Android test cleanup takes too long to time out when device unresponsive
Categories
(Firefox for Android Graveyard :: Testing, defect)
Tracking
(firefox67 fixed)
| Tracking | Status | |
|---|---|---|
| firefox67 | --- | fixed |
People
(Reporter: gbrown, Assigned: gbrown)
References
Details
Attachments
(1 file)
In bug 1532435, there are recent examples of tasks exceeding their max-run-time after multiple 5 minute adb timeouts.
https://treeherder.mozilla.org/logviewer.html#/jobs?job_id=232261308&repo=autoland&lineNumber=1862
https://treeherder.mozilla.org/logviewer.html#/jobs?job_id=232339480&repo=mozilla-inbound&lineNumber=3605
https://treeherder.mozilla.org/logviewer.html#/jobs?job_id=232414336&repo=autoland&lineNumber=1912
Since max-run-time is exceeded, logcats and other diagnostics are lost, which makes it hard to determine what caused the original problem.
It also wastes test time.
I want to reduce timeouts during shutdown and/or avoid additional device interaction after adb timeouts.
Comment 1•7 years ago
|
||
I think that once we get an ADBTimeoutError we should quit trying to run the test and exit as soon as possible. It would be great if we could just catch ADBTimeoutError at the top level of the scripts running the test, emit a generic "device disconnected" or something error the sheriffs could use to classify all of the ADBTimeoutError failures to a single bug.
| Assignee | ||
Comment 2•7 years ago
|
||
You basically did that in bug 1477807! That's working fine for many cases, like in
https://taskcluster-artifacts.net/N3Zub3eWRzWePNJKZx1xeQ/0/public/logs/live_backing.log
| Assignee | ||
Comment 3•7 years ago
|
||
ANR/tombstone checks introduced in bug 1531611 are part of the problem.
| Assignee | ||
Comment 4•7 years ago
|
||
Comment 6•7 years ago
|
||
| bugherder | ||
Updated•5 years ago
|
Description
•