Goal: Ensure the OLM 1.x rukpak component that will be used in phase 0 for managing operator bundle content is building in the o/operator-framework-rukpak downstream repository.

This component will only be present downstream when admins enable tech preview features in 4.12 OCP clusters.

Note: downstream repositories cannot use github actions as the openshift organization uses a legacy github account, which requires using prow for CI tooling.

Open Question:

• How to handle kube-rbac-proxy dependency downstream?
• Do we continue the existing downstream process (e.g. commit-for-commit) here?
• Do we need to continue commiting the vendor directory into source control downstream?

AC:

• Create an OWNERS file
• Massage the operator-framework/rukpak upstream repository into the downstream o/operator-framework-rukpak repository
• Create a Dockerfile for the provisioner/webhooks binaries that CI will build
• Bootstrap a CI pipeline by contributing this rukpak repository to o/release, and introduce sanity/unit/etc. rehearsal jobs

Stretch:

• The CI pipeline can install the rukpak provisioners + cert-manager after a cluster has been created. This can be done by creating a bash script that's responsible for installing the rukpak manifests, and substituting the container images with the CI built ones using the environment variables the ci-operator exposes. A makefile target can reference that bash script, and the prow job that runs the rukpak e2e tests can specify that makefile target.