Closed Bug 1060813 Opened 5 years ago Closed 3 years ago

Intermittent browser_social_workercrash.js | leaked until shutdown [nsGlobalWindow #4089 inner about:socialerror?mode=workerFailure&origin=https%3A%2F%2Fexample.com about:socialerror?mode=workerFailure&origin=https%3A%2F%2Fexample.com]

Categories

(Firefox Graveyard :: SocialAPI, defect)

x86_64
Linux
defect
Not set

Tracking

(Not tracked)

RESOLVED WONTFIX

People

(Reporter: RyanVM, Unassigned)

References

(Depends on 1 open bug)

Details

(Keywords: intermittent-failure, memory-leak, Whiteboard: [test disabled][leave open])

Attachments

(1 file)

https://tbpl.mozilla.org/php/getParsedLog.php?id=47074457&tree=B2g-Inbound

Ubuntu ASAN VM 12.04 x64 b2g-inbound opt test mochitest-browser-chrome-1 on 2014-08-29 17:34:57 PDT for push 492bf5f4325d
slave: tst-linux64-spot-286

18:23:17     INFO -  1469 INFO TEST-UNEXPECTED-FAIL | chrome://mochitests/content/browser/browser/base/content/test/social/browser_social_workercrash.js | leaked until shutdown [nsGlobalWindow #4089 inner about:socialerror?mode=workerFailure&origin=https%3A%2F%2Fexample.com about:socialerror?mode=workerFailure&origin=https%3A%2F%2Fexample.com]
This is happening pretty frequently. Any chance we could get some eyes on it soon?
Flags: needinfo?(mixedpuppy)
(In reply to Ryan VanderMeulen [:RyanVM UTC-4] from comment #25)
> This is happening pretty frequently. Any chance we could get some eyes on it
> soon?

Looking at the logs, there are a lot with failures in the popupnotifications tests (prior to socialapi tests) that are leaking beyond that test and showing up during my tests.  There are also a uncaught exceptions in browser_newtab_sponsored_icon_click.js, as well as some from sociallapi tests.  I'm taking a guess that these all are exacerbating some async issue.  I've done a bunch of cleanup on the socialapi async stuff in bug 1014332 (part 4), I could land that part on it's own and see it helps, it certainly cleaned up problems from that bug.

As well, markh might have some further insights as to why that particular test started leaking.
Flags: needinfo?(mixedpuppy) → needinfo?(mhammond)
Even though I can repro this locally, I've no idea :(  I started a thread on firefox-dev - https://mail.mozilla.org/pipermail/firefox-dev/2014-September/002208.html
Flags: needinfo?(mhammond)
Thanks for offering to help with this!
Flags: needinfo?(continuation)
Mark looked at the log locally, and found that the global window was being held alive by a document, which had 1 unknown reference.  Which is not too useful.  I may have to try to look at this locally.
Flags: needinfo?(continuation)
Flags: needinfo?(continuation)
A month has passed and this is currently #6 on OrangeFactor with no activity. Any updates?
I'll make sure to get some more activity here.
Points: --- → 8
QA Whiteboard: [qa-]
QA Whiteboard: [qa-]
Flags: qe-verify-
Flags: firefox-backlog+