Mobile/Testing/09 26 12
From MozillaWiki
Contents
Previous Action Items
- clint to look into bug 785032
Status reports
Dev team
- trying to reduce rc intermittent failures ... mixed success!
- want to have another look at bug 792212 on the tegras we have found that reftests fail a lot...
Rel Eng
Tegra
- Some tegras are being relocated in MTV this Friday
- no downtime is needed
- bug 793091 Treat all verify.py failures as cause for RETRY - deployed.
- Some minor messaging tweaks to sut_tools code, for sheriff sanity reviewed.
- Working on abstracting sut_tools code to remove "tegra" hardcodes.
- Adding back Android NoIon Builds (bug 793994)
Panda
Foopy-less
Other
- x86 onhold for now bug 750366
IT
- Developing a remote re-imaging process for pandas bug 764534 - in progress
- Working on a higher density chassis bug 777393 - moving forward with current H/W reqs to start designs
- IT will be expanding in SCL1 Datacenter for mobile device housing bug 774477 - in progress
- Setups 2nd chassis for smoketest bug 789516 - Pandas imaged, smoke tests running
A Team
- 8.18% total failure rate. m2, rc, and R4 are the biggest causes of failures.
- Panda board smoketests- still problematic, now we are not getting double agents, the board is not accessible (via SUT or ADB or even ping) after a reboot. Power cycles solve the problem each time.
- One option would be to only use the power cycle relay for reboots (via sut_tools instead of 'rebt' from sutagent)
- Looking into both patching fennec and changing the way reftests are run to resolve bug 792212. Seems like the problem is being triggered (probably just coincidentally) by the profile migration code and/or shared preference reading/writing. Thoughts on approach here?
Autophone
- 3 phones have been trundling along well for the last week.
- A couple phones are not rebooting at all.
- Implementing more recovery code so that a disabled phone will still be pinged every 15 minutes, and, if the ping is answered, the phone will be re-enabled. Ideally we'll figure out what all the underlying problems are, but in the time being hopefully we can recover some phones.
- New, much faster dashboard for rawfennecstartup results. Waiting on review, should be up soon.
Eideticker
- More work on synchronizing profiles with videocaptures ongoing: http://benoitgirard.wordpress.com/2012/09/21/video-synced-profiling/
- More work going into cleanup of devices after eideticker runs (profiling was causing some problems on the Galaxy Nexus) and related robustness fixes
- More startup test work still on todo list before end of Q3.
Round Table
Action Items
- clint to look into bug 785032
- wlach to update patch on bug 792212 and request review from gbrown and gcp