Mobile/Testing/10 31 12

From MozillaWiki
Jump to: navigation, search

Previous Action Items

  • [DONE] blassey and joduinn to take offline x86 priority
  • (kmoir) to get info about why some pandas are failing some things intermittently
  • (callek) to followup on the android test orange rate stuff I deferred away from mtg

Status reports

Dev team

    • Talos is looking better but there's still...
    • bug 796914 Intermittent "talosError: 'stack found after process termination ...
    • Trying to get xpcshell tests running again

Rel Eng

Tegra

Panda

Other

IT

  • Pandas - Android
    • Developing a remote re-imaging process for pandas bug 764534 - Ready for integration with a-team & RelEng
      • Integrating Buildbot/Foopies with BMM/MozPool/Lifeguard bug 805016 - New
    • 48 Production ready Pandas bug 800038 - all imaging done, RelEng working on these boards, their tracker bug 799698
  • Pandas - b2g
    • Image 12 Pandas by hand for b2g (Chassis 3) bug 797629 - DC Ops working with Clint on issues
  • Other Updates
    • Tracker bug - 800 production ready panda boards bug 799922
      • 24 Panda chassis will be in SCL1 by end of day today. DC Ops will work to get these racked and ready for configuration over the next several days, probably into next week.
      • These 24 chassis house 264 panda boards.
      • A-Team has instructed IT to install the android image on these boards. This does not necessarily mean that all 264 boards will remain imaged for android, IT will reimage as instructed.
    • Order parts for 73 Panda Chassis bug 799808 - All parts are in but 49 power supplies, looking for alternative vendors and power supplies. The assembly vendor can not ship us any more chassis until we get more power supplies, current power supply is back ordered until January 2013.
    • Working on a higher density chassis bug 777393 - On hold due to recent influx of Panda activity

Jake is out for PTO/Wedding from 10/18-11/2.

A Team

  • 13.18% total failure rate. M2 and M8 are problematic when they usually are not.
  • panda boards are failing in staging
    • appears sutagent/watcher are spinning the cpu even after a fresh reboot causing lots of dalvikvm GC messages, very hard to reproduce locally.
    • could this be related to changes in dns (we added a zero making the name 33 characters long) or other hardware/infra/network changes since the smoketest work that was done?
  • apc.io boards
    • by hand saw robocop tests running on them last week
    • works moderately well with hacked up sutagent to adjust file permissions/owners and using sdcard for test data and profile
    • fails to work on /data/local due to permissions. sdcard won't mount on boot and requires manual physical intervention, this is a no go until we get tests working from /data/local.

Autophone

  • Pulse translator was down but is now back up again. All 6 phones running fine aside from that outage, but they recovered automatically.
  • MozPool work continues but shifting focus to getting it ready for B2G.

Eideticker

  • Current focus is on enabling useful eideticker tests for B2G, but some work/discussion on mobile still occurring.
  • Adjusted sync periods, test order of dash updates for maximum utility of gfx/mobile folks (bug 805867)
  • Various other ideas discussed during a-team work week in Toronto with Benoit Girard, particularly bug 805196 (make panning tests harder) and adding better canvas tests. Will hopefully implement soon (getting something working with B2G is higher priority).
  • In SF for a few weeks; ping me if you're there!

Round Table

Action Items

  • jmaher to help kmoir with the failing pandas
  • kmoir to send blassey an email with specific bugs they could use dev eyes on
  • joduinn to contact apc about ability to keep up with demand.
  • gbrown to look into m2 and M8 failures: bug 792873 and bug 804885
Retrieved from "https://wiki.mozilla.org/index.php?title=Mobile/Testing/10_31_12&oldid=484387"