canonical-ci-engineering team mailing list archive

[Francis Ginther <francis.ginther@xxxxxxxxxxxxx>]

Thomi,

NIcholas and I discussed these failures last week and some options for
debugging. The first step is to create some infrastructure to run all of
the core-apps jobs en masse to more easily identify which ones are
regularly failing and provide a means to do more systematic regression
testing of changes to the jobs themselves. Then it's just the process of
looking at the specific failures.

Is there a specific priority to this work or is it part of a bigger goal? I
expect to have the infrastructure in place this week to collect the
failures and perform regression testing. Anything beyond that depends on
the failures encountered.

Francis

On Thu, Sep 18, 2014 at 4:43 PM, Thomi Richards <
thomi.richards@xxxxxxxxxxxxx> wrote:

> Hi,
>
> On Fri, Sep 19, 2014 at 2:52 AM, Nicholas Skaggs <
> nicholas.skaggs@xxxxxxxxxxxxx> wrote:
>
>> At the moment, clock autolands on the device, but nothing else does. The
>> straightforward job changes resulted in the tests failing, not running etc
>> so they were not enabled. It's unclear why this is occurring inside jenkins
>> and will take some effort for each application to diagnose. The failures
>> themselves seemed related to infrastructure and not actual test failures,
>> though there are some actual test failures as well I'm sure.
>>
>> Switching to autolanding on devices with test failures results in MP's
>> piling up and not landing; the jobs need to work before we can enable them.
>>
>
> That seems fair to me. Can someone from CI comment on this please?
>
> Cheers,
>
>
> --
> Thomi Richards
> thomi.richards@xxxxxxxxxxxxx
>



-- 
Francis Ginther
Canonical - Ubuntu Engineering - Continuous Integration Team