Closed Bug 1167896 Opened 5 years ago Closed 4 years ago

Intermittent browser_tabs_close_beforeunload.js | Test timed out

Categories

(Firefox :: Tabbed Browser, defect)

defect
Not set

Tracking

()

RESOLVED FIXED
Tracking Status
firefox41 --- wontfix
firefox45 --- wontfix
firefox46 --- fixed
firefox47 --- fixed
firefox48 --- fixed

People

(Reporter: philor, Unassigned)

References

Details

(Keywords: intermittent-failure)

No description provided.
Anything special about those runs in terms of timing / infra load? It's somewhat surprising that things go randomorange twice in quick succession well over a week after the test landing. Either something regressed the test (which tbh I thought was unlikely without a relevant code change landing on a weekend, which seems very unlikely), or it's infrequent orange (but the two times in succession makes that less likely, unless bad luck, I guess?). So I'm confused. :-)
Flags: needinfo?(philringnalda)
Intermittent and coincidence seems most likely to me.
Flags: needinfo?(philringnalda)
Flags: needinfo?(gijskruitbosch+bugs)
This seems to be mostly taskcluster linux64 and that's going away. That and the abonimably slow 10.6 stuff, but only rarely. De-needinfo'ing as it doesn't seem worth spending time investigating after all.
Flags: needinfo?(gijskruitbosch+bugs)
When are these platforms going away?  This is currently our top orange.  Can we disable on TC linux64 in some way?
Flags: needinfo?(gijskruitbosch+bugs)
(In reply to Ben Kelly [:bkelly] from comment #63)
> When are these platforms going away?

I don't know the answer to this question, but I *thought* that the TC version of these was supposed to replace the non-TC version? Have you tried asking releng or whoever makes these decisions? I'm not sure why I'm being asked.

>  Can we disable on TC linux64 in some way?

Not specifically, to the best of my knowledge. We could disable on linux64 debug e10s, and keep only the 10.6 orange. But really, it'd be better if we could understand why it was failing...

I'll push some debugging changes.
(In reply to :Gijs Kruitbosch from comment #64)
> I don't know the answer to this question, but I *thought* that the TC
> version of these was supposed to replace the non-TC version? Have you tried
> asking releng or whoever makes these decisions? I'm not sure why I'm being
> asked.

Sorry, I thought you knew the plan due to comment 47.

> I'll push some debugging changes.

Thanks!
(In reply to Ben Kelly [:bkelly] from comment #65)
> (In reply to :Gijs Kruitbosch from comment #64)
> > I don't know the answer to this question, but I *thought* that the TC
> > version of these was supposed to replace the non-TC version? Have you tried
> > asking releng or whoever makes these decisions? I'm not sure why I'm being
> > asked.
> 
> Sorry, I thought you knew the plan due to comment 47.

Eh, hm, sorry - I think by now I am less confident than I was that I have any idea, but ISTR a recent m.d.platform message saying we were close to using these and hiding the "normal" build and test runs. Maybe I'm misremembering?
 
> > I'll push some debugging changes.
> 
> Thanks!

Done!
Flags: needinfo?(gijskruitbosch+bugs)
Keywords: leave-open
\o/

Probably worth uplifting this to aurora and beta. Since the morning of the 17th, the only failures starred to this bug have been on aurora (where it's still fairly frequent).
Status: NEW → RESOLVED
Closed: 4 years ago
Keywords: leave-open
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.