Closed
Bug 1436810
Opened 7 years ago
Closed 7 years ago
browser/components/originattributes/test/browser/browser_cache.js perma leaks windows on a try push when only arbitrarily changed
Categories
(Core :: DOM: Security, defect)
Core
DOM: Security
Tracking
()
RESOLVED
DUPLICATE
of bug 1330046
People
(Reporter: mayhemer, Unassigned)
References
Details
STR:
- make a trivial change to that test like [1]
- push to try (empty) and trigger debug tier 2[TV] test, like [2]
Result:
the test result log claims on every retrigger there is a domwindow leak caused by the browser_cache.js test ; it seems platform independent
[1] https://hg.mozilla.org/try/rev/84979b843d45519868f702a854ee5f03d13c4449
[2] https://treeherder.mozilla.org/#/jobs?repo=try&revision=99cd8c49a350e8b27f560a16f7e8242aaed4f60c&filter-searchStr=tv
Comment 1•7 years ago
|
||
This isn't related to Treeherder (it doesn't run the jobs themselves, determine the result, nor is it involved with the test harnesses). Guessing at the correct component :-)
Component: Treeherder → DOM: Security
Product: Tree Management → Core
Version: --- → unspecified
Comment 2•7 years ago
|
||
Test-verify is a job that explicitly stress-tests the changed tests in a given push. Read the post below:
https://groups.google.com/d/msg/mozilla.dev.platform/sf3SjDAKL60/Cs0ZJPBECAAJ
If you have any more questions on how it works, Geoff can probably help more than I can.
Reporter | ||
Comment 3•7 years ago
|
||
Thanks! Could it then be that bug 1330046 is a perma failure that shows only when we either clobber or touch the file?
Comment 4•7 years ago
|
||
Yes! test-verify will only verify a test when it is modified.
Comment 5•7 years ago
|
||
I notice that this seems to fail on "step 2" of verification:
[task 2018-02-08T19:25:37.935Z] 19:25:37 INFO - :::
[task 2018-02-08T19:25:37.936Z] 19:25:37 INFO - ::: Test verification summary for:
[task 2018-02-08T19:25:37.936Z] 19:25:37 INFO - :::
[task 2018-02-08T19:25:37.937Z] 19:25:37 INFO - ::: browser/components/originattributes/test/browser/browser_cache.js
[task 2018-02-08T19:25:37.938Z] 19:25:37 INFO - :::
[task 2018-02-08T19:25:37.939Z] 19:25:37 INFO - ::: 1. Run each test 10 times in one browser. : Pass
[task 2018-02-08T19:25:37.940Z] 19:25:37 INFO - ::: 2. Run each test 5 times in a new browser each time. : FAIL
[task 2018-02-08T19:25:37.940Z] 19:25:37 INFO - ::: 3. Run each test 10 times in one browser, in chaos mode. : not run / incomplete
[task 2018-02-08T19:25:37.941Z] 19:25:37 INFO - ::: 4. Run each test 5 times in a new browser each time, in chaos mode. : not run / incomplete
[task 2018-02-08T19:25:37.941Z] 19:25:37 INFO - :::
[task 2018-02-08T19:25:37.942Z] 19:25:37 INFO - ::: Test verification FAILED!
[task 2018-02-08T19:25:37.942Z] 19:25:37 INFO - :::
That is roughly equivalent to:
for i = 1 to 5:
mach mochitest browser/components/originattributes/test/browser/browser_cache.js
In a case like this where a leak is seen in test-verify but not in "normal" test runs in continuous integration, usually the important difference is that the test is run by itself, rather than as part of a larger test set. Maybe a later test is cleaning up something for browser_cache.js, or maybe something just needs time to run after browser_cache.js but before the browser is shutdown.
At any rate, remember that this is a tier 2 test, and the fault may have nothing to do with changes you have made. It would be great if you followed up and got this passing in test-verify, but it's certainly not required.
Reporter | ||
Comment 6•7 years ago
|
||
(In reply to Geoff Brown [:gbrown] from comment #5)
thanks for confirming this.
> It would be great if you followed
> up and got this passing in test-verify, but it's certainly not required.
I did spend some time on this (maybe too much) w/o any results..
Anyway, what should we do with this bug? WONTFIX? INVALID?
Let you know that (sorry for my possible ignorance here) this really took some of my time until I found this was actually only intermittent. Same trap for Natalia in bug 1152332 again.
Comment 7•7 years ago
|
||
looks like bug 1330046 is still open and covers this test causing leaks.
Status: NEW → RESOLVED
Closed: 7 years ago
Resolution: --- → DUPLICATE
You need to log in
before you can comment on or make changes to this bug.
Description
•