Hi Martins,
Oh - this test has a dependence on the number of CPUs on the
test machine, I missed that earlier. We could either filter
so that we check the percpu inst count matches hinv.ncpu, or
maybe we could use $LINUX_STATSPATH to setup a deterministic
output? (bit like tests 732, 747 or 885)
cheers.
----- Original Message -----
> Nathan,
>
> On 12/15/14 12:53 AM, Nathan Scott wrote:
> >
> > Looks good, thanks Martins. Do you want to add this to your working
> > branch?
> > And where did we get up to with hotproc - are you happy with that & ready
> > to
> > merge? (could you send me the git URL for the latest version of that code,
> > if so?)
> >
> commit 6cb6b341e5a3d0dd37011ee79ecc9a5194b0ac1f
> Author: Martins Innus <minnus@xxxxxxxxxxx>
> Date: Mon Dec 15 15:26:44 2014 -0500
>
> Add qa test 955
> Test that mixing fetches for static and dynamic metrics work as
> expected
> The first test replicates a failure from a bug introduced during
> dymanic metric core changes
>
> qa/955 | 84
> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> qa/955.out | 44 +++++++++++++++++++++++++++++++
> qa/group | 1 +
> 3 files changed, 129 insertions(+), 0 deletions(-)
>
>
>
> This qa test is in martins_working. Hotproc is pretty close. Writing a
> qa test now.
>
> Thanks
>
> Martins
>
|