[00:53:57] (03CR) 10Dzahn: [C: 04-1] "apache-config repo has been abandoned per file README_BEFORE_EDITING , please abandon or convert to a change in operations/puppet" [operations/apache-config] (betacluster) - 10https://gerrit.wikimedia.org/r/148741 (owner: 10Ori.livneh) [00:57:01] (03Abandoned) 10Ori.livneh: mod_proxy_balancer configs [operations/apache-config] (betacluster) - 10https://gerrit.wikimedia.org/r/148741 (owner: 10Ori.livneh) [04:45:40] 3Wikimedia / 3Continuous integration: remove all tab characters from DNS repository - 10https://bugzilla.wikimedia.org/68769#c8 (10Daniel Zahn) https://gerrit.wikimedia.org/r/143208 https://gerrit.wikimedia.org/r/143212 https://gerrit.wikimedia.org/r/148437 https://gerrit.wikimedia.org/r/150067 https://gerri... [06:34:06] Hmpf. It seems that to finish my unit tests I'll really need to destroy my instance and create one with 4 GB RAM or more [07:55:53] [32447.524163] Out of memory: Kill process 9563 (hhvm) score 778 or sacrifice child [07:55:56] [32447.525426] Killed process 9563 (hhvm) total-vm:6140528kB, anon-rss:1531916kB, file-rss:1272kB [08:11:30] Well, better than php5 which hung [08:37:49] hello [08:58:06] (03CR) 10Zfilipin: The first job for Math extension (032 comments) [integration/jenkins-job-builder-config] (cloudbees) - 10https://gerrit.wikimedia.org/r/149323 (https://bugzilla.wikimedia.org/67221) (owner: 10Cmcmahon) [08:58:12] (03PS4) 10Zfilipin: The first job for Math extension [integration/jenkins-job-builder-config] (cloudbees) - 10https://gerrit.wikimedia.org/r/149323 (https://bugzilla.wikimedia.org/67221) (owner: 10Cmcmahon) [08:58:15] (03PS5) 10Zfilipin: The first job for Math extension [integration/jenkins-job-builder-config] (cloudbees) - 10https://gerrit.wikimedia.org/r/149323 (https://bugzilla.wikimedia.org/67221) (owner: 10Cmcmahon) [09:12:12] hashar: around? [09:12:27] do you have a minute (literally) for jjb question? [09:17:55] zeljkof: just ask! :-D [09:19:07] hashar: I already did once or twice, but then I forgot the check irc if you have replied :) [09:19:11] ok, here it comes [09:20:09] so, when updating jjb (jenkins-jobs) itself, is is enough to just update the jjb repo? [09:20:18] or should this also be done? [09:20:18] sudo python setup.py install [09:20:23] https://github.com/zeljkofilipin/dotfiles/blob/master/.mrconfig#L40 [09:21:46] hmm [09:21:58] no clue what .mrconfig is [09:22:06] but yeah basically: [09:22:08] hashar: ignore that [09:22:16] git pull the repository integration/jenkins-job-builder.git [09:22:17] mr is multiple repository tool [09:22:18] and pip install it [09:22:33] I update integration/jenkins-job-builder.git from time to time [09:22:41] when I am sure it is not going to cause any issue with our jobs [09:22:46] there are sometime regression in upstream repo [09:23:45] mr helps you update multiple repositories [09:23:48] great tool [09:23:51] but anyway [09:24:10] hashar: I think I have installed jjb following instructions here [09:24:11] http://ci.openstack.org/jenkins-job-builder/installation.html [09:24:21] with: sudo python setup.py install [09:24:37] now I see that our docs say https://www.mediawiki.org/wiki/Continuous_integration/Jenkins_job_builder [09:24:43] yeah [09:24:45] sudo pip install -e . [09:24:48] if you use our repo that will be fine [09:25:01] pip install -e adds some virtual definitions in your python modules repository [09:25:07] which points the module to your local copy [09:25:10] so you can edit your local copy [09:25:17] and other soft will be made to use that local copy [09:25:31] -e stands for edit [09:25:44] hashar: so, pip install is the preferred way? [09:26:06] :( https://wikitech.wikimedia.org/w/index.php?title=Labs-vagrant&diff=121700&oldid=120431 [09:32:41] 3Wikimedia / 3Continuous integration: Have unit tests of all wmf deployed extensions pass when installed together, in both PHP-Zend and HHVM (tracking) - 10https://bugzilla.wikimedia.org/67216 (10Antoine "hashar" Musso) [09:35:23] Other than that, last hhvm seems to run full test suite beyond the point where it crashed till yesterday, so it's good :) [09:36:37] (03Abandoned) 10Phuedx: Add browser tests for the GettingStarted and GuidedTour extensions [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/144456 (owner: 10Phuedx) [09:46:27] (03PS1) 10Zfilipin: WikidataBrowserTests now has tests in tests/browser [selenium] - 10https://gerrit.wikimedia.org/r/150170 [09:56:09] (03PS1) 10Phuedx: Enable browser tests for the GettingStarted extension [integration/jenkins-job-builder-config] (cloudbees) - 10https://gerrit.wikimedia.org/r/150172 [09:56:59] (03CR) 10Phuedx: [C: 04-2] "This change is blocked on I79d9468e56bc7941c7c5bb27a77d1622bc7d3836." [integration/jenkins-job-builder-config] (cloudbees) - 10https://gerrit.wikimedia.org/r/150172 (owner: 10Phuedx) [10:10:17] zeljkof: yeah pip install is prefered way [10:10:29] hashar: thanks [10:10:35] zeljkof: I use pip install -e ~/projectclones/integration/jenkins-job-builder [10:10:40] this way when I change some python files there [10:10:52] the command jenkins-jobs points to that dir [10:11:00] pip install . [10:11:02] creates a copy [10:18:13] (03PS1) 10Zfilipin: Deleting all qa/browsertests jobs [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/150174 [10:32:32] (03PS1) 10Zfilipin: Running Ruby linter for ArticleFeedbackv5 [integration/zuul-config] - 10https://gerrit.wikimedia.org/r/150177 [10:36:21] hashar: is there a reason several extensions are listed two or more times here? https://github.com/wikimedia/integration-jenkins-job-builder-config/blob/master/mediawiki-extensions.yaml [10:36:24] example [10:36:24] https://github.com/wikimedia/integration-jenkins-job-builder-config/blob/master/mediawiki-extensions.yaml#L376-L377 [10:36:34] https://github.com/wikimedia/integration-jenkins-job-builder-config/blob/master/mediawiki-extensions.yaml#L387-L393 [10:49:37] It was a good thing that I created a m1.large instance this time. Puppet is happily taking all 8 GB of memory (for cache) :D [10:58:18] (03PS1) 10Zfilipin: Deleted duplicated lines with extension names [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/150179 [10:59:52] zeljkof: seems like a bug :-D [10:59:55] git blame would know [11:00:07] hashar: https://gerrit.wikimedia.org/r/#/c/150179/ [11:00:16] zeljkof: https://github.com/wikimedia/integration-jenkins-job-builder-config/commit/f43c2c0ba92ded429f0fec8bf6991260356855a0 :D [11:00:55] \O/ [11:00:56] (03CR) 10Hashar: [C: 032] "Good cleanup, thanks!" [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/150179 (owner: 10Zfilipin) [11:02:09] (03Merged) 10jenkins-bot: Deleted duplicated lines with extension names [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/150179 (owner: 10Zfilipin) [11:06:27] 3Wikimedia / 3Quality Assurance: Move Selenium tests from browsertests repository to the repository they test - 10https://bugzilla.wikimedia.org/62144#c2 (10Željko Filipin) 5NEW>3RESO/FIX qa/browsertests repository is now empty. [11:08:11] 3Wikimedia / 3Quality Assurance: Ruby linter should run after patch set submission to every repo that has Ruby code - 10https://bugzilla.wikimedia.org/63469 (10Željko Filipin) 5NEW>3ASSI a:3Željko Filipin [11:11:53] (03PS2) 10Zfilipin: Deleting all qa/browsertests jobs [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/150174 [11:12:28] (03PS3) 10Zfilipin: Deleting all qa/browsertests jobs [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/150174 (https://bugzilla.wikimedia.org/63469) [11:12:57] (03PS2) 10Zfilipin: Running Ruby linter for ArticleFeedbackv5 [integration/zuul-config] - 10https://gerrit.wikimedia.org/r/150177 (https://bugzilla.wikimedia.org/63469) [11:31:08] (03PS1) 10Zfilipin: Created Ruby lint job for ArticleFeedbackv5 [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/150189 (https://bugzilla.wikimedia.org/63469) [11:31:41] (03PS3) 10Zfilipin: Running Ruby linter for ArticleFeedbackv5 [integration/zuul-config] - 10https://gerrit.wikimedia.org/r/150177 (https://bugzilla.wikimedia.org/63469) [11:33:04] hashar: is this the way to run ruby linter for an extension? [11:33:08] https://gerrit.wikimedia.org/r/#/c/150177/3 [11:33:13] https://gerrit.wikimedia.org/r/#/c/150189/1 [11:39:04] * zeljkof is out to lunch [13:21:23] 3Wikimedia / 3Quality Assurance: Merge master and cloudbees branch of integration/jenkins-job-builder-config repository - 10https://bugzilla.wikimedia.org/68801 (10Željko Filipin) 3NEW p:3Unprio s:3normal a:3None cloudbees branch was a step in moving jobs from Cloudbees. Now that the move is finished... [13:34:02] hello zeljkof! thank you merging the commits :) [13:34:09] hi vikasyaligar [13:34:21] I was pairing with aharoni [13:34:35] so he has some credit too :) [13:34:48] did you get the bot approved at production commons? [13:35:56] zeljkof: they have some more questions lined up. One of them was solved with yesterdays commit and I will reply to another one now [13:36:08] vikasyaligar: great [13:36:36] zeljkof: But I am more worried about this bug => https://bugzilla.wikimedia.org/show_bug.cgi?id=68760 [13:37:04] vikasyaligar: hm [13:37:19] I have noticed that today, new images were not uplaoded [13:37:25] you should ping hashar [13:37:31] he will now [13:37:35] maybe even bd808|BUFFER [13:38:00] and let aharoni and James_F know, they might know who to ask [13:39:26] aharoni told me to put it in wikitech mailing list [13:39:54] So I will put it now right away [13:41:51] zeljkof: do you have any advice for asserting a user is returned to a previous page? context: yesterdays pairing session on https://gerrit.wikimedia.org/r/#/c/123873/ [13:42:05] vikasyaligar: go ahead [13:42:08] phuedx: let me see [13:42:30] ta [13:43:10] phuedx: where would you like to check? [13:43:11] here? https://gerrit.wikimedia.org/r/#/c/123873/10/tests/browser/features/step_definitions/editable_page_steps.rb,unified [13:43:24] line 10? [13:43:30] zeljkof: yep, line 10 [13:43:49] phuedx: what is wrong with line 10 now? :) [13:44:00] you would like to make it more generic? [13:44:05] ^ that [13:44:32] what if the wiki we're running the test against doesn't have that page? [13:45:04] ideally an editable page is one that is editable and is in the "all articles needing copy edit" category [13:45:18] or am i thinking about this too much? [13:46:33] phuedx: ^ that :) [13:46:41] ;) [13:46:48] well, long term, the tests should be as robust as possible [13:47:05] and your test assumes the page exists, which is bad [13:47:30] so in the given step, you should actually create a random page, that will make it editable [13:47:42] https://github.com/wikimedia/mediawiki-ruby-api [13:48:06] there is a ruby gem that will make it easy for you to create the page [13:48:40] \o/ [13:48:47] phuedx: ping me if you need help [13:49:01] time for more liberal use of @random_string [13:49:22] phuedx: you can even reuse it for both user and page name, right [13:49:25] ? [13:49:33] yup [13:49:48] hashar: Hello ! did you take a look at this => https://bugzilla.wikimedia.org/show_bug.cgi?id=68760 [13:50:04] and we can make @random_page a proper random string, if needed [13:56:10] (03PS4) 10Zfilipin: Running Ruby linter for all repos that have Ruby code [integration/zuul-config] - 10https://gerrit.wikimedia.org/r/150177 (https://bugzilla.wikimedia.org/63469) [14:01:52] (03PS2) 10Zfilipin: Running Ruby linter for all repos that have Ruby code [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/150189 (https://bugzilla.wikimedia.org/63469) [14:02:36] hashar: when you have a minute, take a look at these [14:02:37] https://gerrit.wikimedia.org/r/#/c/150177/ [14:02:41] https://gerrit.wikimedia.org/r/#/c/150189/ [14:02:48] I am not sure if I am doing it right [14:03:05] (03PS3) 10Zfilipin: WIP Running Ruby linter for all repos that have Ruby code [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/150189 (https://bugzilla.wikimedia.org/63469) [14:03:12] (03PS5) 10Zfilipin: WIP Running Ruby linter for all repos that have Ruby code [integration/zuul-config] - 10https://gerrit.wikimedia.org/r/150177 (https://bugzilla.wikimedia.org/63469) [14:04:36] zeljkof: have you created the jobs in Jenkins? [14:04:57] hashar: did not push anything yet, but I did this https://gerrit.wikimedia.org/r/#/c/150189/ [14:05:18] +2 :D [14:05:22] push jobs! [14:05:33] wait! [14:05:38] hashar: ^ [14:05:45] there is more jobs to come, it is WIP [14:05:48] 00:02:05.182 Only in output-proposed: mwext-ArticleFeedbackv5-ruby1.9.3lint [14:05:49] 00:02:05.182 Only in output-proposed: mwext-Math-ruby1.9.3lint [14:05:49] :-) [14:06:00] ahh [14:06:03] anyway [14:06:05] what you could instead [14:06:10] hashar: did you merge already? [14:06:16] is add the ruby lint template to the Zuul project template extensions-unittest [14:06:20] err [14:06:27] to Jenkins job-group [14:06:39] and generate ruby lint jobs for all extensions :D [14:06:43] that is evil [14:06:49] but that will be created once for all [14:06:57] or [14:07:02] we could even create a single job [14:07:06] hashar: it will not run if the repo does not have ruby files anyway, right? [14:07:06] lint-ruby1.9.3 [14:07:10] and have all repos trigger that job [14:07:15] exactly [14:07:24] hashar: that would be doable, right [14:07:34] Zuul has been instructed to only trigger those jobs if the patch change some file matching *.rb (or something similar) [14:07:40] brb nature's call [14:11:56] zeljkof: this way we would have the job shared among repositories [14:12:01] and will just have to add it in zuul [14:12:09] hashar: yeah [14:12:12] way better [14:12:56] (03CR) 10Hashar: [C: 031] "Or we come up with a common job shared by all repositories. This way we will just have to adjust the Zuul config :-)" [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/150189 (https://bugzilla.wikimedia.org/63469) (owner: 10Zfilipin) [14:13:27] zeljkof: the main reason we have a job per repo [14:13:34] is to let us finely adjust whether they are voting or not [14:13:46] but for syntax check, I guess we always want them to vote [14:13:51] yes [14:14:03] another problem [14:14:10] is if we have several repositories being +2 [14:14:17] that lands in the gate-and-submit pipeline [14:14:28] that will make all jobs being tested as a dependent queue [14:14:37] A -> B -> C -> D -> E [14:14:43] so if something fail on A everything is retriggered [14:14:49] hashar: [14:14:51] hm [14:14:52] even if changes have been on entirely different repositories [14:15:07] in that case, should I continue as I was doing? [14:15:19] i.e. a job shared by different repos flag the repos as being dependent on each other [14:15:30] which is exactly what I am working on :-] [14:15:46] i.e. making sure patches against extensions do not conflict whith each other [14:16:32] hashar: what should I do? [14:16:41] should I continue as I was doing for now? [14:16:50] or should I create one job to rule them all? [14:19:01] I guess continue what you were doing [14:19:07] and we can discuss with Timo about a common job [14:19:16] Hi [14:19:20] hashar: ok, we can revise the decision later, when you fix the problem [14:19:22] hashar: How was the meeting yesterday :P [14:20:20] Krinkle: hooooo [14:20:32] Krinkle: haven't you told me you were skipping them while at Wikimania? :-/////////// [14:20:48] Jenkins jobs should be per repository. This ensures various integrity and logic inside our current pipeline. I don't think we should aim for a global job, that is not useful and not valuable. [14:21:07] Because it forces thigns to run separate, or rather makes it impossible to do anything inside the job or perform locally in a non-intrusive way. [14:21:16] For example jshint or phpcs configuration. [14:21:23] or version dependencies for that matter [14:21:25] Grunt pipelines etc. [14:21:41] the use case was running ruby syntax check [14:21:57] I do think we should persue combining jobs within a project into one (e.g. "mediawiki-core-test" like visualeditor-npm), but that's a separate quest. [14:22:03] though that would mean cleaning up the workspace in each run [14:22:19] yep, that too. Environment is not optimised. [14:22:43] And concurrency and build history / build trend time tracking becomes useless. [14:23:02] they are always useless since subsequent patches can have different parent anyway [14:23:04] What's to gain by having it be one job? Is jenkins having issues scaling the number of jobs? [14:23:23] that is why I originally crafted some jobs that only run on post merge and per branch, that provides a trend [14:23:27] hashar: True, but they still help getting an overal trend. It's not reliable, but in reality it tends to be good enough to find problems that way. [14:23:34] Yep [14:23:35] agree [14:24:20] Note though, that when we're moving to VM testing, we shoudl really only have 1 job per pipeline/project. You don't want to have a dozen VMs for 1 patchset. I think it's a good thing to force us to combine the different tests into one script, this makes them more mobile/flexible/reusable and also runnable locally [14:24:50] And when we're in VM testing, we won't have a persistent workspace either. So we should learn to optimise that git stuff in a different way. [14:25:01] https://integration.wikimedia.org/ci/job/mediawiki-core-extensions-integration/ :D [14:25:15] maybe have Jenkins send the work tree directory itself instead of a command to git clone it. [14:25:19] that job is going to be run for all extensions deployed on wikimedia [14:25:24] (to the VM that is) [14:26:00] e.g. zuul performs the merge somewhere safe where we have existing clones that we don't need to re-create. And then rsync the directory itself instead of using git to fetch it. [14:26:02] Just an idea :) [14:36:36] (03PS28) 10Hashar: (WIP) Experiment zuul-cloner with extensions [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/141846 [14:37:06] (03CR) 10Hashar: "- intersection : parser tests are broken" [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/141846 (owner: 10Hashar) [14:45:00] (03PS6) 10Zfilipin: WIP Running Ruby linter for all repos that have Ruby code [integration/zuul-config] - 10https://gerrit.wikimedia.org/r/150177 (https://bugzilla.wikimedia.org/63469) [14:49:27] beta labs not answering at right now [14:49:48] at all [14:50:28] (03PS4) 10Zfilipin: WIP Running Ruby linter for all repos that have Ruby code [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/150189 (https://bugzilla.wikimedia.org/63469) [14:53:18] (03PS7) 10Zfilipin: WIP Running Ruby linter for all repos that have Ruby code [integration/zuul-config] - 10https://gerrit.wikimedia.org/r/150177 (https://bugzilla.wikimedia.org/63469) [14:55:17] (03PS5) 10Zfilipin: WIP Running Ruby linter for all repos that have Ruby code [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/150189 (https://bugzilla.wikimedia.org/63469) [14:56:13] chrismcmahon: looking. [14:56:38] chrismcmahon: looks like hhvm <-> apache communication locked up again just like yesterday morning [14:57:12] weird [14:57:20] I'm going to poke a bit to see if I can find a cause before I restart everything [15:00:15] bd808: sure, wfm [15:01:17] chrismcmahon: I pinged _joe_. He'd like a chance to look before we restart things too. We need to see if we can get to the bottom of this lockup. [15:02:19] Zeljko : hi [15:04:18] bd808: fwiw, it seems to be a gradual degradation of service. overnight browser tests show a pattern of long response times (90+) seconds, but by this morning beta was not responding at all [15:06:50] bd808: e.g. here's one where beta is not quite completely dead: https://saucelabs.com/jobs/f25c8d8a56114bbd863d5d020c54c692 [15:13:59] is beta known borked at this point? [15:27:28] manybubbles: Yes. [15:27:50] We have one hhvm server working and the other is stuck somehow [15:28:16] _joe_ and godog are looking to see if they can capture clues as to the cause [15:28:51] thanks [15:49:42] (03CR) 10Physikerwelt: "Ok. I'd like to see this merged." [integration/jenkins-job-builder-config] (cloudbees) - 10https://gerrit.wikimedia.org/r/149323 (https://bugzilla.wikimedia.org/67221) (owner: 10Cmcmahon) [15:56:11] I am in the hangout [15:56:48] g'morning :) [15:58:57] manybubbles, chrismcmahon: beta seems up again at least for now. [15:59:55] The ops folks got a lot of stack traces that hopefully will help figure out where hhvm is getting stuck [16:01:22] (03PS1) 10Legoktm: Whitelist rillke (GSoC student) [integration/zuul-config] - 10https://gerrit.wikimedia.org/r/150242 [16:01:37] zeljkof: ping ? :-] [16:02:00] hashar: pong :) [16:03:26] (03CR) 10Hashar: [C: 032] "wikipedia.de != wikimedia.de" [integration/zuul-config] - 10https://gerrit.wikimedia.org/r/150242 (owner: 10Legoktm) [16:03:33] (03Merged) 10jenkins-bot: Whitelist rillke (GSoC student) [integration/zuul-config] - 10https://gerrit.wikimedia.org/r/150242 (owner: 10Legoktm) [16:03:46] (03CR) 10Rillke: "Thank you!" [integration/zuul-config] - 10https://gerrit.wikimedia.org/r/150242 (owner: 10Legoktm) [16:50:34] (03CR) 10Cmcmahon: [C: 032] "This is already running in production (and it's green) https://integration.wikimedia.org/ci/view/BrowserTests/job/browsertests-Math-en.wik" [integration/jenkins-job-builder-config] (cloudbees) - 10https://gerrit.wikimedia.org/r/149323 (https://bugzilla.wikimedia.org/67221) (owner: 10Cmcmahon) [16:51:53] (03Merged) 10jenkins-bot: The first job for Math extension [integration/jenkins-job-builder-config] (cloudbees) - 10https://gerrit.wikimedia.org/r/149323 (https://bugzilla.wikimedia.org/67221) (owner: 10Cmcmahon) [17:04:51] 3Wikimedia / 3Quality Assurance: create build for Math browser test - 10https://bugzilla.wikimedia.org/67221#c3 (10Chris McMahon) 5PATC>3RESO/FIX Patch is merged [17:10:17] (03CR) 10Cmcmahon: [C: 032] Deleting all qa/browsertests jobs [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/150174 (https://bugzilla.wikimedia.org/63469) (owner: 10Zfilipin) [17:10:45] (03Merged) 10jenkins-bot: Deleting all qa/browsertests jobs [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/150174 (https://bugzilla.wikimedia.org/63469) (owner: 10Zfilipin) [17:13:22] djvutxt: [1-11711] Failed to open 'some-nonexistent-file': No such file or directory. [17:21:01] bd808: the daytime browser test builds on beta start up in 40 minutes. I suggest they might cause the HHVM problems again. [17:21:57] chrismcmahon: Thanks for the heads up [17:48:35] greg-g: promise to use your new powers on beta labs only for good not evil. [17:48:51] * greg-g crosses fingers behind back [17:48:52] greg-g: No! Come to the Dark Side! [17:48:53] promise [17:56:49] chrismcmahon: FYI Flow browsertests on beta labs failed with Ruby timeouts, e.g. Test did not see a new command for 90 seconds. Timing out. [17:56:53] error [17:56:59] in https://saucelabs.com/jobs/a52f2ae317c64239ae829c08931790b6 [17:57:21] (03PS1) 10Mwalker: Add Matt Walker's new email address [integration/zuul-config] - 10https://gerrit.wikimedia.org/r/150272 [17:57:42] spagewmf: yeah, sorry, HHVM died overnight. bd808 ^^ [17:57:49] (03Abandoned) 10Mwalker: Add Matt Walker's new email address [integration/zuul-config] - 10https://gerrit.wikimedia.org/r/150272 (owner: 10Mwalker) [17:58:32] flakey beta is flakey :( [17:59:26] spagewmf: sorry for the noise as we transitioned beta to hhvm :/ I promise things will get better. [17:59:32] Brett is trying to get setup in labs now to see things go sideways first hand. He's been great about fixing problems where he can see what the code is doing wrong so I have high hopes. [17:59:57] (03PS1) 10Mwalker: Add mwalker's new email address [integration/zuul-config] - 10https://gerrit.wikimedia.org/r/150275 [18:00:22] bd808 I see "HHVM in production" meeting in an hour. I suggest someone pound the table and say "Facebook isn't getting a dime from us until you fix up your product!!" [18:01:11] beat up the vendor then make them buy you a three martini lunch :) [18:01:40] :) Brett is too nice to get mad at. I'm looking forward to the talk [18:13:56] jenkins is running a bunch of db-update jobs that have been around for some time [18:21:35] bd808: slow start but the browser builds are running high volume right now, about 8 simultaneous sessions [18:41:24] chrismcmahon: [Regarding: https://bugzilla.wikimedia.org/show_bug.cgi?id=68760]: I think deleting the older images will solve problems for now. The new images are any how uploaded automatically :) [18:41:47] vikasyaligar: sounds good [18:44:56] chrismcmahon: Do I permission to delete it ? [18:50:07] *I have [19:01:41] vikasyaligar: so what exactly do you need right now for privileges? I am a little confused... [19:02:39] chrismcmahon: deleting the images(as there are lots of images) [19:03:05] vikasyaligar: as which user? [19:04:13] LanguageScreenshotBot and test1 [19:06:31] vikasyaligar: ok, LanguageScreenshotBot and test1 users now have "importer" and "file mover" privileges, let me know if that does the job [19:06:57] chrismcmahon: thank you :) Let me look into it [19:13:30] (03PS1) 10Jforrester: Add new extension repo: CiteThisPage [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/150298 [19:15:29] (03PS9) 10Jforrester: Create a demos-publish task for OOjs UI [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/125543 [19:17:06] (03PS4) 10Jforrester: [WIP] Generic banana linter job template [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/129826 (https://bugzilla.wikimedia.org/64045) [19:29:15] (03CR) 10Krinkle: [C: 032] "Compiled and created in Jenkins (not yet activated in Zuul) mwext-CiteThisPage-jslint, mwext-CiteThisPage-lint, mwext-CiteThisPage-phpcs-" [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/150298 (owner: 10Jforrester) [19:29:29] Krinkle: Thanks! [19:30:13] James_F: Should I activate lint check or unit test pipeline in Zuul? [19:30:21] e.g. are there phpunit tests that pass? [19:30:27] or just lint/jshint for now [19:30:36] (03Merged) 10jenkins-bot: Add new extension repo: CiteThisPage [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/150298 (owner: 10Jforrester) [19:30:41] Krinkle: Looking. [19:31:23] (03PS1) 10Krinkle: Enable CiteThisPage jobs [integration/zuul-config] - 10https://gerrit.wikimedia.org/r/150303 [19:31:23] Krinkle: Right now it's just PHP and CSS for linting; no actual tests AFAICS. [19:31:30] (03CR) 10Krinkle: "Enabled in I7f80ee0b85d29791b7." [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/150298 (owner: 10Jforrester) [19:31:43] James_F: There is no css lint in this template, only phplint and jslint [19:31:54] (the one you submitted to jjb-config) so Ill activate those. [19:31:58] Krinkle: Oh well. Next follow-up. [19:32:11] Krinkle: (Shouldn't we enable csslint by default?) [19:32:31] (03PS2) 10Krinkle: Enable CiteThisPage jobs [integration/zuul-config] - 10https://gerrit.wikimedia.org/r/150303 [19:33:16] James_F: No because it requires configuration, and it's quite biased. Very few rules are actually valuable. Near nothing actually passes it or produces useful results. [19:33:49] Krinkle: Could we make a Wikimedia-standard configuration that we like, though? [19:34:01] And, while there are legacy global jobs for phplint and jshint, for the rest I'd like to encourage a Gruntfile (like we do with banana and jscs) so that we don't get stuck unable to ugprade (upgrade the global job and you break everything, + no ability to test locally) [19:34:42] So there can be a convention, but still, it should be coded into the repo (just line jscsrc and jshintrc) so that it is also picked up by Sublime, and Grunt etc. [19:34:56] like on [[mw:CC/JS#Linting]] [19:35:08] Yeah, we could add a section like that for CSS [19:36:06] We should probably adopt it in mediawiki-core first before promoting to a convention. csslint is very under used right now, I"m not convinced it's good enough (not well maintained, and only used for oojs-ui/ve so far) [19:36:23] to find out if it's sane enough to take on core [19:36:29] Is there a way we can write a template Gruntfile system so it's less work for each repo? [19:36:58] Also, is there a node phpcs implementation so we can simplify the linting system? :-) [19:37:01] copy/paste from a repo you like? [19:37:51] Gruntfile tends to be minimal boiler plate. It's quite expressive. Most of it is specific to the repo (which files to scan etc.) [19:39:40] 3Wikimedia / 3Continuous integration: Add Jenkins jobs for OggHandler extension repo - 10https://bugzilla.wikimedia.org/68816 (10Florian) 3NEW p:3Unprio s:3normal a:3None Can you add a phpcs jenkins job for extension OnlineStatus, please? https://gerrit.wikimedia.org/r/#/admin/projects/mediawiki/ext... [19:39:41] 3Wikimedia / 3Continuous integration: Add Jenkins jobs for OggHandler extension repo - 10https://bugzilla.wikimedia.org/64990 (10Florian) [19:39:56] 3Wikimedia / 3Continuous integration: Add Jenkins jobs for OggHandler extension repo - 10https://bugzilla.wikimedia.org/64990 (10Florian) [19:39:56] 3Wikimedia / 3Continuous integration: Add Jenkins jobs for OggHandler extension repo - 10https://bugzilla.wikimedia.org/68816 (10Florian) [19:41:07] 3Wikimedia / 3Continuous integration: Add Jenkins jobs for OggHandler extension repo - 10https://bugzilla.wikimedia.org/68816#c1 (10Brion Vibber) OggHandler or OnlineStatus? [19:41:54] 3Wikimedia / 3Continuous integration: Add Jenkins jobs for OnlineStatus extension repo - 10https://bugzilla.wikimedia.org/68816 (10Siebrand Mazeland) [19:47:17] Krinkle: But most repos should have their assets in the same places, ideally. [19:47:37] Krinkle: JS and CSS in modules/**; PHP in * and src/**; etc. [19:47:58] Yes, but you don't want to have a jenkins-global setting that says where it is. that jus creates a mess, and again, makes it unusable locally. [19:48:52] And whether they use jscs or not, and watch, and qunit, and which settings. There's always small variations. So really easiest I think is to copy/paste from another project (or be inspired otherwise) and type it out. [19:48:52] * James_F shrugs. [19:48:58] package.json, jscsrc, jshintrc, as well [19:49:22] Sure, but it means that introducing a new recommended way to do something involves updated 200 repos seperately. [19:49:59] Whee. NOT_REGISTERED. [19:52:55] (03CR) 10Krinkle: [C: 032] Enable CiteThisPage jobs [integration/zuul-config] - 10https://gerrit.wikimedia.org/r/150303 (owner: 10Krinkle) [19:53:00] (03Merged) 10jenkins-bot: Enable CiteThisPage jobs [integration/zuul-config] - 10https://gerrit.wikimedia.org/r/150303 (owner: 10Krinkle) [19:55:04] James_F: indeed, which is how it has to be because adding new rules will require code to be adapted as well. [19:55:29] this is very much by design. Up until january this year, every time I updated jshint broke stuff everywhere and nobody could merge anything. [19:55:39] and people never had the right versions locally [19:56:14] If we change conventions, we tell people, that way they both know about it ahead of time, and can decide when/how to adapt their own infrastructure. [19:56:29] as well as override rules etc. [19:56:35] it's too inflexible to do globally [19:56:40] Tried that, very hard, didn't work. [19:57:00] Krinkle: No no no, I don't mean new rules. [19:57:03] same at jquery, found out the hardway that doing curl shared/jshintrc > .jshintrc from git-clone-hook is not what you want. [19:57:15] Goes for new anything or change. [19:57:21] Krinkle: I mean templated ways to use jscs or whatever. [19:57:22] package update, or added package. [19:57:37] You mean a script that sets up the initial one for a new repo? [19:57:37] Krinkle: Rather than manually having to know you can add jscs now (or that there's a new version), being prompted. [19:57:53] Initial set-up and also maybe a monthly 'there's a new way to do it' check. [19:58:02] 'Cos manual checking doesn't scale. [19:58:22] How many repos are using jscs right now, but aren't using the wikimedia preset, for example? [19:58:31] I don't think we can auto-detect that there's something they should update (other than the package versions themselves, which you can let npm or david-dm inform you about already) [19:59:29] Well, a script can certainly look at .jscsrc for the string '"preset": "wikimedia"'. [19:59:38] But yeah, maybe it'd be too fragile to do well. [19:59:38] Writing a system that would look in their jscsrc for wikimedia seems not justiable, that was a single change we did. Every single one of these changes in the future is going to not justify the cost of writing a detect script I think. An announcement by e-mail saying there's new stuff they should/can be using seems sufficinet. [19:59:57] And yet there is no announcement e-mail, I think? [19:59:58] no, but I mean writing that costs time, and someone would have to run it. When tey could just read an e-mail and find out. [20:00:08] Sure, but there's also no detect script yet. [20:00:15] I'm saying lets go with a simple e-mail? [20:00:25] The simple e-mail that's so simple it doesn't get written? [20:00:26] :-) [20:09:46] (03PS1) 10Dduvall: Better handling of API error responses [ruby/api] - 10https://gerrit.wikimedia.org/r/150371 [20:18:27] (03CR) 10Cmcmahon: [C: 031] Better handling of API error responses [ruby/api] - 10https://gerrit.wikimedia.org/r/150371 (owner: 10Dduvall) [20:30:27] chrismcmahon: I tried updating the image, but still the same error. Also I cannot see the delete action anywhere [20:31:04] vikasyaligar: do you have a link? I'll try deleting [20:32:03] chrismcmahon: http://commons.wikimedia.beta.wmflabs.org/wiki/File:VisualEditor_save_dialog-en.png [20:35:18] idle question: what privileges does it take to delete a file on Commons? I guessed at "importer" and "file mover" but I seem to be wrong. [20:35:28] ww... [20:37:07] chrismcmahon: I can see it as Administrator in https://commons.wikimedia.org/wiki/Special:ListGroupRights [20:39:28] vikasyaligar: yes, one moment [20:40:53] (03PS1) 10Dduvall: Watching and unwatching pages [ruby/api] - 10https://gerrit.wikimedia.org/r/150384 [20:40:55] vikasyaligar: OK, test1 user is now an admin on beta. I'm not sure I want to give that right to a bot, though. [20:43:20] chrismcmahon: that makes sense ! perhaps I think I should have created user with my name. Never mind I write it in the User page :) [20:43:33] *I will [20:43:49] chrismcmahon: thank you ! let me try it out now [20:46:34] vikasyaligar: try once more, I might not have updated you on commons completely in time [20:48:06] chrismcmahon: I can see it now. Thank you :) [21:00:23] (03CR) 10Cmcmahon: [C: 031] "Neato. Note that we still want to test flipping "watched" on and off via the UI, but this will be useful." [ruby/api] - 10https://gerrit.wikimedia.org/r/150384 (owner: 10Dduvall) [21:02:16] (03CR) 10Dduvall: "Absolutely. This was only needed for reliably setting the initial "Given" state." [ruby/api] - 10https://gerrit.wikimedia.org/r/150384 (owner: 10Dduvall) [21:32:53] Krinkle|detached: When you're back, why did Jenkins vote V+1 and not V+2 on https://gerrit.wikimedia.org/r/#/c/149592/ ? [22:06:54] English beta labs API is returning 503 error [22:07:31] bd808: ^^ [22:09:08] 3Wikimedia / 3Continuous integration: remove all tab characters from DNS repository - 10https://bugzilla.wikimedia.org/68769#c9 (10Daniel Zahn) all merged, all clean, plus this one that fixes the templates: https://gerrit.wikimedia.org/r/#/c/150397/ and done :) ~/wmf/dns/templates$ grep -l -P '\t' *.* ~/... [22:09:49] y'all are fast [22:10:36] bd808: is Brett on IRC somewhere? [22:10:55] chrismcmahon: swtaarrs in #mediawiki-core [22:13:00] maybe just a glitch? beta seems to be back now [22:13:42] The hhvm instances were down. Possibly because we had just installed a new hhvm related package [22:14:17] chrismcmahon: Come sit in #-core all the cool kids are there :) [22:17:02] ok! Let me language screenshot job again [23:21:08] 3Wikimedia / 3Continuous integration: Install PHP xdiff extension on gallium (PHPUnit skipping tests) - 10https://bugzilla.wikimedia.org/41683#c9 (10Rainer Rillke @commons.wikimedia) well i got it installed locally and it even worked -- https://www.mediawiki.org/wiki/Manual_talk:PHP_unit_testing/Installing_P... [23:43:37] 3Wikimedia / 3Continuous integration: remove all tab characters from DNS repository - 10https://bugzilla.wikimedia.org/68769 (10Daniel Zahn) a:3Daniel Zahn [23:44:37] 3Wikimedia / 3Continuous integration: remove all tab characters from DNS repository - 10https://bugzilla.wikimedia.org/68769#c10 (10Daniel Zahn) 5PATC>3RESO/FIX ..and langlist template: https://gerrit.wikimedia.org/r/#/c/150397/ ..and config-geo https://gerrit.wikimedia.org/r/#/c/150406/ resolved :... [23:45:39] 3Wikimedia / 3Continuous integration: remove all tab characters from DNS repository - 10https://bugzilla.wikimedia.org/68769#c11 (10Daniel Zahn) 5RESO/FIX>3REOP eh, reopening for hashar to make the jenkins change, wasn't sure if that should have been a separate bug now [23:45:53] 3Wikimedia / 3Continuous integration: remove all tab characters from DNS repository - 10https://bugzilla.wikimedia.org/68769 (10Daniel Zahn) a:5Daniel Zahn>3Antoine "hashar" Musso [23:53:37] 3Wikimedia / 3Continuous integration: remove all tab characters from DNS repository, then enable jenkins check - 10https://bugzilla.wikimedia.org/68769 (10Daniel Zahn)