On Thu, Oct 11, 2018 at 4:26 AM Cody <codeology.lab@gmail.com> wrote:
>
> Hi everyone,
>
> If I wish to define CPU and RAM allocation ratios at the time of deployment for the Queens release, which one of the following methods is correct?
>
> # Method 1
> parameter_defaults:
>     ComputeExtraConfig:
>         nova::compute::cpu_allocation_ratio: 10.0
>         nova::compute::ram_allocation_ratio: 1.0

This is the right approach but the hiera keys you're using are wrong I
think - check:

https://github.com/openstack/tripleo-heat-templates/blob/stable/queens/puppet/services/nova-scheduler.yaml
https://github.com/openstack/puppet-nova/blob/master/manifests/init.pp#L385

So something like:

  parameter_defaults:
     ComputeExtraConfig:
         nova::cpu_allocation_ratio: 10.0
         nova::ram_allocation_ratio: 1.0

is probably neeeded - for *ExtraConfig overrides the hiera keys must
always align with the interfaces exposed by the underlying puppet
modules.

>
> # Method 2
> parameter_defaults:
>     NovaCpuAllocationRatio: 10.0
>     NovaRamAllocationRatio: 1.0

I don't think this will work because no parameters with these names
are exposed via tripleo-heat templates (see the link to
nova-scheduler.yaml above for example).

Such an interface could be added though, but we don't expose every
puppet/hiera option as heat parameters (a case can probably be made
for these being common enough to add though).

Hope that helps,

Steve Hardy