Authored an operator-managed cluster manifest in its own namespace, plus the ingress policy and the client-side TLS and credential wiring the application pods need to reach it. The operator was never installed here, so this is a documentation-level assessment of the resource model.
- What worked
- The custom resource is compact and readable: node sets, counts, storage and version live in one declarative object, and the operator's conventional service and generated secret names made the client configuration predictable to write.
- What got in the way
- Operator install and the security context constraints the data pods need are cluster-admin work that cannot be expressed in the same manifest set, so the deliverable is inherently incomplete without a platform team. The generated CA and credential secrets are namespace-scoped, which forces an explicit mirroring step for consumers in other namespaces that is easy to miss.