Closed Bug 1185486 Opened 7 years ago Closed 7 years ago

Intermittent Linux Marionette "ERROR - timed out after 1000 seconds of no output"

Categories

(Testing :: Marionette, defect)

Product:
Testing
Component:
Marionette
Platform:
Unspecified
Linux
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

(firefox44 fixed)

Status:
RESOLVED FIXED
Milestone:
mozilla44
Tracking Flags:
Tracking Status
firefox44 --- fixed

People

(Reporter: RyanVM, Unassigned)

Details

(Keywords: intermittent-failure, pi-marionette-intermittent)

Attachments

(2 files)

MozReview Request: Bug 1185486 - Part 1: Use a timeout when waiting on the browser process after a timed out command. r=jgriffin
40 bytes, text/x-review-board-request
jgriffin
: review+
Details
MozReview Request: Bug 1185486 - Part 2: Prevent message listeners from a failed command from causing an out of order response when a content command hangs. r=ato
40 bytes, text/x-review-board-request
ato
: review+
Details 
First instance was Friday on mozilla-central. Affects both e10s and non-e10s runs.
Hey Julien, I saw a few changes of yours land on Friday, could this be related?
Flags: needinfo?(j.parkouss)
Well, this is not obvious. My changes were on the client (and python runner only) side, so I would be surprised if that could produce intermittents.
Flags: needinfo?(j.parkouss)
(In reply to Julien Pagès from comment #16)
> Well, this is not obvious. My changes were on the client (and python runner
> only) side, so I would be surprised if that could produce intermittents.

Sorry, I should have noticed before but this started before your landings.
Bug 1185486 - Part 1: Use a timeout when waiting on the browser process after a timed out command. r=jgriffin
Attachment #8665804 - Flags: review?(jgriffin)
Bug 1185486 - Part 2: Prevent message listeners from a failed command from causing an out of order response when a content command hangs. r=ato
Attachment #8665805 - Flags: review?(ato)
Comment on attachment 8665805 [details]
MozReview Request: Bug 1185486 - Part 2: Prevent message listeners from a failed command from causing an out of order response when a content command hangs. r=ato

https://reviewboard.mozilla.org/r/20363/#review18303

This looks okay to me, but this code is known to be quite brittle.  Can you do a try run before you land?
Attachment #8665805 - Flags: review?(ato) → review+
Attachment #8665804 - Flags: review?(jgriffin) → review+
Comment on attachment 8665804 [details]
MozReview Request: Bug 1185486 - Part 1: Use a timeout when waiting on the browser process after a timed out command. r=jgriffin

https://reviewboard.mozilla.org/r/20361/#review18309

::: testing/marionette/driver/marionette_driver/marionette.py:691
(Diff revision 1)
> -                returncode = self.instance.runner.wait()
> +                returncode = self.instance.runner.wait(timeout=self.socket_timeout)

We could actually use a much shorter timeout here, probably something like 60s (the default socket timeout is 360 in Marionette, I believe).
https://hg.mozilla.org/mozilla-central/rev/9ffc3758f4ba
https://hg.mozilla.org/mozilla-central/rev/1370508d6c1e
Status: NEW → RESOLVED
Closed: 7 years ago
status-firefox44: --- → fixed
Resolution: --- → FIXED
Target Milestone: --- → mozilla44
You need to log in before you can comment on or make changes to this bug.