1194280 - [raptor] Write a Marionette test with Raptor to set mozSettings will freeze the device (Flame)

Status: RESOLVED WORKSFORME

(Firefox OS Graveyard :: Gaia::PerformanceTest, defect)

Description

[Greg Weng [:snowmantw][:gweng][:λ]]

Attachment: Test will freeze the device

I've found that if I port the existing Marionette test with setting some mozSettings, the device will be frozen (just like my attachment). And the console reports it's 

> Polling socket recv() timeout!

I've test it with only other behaviors without setting mozSettings. It worked fine, even with the `execAsyncScript`. So I really don't know where I could star to fix it.

Comment 1

[Greg Weng [:snowmantw][:gweng][:λ]]

And another reason I feel it's hard to debug is because I don't know how to set the VERBOSE flag of marionette as true, which can print extra `console.log` to help us to debug. From Raptor's MDN page and the attempts I've tried it just doesn't work.

Comment 2

[Greg Weng [:snowmantw][:gweng][:λ]]

CC Bobby and NI Eli.

Comment 3

[Greg Weng [:snowmantw][:gweng][:λ]]

I may prepare more logs and video ASAP. Meanwhile it's really easy to reproduce it with the test I've attached. Maybe the test is with some flaw to freeze the device, but I'm not sure. Partly because it's from a marionette test that works well (or the "well" actually means the problems are covered).

Comment 4

[:Eli Perelman]

Using the following gist I can get past the socket timeout error, but once the lock is set, the screen locks, and the result never seems to hit onsuccess/onerror, and everything hangs.

https://gist.github.com/eliperelman/dc7c55d3ff9eb4bd63e4

Comment 5

[Greg Weng [:snowmantw][:gweng][:λ]]

Thanks Eli. Since the alternative works well and doesn't reproduce the bug, I close it now.