In preparation for the deploying the updates to Autophone, I've turned off pulse notifications (which also implies no try build processing) and shutdown the mochitest-media tests to allow the nexus-4-jdq39-3 device to catch up on s1s2 and webapp tests before updating with the latest patches. Patches from bug 1150111, bug 1065599, and bug 1118751 will be deployed. I will also attempt to bring the nexus-5-kot49h-3 device back on line testing mozilla-inbound in parallel with nexus-5-kot49h-1 to see if we can replace the failing nexus 5 device and to get a comparative baseline. I've had to reset the /data/misc/wifi/wpa_supplicant.conf files on nexus-s-3 and nexus-5-kot49h-1. Not sure why they have been changed/corrupted. Once nexus-4-jdq39-3 finishes its perf related test queue, I'll complete the deployment. ETA is 6-7 hours.

Note that after the restart the nexus-5-kot49h-1 reverted to the old behavior before the factory reset for builds beginning with 2015-04-23 08:19:57. The job run time increased back to approximately an hour and the throbber stop time also decreased again.

This is the temporary test manifest used until we can land Bug 1153992 and enable remote tests, nytimes and further partition the tests across the devices. Restarted at 2015-04-26 17:23. Back filled builds for all repos to 2015-04-26 02:00:00. Back filled nexus-5-kot49h-3 (which replaced nexus-5-kot49h-1) and nexus-7-jss15q-2 (which replaced nexus-7-jss15q-1) for a limited set of mozilla-inbound builds for 2015-05-{17,19,20,21,22,23,24,25,26}. Note the back filled builds will all use the new preferences and may change behavior compared to the earlier runs which did not have the updated preferences.

Note that Android 2.3 no longer fails with the missing minidump directory but we do have a significant regression in throbber start, throbber stop in First run and at least throbber start for Second run for blank, twitter and webappstartup. The prefs change and the root=True when accessing the profile directory shouldn't have affected the webappstartup since we don't yet (until bug 1152392) control the webapp profile. I'm not sure what is up unless the s1s2test is somehow affecting the webappstartup test. We'll know more as we get more data.

nexus-5-kot49h-3 appears to be very noisy for local-twitter second run throbber start and somewhat noisy for local-twitter second run throbber stop. This causes the measurements to be rejected and a second run of 8 iterations is performed. If measurements appear to be missing from phonedash, select "Include rejected results".