Market Problem

• As a Managed OpenShift cluster administrator, I want to use the Priority based expander for cluster-autoscaler to select instance types based on priorities assigned by a user to scaling groups.
  The Configuration is based on the values stored in a ConfigMap
• flags need to be set on the cluster-auto-scaler --expander=priority .
  The user can give a list of expanders in order to have the autoscaler make better choices. for example, we used this in our cost analysis testing --expander=priorty,least-waste
  this would first try to use the priority configmap, and if it finds multiple groups it then tries to find the group with the least waste.

apiVersion: v1
kind: ConfigMap
metadata:
  name: cluster-autoscaler-priority-expander
  namespace: kube-system
data:
  priorities: |-
    10: 
      - .*t2\.large.*
      - .*t3\.large.*
    50: 
      - .*m4\.4xlarge.*

• OCM will need to expose the option to do that.
• Hypershift just sets --expander=priorty flag on the autosacler, and OCM will be responsible of creating the cluster-autoscaler-priority-expander configMap , based on customer configuration.
• This feature should be available in CAPA.

Expected Outcomes

• When creating a MachineSet, customers would like to define their preferred instance types with priorities so that if one is not available a fallback instance type option is available down the list.

https://github.com/kubernetes/autoscaler/blob/master/cluster-autoscaler/expander/priority/readme.md

Documentation Considerations

Documentation will need to be updated to point out the new maximum for ROSA HCP clusters, and any expectations to set with customers.