Re: LAVA vs. manual test discrepancy

On 06/19/2012 09:24 AM, Vishal Bhoj wrote:

On 19 June 2012 19:32, Andy Doan <andy.doan@linaro.org mailto:andy.doan@linaro.org> wrote:

On 06/19/2012 08:43 AM, Mathieu Poirier wrote:

    On 12-06-19 07:36 AM, Zach Pfeffer wrote:

        Found a descrepancy between the LAVA results of:

        https://android-build.linaro.__org/builds/~linaro-android/__snowball-ics-gcc46-igloo-__stable-blob/#build=327
        <https://android-build.linaro.org/builds/%7Elinaro-android/snowball-ics-gcc46-igloo-stable-blob/#build=327>

        and Abhishek's results captured in bug:

        https://bugs.launchpad.net/__igloocommunity/+bug/1014650
        <https://bugs.launchpad.net/igloocommunity/+bug/1014650>

        bootloader difference?


    Humm... The problem is in userspace - I don't think the
    bootloader is to
    blame in this case.  If there is a difference, it's probably in the
    steps taken in the manual vs. automated test.  Reporting could
    also be
    wrong - those are just possibilities that are coming to mind.


Could it be somehow to related to where the shell script is run (ie
ADB versus serial)? The only shell scripts we execute in LAVA are
done via ADB.

Even I think it is the adb issue for LAVA to fail. ADB over USB is broken in today's build,it may be broken over network also: https://bugs.launchpad.net/igloocommunity/+bug/1015085

ADB over ethernet works according to the LAVA logs for today. http://validation.linaro.org/lava-server/scheduler/job/22725