[00:33:19] 10Continuous-Integration-Infrastructure, 10Ops-Access-Requests, 6operations, 5Patch-For-Review: Let contint-admins force run puppet with /usr/local/sbin/puppet-run - https://phabricator.wikimedia.org/T110943#1643666 (10Dzahn) @hashar yes, i will do that [05:36:10] (03CR) 10Florianschmidtwelzow: "> So maybe we no more need htmlspecialchars , or just replace < and >." [tools/release] - 10https://gerrit.wikimedia.org/r/231317 (owner: 10Florianschmidtwelzow) [05:37:36] Yippee, build fixed! [05:37:36] Project browsertests-MultimediaViewer-en.wikipedia.beta.wmflabs.org-windows_7-internet_explorer-11-sauce build #543: FIXED in 35 min: https://integration.wikimedia.org/ci/job/browsertests-MultimediaViewer-en.wikipedia.beta.wmflabs.org-windows_7-internet_explorer-11-sauce/543/ [08:26:51] (03Abandoned) 10Hashar: Import of legoktm/tools-ci [integration/config] - 10https://gerrit.wikimedia.org/r/238445 (https://phabricator.wikimedia.org/T111758) (owner: 10Hashar) [08:36:01] 10Continuous-Integration-Infrastructure, 5Patch-For-Review: Move https://github.com/legoktm/tools-ci to gerrit - https://phabricator.wikimedia.org/T111758#1644176 (10hashar) 5Open>3Resolved a:3hashar I have created in Gerrit the integration/dashboard repository and imported https://github.com/legoktm/to... [08:45:08] (03CR) 10Hashar: [C: 031] "So all repos having the ruby2.0lint would be considered to be dependent in gate-and-submit. So a patch to mediawiki/selenium would have t" (032 comments) [integration/config] - 10https://gerrit.wikimedia.org/r/237680 (https://phabricator.wikimedia.org/T112272) (owner: 10Zfilipin) [08:47:46] (03CR) 10Hashar: [C: 04-1] WIP Run ruby2.0lint instead of ruby1.9.3lint (031 comment) [integration/config] - 10https://gerrit.wikimedia.org/r/237680 (https://phabricator.wikimedia.org/T112272) (owner: 10Zfilipin) [08:54:08] (03PS1) 10Hashar: integration/dashboard runs tox on trusty [integration/config] - 10https://gerrit.wikimedia.org/r/238687 (https://phabricator.wikimedia.org/T111758) [08:54:26] (03CR) 10Hashar: [C: 032] integration/dashboard runs tox on trusty [integration/config] - 10https://gerrit.wikimedia.org/r/238687 (https://phabricator.wikimedia.org/T111758) (owner: 10Hashar) [08:55:12] (03Merged) 10jenkins-bot: integration/dashboard runs tox on trusty [integration/config] - 10https://gerrit.wikimedia.org/r/238687 (https://phabricator.wikimedia.org/T111758) (owner: 10Hashar) [08:56:57] (03PS1) 10Hashar: Use lib to determine paths and os.path.join() [integration/dashboard] - 10https://gerrit.wikimedia.org/r/238688 [08:56:59] (03PS1) 10Hashar: Introduce config file to set the various paths [integration/dashboard] - 10https://gerrit.wikimedia.org/r/238689 [08:57:01] (03PS1) 10Hashar: Support different zuul server command [integration/dashboard] - 10https://gerrit.wikimedia.org/r/238690 [08:57:03] (03PS1) 10Hashar: Update repositories on developer machine [integration/dashboard] - 10https://gerrit.wikimedia.org/r/238691 [08:58:02] 10Continuous-Integration-Infrastructure, 5Patch-For-Review: Move https://github.com/legoktm/tools-ci to gerrit - https://phabricator.wikimedia.org/T111758#1644216 (10hashar) flake8 runs with python3.4 :-} [08:58:18] legoktm: zeljkof: tools-ci is now on Gerrit under integration/dashboard [08:58:35] nice [08:59:30] (03CR) 10Hashar: "Still using globals in lib, but I think I got rid of hardcoded paths. Still uses legoktm specific laptop and labs project context though" [integration/dashboard] - 10https://gerrit.wikimedia.org/r/238688 (owner: 10Hashar) [09:01:12] (03CR) 10Hashar: [C: 04-1] "config file for the win. I am not happy with:" [integration/dashboard] - 10https://gerrit.wikimedia.org/r/238689 (owner: 10Hashar) [09:01:49] (03CR) 10Hashar: "I must run Zuul within a venv." [integration/dashboard] - 10https://gerrit.wikimedia.org/r/238690 (owner: 10Hashar) [10:04:49] Hallo. [10:04:51] Any idea why is https://gerrit.wikimedia.org/r/#/c/238703/ failing? [10:09:14] hashar, zeljkof ^ ? [10:10:45] aharoni: will take a look in a few mins [10:12:24] 09:53:10 Chromium 44.0.2403 (Ubuntu 0.0.0) ext.cx.tools.categories Get categories FAILED [10:12:24] 09:53:10 Expected: 4 [10:12:24] 09:53:10 Actual: 5 [10:12:28] almost looks unrelated [10:13:40] hmmmmmmmmmmmmm [10:13:56] It does look like a possible issue, though. [10:14:24] does it pass locally? [10:15:22] oh crap, that's what happens when you rely on the content of production Wikipedia. [10:15:39] It counts the categories of https://es.wikipedia.org/wiki/Han_Feizi [10:15:48] and, well - https://es.wikipedia.org/w/index.php?title=Han_Feizi&type=revision&diff=85135419&oldid=74311250 [10:16:01] lolol [10:16:05] Awesome :D [10:17:30] 10Beta-Cluster, 10Continuous-Integration-Infrastructure, 5Patch-For-Review: deployment-logstash2 puppet certificate - https://phabricator.wikimedia.org/T112537#1644476 (10hashar) 5Open>3Resolved puppet masters self sign again. [10:18:16] 10Continuous-Integration-Config: Define an entry point for ruby material - https://phabricator.wikimedia.org/T104024#1644479 (10hashar) [10:18:58] 10Continuous-Integration-Config: Define an entry point for ruby material - https://phabricator.wikimedia.org/T104024#1644480 (10hashar) p:5Normal>3Low @qgil being discussed in the #releng team by @dduvall and @zeljkofilipin . No strong agreement yet and it is not that much of a priority for us right now. [10:38:30] aharoni: problem solved? :) [10:41:01] zeljkof: damned wikis changing pages [10:41:02] ;D [10:46:25] Reedy: those wikis... ;) [10:48:05] aharoni: I'll make you a patch to put eswiki in readonly mode [11:05:07] 10Beta-Cluster, 7HHVM, 5Patch-For-Review: Upgrade Beta Cluster tmh* host(s) to HHVM and Trusty - https://phabricator.wikimedia.org/T110707#1644560 (10Joe) So, for the jobrunner count issue: that has nothing to do with puppet, but with the jobrunner code directly. I restarted the jobrunner service and it rep... [11:15:33] 10Browser-Tests: issues with QUnit tests on Firefox - https://phabricator.wikimedia.org/T112717#1644565 (10Reedy) 5stalled>3Open [11:15:47] 10Browser-Tests: issues with QUnit tests on Firefox - https://phabricator.wikimedia.org/T112717#1644567 (10Reedy) 5Open>3stalled [11:16:08] ffs [11:27:56] 10Browser-Tests, 5Patch-For-Review: Improve password fallback for mediawiki_selenium - https://phabricator.wikimedia.org/T112279#1644583 (10zeljkofilipin) p:5Triage>3Normal [12:54:11] 10Continuous-Integration-Config, 5Patch-For-Review, 7Puppet: Setup rubocop for operations/puppet ruby code lints - https://phabricator.wikimedia.org/T102020#1644695 (10zeljkofilipin) Well, we do not even need to explicitly ignore git submodules, since the Jenkins job does not even clone them. :D https://int... [12:54:51] Yippee, build fixed! [12:54:51] Project browsertests-GettingStarted-en.wikipedia.beta.wmflabs.org-linux-firefox-sauce build #597: FIXED in 50 sec: https://integration.wikimedia.org/ci/job/browsertests-GettingStarted-en.wikipedia.beta.wmflabs.org-linux-firefox-sauce/597/ [12:58:20] 10Browser-Tests: All repositories that have browser tests should be updated to the latest version of watir-webdriver. - https://phabricator.wikimedia.org/T112748#1644704 (10zeljkofilipin) 3NEW a:3zeljkofilipin [13:02:20] 10Browser-Tests: All repositories that have browser tests should be updated to the latest version of watir-webdriver - https://phabricator.wikimedia.org/T112748#1644720 (10zeljkofilipin) [13:05:38] 10Browser-Tests: All repositories that have browser tests should be updated to the latest version of watir-webdriver - https://phabricator.wikimedia.org/T112748#1644724 (10zeljkofilipin) |Repository| |--| |mediawiki| |ArticleFeedbackv5| |CentralAuth| |CentralNotice| |CirrusSearch| |ContentTranslation| |Echo| |Fl... [13:06:06] 10Browser-Tests: All repositories that have browser tests should be updated to the latest version of watir-webdriver - https://phabricator.wikimedia.org/T112748#1644729 (10zeljkofilipin) [13:06:58] 10Browser-Tests: All repositories that have browser tests should be updated to the latest version of watir-webdriver - https://phabricator.wikimedia.org/T112748#1644704 (10zeljkofilipin) [13:32:59] 10Browser-Tests: All repositories that have browser tests should be updated to the latest version of watir-webdriver - https://phabricator.wikimedia.org/T112748#1644770 (10zeljkofilipin) [13:43:24] 10Browser-Tests: All repositories that have browser tests should be updated to the latest version of watir-webdriver - https://phabricator.wikimedia.org/T112748#1644783 (10zeljkofilipin) [14:17:21] 5Continuous-Integration-Scaling, 5Patch-For-Review: Gearman plugin doesn't put nodes offline when given parameter OFFLINE_NODE_WHEN_COMPLETE - https://phabricator.wikimedia.org/T103551#1644829 (10hashar) 5Open>3declined a:3hashar OFFLINE_NODE_WHEN_COMPLETE has to be injected by Zuul since it is part of t... [14:25:35] 5Continuous-Integration-Scaling: Switch a job to Nodepool instances - https://phabricator.wikimedia.org/T112750#1644858 (10hashar) 3NEW a:3hashar [14:26:16] (03CR) 10Zfilipin: WIP Run ruby2.0lint instead of ruby1.9.3lint (031 comment) [integration/config] - 10https://gerrit.wikimedia.org/r/237680 (https://phabricator.wikimedia.org/T112272) (owner: 10Zfilipin) [14:26:44] (03PS3) 10Zfilipin: Run ruby2.0lint instead of ruby1.9.3lint [integration/config] - 10https://gerrit.wikimedia.org/r/237680 (https://phabricator.wikimedia.org/T112272) [14:26:54] (03PS4) 10Zfilipin: Run ruby2.0lint instead of ruby1.9.3lint [integration/config] - 10https://gerrit.wikimedia.org/r/237680 (https://phabricator.wikimedia.org/T112272) [14:43:21] 5Continuous-Integration-Scaling: Switch a job to Nodepool instances - https://phabricator.wikimedia.org/T112750#1644904 (10hashar) [14:45:32] (03PS1) 10Hashar: Switch integration-jjb-config-diff to Nodepool [integration/config] - 10https://gerrit.wikimedia.org/r/238752 (https://phabricator.wikimedia.org/T112750) [14:47:02] 10Continuous-Integration-Infrastructure, 5Continuous-Integration-Scaling, 7Tracking: tracking: Cleanup contint puppet manifests so they are easier to reuse / split slaves by functions - https://phabricator.wikimedia.org/T110864#1644924 (10hashar) p:5Triage>3Normal [14:49:38] zeljkof: I am migrating the jjb diff job to Nodepool instances [14:49:41] and see what will happen [14:49:52] hashar: woohoo! :D [14:50:05] I can see the flames from here ;) [14:50:25] good luck, by the way [14:50:25] !log Move integration-jjb-config-diff Jenkins job to Nodepool instances. https://gerrit.wikimedia.org/r/#/c/238752/ and https://phabricator.wikimedia.org/T112750 [14:52:00] (03CR) 10Hashar: [C: 032] "Deploying zuul change first, then the Jenkins job." [integration/config] - 10https://gerrit.wikimedia.org/r/238752 (https://phabricator.wikimedia.org/T112750) (owner: 10Hashar) [14:53:56] (03Merged) 10jenkins-bot: Switch integration-jjb-config-diff to Nodepool [integration/config] - 10https://gerrit.wikimedia.org/r/238752 (https://phabricator.wikimedia.org/T112750) (owner: 10Hashar) [14:58:39] (03PS1) 10Hashar: Test integration-jjb-config-diff on Nodepool [integration/config] - 10https://gerrit.wikimedia.org/r/238762 (https://phabricator.wikimedia.org/T112750) [14:59:46] job is running https://integration.wikimedia.org/ci/job/integration-jjb-config-diff/3717/console ! [14:59:56] and Nodepool is booting a new instance to replenish the pool [15:01:31] (03Abandoned) 10Hashar: Test integration-jjb-config-diff on Nodepool [integration/config] - 10https://gerrit.wikimedia.org/r/238762 (https://phabricator.wikimedia.org/T112750) (owner: 10Hashar) [15:37:29] 6RelEng-Admin, 6Team-Practices: One year review of RelEng offsite outcomes (April 2016) - https://phabricator.wikimedia.org/T112763#1645224 (10greg) 3NEW [15:37:56] 6RelEng-Admin, 6Team-Practices: Do 90 day review of post-offsite actions in early August '15 - https://phabricator.wikimedia.org/T103917#1402373 (10greg) >>! In T103917#1642075, @demon wrote: >>>! In T103917#1642050, @demon wrote: >> Re-check with the outside stakeholders of Release Engineering in one year. Th... [15:40:34] 10Browser-Tests: issues with QUnit tests on Firefox - https://phabricator.wikimedia.org/T112717#1645248 (10TasneemLo) [15:40:36] 5Continuous-Integration-Scaling, 5Patch-For-Review: Switch a job to Nodepool instances - https://phabricator.wikimedia.org/T112750#1645247 (10hashar) Interesting parts from `/var/log/nodepool/debug.log` ZeroMQ notification sent by Jenkins to Nodepool: Start of job ------------- ``` lang=json 2015-09-16 14:5... [15:41:55] 10Browser-Tests: Remove regular expressions in browser tests page element selectors - https://phabricator.wikimedia.org/T112291#1645251 (10zeljkofilipin) [15:42:55] stupid Jenkins [15:47:25] 10Browser-Tests, 7Tracking: Remove regular expressions in browser tests page element selectors - https://phabricator.wikimedia.org/T112291#1631130 (10zeljkofilipin) [15:48:00] 10Browser-Tests, 7Tracking: Remove regular expressions in browser tests page element selectors - https://phabricator.wikimedia.org/T112291#1631130 (10zeljkofilipin) [15:48:45] 10Continuous-Integration-Infrastructure, 6Release-Engineering-Team, 7Tracking: Repositories with Ruby code should be documented and appropriate Jenkins jobs should be running - https://phabricator.wikimedia.org/T1361#1645279 (10zeljkofilipin) [16:11:25] (03PS1) 10Chad: Allow top-level logger to track lower level git operations [tools/scap] - 10https://gerrit.wikimedia.org/r/238777 [16:11:53] (03CR) 10jenkins-bot: [V: 04-1] Allow top-level logger to track lower level git operations [tools/scap] - 10https://gerrit.wikimedia.org/r/238777 (owner: 10Chad) [16:14:18] (03PS2) 10Chad: Allow top-level logger to track lower level git operations [tools/scap] - 10https://gerrit.wikimedia.org/r/238777 [16:19:10] (03PS3) 10Chad: Allow top-level logger to track lower level git operations [tools/scap] - 10https://gerrit.wikimedia.org/r/238777 [16:22:01] (03PS4) 10Chad: Allow top-level logger to track lower level git operations [tools/scap] - 10https://gerrit.wikimedia.org/r/238777 [16:22:14] (03CR) 10jenkins-bot: [V: 04-1] Allow top-level logger to track lower level git operations [tools/scap] - 10https://gerrit.wikimedia.org/r/238777 (owner: 10Chad) [16:23:18] (03PS5) 10Chad: Allow top-level logger to track lower level git operations [tools/scap] - 10https://gerrit.wikimedia.org/r/238777 [16:27:33] (03PS6) 10Chad: Allow top-level logger to track lower level git operations [tools/scap] - 10https://gerrit.wikimedia.org/r/238777 [16:27:36] (03PS1) 10Chad: Use HHVM's logger in its sudo_check_call [tools/scap] - 10https://gerrit.wikimedia.org/r/238782 [16:29:08] andrewbogott: thanks the nodepool quota approval :-) [16:29:24] andrewbogott: now nodepool maintains a pool of at least 10 instances. Turns out that entirely fills the cloudint labs project quota [16:29:38] oh yeah, I guess that’s an issue :) [16:29:41] I’ll raise the quotas [16:29:56] if it is not threatening the labs infra yeah [16:30:17] should be 20 instances, 40 CPU, 80 GB RAM to match the 20 instances upper limit [16:30:30] there is one job consuming them, so I am not worried for today [16:30:34] will migrate a few more jobs tomorrow [16:30:56] seems labs spawn instances very fast [16:31:15] hashar: ‘contintcloud’ right? [16:31:19] (03CR) 1020after4: [C: 032] Execute distinct stages of deployment separately [tools/scap] - 10https://gerrit.wikimedia.org/r/238631 (https://phabricator.wikimedia.org/T109861) (owner: 10Dduvall) [16:31:25] andrewbogott: yes [16:34:54] andrewbogott: missing ram quota from 50 to 80GB :D [16:35:31] I will look for a way to track instances memory consumption and we can probably come with a flavor that uses less memory [16:40:26] 5Continuous-Integration-Scaling, 5Patch-For-Review: Switch a job to Nodepool instances - https://phabricator.wikimedia.org/T112750#1645486 (10hashar) The task is essentially complete. Have to fill a few tasks following the log investigation above. Namely: * look at speeding up the time to get instance ready... [16:40:36] 5Continuous-Integration-Scaling: Switch a job to Nodepool instances - https://phabricator.wikimedia.org/T112750#1645487 (10hashar) [16:42:24] I am offf [16:42:45] thank you andrewbogott ! [16:44:00] hashar: yay! [16:44:08] hashar: and good night ;) [16:44:12] (03PS1) 10Chad: sudo_check_call: Improve logging on failures [tools/scap] - 10https://gerrit.wikimedia.org/r/238786 [16:44:30] legoktm: good morning :-} we have migrated tools-ci to gerrit under integration/dashboard [16:44:55] marktraceur, thcipriani: My logging patches are up for review. [16:44:56] legoktm: I sent a few lame patches for you to complain about [16:45:08] All culminates in 238786 [16:46:00] s/marktraceur/marxarelli [16:46:24] ostriches: excellent! [16:46:39] i'm pretty close to a rollback patch [16:47:20] (03CR) 10Chad: [C: 032] Rename and simplify some git deploy functions [tools/scap] - 10https://gerrit.wikimedia.org/r/236241 (https://phabricator.wikimedia.org/T109514) (owner: 10Dduvall) [16:47:43] I'm actually surprised that ^ didn't conflict with my patches. [16:48:07] yeah, me too [16:50:06] (03Merged) 10jenkins-bot: Rename and simplify some git deploy functions [tools/scap] - 10https://gerrit.wikimedia.org/r/236241 (https://phabricator.wikimedia.org/T109514) (owner: 10Dduvall) [16:50:07] (03Merged) 10jenkins-bot: Execute distinct stages of deployment separately [tools/scap] - 10https://gerrit.wikimedia.org/r/238631 (https://phabricator.wikimedia.org/T109861) (owner: 10Dduvall) [16:58:55] (03PS7) 10Chad: Allow top-level logger to track lower level git operations [tools/scap] - 10https://gerrit.wikimedia.org/r/238777 [16:58:58] (03PS2) 10Chad: Use HHVM's logger in its sudo_check_call [tools/scap] - 10https://gerrit.wikimedia.org/r/238782 [16:59:01] (03PS2) 10Chad: sudo_check_call: Improve logging on failures [tools/scap] - 10https://gerrit.wikimedia.org/r/238786 [16:59:10] marxarelli: Rebased everything on top so it's nice and clean [17:00:02] ostriches: fantastic. i'll take a look right after SoS [17:00:04] ostriches: yay! Will poke at. [17:01:08] https://phabricator.wikimedia.org/P2050 [17:01:35] The stacktrace below was what you got before (and still get), but now you get that nice output so you can know what caused the exit code 1 :p [17:04:03] 10Browser-Tests: All repositories that have browser tests should be updated to the latest version of watir-webdriver - https://phabricator.wikimedia.org/T112748#1645605 (10zeljkofilipin) [17:06:51] ostriches: nice. So now I don't have to copy and paste into vim with :%s/,//g :%s/[// :%s/]// :%s/'//g [17:39:05] 10Continuous-Integration-Infrastructure: Upgrade PHPUnit to 4.0+ - https://phabricator.wikimedia.org/T99982#1645951 (10Anomie) [18:30:03] 10Deployment-Systems, 10MediaWiki-extensions-LocalisationUpdate, 7I18n, 7Wikimedia-log-errors: l10n-update not updating Vector - https://phabricator.wikimedia.org/T103879#1646277 (10Amire80) @nikerabbit, @KartikMistry, any idea about this? [18:33:40] (03CR) 1020after4: [C: 032] Use HHVM's logger in its sudo_check_call [tools/scap] - 10https://gerrit.wikimedia.org/r/238782 (owner: 10Chad) [18:34:04] (03Merged) 10jenkins-bot: Use HHVM's logger in its sudo_check_call [tools/scap] - 10https://gerrit.wikimedia.org/r/238782 (owner: 10Chad) [18:35:11] (03CR) 1020after4: [C: 031] sudo_check_call: Improve logging on failures [tools/scap] - 10https://gerrit.wikimedia.org/r/238786 (owner: 10Chad) [18:41:54] twentyafterfour: "aka: CalledProcessError isn't all that useful how we use it" :p [18:43:24] (03PS8) 10Greg Grossmeier: Allow top-level logger to track lower level git operations [tools/scap] - 10https://gerrit.wikimedia.org/r/238777 (https://phabricator.wikimedia.org/T109858) (owner: 10Chad) [18:44:00] (03PS3) 10Greg Grossmeier: sudo_check_call: Improve logging on failures [tools/scap] - 10https://gerrit.wikimedia.org/r/238786 (https://phabricator.wikimedia.org/T109858) (owner: 10Chad) [18:44:37] I was like greg's amending my patches? [18:44:40] Oh, commit summary :p [18:44:53] :) :) [18:45:00] I may have ef'd up the dependency? [18:45:25] Eh, it does that. [18:45:37] sha1 changed. [18:45:39] (obvs) [18:45:41] ostriches: it might be nice to pass the 'fallout' to the CalledProcessError as well, instead of just logging it? I'm not sure if it would be useful but it seems potentially it would. [18:45:44] bugger, sorry [18:46:01] twentyafterfour: CalledProcessError only takes errCode, cmd [18:46:15] greg-g: nbd. Merge parent, rebase child, done. [18:46:24] Hmm maybe we should use a different exception class then [18:46:39] It's pretty much the standard for subprocess.* stuff afaict. [18:47:07] Problem isn't so much the exception as we don't catch exceptions higher up and just let them bubble. If you caught it, you should be able to extract it pretty easily. [18:47:20] But there's a bunch of code paths that lead here, so I wasn't sure how best to do the try/catch [18:47:28] 10Continuous-Integration-Infrastructure, 6Editing-Department, 3Reading-Web, 5Testing Initiative 2015, 7Tracking: [EPIC] Wikimedia should use a standard set of tools for managing code quality - https://phabricator.wikimedia.org/T111396#1603051 (10greg) [18:47:37] (Hence, pass the loggers and shove the output before blowing up) [18:48:47] Long as the logger is required param (which it is now), it didn't seem too janky. [18:49:04] wth is "19807 Unknown modifier '\': [([^\s,]+)\s*=\s*([^\s,]+)[\+\-]]" in the fatalmonitor output? is that a real log message or is it a bug in fatalmonitor script? [18:49:17] Sounds like the latter. [18:49:42] ostriches: I'm fine with just logging the output and throwing the exception, it's an improvement [19:01:05] (03CR) 10Chad: "Bump. What's the status here?" [tools/scap] - 10https://gerrit.wikimedia.org/r/224629 (owner: 10Ori.livneh) [19:02:00] Aw no bd808 [19:03:45] Project beta-scap-eqiad build #70310: FAILURE in 19 min: https://integration.wikimedia.org/ci/job/beta-scap-eqiad/70310/ [19:03:46] Ah, vacation [19:04:18] Looking at ^ [19:04:26] It actually looks like a pretty easily fixed bug. [19:05:31] 10Browser-Tests, 10Continuous-Integration-Infrastructure, 6Release-Engineering-Team, 7Epic, 7Tracking: [EPIC] trigger browser tests from Gerrit (tracking) - https://phabricator.wikimedia.org/T55697#1646484 (10Jdlrobson) [19:07:25] Well that isn't a fatalmonitor bug, it's really in the logs [19:08:00] 10Browser-Tests, 10Continuous-Integration-Infrastructure, 6Release-Engineering-Team, 7Epic, 7Tracking: [EPIC] trigger browser tests from Gerrit (tracking) - https://phabricator.wikimedia.org/T55697#1646493 (10Jdlrobson) The description is a little confusing. Does VisualEditor do this already? I can't see... [19:09:10] ostriches: are you referring to the beta-scap-eqiad failure? [19:09:44] (03Abandoned) 10Ori.livneh: Expect l10n_cache-en.php, not l10n_cache-en.cdb [tools/scap] - 10https://gerrit.wikimedia.org/r/224629 (owner: 10Ori.livneh) [19:12:47] 6RelEng-Admin, 15User-greg: Write draft/strawman code-hosting exception guideline - https://phabricator.wikimedia.org/T109920#1646514 (10greg) Upon review/thinking, I'm not going to detangle them now because we are going to use the same platform for both (Phab) and that is actually one of the main reasons for... [19:14:11] twentyafterfour: so this interface got 100% more awesome https://phabricator.wikimedia.org/mail/query/outbox/ [19:14:36] (03PS1) 10Chad: Avoid race condition where lock file disappeared from under us [tools/scap] - 10https://gerrit.wikimedia.org/r/238828 [19:14:52] twentyafterfour: Yes. ^ may or may not help us there. [19:15:26] chasemp: very nice [19:16:21] (03CR) 10Chad: "Saw this manifest as https://integration.wikimedia.org/ci/job/beta-scap-eqiad/70310/console" [tools/scap] - 10https://gerrit.wikimedia.org/r/238828 (owner: 10Chad) [19:16:57] ostriches: I thought I added code for that exact problem before - I remember it exploding because of a missing lock file before, and bryon said it should be tolerant to the file missing so I added exception handling. Maybe it deals with the lock file in more than one location and I missed a spot. [19:17:21] maybe? I just followed the stacktrace :) [19:17:57] (03CR) 1020after4: [C: 032] Avoid race condition where lock file disappeared from under us [tools/scap] - 10https://gerrit.wikimedia.org/r/238828 (owner: 10Chad) [19:18:12] (03Merged) 10jenkins-bot: Avoid race condition where lock file disappeared from under us [tools/scap] - 10https://gerrit.wikimedia.org/r/238828 (owner: 10Chad) [19:26:58] (03CR) 10Hashar: [C: 04-1] "I find it very horrible to have to pass the logger everywhere. The cli.Application class takes care of setting up the root logger which i" [tools/scap] - 10https://gerrit.wikimedia.org/r/238777 (https://phabricator.wikimedia.org/T109858) (owner: 10Chad) [19:27:52] ostriches: hi :-} [19:27:53] ostriches: I think I found a better strategy to use the scap.Application logger from within the git functions [19:27:54] https://gerrit.wikimedia.org/r/#/c/238777/ :D [19:27:58] commented there [19:28:23] uses a trick which is that the logger setup by Application is what is yield when invoking logging.getLogger() (aka that is the root logger) [19:28:38] and from there you can get a child logger that is setup alike, but can pass it a suffix [19:28:40] Yippee, build fixed! [19:28:40] Project beta-scap-eqiad build #70311: FIXED in 23 min: https://integration.wikimedia.org/ci/job/beta-scap-eqiad/70311/ [19:28:43] Hmm [19:28:48] so logging.getLogger().getChild('git_fetch') [19:28:53] should log: [19:29:01] WARNING:git_fetch:sometstufff [19:29:20] tried it locally with some dummy code and it seems to work. Copy pasted at http://labs.codecademy.com/CwVf#:workspace [19:29:38] it is worth a short and can probably be reused everywhere to avoid having to pass logger to functions [19:30:08] someone smarter would craft a decorator :-D (such as @use_app_logger or whatever, that would inject the logger magically [19:30:36] python magic I don't understand :p [19:30:51] I think the trick is when you setup the logging module [19:30:53] it is global [19:31:03] so logging.basicConfig() set it up for everyone [19:31:08] and logging.getLogger() get the root [19:31:13] with whatever config you passed [19:31:17] yeah globals are evil [19:31:36] it is more or less as if you were doing : $wgLogger->basicConfig() [19:31:40] $wgLogger->getLogger(); [19:31:45] yields you the same thing [19:31:50] (I think) [19:36:19] (03PS1) 10Dduvall: Support atomic promotion and rollback [tools/scap] - 10https://gerrit.wikimedia.org/r/238839 (https://phabricator.wikimedia.org/T109514) [19:38:43] ostriches: only one conflict! [19:38:52] and it was an easy one [19:42:16] (03CR) 10Dduvall: "Note that to test this patch with scap-vagrant, you'll need to pull down the latest version, nuke your containers (`lxc-ls | while read c;" [tools/scap] - 10https://gerrit.wikimedia.org/r/238839 (https://phabricator.wikimedia.org/T109514) (owner: 10Dduvall) [20:00:34] hashar, ostriches: re passing around `logger` (and `user`, and a working directory) it seems like we're maybe missing some kind of "executor" or "context" abstraction [20:00:44] Yah [20:03:13] (03PS1) 10Jforrester: Add tox testing for labs/tools/forrestbot [integration/config] - 10https://gerrit.wikimedia.org/r/238847 [20:10:37] (03CR) 10Hashar: [C: 04-1] "You are appending a newline while it has been strip() just a line above." (032 comments) [tools/scap] - 10https://gerrit.wikimedia.org/r/238786 (https://phabricator.wikimedia.org/T109858) (owner: 10Chad) [20:11:01] ostriches: poor chad. I found a bug in the way sudo_check_call() consumes the command output [20:11:20] it does not log all the command output. That is fairly annoying :} [20:13:13] marxarelli: ostriches: the context, executor would be achieved by wrapping all those def functions in a class. Then we would get the method name with __name__ . Or have the Application to initialize a global logger in scap.utils. [20:13:43] 10Browser-Tests, 5Patch-For-Review: All repositories that have browser tests should be updated to the latest version of watir-webdriver - https://phabricator.wikimedia.org/T112748#1646791 (10zeljkofilipin) [20:14:14] ostriches: are you using vim for python coding ? [20:14:47] bunch of pythonist are using https://www.jetbrains.com/pycharm/ [20:15:24] for vim you definitely want https://github.com/davidhalter/jedi-vim [20:17:00] Nope, heh [20:17:32] don't tell me emacs [20:18:43] I don't like editing from the cli unless I'm on terminal :p [20:29:38] hashar, ostriches: yeah, that's what i was thinking. the cli.Application classes would each instantiate their own `Executor` or whatever with the right user, default working directory, logger, etc. [20:29:50] +1 [20:30:30] marxarelli: my trick might be possible with a decorator [20:30:49] I haven't looked whether a decorator can inject arbitrary named parameters to a function which does not define them [20:30:51] something like: [20:30:57] @inject_logger_param [20:31:08] def git_fetch(): [20:31:12] hmm... i'm just learning python so no idea [20:31:22] but it seems terrible to inject a parameter that isn't declared [20:31:22] (03PS1) 10Chad: Fix logging output from sudo_check_call [tools/scap] - 10https://gerrit.wikimedia.org/r/238858 [20:32:00] and i don't see how a decorator would get the right logger instance, without cli.Application exposing it as a global [20:32:11] which i would also avoid [20:33:25] does python have currying? [20:35:45] 238858 can probably land alone, it doesn't really restructure the logging yet. [20:35:54] But does give us the output plus the bugfix hashar spotted. [20:35:54] why not declare the logger parameter? [20:36:09] you could inject it with a decorator but have it declared all over the place [20:36:16] ostriches: yeah, i don't think we need to refactor this stuff quite yet. your patches look good so far [20:36:48] (03CR) 10Chad: [C: 04-2] sudo_check_call: Improve logging on failures [tools/scap] - 10https://gerrit.wikimedia.org/r/238786 (https://phabricator.wikimedia.org/T109858) (owner: 10Chad) [20:36:52] (03CR) 10Chad: [C: 04-2] Allow top-level logger to track lower level git operations [tools/scap] - 10https://gerrit.wikimedia.org/r/238777 (https://phabricator.wikimedia.org/T109858) (owner: 10Chad) [20:36:52] once we prove this new prototype we can refactor and remove cruft [20:37:15] (03CR) 10Hashar: [C: 04-1] "Atomic deploy and rollback. That is a killer feature :-}" (033 comments) [tools/scap] - 10https://gerrit.wikimedia.org/r/238839 (https://phabricator.wikimedia.org/T109514) (owner: 10Dduvall) [20:37:20] and yet another not serious review [20:37:34] marxarelli: i haven't read your scap atomic deploy code. Just commented on the commit summary. [20:38:07] hashar: what's the -1 about? :) moving it all to docs? [20:38:08] marxarelli: meanwhile, nodepool is in prod now [20:38:13] hashar: nice! [20:38:22] marxarelli: oh -1 is just to get rid of your change from my dashboard :-} [20:38:30] haha [20:38:38] https://gerrit.wikimedia.org/r/#/q/is:open+reviewer:self+label:code-review%253D0%252Cself,n,z <-- works for anyone logged in [20:38:38] ok ... [20:38:52] list any open change for which I am a reviewer and haven't voted code-review yet [20:39:10] cleanliness ain't free i guess [20:39:14] so in theory one can go through all this change and +1/-1 frenzy vote and it ends up empty [20:39:16] cost folks like you and me [20:39:59] The problem right now is that stuff in utils/tasks are basically all static, so they either need a logger /passed/ to them or they have to deal with getting a global logger from somewhere. [20:40:02] but ... doesn't -1 mean you disagree with the implementation or have some nitpick? :) [20:40:54] ostriches: logging.getLogger().getChild('git_fetch') would do [20:40:59] the root logger is already global [20:41:06] Ya [20:44:04] isn't this the thing that context managers are for? [20:44:41] https://docs.python.org/2/library/contextlib.html [20:47:25] twentyafterfour: it is more like a decorator that wraps before and after [20:47:34] with yield being the entry point for the decorated function [20:48:44] 10Browser-Tests, 5Patch-For-Review: All repositories that have browser tests should be updated to the latest version of watir-webdriver - https://phabricator.wikimedia.org/T112748#1646935 (10zeljkofilipin) [20:48:52] at worth [20:48:59] we could do something like cap.utils.basicConfig() [20:49:08] that would setup global variables [20:49:15] in scap.utils [20:49:24] and the function would refer to it. But that is ugly [20:49:38] tbh I think that's uglier than passing the parent's logger. [20:49:56] yeah :-} [20:50:17] anyway in this case the logger can be get directly and passing it as a param is not needed [20:51:12] the logger can only be retrieved globally if you know its name, though [20:51:13] right? [20:51:24] with a context manager, you basically maintain global state on a stack and then you can get the top of the stack to get the right logger for the current context [20:51:52] so, having a module level function know the name you're using within the instance of some other class is pretty ugly [20:52:06] marxarelli: that's why I said contextmanager [20:52:27] you store the global state in some context that the called function doesn't need to be concerned about [20:52:39] but it can retrieve the 'current' logger from the context when needed [20:52:51] and it can be nested without problems [20:53:12] * marxarelli reads twentyafterfour's link [20:53:31] we need the loggers to be very flexible if it's going to work with iscap [20:53:46] 10Beta-Cluster, 10MediaWiki-extensions-SecurePoll: Extension:Secure Poll gets Database error, when try to vote at beta cluster - https://phabricator.wikimedia.org/T112818#1646951 (10Luke081515) 3NEW [20:54:05] and context managers seem like the idiomatic way to do things like this [20:54:09] 10Beta-Cluster, 10MediaWiki-extensions-SecurePoll: Extension:Secure Poll throws database error, when try to vote at beta cluster - https://phabricator.wikimedia.org/T112818#1646959 (10Luke081515) [20:54:34] it's like dependency injection but built into the language [20:56:07] they seem similar to Ruby's blocks [20:56:14] so yeah, lightweight dependency injection [20:56:19] (03PS1) 10Chad: Simplify logging in ssh module [tools/scap] - 10https://gerrit.wikimedia.org/r/238959 [20:57:17] Although in the context of this discussion that patch is probably totally wrong. [20:57:24] i'm not quite sure how we'd apply that pattern in this case, though, not without introducing some "executor" class like ostriches and i propose [20:58:23] Well the thing is, cli.Application and scap.* mostly get it right. [20:58:32] They have a get_logger() they make use of [20:58:39] Which keeps track of the correct log. [20:58:59] The trick is getting the static-y stuff (utils, tasks, ssh) to make use of /that/ log [21:00:39] (03Abandoned) 10Chad: Allow top-level logger to track lower level git operations [tools/scap] - 10https://gerrit.wikimedia.org/r/238777 (https://phabricator.wikimedia.org/T109858) (owner: 10Chad) [21:00:59] 10Browser-Tests, 5Patch-For-Review: All repositories that have browser tests should be updated to the latest version of watir-webdriver - https://phabricator.wikimedia.org/T112748#1646984 (10zeljkofilipin) [21:01:05] hashar: Meh, is anyone else getting lots of Chrome disconnects when running npm karma? [21:01:33] (03Abandoned) 10Chad: sudo_check_call: Improve logging on failures [tools/scap] - 10https://gerrit.wikimedia.org/r/238786 (https://phabricator.wikimedia.org/T109858) (owner: 10Chad) [21:04:08] ostriches, twentyafterfour: i tend to do things very OO, but here's what i see as one possibility https://phabricator.wikimedia.org/P2052 [21:04:13] marxarelli: I think there should be some sort of context instance object ("executor" or whatever) [21:05:01] marxarelli: exactly [21:06:46] if we used context managers, the static-y stuff could use a logger that was built by whatever part of the code calls it [21:08:28] twentyafterfour: right, so DeployLocal would add a "standard" logger (stdout/stderr) while Deploy might add additional loggers for sending to the verbose tmux window, summary window, logstash, etc.? [21:08:45] marxarelli: exactly [21:08:57] that sounds like a pretty clean design to me [21:10:05] ostriches: why'd you -2 those patches. do you think we should tackle this design change now? [21:10:37] i guess we'd have to if we want to support the multiple logger approach [21:10:55] Those patches didn't help really towards improving this [21:11:21] So I broke out the important bit on its own until we decide [21:11:41] hmm... maybe we should all pair on solidifying this initial logger/context thingy design [21:13:22] it's pretty simple if the context manager wraps a global logger then all the code can just treat it the same [21:14:03] also I've already built a sort of executor thing similar to marxarelli's paste [21:14:16] but my iscap code is a bit of a mess still [21:14:46] (03CR) 10Hashar: [C: 04-1] "Almost! :-}" (032 comments) [tools/scap] - 10https://gerrit.wikimedia.org/r/238858 (owner: 10Chad) [21:14:52] my main concern is that we get that merged or at least up for review so we can build better output/logging from it [21:15:10] even though it might appear simple it's important that we don't diverge in our implementations [21:18:44] (03CR) 10Hashar: [C: 031] "Should be fine." [tools/scap] - 10https://gerrit.wikimedia.org/r/238959 (owner: 10Chad) [21:19:43] (03PS2) 10Chad: Fix logging output from sudo_check_call [tools/scap] - 10https://gerrit.wikimedia.org/r/238858 [21:19:49] hashar: Fixed ^ :) [21:20:03] Funny, my version also worked, just with some extra newlines thrown in :p [21:22:08] (03CR) 10Chad: "While it does attach to the root logger, it doesn't attach to the logger I want...the one for our cli.Application. Per IRC, we're likely t" [tools/scap] - 10https://gerrit.wikimedia.org/r/238959 (owner: 10Chad) [21:22:21] 10Browser-Tests, 10QuickSurveys, 6Release-Engineering-Team, 3Reading-Web-Sprint-56-Four Lions: QA: Setup browser tests on beta cluster so we can share test articles - https://phabricator.wikimedia.org/T112204#1647052 (10Jdlrobson) a:3Jdlrobson [21:23:20] (03CR) 1020after4: [C: 031] Fix logging output from sudo_check_call [tools/scap] - 10https://gerrit.wikimedia.org/r/238858 (owner: 10Chad) [21:24:29] (03CR) 1020after4: "I will write up the context manager to handle this." [tools/scap] - 10https://gerrit.wikimedia.org/r/238959 (owner: 10Chad) [21:24:38] ostriches: haven't tested the logging thing but should work [21:24:47] the patch that consumes the rest of stdout I am pretty sure it is fine [21:25:06] the ssh one "works" but isn't right. [21:25:14] (03PS1) 10Jdlrobson: Setup QuickSurveys jenkins job [integration/config] - 10https://gerrit.wikimedia.org/r/238974 (https://phabricator.wikimedia.org/T112204) [21:25:17] Probably will abandon in favor of our context thingie. [21:25:20] hashar: use https://github.com/marxarelli/scap-vagrant if you want to do a quick test :) [21:25:23] ^ marxarelli :) [21:25:29] NO WAY [21:25:31] marxarelli: i assume that's eough? [21:25:46] If I want to test I would run the integration test suite :-} [21:25:54] 10Browser-Tests, 5Patch-For-Review: All repositories that have browser tests should be updated to the latest version of watir-webdriver - https://phabricator.wikimedia.org/T112748#1647058 (10zeljkofilipin) [21:26:22] marxarelli: maybe we could reuse that scap-vagrant as a jenkins job against a proposed patch ? [21:26:31] not sure whether we have any usable vagrant on Jessie [21:26:57] marxarelli: On a related (but not) note, I still get permission denied errors trying to do the initial clone to the target containers. [21:27:27] ostriches: for reals? super weird. maybe there's something different about the parallels base box [21:27:38] marxarelli: kudos on the vagrant-scab box [21:28:19] hashar: it's a bit heavyweight for a CI job, but maybe if we can get it working with an lxc provider [21:28:21] i.e. lxc in lxc [21:28:23] :) [21:28:39] on kvm [21:28:55] marxarelli: That's my best guess :\ [21:29:00] wait until we get an openstack cloud on top of ganeti :} [21:29:18] zeljkof: are you around and have time to take a look at jdlrobson's CI/browsertest patch? [21:29:43] marxarelli: just thinking it is time to get some sleep :) [21:29:47] which patch? [21:29:51] * zeljkof is looking... [21:30:26] marxarelli: this? https://gerrit.wikimedia.org/r/#/c/238974/ [21:30:33] marxarelli: Who should own /srv/deployment/mockbase/deploy? [21:30:35] zeljkof: sorry... take a look if you want but i can review it if you need to sleep [21:30:42] (which is an empty dir root:root for me) [21:30:45] yeah, that one [21:30:56] thanks guys [21:30:59] marxarelli: I can take a qucik look [21:31:00] ostriches: oh, it should be owned by vagrant:vagrant [21:31:23] That was my thought, hm [21:31:31] ostriches: setup.sh:201 [21:31:33] jdlrobson: at worth you can deploy new browser tests jobs using your wmf labs account ( http://www.mediawiki.org/wiki/CI/JJB ) :D [21:31:43] o rllyyy? [21:31:57] oh men I sucks at communicating :D [21:31:58] jdlrobson: yeah :) [21:32:06] wait so i don't have to do code review? [21:32:12] I am pushing the job to jenkins now, to test it [21:32:15] i can just use the script and add it? confused. [21:32:21] /srv/deployment/mockbase/deploy/.git: Permission denied [21:32:21] can you send a mail to the QA list? [21:32:22] jdlrobson: yeah I wrote that tutorial 3 years ago and ton of folks managed to install JJB and deploy random jobs [21:32:24] Lies. [21:32:44] jdlrobson: anyone in the wmf LDAP group has rights to create jobs on Jenkins [21:32:49] jdlrobson: well, you can deploy a jenkins job before the change is merged into master :) [21:32:53] jdlrobson: JJB is just a DSL for it. [21:33:01] Ahhh, there's a setup.sh [21:33:04] okay i'll have to have a read through that later. That's news to me [21:33:14] so what we do is we deploy, test, amend change, deploy, test, amend until happy [21:33:19] then we +2 [21:33:24] hashar: what are you doing up so late? ;) [21:33:31] you don't want to know [21:33:56] (calculating degree of separation between enwiki articles [[Brainstormers]] and [[Masturbation]], that is a challenge by my local hacker group) [21:34:14] :D [21:34:50] the topic is: how close are "brainstormers" and "masturbation"? Free choice of language implementation, but make it unnecessary complicated as usual. [21:35:12] (03PS2) 10Zfilipin: Setup QuickSurveys jenkins job [integration/config] - 10https://gerrit.wikimedia.org/r/238974 (https://phabricator.wikimedia.org/T112204) (owner: 10Jdlrobson) [21:35:21] jdlrobson: spread the word! :-) and feel free to announce it on QA list [21:35:27] jdlrobson: sorry you weren't aware about it [21:35:46] 10Browser-Tests, 10QuickSurveys, 5Patch-For-Review, 3Reading-Web-Sprint-56-Four Lions: QA: Setup browser tests on beta cluster so we can share test articles - https://phabricator.wikimedia.org/T112204#1647074 (10zeljkofilipin) [21:35:56] hashar: next time i need to do this I'll try and pair with one of you to make sure i get it done easily [21:36:12] sure! [21:36:14] Should be a need next quarter for hover cards and a read more extension (plus i want to set some up for WikidataPageBanner soon) [21:36:38] we can even start a rolling training session for browertests to train everyone [21:36:41] jdlrobson: the job is deployed https://integration.wikimedia.org/ci/view/BrowserTests/view/-All/job/browsertests-QuickSurveys-en.m.wikipedia.beta.wmflabs.org-linux-chrome-sauce/ [21:36:44] running it now [21:36:51] including jjb / browser / cucumber etc [21:37:01] I will merge it when the job is green [21:37:11] I thought about recording some short tech talks. Haven't gotten to it yet [21:37:51] anyway *kisses* and have a good afternoon. self.sleep(7.5 * 3600) [21:38:32] marxarelli: My cpu is so hot. It's like "STOP MAKING AND KILLING THESE VMS. I CANT TAKE IT ANYMOAR" [21:39:11] jdlrobson: the job is running fine, but some scenarios are failing https://integration.wikimedia.org/ci/view/BrowserTests/view/-All/job/browsertests-QuickSurveys-en.m.wikipedia.beta.wmflabs.org-linux-chrome-sauce/lastBuild/console [21:40:35] Project browsertests-QuickSurveys-en.m.wikipedia.beta.wmflabs.org-linux-chrome-sauce build #1: FAILURE in 3 min 15 sec: https://integration.wikimedia.org/ci/job/browsertests-QuickSurveys-en.m.wikipedia.beta.wmflabs.org-linux-chrome-sauce/1/ [21:40:42] (03CR) 10Dduvall: [C: 031] Fix logging output from sudo_check_call [tools/scap] - 10https://gerrit.wikimedia.org/r/238858 (owner: 10Chad) [21:41:24] ostriches: haha [21:41:36] you should have tried it when it was using only 512M memory [21:41:43] major thrashing [21:42:19] turns out btrfs snapshots on a loopback device is a little memory intensive [21:42:38] zeljkof: yeh so i think I need to make a config change to beta labs... [21:42:47] (03CR) 10Zfilipin: [C: 031] "The job is running fine, but some scenarios are failing:" [integration/config] - 10https://gerrit.wikimedia.org/r/238974 (https://phabricator.wikimedia.org/T112204) (owner: 10Jdlrobson) [21:42:53] we have some config in tests/browsers/LocalSettings.php [21:42:54] lmao [21:43:14] i'm not sure whether it's best to copy pasta those or if there is a more sophisticated way [21:43:19] thoughts marxarelli / zeljkof ? [21:43:41] Does 2 +1s make a +2? [21:43:42] :p [21:43:46] http://git.wikimedia.org/blob/mediawiki%2Fextensions%2FQuickSurveys.git/8b923dd0427ecb5768dd7e5e234e1049e2eb2ec0/tests%2Fbrowser%2FLocalSettings.php#L2 < this is what we need [21:43:47] jdlrobson: sorry, my brain is shutting down [21:44:06] can you create a task and cc marxarelli, hashar and me, please? [21:44:19] jdlrobson: overriding LocalSettings.php only works for the ci job, not for periodic tests against beta [21:44:43] sure. I get that.. i'm just asking if the best solution is to maintain 2 configs.. [21:44:57] or if there is something cleverer I can do to share the configs [21:45:12] i see [21:45:20] i can't think of anything more cleaver :) [21:45:36] clever [21:45:42] also cleaver [21:45:46] 10Continuous-Integration-Infrastructure: colordiff: command not found - https://phabricator.wikimedia.org/T112821#1647107 (10zeljkofilipin) 3NEW a:3zeljkofilipin [21:46:37] cleeavverr [21:48:27] good night everybody [21:48:30] * zeljkof-away zzz [21:48:31] zeljkof-away: night! [21:49:36] marxarelli: Blew away the whole vagrant, restarted (in case I was missing some setup.sh fixes or somesuch), no dice. [21:50:13] ostriches: weak sauce [21:50:27] maybe it's the uid mappings [21:51:22] who's the owner of /var/lib/lxc/scap-target-base/rootfs//srv/deployment/mockbase/deploy ? [21:51:28] outside of the container [21:53:10] (03CR) 10Jdlrobson: [C: 031] "It will fail till https://gerrit.wikimedia.org/r/238981 is merged and deployed and I'm happy to have the job failing till then if that's " [integration/config] - 10https://gerrit.wikimedia.org/r/238974 (https://phabricator.wikimedia.org/T112204) (owner: 10Jdlrobson) [21:55:41] marxarelli: drwxr-xr-x 1 vagrant vagrant 0 Sep 16 14:38 . [21:55:57] legoktm: Want to push https://gerrit.wikimedia.org/r/238847 ? :-) [21:56:16] ostriches: whaaa [21:56:26] but inside the container it's root:root? [21:56:55] 10Browser-Tests, 10QuickSurveys, 5Patch-For-Review, 3Reading-Web-Sprint-56-Four Lions: QA: Setup browser tests on beta cluster so we can share test articles - https://phabricator.wikimedia.org/T112204#1647153 (10Jdlrobson) [21:57:05] https://phabricator.wikimedia.org/P2054 [21:57:13] ^ this is what I was talking about ... [21:57:21] marxarelli: Yerp [22:00:59] ostriches: what does your /etc/lxc/default.conf and /etc/libvirt/lxc.conf look like? [22:02:15] https://phabricator.wikimedia.org/P2055 [22:04:53] (03CR) 10Legoktm: [C: 04-1] "https://gerrit.wikimedia.org/r/#/c/238837/3/tox.ini,cm doesn't define flake8, so just py34 for now." [integration/config] - 10https://gerrit.wikimedia.org/r/238847 (owner: 10Jforrester) [22:04:57] ostriches: ok ... [22:05:02] ostriches: what about `getent passwd vagrant`? [22:05:04] ostriches: marxarelli take a look at P2054 and let me know what you think. [22:05:06] on the host [22:05:08] legoktm: OK. [22:05:25] legoktm: Or should I add flake8? [22:06:11] (03PS2) 10Jforrester: Add tox testing for labs/tools/forrestbot [integration/config] - 10https://gerrit.wikimedia.org/r/238847 [22:06:31] marxarelli: vagrant:x:900:900:vagrant,,,:/home/vagrant:/bin/bash [22:06:59] ostriches: aha. i think that's the issue [22:07:12] in the vbox base box, the uid is 1000 [22:07:19] i assumed that was the case for all vagrant base boxes [22:07:37] so setup.sh creates a vagrant user inside the base container with 1000 as well [22:07:43] a-ha! [22:08:16] * ostriches wonders how much time would've been saved if he just downloaded the ~150MB install for VirtualBox yesterday :p [22:08:51] ostriches: but then you'd be somehow supporting oracle by using their software [22:09:09] and oracle is pretty much the devil [22:10:13] James_F: lets add flake8 in a follow up [22:10:22] legoktm: Kk. [22:11:03] (03PS1) 10Jforrester: Add flake8 tox testing for labs/tools/forrestbot [integration/config] - 10https://gerrit.wikimedia.org/r/238985 [22:11:17] (03CR) 10Jforrester: [C: 04-1] "Needs a patch in the repo first. :-)" [integration/config] - 10https://gerrit.wikimedia.org/r/238985 (owner: 10Jforrester) [22:11:54] twentyafterfour: i'll have to read up on contextmanager a bit to totally understand that :) [22:11:57] looks cool though [22:12:02] (03CR) 10Legoktm: [C: 032] Add tox testing for labs/tools/forrestbot [integration/config] - 10https://gerrit.wikimedia.org/r/238847 (owner: 10Jforrester) [22:12:27] all the context manager does is to make sure that the stack.pop() happens after the `with` block exits [22:12:45] kinda like a try catch block but no errors involved [22:12:49] err, like a try finally [22:13:12] so what's different about it than a normal `with` block in python [22:13:17] or are they the same thing? [22:13:34] the same thing [22:13:53] twentyafterfour: Not much better, I use parallels :p [22:13:53] the context manager is just a function that is used in conjunction with the `with` block [22:14:14] ooooh [22:14:33] makes sense now :) [22:15:02] it manages the global state in the logger_stack[] list [22:15:19] so why the logger stack over just a list of loggers that's setup by a cli.Application? [22:15:21] (note I forgot the [] around the first line there....fixed now) [22:15:34] i.e. why do we need a global stack? [22:15:48] marxarelli: If I switch my UID in the host to 1000 as well...should work? [22:15:58] marxarelli: just to avoid passing logger instances around everywhere [22:16:03] ostriches: one sec. i'm a gonna patch it [22:16:06] legoktm: Eurgh, Gather is hogging the gate-and-submit pipeline. [22:16:12] k [22:16:24] legoktm: Which of course is CI's fault, and not Gather's. :-) [22:17:00] I thought I dragged integration/config out of the mw pipeline [22:19:30] no, that was ops/mw-config [22:20:56] (03PS1) 10Legoktm: Get integration/config out of the 'mediawiki' queue [integration/config] - 10https://gerrit.wikimedia.org/r/238988 [22:23:03] umm [22:23:08] (03CR) 10Thcipriani: [C: 04-1] "This is awesome." (034 comments) [tools/scap] - 10https://gerrit.wikimedia.org/r/238839 (https://phabricator.wikimedia.org/T109514) (owner: 10Dduvall) [22:23:16] ci-jessie-wikimedia-51 doesn't have colordiff installed [22:23:26] ostriches: ok, pushed up a fix [22:24:49] (03CR) 10Jforrester: [C: 031] ":-)" [integration/config] - 10https://gerrit.wikimedia.org/r/238988 (owner: 10Legoktm) [22:25:08] (03PS2) 10Dduvall: Support atomic promotion and rollback [tools/scap] - 10https://gerrit.wikimedia.org/r/238839 (https://phabricator.wikimedia.org/T109514) [22:26:06] (03CR) 10Dduvall: "Rebased" [tools/scap] - 10https://gerrit.wikimedia.org/r/238839 (https://phabricator.wikimedia.org/T109514) (owner: 10Dduvall) [22:28:48] 10Continuous-Integration-Infrastructure: colordiff: command not found - https://phabricator.wikimedia.org/T112821#1647307 (10Legoktm) https://integration.wikimedia.org/ci/job/integration-jjb-config-diff/3720/console is totally useless without colordiff :( I can't seem to even log into that slave either... ``` k... [22:28:49] (03CR) 10Dduvall: Support atomic promotion and rollback (032 comments) [tools/scap] - 10https://gerrit.wikimedia.org/r/238839 (https://phabricator.wikimedia.org/T109514) (owner: 10Dduvall) [22:31:10] 10Beta-Cluster, 6Discovery, 10MediaWiki-API, 10MediaWiki-Search, 7Pywikibot-tests: test.wikipedia.org search is timing out - https://phabricator.wikimedia.org/T112828#1647319 (10jayvdb) 3NEW [22:31:41] marxarelli: afk for a few...will check when I'm back at home [22:32:28] * marxarelli crosses his fingers [22:33:02] thcipriani: thanks for the comments [22:33:17] so what's the difference between lambdas and local-scope functions in python? [22:34:47] lambdas look cooler [22:34:49] i mean `foo = lambda ...` vs `def foo:` [22:34:51] (03Merged) 10jenkins-bot: Add tox testing for labs/tools/forrestbot [integration/config] - 10https://gerrit.wikimedia.org/r/238847 (owner: 10Jforrester) [22:34:58] since they both have a name [22:35:06] yeah, not too sure internally, I guess the outcome is the same overall. [22:35:31] legoktm: that works for me :) [22:35:59] i like the fact that functions are first order objects [22:36:14] Ruby has always bugged me in that regard [22:36:51] e.g. you can't do `f = obj.f; f(x)` [22:37:04] you'd have to do `f = obj.method(:f); f.call(x)` [22:38:40] thcipriani: and they both create a closure? [22:39:19] sorry, i should probably rtfm :) i like learning languages! [22:41:11] one gotcha in python is that function parameters are late-binding [22:42:07] so you can't use them the way you would in javascript closures to make function instances with pre-populated state vars [22:44:09] marxarelli: oops, I was being sarcastic. [22:44:17] lambdas are much harder to read and follow imo [22:44:23] legoktm: me too [22:44:25] I know some people like python one-liners, but I don't [22:44:30] ok :D [22:45:00] http://benkurtovic.com/2014/06/01/obfuscating-hello-world.html how to say "Hello world" using lambdas [22:45:44] those are some descriptive parameter names! [22:48:04] legoktm: in all seriousness though, i do think one liners are fine as long as it's intelligible [22:48:16] well, I don't know, I've been using them to create small functions that reduce the airity of other functions and I think it makes a readable outcome. [22:49:05] e.g.: https://github.com/wikimedia/mediawiki-tools-scap/blob/master/tests/utils_test.py#L51 [22:50:08] but it probably wouldn't have hurt anybody to use: def gth(x): there—I am surprised when I see that in a function, though [22:50:29] throws me off of what I was looking at, "wait...am I in a function...?" etc. [22:52:48] thcipriani: oh yeah, ain't nothing wrong with that [22:53:23] i do that in javascript a lot as well https://github.com/wikimedia/integration-raita/blob/master/src/raita.js#L577 [22:53:33] I mean, if it stays as a *one-liner* thats fine [22:53:42] but I've seen enough lambdas become multi-line quickly :P [22:53:54] !log deploying https://gerrit.wikimedia.org/r/238847 [22:54:29] sure, 'it depends' is where i stand :) [22:55:22] yeah, I suppose the problem here is that lambdas are kinda neat, and as with most things that are neat, they're way over-used (see: regex) :) [22:55:42] see: instance_eval in Ruby [22:55:57] that turns into a hellscape real quick [22:56:07] but people just have to do it [23:02:41] (03PS3) 10Dduvall: Support atomic promotion and rollback [tools/scap] - 10https://gerrit.wikimedia.org/r/238839 (https://phabricator.wikimedia.org/T109514) [23:03:45] (03CR) 10Dduvall: Support atomic promotion and rollback (033 comments) [tools/scap] - 10https://gerrit.wikimedia.org/r/238839 (https://phabricator.wikimedia.org/T109514) (owner: 10Dduvall) [23:12:24] (03PS4) 10Dduvall: Support atomic promotion and rollback [tools/scap] - 10https://gerrit.wikimedia.org/r/238839 (https://phabricator.wikimedia.org/T109514) [23:21:30] (03PS5) 10Dduvall: Support atomic promotion and rollback [tools/scap] - 10https://gerrit.wikimedia.org/r/238839 (https://phabricator.wikimedia.org/T109514) [23:24:12] (03PS6) 10Dduvall: Support atomic promotion and rollback [tools/scap] - 10https://gerrit.wikimedia.org/r/238839 (https://phabricator.wikimedia.org/T109514) [23:25:55] (03PS7) 10Dduvall: Support atomic promotion and rollback [tools/scap] - 10https://gerrit.wikimedia.org/r/238839 (https://phabricator.wikimedia.org/T109514) [23:27:46] (03CR) 10Dduvall: [C: 031] "Implemented Thcipriani's suggestions and fixed a few minor issues." [tools/scap] - 10https://gerrit.wikimedia.org/r/238839 (https://phabricator.wikimedia.org/T109514) (owner: 10Dduvall) [23:56:15] marxarelli: Possibly stupid question. Is there such a thing as a ~/.vagrantrc or something? [23:56:25] So I can specify --provider=parallels and not have to retype every dang time :p [23:57:10] ostriches: looks like it [23:57:11] https://docs.vagrantup.com/v2/providers/default.html [23:57:33] to the bash_profile I go! [23:59:21] Aww, "==> default: groupadd: invalid group ID 'vagrant'" [23:59:25] On a fresh provision [23:59:41] That's stupid.