Outcome Overview

Stop maintaining existing Machine API controllers

This will reduce the total maintenance cost of machine management in OpenShift/HyperShift and related products. A unified Machine management approach will accelerate both OpenShift and HyperShift teams by sharing resources contributing to the same upstream projects. With OpenShift teams also contributing to Cluster API, HyperShift can spend less time maintaining infrastructure management and focus more on the core project.

Provide a seamless migration of resources to Cluster API

We have (at time or writing) around 157k Machines across 14k clusters reporting into telemetry. These existing users must be migrated to Cluster API to allow us to deprecate the Machine API.

We want to minimise the impact to end users and in particular, service delivery, so we must make the migration process easy, and ideally, transparent to end users. They should either not notice, or not care.

Users with automation built on top of Machine API will be provided with migration documentation and will be expected to migrate over a period of at least 3 minor (Y) releases. The deprecation cycle should start on an EUS (even) release to provide the most time for EUS consumers to migrate. 

Deprecate Machine API

At some point, we want to stop installing new clusters with Machine API. We want to promote users to migrate to Cluster API so new clusters should use the newer APIs by default. No new features will be added to Machine API once it has been declared deprecated.

Leverage CAPI to introduce new features to OpenShift

CAPI provides support for extensions such as MachineDeployments and MachinePools. These features could be added to OpenShift to improve the scope of Machine management within OpenShift.

Other features such as deploying additional clusters via CAPI are strictly out of scope for this project.

Success Criteria

Customer can use CAPI to lifecycle ROSA+HCP, ARO+HCP
Cluster-admin can install and managed CAPI infra providers(AWS, Azure) through MCE.

Expected Results (what, how, when)

CAPI for machine management(in standalone OCP) and lifecycle management(ARO, ROSA HCP) will help drive the adoption of ARO HCP, CAPI driven deployments in multiple regions. Increase the adoption of ARO HCP. 

Post Completion Review – Actual Results

After completing the work (as determined by the "when" in Expected Results above), list the actual results observed / measured during Post Completion review(s).