Background

The main problem with the component architecture right now is that cluster api providers (core and infrastructure) code and manifests (Core CAPI CRDs, RBACs, Deployments…) are actually in different repositories. The code lives in the `openshift/cluster-api-provider-*` forks (and branches), while the manifest templates are pulled from those repos manually (via the ./hack/assets tool) and are kustomised (+annotated etc.) once in a while. This leads to the fact that it is impossible to perform an atomic update of the system without disrupting its functionality.

Description

Write an enhancement to redesign the CAPI on OCP architecture to improve the assets management issues we are currently facing.

See this document for more details on the possible pros and cons for the various redesign approaches we took into consideration.

Possible Solutions/Areas to explore

• Investigate the pros/cons of the cluster-api-operator and determine if it can be removed/what this would mean for our other operators
• Can Kustomize be used to modify manifests in the provider repositories, if so, what benefit does this provide for us (POC)
• Using configmaps to transport manifests from providers into the operator and the operator deploying from those configmaps (POC around CRDs)
• Investigate could operators/deployments be applied via OLM rather than CAPI operator (discussion of pros/cons)
• Write enhancement for plan going forward (based on other tasks)

Definition of Done

• Write and post for review an enhancement proposal for the CAPI assets management redesign.
• Make sure to provide alternatives section with valid alternatives in the enhancement
•  

(Note addressing feedback and merging this will be covered in a separate card)