1. Proposed title of this feature request
Adding the most-pods expander to the Cluster Autoscaler in RHOCP4

2. What is the nature and description of the request?

• The most-pods expander is an optional configuration for the Cluster Autoscaler. It evaluates which node group can accommodate the highest number of pending pods during a scale-up operation. It is particularly effective in environments with:

  - Diverse workload requirements (e.g., nodeSelector or taints and tolerations).
  - Mixed-instance node pools with varying configurations.

• The most-pods logic aligns with existing expanders like random or least-waste. This addition would extend the expander functionality in the Cluster Autoscaler for OCP 4.

• The expander would be an optional feature, ensuring no disruption to existing autoscaler configurations.

3. Why does the customer need this? (List the business requirements here)

• Many enterprise applications require specific nodes for their workloads due to factors like hardware constraints, licensing, compliance, or performance optimizations. The most-pods expander would ensure that the Cluster Autoscaler prioritizes scaling strategies that maximize the capacity of node groups to host these pods, improving resource utilization and workload placement.

• The most-pods expander provides a more workload-aware scaling mechanism, reducing scheduling failures and retry cycles.

• By explicitly supporting this expander, OpenShift enhances its flexibility for diverse enterprise scenarios.

• Other Kubernetes-based solutions, such as vanilla Kubernetes or GKE, allow flexible expander configurations. By including the most-pods expander, OpenShift reinforces its reputation for enterprise-readiness and user-centric design.

4. List any affected packages or components.

machine-api