Closed Bug 970680 Opened 10 years ago Closed 10 years ago

"Timed out after 5.06385302544 seconds" is both very precise and very fast

Categories

(Remote Protocol :: Marionette, defect)

defect
Not set
normal

Tracking

(Not tracked)

RESOLVED FIXED
mozilla30

People

(Reporter: philor, Assigned: ato)

Details

Attachments

(1 file, 1 obsolete file)

In https://tbpl.mozilla.org/php/getParsedLog.php?id=34447794&tree=Mozilla-Inbound test_clock_set_alarm.py test_clock_set_alarm.TestClockSetAlarm.test_clock_set_alarm timed out, after 5.06385302544 seconds.

That's a wee bit more precision than we actually need, and although tbpl isn't currently trying to do any matching on that, it would be really inconvenient if something thought two failures were different things because one timed out after 5.06385302544 seconds and another timed out after 5.06385302543 seconds.

That also strikes me as being far far shorter than any timeout ought to be - if there's one thing looking at hundreds of thousands of our test failures has taught me, it's "yes, as a matter of fact, our test infra sometimes *can* be that slow."
Two things here; you want the number of seconds rounded? (that's in Marionette, for :ato to fix).

5 seconds is the Marionette default and I don't think this is the problem. We can override it on the Gu TBPL command line to be 10 but I don't think it will change much. When we know something will take a long time we will have coded in a larger timeout, but this 5 seconds is the timeout for something basic like closing a dialog.

We did make a big change to the waits in Gu last week and it has unexpectedly introduced a couple of intermittents which we are bashing through.
philor or ato can one of you round this up to a smaller number?
Component: Gaia::UI Tests → Marionette
Flags: needinfo?(philringnalda)
Flags: needinfo?(ato)
Product: Firefox OS → Testing
Flags: needinfo?(philringnalda)
I can certainly do that!
Assignee: nobody → ato
Status: NEW → ASSIGNED
Flags: needinfo?(ato)
Attachment #8383856 - Flags: review?(florin.strugariu)
Comment on attachment 8383856 [details] [diff] [review]
0001-Bug-970680-Limit-number-of-decimals-used-in-TimeoutE.patch

Review of attachment 8383856 [details] [diff] [review]:
-----------------------------------------------------------------

r+ on the w variable being updated.

::: testing/marionette/client/marionette/tests/unit/test_wait.py
@@ +260,4 @@
>              self.w.until(lambda x: x.true(wait=4), is_true=at_third_attempt)
>  
> +    def test_timeout_elapsed_rounding(self):
> +        w = Wait(self.m, clock=SequenceClock([1, 0.01, 1]), timeout=0)

w is a special character in pdb[1] so since you are editing this file can you update the rest of the file

[1] http://docs.python.org/2/library/pdb.html
Attachment #8383856 - Flags: review?(florin.strugariu) → review+
Good points!  Carrying on r+ from dburns.
Attachment #8383856 - Attachment is obsolete: true
Attachment #8383904 - Flags: review+
Keywords: checkin-needed
https://hg.mozilla.org/mozilla-central/rev/cff670cb1499
Status: ASSIGNED → RESOLVED
Closed: 10 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla30
Product: Testing → Remote Protocol
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: