+++ This bug was initially created as a clone of Bug #1293078 +++ https://treeherder.mozilla.org/logviewer.html#?job_id=81559916&repo=autoland [task 2017-03-03T22:43:42.777671Z] 22:43:42 INFO - 267 INFO Running WINDOW_LOCATION_RELOAD script to check for a submission [task 2017-03-03T22:43:42.778047Z] 22:43:42 INFO - 268 INFO TEST-FAIL | toolkit/components/passwordmgr/test/mochitest/test_formless_submit_navigation_negative.html | The author of the test has indicated that flaky timeouts are expected. Reason: Testing that a message doesn't arrive [task 2017-03-03T22:43:42.778181Z] 22:43:42 INFO - Buffered messages logged at 22:43:29 [task 2017-03-03T22:43:42.778500Z] 22:43:42 INFO - 269 INFO TEST-PASS | toolkit/components/passwordmgr/test/mochitest/test_formless_submit_navigation_negative.html | Done waiting for captures [task 2017-03-03T22:43:42.778717Z] 22:43:42 INFO - 270 INFO Starting testcase with script HISTORY_BACK: {"document":"<input type=password value=\"pass2\">","wouldCapture":true} [task 2017-03-03T22:43:42.778896Z] 22:43:42 INFO - Buffered messages finished [task 2017-03-03T22:43:42.779142Z] 22:43:42 INFO - 271 INFO TEST-UNEXPECTED-FAIL | toolkit/components/passwordmgr/test/mochitest/test_formless_submit_navigation_negative.html | No formSubmissionProcessed should occur in this test [task 2017-03-03T22:43:42.779449Z] 22:43:42 INFO - submissionProcessed@toolkit/components/passwordmgr/test/mochitest/test_formless_submit_navigation_negative.html:29:3 [task 2017-03-03T22:43:42.779832Z] 22:43:42 INFO - 272 INFO got: [{"hostname":"http://test1.mochi.test:8888","formSubmitURL":"http://test1.mochi.test:8888","usernameField":null,"newPasswordField":{"name":"","value":"pass2"},"oldPasswordField":null}] [task 2017-03-03T22:43:42.780170Z] 22:43:42 INFO - 273 INFO Running HISTORY_BACK script to check for a submission

This is almost always on Android Debug, and the same failures as bug 1293078, where I increased the timeout. That change seemed to work for a few months.

Matt, do you have thoughts on how to make this more reliable? It occurs to me we could wait longer, skip the test on Android Debug (other platforms are all mis-stars), or try to eliminate the timeout somehow...I'm tempted to skip on Android Debug.

I would prefer increasing the timeout again only for (Android) debug builds since bug 1293078 made running the pwmgr tests much slower. Disabling for Android Debug would be my next preference.

This increases the total Android run time for the test to about 210 seconds. https://treeherder.mozilla.org/#/jobs?repo=try&revision=04a6bb5a0330c6ef915f0c210672214d14c93cca

Comment on attachment 8857152 [details] [diff] [review] increase android wait time Review of attachment 8857152 [details] [diff] [review]: ----------------------------------------------------------------- r=me if it works

Pushed by gbrown@mozilla.com: https://hg.mozilla.org/integration/mozilla-inbound/rev/3fb2e54c398e Increase wait time in test_formless_submit_navigation_negative, on Android only; r=mattn