A managed Kubernetes cluster with a node group, an RBAC user role, and a cluster user, then the user's kubeconfig read off status and saved to a file. Reuses client and the real folder_id / subnet_id from the Networking example.

The API always takes real ULIDs, so controlPlaneLocations needs no special handling here — that gotcha applies only to the kc CLI, which resolves names for top-level spec fields but not inside nested lists. Note the 3600-second waits: a control plane takes tens of minutes, and Kubernetes changes are serialized per organization.

As a single transaction

The same stack created atomically. The cluster and the role get client-generated ULIDs up front so the node group and the user can reference resources that don't exist yet (read the kubeconfig separately afterwards, once it has reconciled):

Related pages

Resource docs

• Kubernetes Cluster resource docs
• Node Group resource docs
• User Role resource docs
• Cluster User resource docs
• VPC Subnet resource docs
• Transaction resource docs

More in this section

• Networking
• VMs
• VPN
• Load Balancing
• S3
• Examples

Reference

• Interactive API reference (Swagger)