Stateful node rules are for the internal nodes of the metadata and storage nodes, and come in two forms:

• nodeSet, where each GMS and DN adhere to the rules of nodeSet to deploy internal nodes
• rolling, which is only for DN and will deploy internal nodes in a stacked manner on all available nodes within the Kubernetes cluster (for testing purposes), maximizing resource utilization.

spec:
  topology:
    rules:
      components:
        #  **Optional**
        #
        # GMS deployment rules, defaults to DN
        gms:
          # Stacked deployment structure, the operator attempts to stack deploy in the nodes specified by the node selector
          # Each child node of the storage node to achieve higher resource utilization, for testing purposes only
          rolling:
            replicas: 3
            selector:
              reference: zone-a
          # Node set deployment structure, you can specify the node group and node selector of each DN's child node,
          # to achieve high availability deployment structures such as cross-region, cross-city, etc.
          nodeSets:
            - name: cand-zone-a
              role: Candidate
              replicas: 1
              selector:
                reference: zone-a
            - name: cand-zone-b
              role: Candidate
              replicas: 1
              selector:
                reference: zone-b
            - name: log-zone-c
              role: Voter
              replicas: 1
              selector:
                reference: zone-c