:55] so to me thats clearly a regression [14:55] even with 1.0 [14:55] maybe it's more often now [14:55] but would be interesting to know exactly how you trigger it [14:55] so that upstream can fix [14:55] i didnt do anything special [14:56] I don't do anything special and have it [14:56] ogra_: is your system in German? [14:56] my mako was ugraded to 32 when we released that ... i must admit i didnt use it much since ... but it constantly ran ... today i tried to search in it for the first time [14:56] yes [14:57] (thats why i asked if you guys use french setup) [14:57] ogra_: ok, yeah, on that one, but it was even before image 10 [14:57] like I saw it on 1.0 [14:57] I reswitched to english since [14:57] i have never had it [14:57] let me try to switch to French [14:57] on a stable image [14:57] I got it, not 100%, but got this [14:58] (i have seen it on -proposed images before for sure ... ) [14:58] ok, switched to French [14:58] no issue [14:58] let me reboot [15:01] (long to boot…) [15:01] ogra_: english setup here [15:01] apport ... [15:01] ogra_: confirming in French [15:01] let me see if there is a crash file [15:02] my maliit log is empty :( [15:02] yeah, no crash either :/ [15:02] maliit-server is running [15:02] here as well [15:04] ogra_: WARNING: no dictionary to turn on spellchecking [15:04] I wonder if that can be it [15:04] oh, where is that from ? [15:04] ogra_: ~phablet/.cache/upstart/maliit-server.log [15:04] do you confirm? [15:04] no [15:04] as i said above [15:04] mine is empty [15:05] root@ubuntu-phablet:/# cat /home/phablet/.cache/upstart/maliit-server.log [15:05] __pthread_gettid -2 [15:05] __pthread_gettid -2 [15:05] __pthread_gettid -2 [15:05] I have that one top [15:05] before the classic: [15:05] WARNING: QOpenGLShader::link: "--From Fragment Shader: [15:05] --From Vertex Shader: [15:05] Link was successful. [15:05] " [15:05] well, thats all i have [15:05] hence "classic" ;) [15:06] no, i mean the above is all thats in there [15:06] only the hybris noise [15:06] ah… [15:06] ogra_: ok, clearly, local-dependant [15:07] right [15:07] ogra_: adding infos to the bug report [15:08] done [15:09] thx [15:10] ogra_: nice catch btw, I think it's not the same bug that vila and I were talking about [15:10] like, sometimes, the keyboard will never ever come up again [15:10] and you have to reboot [15:10] havent had that on the stable one [15:10] my mako explicitly only runs trusty, not proposed [15:10] you were lucky, I was using French all the time, and after 10 minutes of playing, I clearly had that [15:11] wow [15:11] but you also have that dictionary error [15:11] yeah, I wonder if that's really linked or the previous session [15:11] i wonder how german and french langpacks differ here [15:11] do you know about upstart session job? [15:11] well, rm the log and reboot [15:12] like, the logs are reset at each login? [15:12] yeah [15:12] i dont think they are [15:12] doing that, hoping that upstart won't be upset by that [15:12] they are just adding up until logrotate rotates them i think [15:12] ok [15:12] so no handler on the file? [15:12] (done, and rebooting) [15:12] probably while the session runs [15:13] but not before or after [15:13] ok [15:13] ogra_: yeah, that came from an english session I guess [15:13] nothing anymore apart from the libhybris one [15:14] right, so the same as i see [15:14] yep [15:14] so, let's extrapolate anything != english [15:14] that would be easy for them to reproduce [15:14] yeah [15:14] btw, does your clock show a proper 24h time ? [15:15] no, it's 4:15 [15:15] I didn't konw I worked that long :p [15:15] yeah :( [15:15] a recent change made it drop the AM/PM ... [15:16] but the format is still wrong it seems [15:16] right [15:16] indicator-datetime maybe? [15:16] which now makes it look really weird [15:16] we really need to have tests for those [15:16] i had a bug open and just transferred it into a whishlist for session-migration today [15:16] seems i was to fast [15:17] oh! let me look at it [15:18] ogra_: ah, session-migration is just a helper [15:18] bug 1255530 [15:18] bug 1255530 in session-migration (Ubuntu) "/home/phablet/.pam_environment does not get updated on OTA upgrades" [Wishlist,New] https://launchpad.net/bugs/1255530 [15:18] ogra_: it doesnt' contain any script [15:18] so indicator-datetime contains the script and use session-migration [15:18] it uses scripted backends, doesnt it ? [15:18] yeah, and those are just scripts :) [15:18] right [15:19] so, indicator-datetime just ship a script [15:19] call dh_migration [15:19] well, that wont help the above case [15:19] unless pam starts shipping such a script [15:19] s/case/bug/ [15:20] yeah, that's the idea [15:20] hum, btw, session-migration won't work on the phone, that's the issue [15:20] we need to hook it into upstart [15:20] we have it installed :) [15:20] (it was hooked in gnome-session) [15:20] I doesn't do a lot I guess then :p [15:21] well, we should have another bug for that one then ;) [15:21] yep ;) [15:21] ogra_: we should rewrite it in Go as well ;) [15:22] ah no, it's C, it's fine :p [15:22] i dont think it is overly important right now [15:22] (with, hem, perl, for debhelper) [15:22] but we should look at it before the next stable goes out [15:22] (before release) [15:22] so that such bits are catched when people do stable -> stable upgrades [15:24] yeah [15:25] ogra_: but soon systemd will implement it ;) [15:25] shudder [15:26] ogra_: if you don't mind, can you open the other bug so that I hook up session-migration into upstart? [15:28] ok [15:33] hello [15:34] could someone please update the "Proposed" link on http://ci.ubuntu.com/ for s/Saucy/Trusty/ ? [15:35] cihelp, ^ [15:39]