mozilla :: #ateam

 
27 Apr 2017
05:52whimbooTomcat|sheriffduty: good morning... err, hm? https://bugzilla.mozilla.org/show_bug.cgi?id=1335778#c98
05:52bugbotBug 1335778: Marionette, major, hskupin, REOPENED , Make element#click command check for page load and wait
05:52whimboowhy is autoland approval required?
05:52Tomcat|sheriffdutygood morning whimboo
05:53whimbooi will just try again
05:53whimbooand now it works
05:53Tomcat|sheriffdutywhimboo: because KWierso|afk marked autoland approval required during a job backlog
05:54whimboomaybe it was related to the other close time
05:54Tomcat|sheriffdutyi guess to let catch things up
05:54whimboono, it was closed
05:54whimbooso the landing on autoland was queued
05:54whimboofor a couple of hours
05:54Tomcat|sheriffdutyat 2017-04-26T22:23:38.930098+00:00 it was approval required :)
05:55Tomcat|sheriffdutyon 2017-04-26T22:26:23.961069+00:00 the tree was set to green
05:55Tomcat|sheriffdutyand before on 2017-04-26T16:27:28.039981+00:00 autoland was closed
05:55whimboomaybe by accident
05:55whimbooit explains it
05:55whimboothanks
05:55whimbooworked now
05:55Tomcat|sheriffduty:)
06:53KWierso|afkI briefly switched it to approval required to get a backout patch landed while autoland kept push-racing me with all of the queued up autoland pushes from the previous five hours
06:53KWierso|afkbut I switched it back to open three minutes later
06:54whimbooKWierso|afk: so i assume all of the queued items were not landed
06:54KWierso|afkthe pushlog disagrees
06:55whimbooor at least those between the approval required and open changes
06:55KWierso|afkit still should've kept retrying
06:55KWierso|afkunless the autoland code only queues up when the tree is fully closed?
06:55KWierso|afkdunno
06:55whimbooi think so
06:56whimboothen it tried landing and failed
06:56whimboosame as for merge conflicts it aborted
06:56whimbooso sheriffs also cannot land / backout patches in clsoed state?
06:57KWierso|afkwe can
06:57KWierso|afkbut I would have had to amend the commit message to include "CLOSED TREE" to get it pushed. I already had "a=bustage" in the original message, so I just switched it to approval required to save me some work
06:58whimboosure, but which actually delayed other patches to land, and people have to request landing on autoland
06:58whimbooagain
07:04KWierso|afkwhimboo: filed bug 1360116 for this
07:04bugbotBug https://bugzilla.mozilla.org/show_bug.cgi?id=1360116 Autoland, normal, nobody, NEW , Autoland seems to fail when the tree is set to approval required rather than closed.
07:04whimbooKWierso|afk: thanks
07:05KWierso|afkdefinitely not my intent to pull you out of the autoland queue and toss you aside :)
07:05whimboono problem
07:05whimboowas hoping to have it on mc soon
07:06whimboobut I have other bugs to work on in the meantime
07:59whimbooStandard8: hello. do you know if there is way to control the default logging level for modules using Log.jsm already during startup? Maybe a pref or something?
08:00Standard8whimboo: I think thats generally defined per instance that is using it
08:01whimboowhen I enable extensions.logging.enabled it logs all up to DEBUG
08:01whimboobut I don't see a way to say only INFO
08:02Standard8whimboo: I think the only way to set info would be to change the code
08:02whimboofor marionette i tried to set the level of the root logger when it initializes, but this even doesn't work
08:03Standard8or somehow get the logger
08:03Standard8yeah, well youre fighting this code:
08:03Standard8https://dxr.mozilla.org/mozilla-central/source/toolkit/mozapps/extensions/AddonManager.jsm#124
08:03Standard8https://dxr.mozilla.org/mozilla-central/source/toolkit/mozapps/extensions/AddonManager.jsm#166
08:06whimbooStandard8: bah! :( looks like its not really possible to control. it might take the rootlogger level into account
08:06whimbooas max value at least
08:06Standard8whimboo: the only thing you might be able to do is Log.repository.getLogger(addons).level = Log.Level.Info;
08:06Standard8thats a guess
08:06Standard8err Log.repository.getLogger(addons).level
08:07whimbooStandard8: all that doesn't help during startup
08:07Standard8ah true
08:07whimboountil marionette is actually setting this
08:07whimbooso lots of unnecessary output
08:08whimboolooks like we will have to force extensions.logging.enabled to be false
08:08whimbooor set this by default
08:09whimbooStandard8: maybe i can file a bug for log.jsm in general
08:10whimbooto allow a rootlogger level configuration, which all loggers have to follow
08:11whimboobest as a preference so it will work at any time
12:47whimbooato: so the listener reload due to file:// is strange
12:47whimbooi cannot use Services.wm.getOuterWindowWithId() for the new window id
12:47whimbooit returns null
12:54atowhimboo: Can you ask bz or smaug if thats expected?
12:54whimboosure
14:30philorarmenzg: odd "Trigger missing jobs" behavior on https://treeherder.mozilla.org/#/jobs?repo=autoland&revision=ea481e3d0cda15e328b838dd8637ee768310eddb&group_state=expanded - I've run it three or four times, the Sch job shows as running for a bit, and triggers a handful of jobs, but disappears without ever showing as finished so I can look at its log to see why it only triggered
14:30philorand handful
14:31armenzgphilor: let me look at the logs
14:31armenzgI believe there's a bug on Sch/TH work
14:32armenzgif the job fails in any way it won't show up
14:32armenzgor at least that is my theory
14:46armenzgphilor: I found the log https://tc-gp-public-31d.s3-us-west-2.amazonaws.com/ateam/pulse-action-dev/02abf623-5434-49be-8011-5d05057daac3
14:48philoroh noes, no builder None! :)
14:50philorthat sounds a lot like a buildbot Linux64 gtest, no wonder there's nothing for it
14:52philorand indeed, "Add new jobs" also thinks that both linux64 opt and linux64 pgo could have buildbot gtest jobs
15:04armenzgphilor: I'm hoping I can soon ask to go back to fixing that as I've been refocused for Quantum perf work
15:04armenzgI hope that by next week the project will start going in the right direction
15:05philorthanks for looking up the log, good luck with the perf
15:09* philor decides it's too early in the day for looking at buildbot-configs to see why some gross hack involving gtests is still creating an orphan job
15:09philora decision which doesn't even require looking at a clock, or at whether the sun is up or down
15:10atoAutomatedTester: I need security review on https://bugzilla.mozilla.org/show_bug.cgi?id=1355888. Care to suggest someone?
15:10bugbotBug 1355888: Marionette, normal, ato, ASSIGNED , Prevent Marionette TCP server from starting/stopping on flipping marionette.enabled
15:11AutomatedTesterato: can you need-info pauljt
15:11AutomatedTesterhe will tell you what to do
15:11AutomatedTesterbbiab
15:11atoThanks.
17:14RyanVMjmaher: your comment about OSX debug SETA bugs - would that explain why we aren't scheduling OSX debug tests on Beta since 54 was uplifted to it?
17:15jmaherRyanVM: ?
17:16jmaherI wasn't aware of that
17:16RyanVMnow you know :P
17:16jmaherSETA only affects inbound/autoland
17:16RyanVMso yeah, we haven't had any OSX debug test coverage on 54 since it was uplifted to Beta
17:17jmaherRyanVM: I assume this is related to buildbotbridge
17:17RyanVMi know philor was poking around in buildbot-configs looking for plausible explanations and wasn't getting anywhere - BBB could explain why then
18:16RyanVMjmaher: &quot;as a note, lets ensure we update to build 1703&quot; <3<3<3<3<3<3<3<3
18:24RyanVMjmaher: so far, ~3% of our release/beta users are on 1703, nightly users are at 13%
18:26jmaherRyanVM: cool- by the time we are testing on this, we should be double that
18:27jmaherI am thinking keeping this stable until 57 releases
18:27RyanVMthat sounds entirely reasonable
18:27jmaherRyanVM: we are using that for the win10 hardware that talos will run on in the coming weeks
18:27RyanVMand yeah, I&#39;m still expecting MS to start opening the floodgates in the next few weeks
18:52dylanemorley: if it wasn&#39;t clear in teh bug, experimentally most things are immutable on https://bugzilla-dev.allizom.org/ right now.
18:53dylanalso, http2.
19:56atowhimboo: Congratulations (-:
20:10jgraham?
20:11atoThe wait-for-document-to-finish-loading-on-element-click patch landed (-:
20:11atoThis is quite a milestone.
20:12AutomatedTesternext up... solving the rest of the halting problem
20:12atoIt involved a lot of nasty work with the web progress listener, Marionettes ugly chrome-to-listener listeners, proxy work, remoteness debugging, and general sweat and toil.
20:14atojgraham: OK, so thanks to you Im now modifying http://searchfox.org/mozilla-central/source/browser/base/content/browser.xul and the code last touched a decade ago. Thank you very much.
20:14atos/the//
20:19jgrahamato: No problem
20:19jgraham:p
20:20atoWhoever thought inline chrome JS in CDATA was a good idea!?
 
27 Apr 2017
   
Last message: 14 minutes and 9 seconds ago