Closed Bug 1243103 Opened 4 years ago Closed 2 years ago

Intermittent browser_newtab_drag_drop_ext.js | grid status = 99p,0,1,2,3,4,5,7p,8p - Got ,0,1,2,3,4,5,7p,8p, expected 99p,0,1,2,3,4,5,7p,8p

Categories

(Firefox :: New Tab Page, defect, P3)

Unspecified
Linux
defect

Tracking

()

RESOLVED FIXED
Firefox 60
Tracking Status
firefox48 --- disabled
firefox49 --- disabled
firefox60 --- fixed

People

(Reporter: KWierso, Assigned: enndeakin)

References

Details

(Keywords: intermittent-failure, leave-open, Whiteboard: [test disabled on Linux])

Attachments

(2 files)

This has been failing intermittently since it was re-enabled by bug 1240014.
Blocks: 1240014
Flags: needinfo?(enndeakin)
Logs suggest this is Linux only but happens with e10s on or off. The failure always happens in checkGrid, but doesn't always happen in the call to it.
Flags: needinfo?(enndeakin)
OS: Unspecified → Linux
I think I know what it happening here.
Assignee: nobody → enndeakin
Status: NEW → ASSIGNED
Blocks: 1243398
Blocks: 1244554
Blocks: 1244713
Blocks: 1244406
Blocks: 1247528
This patch does two things:

1. Adds a notificationsEnabled flag so that multiple notifications get ignored. This was causing tests to run after the wrong update call and have the wrong data.

2. Makes setPinnedLinks wait for an update before returning.

The once concern is whether updateGrid should have error checking to reset notificationsEnabled if an error occurs.
Any news here, Neil? This is still one of the top overall failures.
Flags: needinfo?(enndeakin)
I'll look at this again shortly. The patch here caused another test to fail.
Flags: needinfo?(enndeakin)
I've spent way too much time on this and haven't found a working solution. Can you disable this test on Linux?

Although maybe some OS10.6 failures are starting from the last week logs? Let's keep an eye on that too.
Flags: needinfo?(ryanvm)
Flags: needinfo?(ryanvm)
Keywords: leave-open
Whiteboard: [test disabled on Linux]
browser.ini got an extra line with just "skip-if =" which broke a bunch of tests.
Dropped that line in https://hg.mozilla.org/integration/mozilla-inbound/rev/5806797af921
Assignee: enndeakin → nobody
Status: ASSIGNED → NEW
Attachment #8778382 - Flags: review?(jmaher) → review-
Comment on attachment 8778382 [details]
Bug 1243103 - Move intermittent failure to clibboard suite;

https://reviewboard.mozilla.org/r/69708/#review66774

I calculate a 1.5% failure rate on the existing hardware and a 8% failure rate on AWS- even if more data would change the percentages, that is still large enough to say this is an issue when running in a VM- lets try to figure it out or find a way to reduce this if possible before disabling the test.
Comment on attachment 8778382 [details]
Bug 1243103 - Move intermittent failure to clibboard suite;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/69708/diff/1-2/
Attachment #8778382 - Flags: review- → review?(jmaher)
Comment on attachment 8778382 [details]
Bug 1243103 - Move intermittent failure to clibboard suite;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/69708/diff/2-3/
Attachment #8778382 - Attachment description: Bug 1243103 - Disable intermittent failure on win debug; → Bug 1243103 - Move intermittent failure to clibboard suite;
Comment on attachment 8778382 [details]
Bug 1243103 - Move intermittent failure to clibboard suite;

https://reviewboard.mozilla.org/r/69708/#review66794

please add a comment indicating this is temporary until we can determine why this is more intermittent on a VM.  Also we will need to run this on try as removing this test can cause problems (retrigger 20 times for the chunk that ends up running this browser/base/content/test directory), likewise adding this test to 'cl' (20 times here is good) can cause problems there.

as a note, this will be on all platforms, that is OK, but it requires testing on all platforms prior to landing it!
Attachment #8778382 - Flags: review?(jmaher) → review+
Comment on attachment 8778382 [details]
Bug 1243103 - Move intermittent failure to clibboard suite;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/69708/diff/3-4/
(In reply to Joel Maher ( :jmaher ) from comment #49)
> Comment on attachment 8778382 [details]
> Bug 1243103 - Move intermittent failure to clibboard suite;
> 
> https://reviewboard.mozilla.org/r/69708/#review66794
> 
> please add a comment indicating this is temporary until we can determine why
> this is more intermittent on a VM.  Also we will need to run this on try as
> removing this test can cause problems (retrigger 20 times for the chunk that
> ends up running this browser/base/content/test directory), likewise adding
> this test to 'cl' (20 times here is good) can cause problems there.
> 
> as a note, this will be on all platforms, that is OK, but it requires
> testing on all platforms prior to landing it!

Looks good on try:

https://treeherder.mozilla.org/#/jobs?repo=try&revision=56c3e5f72f794d88a3007af50c32489fa202c3b2
Pushed by rwood@mozilla.com:
https://hg.mozilla.org/integration/autoland/rev/d40036e732c8
Move intermittent failure to clibboard suite; r=jmaher
Bulk assigning P3 to all open intermittent bugs without a priority set in Firefox components per bug 1298978.
Priority: -- → P3
(In reply to Carsten Book [:Tomcat] from comment #54)
> https://hg.mozilla.org/mozilla-central/rev/d40036e732c8

Oops, this tickled a bug in our test manifest parsing and as a result, this test has not been running - see bug 1333506.

Once bug 1333506 lands and this test starts running again, if it passes consistently, we should consider moving other tests that fail in a similar way - bug 1243398 - to a subsuite.
Flags: needinfo?(gbrown)
I'll try the same "fix" in bug 1243398; there are other open bugs with a similar signature, but 1243398 is the only one currently failing.
Flags: needinfo?(gbrown)
Test was removed in bug 1433133 https://hg.mozilla.org/mozilla-central/rev/fc80dd3e4fae
Status: NEW → RESOLVED
Closed: 2 years ago
Depends on: 1433133
Resolution: --- → FIXED
Target Milestone: --- → Firefox 60
Assignee: nobody → enndeakin
You need to log in before you can comment on or make changes to this bug.