Summary:

• Investigate what work need to be done to simplify maintenance by merging the codebase, enabling easier application of common patches (e.g., dependency updates, shared features like the Topology CR), while preserving the separation of the placement controller to avoid changes to core functionality.

Goal:

Answer following questions:

• Do we want to use this step to rename the nova-operator repo to compute-operator?
  • Do we transform nova-operator to a multi group operator to hold both nova.openstack.org and placement.openstack.org API groups? Or do we rename the group to compute.openstack.org and move Nova* CRDs and PlacementAPI CRD to that new group?

Investigate and create following tasks:

•  agree on the new directory structure. Will we rename /api to /apis and have a separate subdir one for nova and one for placement. Similarly how https://github.com/openstack-k8s-operators/openstack-operator/tree/main/apis looks like. Or we just move the placement CRD def into the existing api/v1beta1 dir? Similarly about pkg, tests and other directories
• agree on testing strategy: do we run all tests (envtest, kuttl, tempest) for every commit, or we will be more selective (e.g. only run placement envtest, placement kuttl, and global tempest on changes that are placement controller only?)?
• modify openstack-operator not to depend on placement-operator any more (move the api def refs, change bindata handling, change operator install logic)
• create automation or documentation that cleans up placement-operator from a running deployment after upgrading to a version where placement-operator logic is moved to nova-operator.
• create a story to indicate early to rel-del that the will not need to package and deliver placement-operator when we are ready.

TimeBox:

• One sprint

Deliverables/Outcomes:

• Create all tasks in epic https://issues.redhat.com/browse/OSPRH-14957 and discuss this with brother audiance on round table meeting