Advanced Operations

Operating a fleet

When you manage several clusters, look at them together rather than one by one. kubeopera-api's federation API aggregates data across every registered cluster:

Endpoint	Returns
GET /api/v1/federation/overview	Health, nodes, usage and cost for every cluster.
GET /api/v1/federation/health	Health scores and issues, fleet-wide.
GET /api/v1/federation/cost	Cost by cluster, namespace and workload.
GET /api/v1/federation/anomalies	Recent anomalies across the fleet.
GET /api/v1/federation/optimizer	Right-sizing opportunities across the fleet.

The Multi-Cluster view is built on these endpoints, and to AI agents through the get_multi_cluster_overview tool.

Tips for fleets:

• Name clusters consistently (<env>-<region>) so sorting and filtering stay useful.
• Use the same environment structure for every cluster; let overlays capture the differences.
• Review the Multi-Cluster treemap weekly to spot clusters drifting in cost or usage.

Working with GitOps

Everything Tantilium Cloud Runner (TCR) deploys, its own services and every tenant app, is delivered by Flux from Git. A few habits make this smooth:

• Change Git, not the cluster. A manual kubectl edit is reverted on the next reconcile. Change the dashboard or the repository instead.

• Expect a short delay. Changes apply on Flux's next reconcile (typically within a minute). For an immediate apply, reconcile explicitly:

  flux reconcile kustomization <name> --with-source

• Roll back with Git. Reverting a commit rolls the change back everywhere it applied. Apps also have one-click rollback in the dashboard, which does the same thing for you.

• Pause for maintenance. flux suspend kustomization <name> stops reconciliation while you investigate; flux resume starts it again.

To see how an app's configuration becomes resources in a tenant's vCluster, read app-controller.

Next steps

• Setup: how environments are structured.
• kubeopera-api: the full API.
• Clusters: add and manage clusters.