Management Cluster

Status: sketch — not implemented. Carried over in raw form from the original C5C3 architecture document. Today every component named below that exists runs on the single management cluster next to the control plane; the dedicated cluster and the observability UI do not exist in this repository.

The original document separates a management cluster from the control-plane cluster: GitOps, secrets, and cross-cluster observability live on their own cluster, which then deploys workloads into the control-plane, hypervisor, and storage clusters via kubeconfig Secrets and syncs credentials everywhere through ESO.

In the multi-cluster target picture this cluster is the one labelled CobaltCore Operation and Monitoring.

Sketched components

• Flux Operator + FluxCD as the GitOps hub for all clusters, deploying remotely via kubeconfig Secrets. Implemented today as the in-cluster FluxInstance.
• OpenBao as the central secret store (HA, 3x Raft) for every cluster. Implemented today as the single-cluster proving instance.
• External Secrets Operator running in all clusters, each with a ClusterSecretStore pointing back at the management-cluster OpenBao.
• Greenhouse for centralized monitoring and alerting across clusters.
• Aurora Dashboard as the unified management UI over servers, networks, and volumes.

Open questions

Splitting the implemented single-cluster stack would need a story for where the operators run, which cluster owns the ControlPlane CR, and how the existing target-cluster placement relates to a remote control-plane cluster.

Source

• Management components
• Architecture overview