Mobile/Testing/05 15 13
From MozillaWiki
Contents
Previous Action Items
Status reports
Dev team
- Done:
- bug 842899 Intermittent java-exception | android.database.sqlite.SQLiteDiskIOException: disk I/O error
- bug 872050 remove /data/local/xpcb from tegras in the cleanup step
- bug 870768 Intermittent testHistoryTab | Awesomebar URL stayed the same
- bug 870056 Enable testClearPrivateData
- Working on:
- bug 872244 Collect /data/anr/traces.txt after Android test runs
- bug 869030 Robocop testDoorHanger and testSystemPages cause future reboots on pandas
- bug 867360 Watcher dies sometimes on Pandas
Rel Eng
- (kmoir) working on mozharness for panda talos tests
IT
- Still working on a higher density chassis. Chassis should be delivered tomorrow.
- Tegra move - mrz is working on a proposal to present to the team for review with 2 options; 1) onsite (office), 2) data center .
A Team
- tegra failure rate: [14.64%]
- this is ignore mochitest-gl which has a 30% failure rate
- panda failure rate: [14.07%]
- this is ignoring mochitest-gl which has 100% failure rate
- robocop has a LOT of failures, so does mochitest and reftest, but not talos!?!
- ignoring reboot step errors, they are silenced and not printed out
- lots of tegra problems in general
- lack of space, bug 872050
- com.android.phone is not running, causing rck tests to fail
x86 automation
- Wrote mozharness script to run mochitests and reftests locally
- Next step is to run these against a rel-eng machine remotely (and see how many emulators can run stably at the same time)
- Will work on bringing up robocop tests next.
Autophone
- Almost finished back filling April tinderbox builds.
- Local test throbber start/stop very noisy for some but not all phones, compared to previous results. Total throbber times, though, are relatively smooth.
- gs3 phones continue to exhibit unreliable behavior with regard to pushing files to the phones and other interactions with the sutagent. Will investigate as time permits.
- Will back fill ranges as time permits for
- bug 819984 Regression in "time to throbber start" - Dec 5,
- bug 825612 Regression in "time to throbber start" and "time to throbber stop" - Dec 22,
- bug 827361 Regression in "time to throbber stop" - Jan 3,
- bug 836886 Regression in "time to throbber stop" (times doubled, intermittently, on Nexus S) - Jan 23
Eideticker
- HTTP request times now part of Eideticker startup test results (e.g. this one for the nytimes load test: http://eideticker.wrla.ch/httplogs/http-log-1368569976.74.json)
- Did a bit of network profiling to determine cause of slow browser startup. Noticed that a bunch of safebrowsing requests were being sent during these tests. We may want to turn this off (not just for eideticker, but for autophone and talos as well).
Round Table
- (blassey) does the readermode icon throw off the eideticker page load test?