Closed Bug 1248028 Opened 10 years ago Closed 10 years ago

Bisect bug 1244936

Categories

(Testing :: General, defect)

Product:
Testing
Component:
General
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

(Not tracked)

Status:
RESOLVED DUPLICATE of bug 1227730

People

(Reporter: armenzg, Unassigned)

References

Details

Attachments

(1 file)

interactive.txt
80.95 KB, text/plain
Details
<jmaher> armenzg: ok, maybe if we get an hour or two, we can go in with interactive and run mochitests by hand with special flags <jmaher> armenzg: --bisect-chunk is the cli flag you want <jmaher> armenzg: ideally "python runtests.py ... --bisect-chunk <path>/browser_<testname>.js" <armenzg> jmaher, what does this do? <jmaher> it will verify the error, then see if anything that runs prior to that test case is causing it to happen <jmaher> then... <jmaher> it will remove whatever is failing (either the prior or the current one) and verify the dir runs successfully <jmaher> it will repeat as necessary
I'm trying these steps locally: https://pastebin.mozilla.org/8859916 It appends --bisect-chunk browser/base/content/test/alerts/browser_notification_close.js to the final command. I will be pushing to try with interactive in a bit.
My first two pushes are to be ignored. I can't reproduce the issue locally so I hope we can have some results in the next two hours.
jmaher: I think this job [1] might be doing the bisection with the test we care about: > TEST-UNEXPECTED-FAIL | Bisection | Please ignore repeats and look for 'Bleedthrough' (if any) at the end of the failure list Here's the test failure we care about: > TEST-UNEXPECTED-FAIL | browser/base/content/test/alerts/browser_notification_close.js | leaked 2 window(s) until shutdown [url = http://example.org/browser/browser/base/content/test/alerts/file_dom_notifications.html] However, I don't see "Bisection summary" in the log. [1] https://public-artifacts.taskcluster.net/NbFVM1YfRre5TAbmsIHiLw/0/public/logs/live_backing.log
Running inside of interactive mode gives me errors: https://bugzilla.mozilla.org/show_bug.cgi?id=1248758#c1
Bisecting did not help. I disabled the two tests that have failed for us (_close.js and _open_settings.js) From looking at a job [1], I don't see browser_notifications_*.js test fail anymore. I've pushed again to try since --interactive and bisection makes things hard to analyze. [1] https://treeherder.mozilla.org/#/jobs?repo=try&author=armenzg@mozilla.com&group_state=expanded&revision=8a9e47b7148a,5d196bbe596b,00d6cd6531c1,5cc6b0b6c425&selectedJob=16795431
It seems that disabling both tests got us complete green results. https://treeherder.mozilla.org/#/jobs?repo=try&revision=b3e4832f3edc
Aha! It seems that a week ago disabling the two tests was not sufficient. browser_notification_remove_permission.js was failing in here: https://treeherder.mozilla.org/#/jobs?repo=try&revision=a64ceebb39d4&filter-searchStr=tc&group_state=expanded
Attached file interactive.txt —
jmaher: should I spend time trying to figure out steps for the developers to run the two tests under interactive mode? (since we're talking about disabling them and cannot reproduce locally). I've tried running this through interactive mode and I've got a crash after the suite timed out (see attachment). Interesting piece of log below [1] Here are my steps for interactive mode: https://pastebin.mozilla.org/8860115 [1] TEST-UNEXPECTED-TIMEOUT | automation.py | application timed out after 330 seconds with no output Determining child pids from psutil Found child pids: [1512] Killing process: 1512 TEST-INFO | started process screentopng TEST-INFO | screentopng: exit 0 psutil found pid 1512 dead Killing process: 1453 Not taking screenshot here: see the one that was previously logged TEST-INFO | Main app process: exit 6 TEST-UNEXPECTED-FAIL | automation.py | application terminated with exit code 6 runtests.py | Application ran for: 0:06:11.081510 zombiecheck | Reading PID log: /tmp/tmpll_dxKpidlog ==> process 1453 launched child process 1512 zombiecheck | Checking for orphan process with PID: 1512 mozcrash Downloading symbols from: https://queue.taskcluster.net/v1/task/TmJr0FB3QeWhpq1z1WN4zg/artifacts/public/build/target.crashreporter-symbols.zip mozcrash Copy/paste: /usr/local/bin/linux64-minidump_stackwalk /tmp/tmpfucLc1.mozrunner/minidumps/2e34b71c-f1ec-4c1c-1de9d531-5b91105b.dmp /tmp/tmpyuaVQe mozcrash Saved minidump as /home/worker/workspace/build/blobber_upload_dir/2e34b71c-f1ec-4c1c-1de9d531-5b91105b.dmp mozcrash Saved app info as /home/worker/workspace/build/blobber_upload_dir/2e34b71c-f1ec-4c1c-1de9d531-5b91105b.extra PROCESS-CRASH | automation.py | application crashed [@ libpthread-2.15.so + 0xc565] Crash dump filename: /tmp/tmpfucLc1.mozrunner/minidumps/2e34b71c-f1ec-4c1c-1de9d531-5b91105b.dmp Operating system: Linux 0.0.0 Linux 3.13.0-48-generic #80-Ubuntu SMP Thu Mar 12 11:16:15 UTC 2015 x86_64 CPU: amd64 family 6 model 62 stepping 4 1 CPU Crash reason: SIGABRT Crash address: 0x3e800000566
no, interactive mode is not stable atm for this tests. I wonder if there is a root cause related to alerts/notifications which can fix this and test_alerts.html.
(In reply to Armen Zambrano [:armenzg] - Engineering productivity from comment #11) > It seems that disabling both tests got us complete green results. > https://treeherder.mozilla.org/#/jobs?repo=try&revision=b3e4832f3edc It seems that I might have looked at the wrong revision when I made this comment. I see browser_notification_remove_permission.js fail as in comment 12. https://treeherder.mozilla.org/#/jobs?repo=try&revision=f276199ce3ce&group_state=expanded&selectedJob=16807162 Let me push again disabling that test as well. I'm sorry :/
Since it is not a matter of bisection and it gets fixed by bug 1227730.
Status: NEW → RESOLVED
Closed: 10 years ago
Resolution: --- → DUPLICATE
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: