Mobile/Testing/06 20 12: Difference between revisions

From MozillaWiki
Jump to navigation Jump to search
No edit summary
 
(18 intermediate revisions by 7 users not shown)
Line 5: Line 5:
** done
** done
* ted to get apk rename patch landed {{bug|761454}}
* ted to get apk rename patch landed {{bug|761454}}
** done
* ted to get MAC address issue fixed on pandas {{bug|731670}}
* ted to get MAC address issue fixed on pandas {{bug|731670}}
* joel to figure out what needs to be changed on the automation side for ARMv6
* joel to figure out what needs to be changed on the automation side for ARMv6
Line 11: Line 12:
= Status reports =
= Status reports =
== Dev team ==
== Dev team ==
* investigated bug 756817 Fix and reenable tcheck2 and tcheck3
* bug 764044 Robocop: testNewTab fails silently
* bug 764489 Silence exceptions from SUTAgent so they don't appear in the test logs
* bug 764147 SUT agent: near 100% cpu during StartPrg2
* bug 756264 devicemanager: reduce chmod's
* {{bug|766619}} - Pull in compatibility libraries for android sdk (needed for {{bug|759041}}


== Rel Eng ==
== Rel Eng ==
* Running armv6 builds through staging after packaging bug got fixed
* Staging the following (Likely deploy later today or tomorrow)
** {{bug|766030}} verify.py should run instead of updateSUT/cleanup.py for tegra jobs
*** This also ensures that we actually *always* ping the tegra before anything else relies on it, and fail out (killing buildbot) if it fails.
*** NB: Requires buildbot changes to happen after the foopy changes, so this will not be fully deployed until early next week.
** {{bug|766029}} pingTegra from sut_lib needs to support non-mac machines.
*** Mostly prep for foopy-on-linux, but it does have the added benefit that it requires 5 successful pings to consider the tegra "up", instead of just 1 good in 5 tries.
** {{bug|714101}} remove the fencp.apk and ffxcp.apk from the tegras
*** Really just cleanup, and having less stuff running on a tegra.
* Awaiting Dev-Team to hand over a new SUTAgent apk to stage as soon as it is ready.


== IT ==
== IT ==


== A Team ==
== A Team ==
* [[https://docs.google.com/spreadsheet/ccc?key=0ArS97F99-BEZdE54R1JnOWVZdy1Hd2pfa29SaFQ4X1E 25.47%]] total failure rate!!!!
* [[https://docs.google.com/spreadsheet/ccc?key=0ArS97F99-BEZdDFtSXVrZ1JIQlVpNzg1a3JtQ1B6OEE 34.24%]] total failure rate, last week was [[https://docs.google.com/spreadsheet/ccc?key=0ArS97F99-BEZdE54R1JnOWVZdy1Hd2pfa29SaFQ4X1E 25.47%]]
** should we shoot for 50% next week?!?
** crashes are related to {{bug|763166}} and {{bug|749917}}


* {{bug|761125}} - turned on 250,000+ mochitests
* {{bug|761125}} - turned on 250,000+ mochitests
Line 23: Line 42:


== S1/S2 Automation ==
== S1/S2 Automation ==
* Server and phones moved to new subnet; afterward experienced some big problems with adb stability. No idea if or how it's related to the move.
* Atm only one GS2 is stable: http://mrcote.info/phonedash/#/org.mozilla.fennec/throbberstart/local-twitter/04_46_65_fd_2f_e1_samsung_gs2/2012-06-13/2012-06-20
== Eideticker ==
== Eideticker ==
* Eideticker automatic runs back up: http://wrla.ch/eideticker/dashboard/
* mcote keeping an eye on them until wlach gets back from vacation on July 2.


= Round Table =
= Round Table =
* running tests on debug builds?
= Action Items =
= Action Items =
* Asa to figure out when automated testing for metro will come online
* joel to get a list of crashes that are hurting us in automation
** crashes are related to {{bug|763166}} and {{bug|749917}}
* clint to determine if network congestion is an issue in hazor andto see if there is an easy IT solution if it is
* clint to decide if beagleboards are a reasonable testing platform for ARMv6

Latest revision as of 17:30, 22 June 2012

Previous Action Items

  • joduinn to arrange time for ATeam and RelEng to brainstorm ways to green up test runs
    • done
  • mfinkle to file a bug to shut up SUT agent
    • done
  • ted to get apk rename patch landed bug 761454
    • done
  • ted to get MAC address issue fixed on pandas bug 731670
  • joel to figure out what needs to be changed on the automation side for ARMv6
  • joduinn to figure out what's needed to get ARMv6 tests running on ARMv7

Status reports

Dev team

  • investigated bug 756817 Fix and reenable tcheck2 and tcheck3
  • bug 764044 Robocop: testNewTab fails silently
  • bug 764489 Silence exceptions from SUTAgent so they don't appear in the test logs
  • bug 764147 SUT agent: near 100% cpu during StartPrg2
  • bug 756264 devicemanager: reduce chmod's
  • bug 766619 - Pull in compatibility libraries for android sdk (needed for bug 759041

Rel Eng

  • Running armv6 builds through staging after packaging bug got fixed
  • Staging the following (Likely deploy later today or tomorrow)
    • bug 766030 verify.py should run instead of updateSUT/cleanup.py for tegra jobs
      • This also ensures that we actually *always* ping the tegra before anything else relies on it, and fail out (killing buildbot) if it fails.
      • NB: Requires buildbot changes to happen after the foopy changes, so this will not be fully deployed until early next week.
    • bug 766029 pingTegra from sut_lib needs to support non-mac machines.
      • Mostly prep for foopy-on-linux, but it does have the added benefit that it requires 5 successful pings to consider the tegra "up", instead of just 1 good in 5 tries.
    • bug 714101 remove the fencp.apk and ffxcp.apk from the tegras
      • Really just cleanup, and having less stuff running on a tegra.
  • Awaiting Dev-Team to hand over a new SUTAgent apk to stage as soon as it is ready.

IT

A Team

  • bug 761125 - turned on 250,000+ mochitests
  • bug 754873 - logcat from reftest/mochitest landed, working on talos bits now

S1/S2 Automation

Eideticker

Round Table

  • running tests on debug builds?

Action Items

  • Asa to figure out when automated testing for metro will come online
  • joel to get a list of crashes that are hurting us in automation
  • clint to determine if network congestion is an issue in hazor andto see if there is an easy IT solution if it is
  • clint to decide if beagleboards are a reasonable testing platform for ARMv6