Is your feature request related to a problem? Please describe.
I want to use the Vertical Pod Autoscaler to more intelligently set resource requests/limits for the workloads in my cluster (primarily to optimize workload placement).
When the VPA needs to adjust a workload's resource requests/limits, it will attempt to do an in-place swap of the resource settings - and if that doesn't work, it'll scale down the resource and scale it back up. To do scale ups and scale downs, it leans on the scale subresource to target (ideally) the owning resource for the workload whose resource requests/limits need to be adjusted.
As is, the Grafana resource does not support the scale resource.
Describe the solution you'd like
The Grafana resource should support the scale resource.
This is adjacent - but not quite related to - this issue. That issue relates to HPA - and there were valid concerns as it pertains to the default Grafana deployment and HPA.
However, I feel like the VPA is a separate, and safer (?) use case as it doesn't adjust replica counts beyond what's configured (and 0).
Am more than happy to try and implement this if this is an acceptable change given the context!
Is your feature request related to a problem? Please describe.
I want to use the Vertical Pod Autoscaler to more intelligently set resource requests/limits for the workloads in my cluster (primarily to optimize workload placement).
When the VPA needs to adjust a workload's resource requests/limits, it will attempt to do an in-place swap of the resource settings - and if that doesn't work, it'll scale down the resource and scale it back up. To do scale ups and scale downs, it leans on the
scalesubresource to target (ideally) the owning resource for the workload whose resource requests/limits need to be adjusted.As is, the
Grafanaresource does not support thescaleresource.Describe the solution you'd like
The
Grafanaresource should support thescaleresource.This is adjacent - but not quite related to - this issue. That issue relates to HPA - and there were valid concerns as it pertains to the default
Grafanadeployment and HPA.However, I feel like the VPA is a separate, and safer (?) use case as it doesn't adjust replica counts beyond what's configured (and
0).Am more than happy to try and implement this if this is an acceptable change given the context!