Feature Overview (aka. Goal Summary)  

HyperShift today bubbles up the available upgrade versions for the control plane via the HostedCluster status resource.

For Managed OpenShift we require the ability to expose a list of available upgrade for node pools individually for customer clusters.

Goals (aka. expected user outcomes)

• HyperShift exposes a list of available upgrade versions for a node pool
• Each node pool should maintain its own list of upgrade-able versions

Requirements (aka. Acceptance Criteria):

• HyperShift exposes a list of available upgrade versions for a node pool
• Each node pool should maintain its own list of upgrade-able versions
• Versions listed should take into account control plane skew
• Updates should be fetched from Cincinnati
• Upgrade risks are evaluated for node pools

Questions to Answer (Optional):

• Placement of upgrade edges, NodePool CR feels correct but HostedCluster CR is another option

Customer Considerations

Customers should not be able to see an invalid upgrade version, ie a version that is Y+ 2 from the control plane.

Documentation Considerations

Documentation for HyperShift is updated where appropriate for self hosted HyperShift

Background

Today for ROSA HCP Cluster Service (OCM) is getting available upgrade versions from CIS and comparing those with the control plane version and returning a list of available upgrades.

We should not have logic to determine upgrade versions in OCM, HyperShift/OCP should be making available safe upgrade versions for node pools