Goal:

Find a way to keep long-running OCP cluster instead of deploying it (from base images or from scratch).

The goal is to see if we can:

• clean all resources related to RHOSO
• keep OCP running
• later, deploy RHOSO, then clean it, then ....

This is mostly needed to stabilize the existing pipelines, while making things faster (we may save ~1h per pipeline if we don't have to manage OCP bootstrap anymore).

Acceptance Criteria:

• We don't kill OCP cluster anymore at the end of the job
• We're able to get a long-running OCP cluster
• Procedure to bootstrap the infra for that usage is documented

Open questions:

• Can we properly clear resources by deleting ArgoCD applications?
  • yes, but we need to use some hooks to clean some leftovers
• Can we benefit of what Ella did in the CI Framework to keep the running OCP?
  • no, it's too deeply linked to the automation in ci-framework/architecture
• How to ensure computes are properly cleaned?
  • by moving to baremetalhost, and provisioning them from scratch from within RHOSO dataplane steps
• What resources do we want to keep? OCP, but what other parts (network, operators, etc)
  • we want to remove everything related to RHOSO. We can keep the vault-secrets operator.