Closed Bug 891892 Opened 7 years ago Closed 7 years ago

[MP] Defect - Intermittent Win8 Metro mochiperf/browser_* command timed out: 1200 seconds without output, attempting to kill

Categories

(Firefox for Metro Graveyard :: Tests, defect, P2)

x86
Windows 8.1
defect

Tracking

(firefox24 unaffected, firefox25 unaffected, firefox26 fixed)

RESOLVED FIXED
Firefox 26
Tracking Status
firefox24 --- unaffected
firefox25 --- unaffected
firefox26 --- fixed

People

(Reporter: RyanVM, Assigned: jimm, NeedInfo)

References

Details

(Keywords: intermittent-failure, Whiteboard: [preview] feature=defect c=testing u=developer p=3)

https://tbpl.mozilla.org/php/getParsedLog.php?id=25099529&tree=Mozilla-Inbound

WINNT 6.2 mozilla-inbound opt test mochitest-metro-chrome on 2013-07-09 18:33:05 PDT for push 79d75a5b45cc
slave: t-w864-ix-032

18:36:07     INFO -  *** Start BrowserChrome Test Results ***
18:36:07     INFO -  TEST-INFO | checking window state
18:36:07     INFO -  TEST-START | chrome://mochitests/content/metro/browser/metro/base/tests/mochiperf/browser_deck_01.js
18:36:07     INFO -  TEST-INFO | chrome://mochitests/content/metro/browser/metro/base/tests/mochiperf/browser_deck_01.js | RUN deck offset
18:36:07     INFO -  TEST-INFO | chrome://mochitests/content/metro/browser/metro/base/tests/mochiperf/browser_deck_01.js | Opening about:mozilla in a new tab
18:36:07     INFO -  ### Content.js loaded
18:36:07     INFO -  ### FormHelper.js loaded
18:36:07     INFO -  ### SelectionPrototype.js loaded
18:36:07     INFO -  ### SelectionHandler.js loaded
18:36:07     INFO -  ### ContextMenuHandler.js loaded
18:36:07     INFO -  ### FindHandler.js loaded
18:36:07     INFO -  ### ConsoleAPIObserver.js loaded
18:36:07     INFO -  MetroWidget::GetDPI
18:36:07     INFO -  MetroWidget::GetDPI
18:36:07     INFO -  TEST-PASS | chrome://mochitests/content/metro/browser/metro/base/tests/mochiperf/browser_deck_01.js | about:mozilla is loaded
18:36:07     INFO -  TEST-INFO | chrome://mochitests/content/metro/browser/metro/base/tests/mochiperf/browser_deck_01.js | Wating for 5000ms
18:36:07     INFO -  * delay load started...
18:36:07     INFO -  * delay load complete.
18:36:07     INFO -  TEST-INFO | chrome://mochitests/content/metro/browser/metro/base/tests/mochiperf/browser_deck_01.js | Console message: [JavaScript Error: "DEPRECATION WARNING: nsIFormHistory2 is deprecated and will be removed in a future version
18:36:07     INFO -  You may find more details about this deprecation at: https://bugzilla.mozilla.org/show_bug.cgi?id=879118
18:36:07     INFO -  resource://gre/components/nsFormHistory.js 22 FormHistory
18:36:07     INFO -  resource://gre/modules/XPCOMUtils.jsm 271 null
18:36:07     INFO -  chrome://browser/content/browser-ui.js 151 BrowserUI.init/<
18:36:07     INFO -  chrome://browser/content/browser-ui.js 135 BrowserUI.init/</<
18:36:07     INFO -  null 0 null
18:36:07     INFO -  " {file: "resource://gre/modules/Deprecated.jsm" line: 79}]
18:36:07     INFO -  TEST-INFO | chrome://mochitests/content/metro/browser/metro/base/tests/mochiperf/browser_deck_01.js | Console message: 1373420167387	Sync.Tracker.Clients	WARN	Changed IDs file clients contains non-object value.
18:36:07     INFO -  TEST-INFO | chrome://mochitests/content/metro/browser/metro/base/tests/mochiperf/browser_deck_01.js | Console message: 1373420167394	Sync.Tracker.Bookmarks	WARN	Changed IDs file bookmarks contains non-object value.
18:36:07     INFO -  TEST-INFO | chrome://mochitests/content/metro/browser/metro/base/tests/mochiperf/browser_deck_01.js | Console message: 1373420167396	Sync.Tracker.Forms	WARN	Changed IDs file forms contains non-object value.
18:36:07     INFO -  TEST-INFO | chrome://mochitests/content/metro/browser/metro/base/tests/mochiperf/browser_deck_01.js | Console message: 1373420167398	Sync.Tracker.History	WARN	Changed IDs file history contains non-object value.

command timed out: 1200 seconds without output, attempting to kill
program finished with exit code 1
This appears to be a failure where setTimeout fails to generate a callback. (!)
Blocks: 880298
Blocks: metro-testing, 865451, metrov1defect&change
No longer blocks: 880298
Priority: -- → P3
Whiteboard: feature=defect c=testing u=developer p=0
Blocks: 880298
No longer blocks: metro-testing
Summary: Intermittent Win8 Metro mochiperf/browser_deck_01.js command timed out: 1200 seconds without output, attempting to kill → Defect - Intermittent Win8 Metro mochiperf/browser_deck_01.js command timed out: 1200 seconds without output, attempting to kill
No longer blocks: 865451
Summary: Defect - Intermittent Win8 Metro mochiperf/browser_deck_01.js command timed out: 1200 seconds without output, attempting to kill → Defect - Intermittent Win8 Metro mochiperf/browser_* command timed out: 1200 seconds without output, attempting to kill
Odd that this seems to have spiked so much lately.
(In reply to Ryan VanderMeulen [:RyanVM UTC-4] from comment #17)
> Odd that this seems to have spiked so much lately.

Yes something is going on here. Looking at the logs, affected tests are pretty random.. Also, looks like a browser freeze.
Looking back on b2g-inbound, earliest I've found is

https://tbpl.mozilla.org/?tree=B2g-Inbound&rev=c5946a8bcd5b
Fri Aug 9 16:31:32 2013 PDT

https://hg.mozilla.org/integration/b2g-inbound/pushloghtml?startID=161&endID=162

This has the start tab landing, some misc. metro front end patches, and some omtc gfx work.
If these were hung tests, we would get a mochitest timeout. I've been getting them running longer iterations of individual test to reproduce. So at this point this is looking like something that is specific to our slaves.
No longer blocks: metrov2defect&change
Depends on: 905628
Jimm, this is now the #1 intermittent failure in automation, across all platforms and test suites (http://brasstacks.mozilla.com/orangefactor/?display=OrangeFactor&endday=2013-08-16&startday=2013-08-13&tree=trunk). Please may you take a look? :-)
Flags: needinfo?(jmathies)
(In reply to Ed Morley [:edmorley UTC+1] from comment #102)
> Jimm, this is now the #1 intermittent failure in automation, across all
> platforms and test suites
> (http://brasstacks.mozilla.com/orangefactor/
> ?display=OrangeFactor&endday=2013-08-16&startday=2013-08-13&tree=trunk).
> Please may you take a look? :-)

I'm not sure what to do here. AFAICT this is a browser freeze. It doesn't happen often enough to reproduce reliably on try, so it would be hard to backout patches and try to reproduce there. Will think about it today, see if i can come up with some method of debugging it.

Note, I filed bug 905628 on trying to get stacks but there's been no activity on it.
Flags: needinfo?(jmathies)
Whiteboard: feature=defect c=testing u=developer p=0 → [preview] feature=defect c=testing u=developer p=0
We could try turning off omtc to see if it goes away. A course approach, but it would isolate the gfx code we leverage that other platforms do not.
Summary: Defect - Intermittent Win8 Metro mochiperf/browser_* command timed out: 1200 seconds without output, attempting to kill → [MP] Defect - Intermittent Win8 Metro mochiperf/browser_* command timed out: 1200 seconds without output, attempting to kill
Here's an open ended regression range (the open ended side goes back in time). This excludes everything metro related we landed on fx-team on friday, including the about tab work.

https://tbpl.mozilla.org/?tree=Mozilla-Inbound&showall=1&jobname=winnt%206.2&fromchange=e177c3d6ffc2&tochange=782e74f1c4c6
Thank you for digging into this btw :-)
Depends on: 906147
fingers crossed!
Status: NEW → RESOLVED
Closed: 7 years ago
Resolution: --- → FIXED
Assignee: nobody → jmathies
Target Milestone: --- → Firefox 26
Whiteboard: [preview] feature=defect c=testing u=developer p=0 → [preview] feature=defect c=testing u=developer p=3
Temporarily reopening to add to Iteration #13.
Blocks: metrov1it13
No longer blocks: metrov1backlog
Status: RESOLVED → REOPENED
Priority: P3 → P2
QA Contact: jbecerra
Resolution: FIXED → ---
Status: REOPENED → RESOLVED
Closed: 7 years ago7 years ago
Resolution: --- → FIXED
Is there anything that QA can do here? Could anyone please give any guidelines that can help the QA in verifying this issue? Thanks!
Flags: needinfo?
OS: Windows 8 Metro → Windows 8.1
You need to log in before you can comment on or make changes to this bug.