@Rocky how do i add observer nodes to helm chart?

To add observer nodes to your StarRocks cluster using the Helm chart (which typically uses the StarRocks Kubernetes Operator), you generally manage the number of FE nodes by adjusting the

replicas

count in your

values.yaml

file. In StarRocks, FE nodes can be Followers (participating in leader election and metadata writes) or Observers (read-only nodes that synchronize metadata). For a high-availability production environment, it is common to have 3 Followers and additional Observers to handle read-only query load. Step 1: Update

values.yaml

In your

values.yaml

file, increase the

replicas

in the

starrocksFESpec

section. For example, to have a total of 5 FE nodes (typically 3 Followers and 2 Observers):

yaml
starrocks:
  starrocksFESpec:
    replicas: 5  # Increase this count to add more FE nodes
    resources:
      requests:
        cpu: 4
        memory: 8Gi
    storageSpec:
      name: fe-meta
      storageSize: 10Gi

Step 2: Apply the Changes Run the helm upgrade command to apply the new configuration:

bash
helm upgrade <release_name> starrocks/kube-starrocks -f values.yaml

Step 3: Verify the Node Roles The StarRocks Operator automatically handles the registration of new FE pods. Once the new pods are running, you should check their roles by connecting to your StarRocks cluster via a MySQL client:

sql
-- Connect to the cluster and run:
SHOW PROC '/frontends';

In the output, check the

Role

column: * LEADER: The primary FE handling writes. * FOLLOWER: Nodes that can be elected as leader (usually 3 nodes total for HA). * OBSERVER: Nodes used for scaling read performance. Important Notes: 1. Operator Logic: The StarRocks Operator typically registers the first 1, 3, or 5 nodes as Followers to maintain a quorum. Subsequent nodes are often registered as Observers. 2. Manual Registration: If the Operator does not automatically register the new nodes as observers (depending on your specific Operator version), you can manually add them via SQL:

sql
    ALTER SYSTEM ADD OBSERVER "fe_host:9010";

_(In a Kubernetes environment,

fe_host

would be the FQDN or IP of the specific FE pod.)_ 3. Read Load Balancing: If you have observer nodes, ensure your load balancer or proxy (like ProxySQL or the FE Service) is configured to include these nodes to distribute query traffic. References • quick_start/helm.md • faq/operation_maintenance_faq.md • deployment/plan_cluster.md • deployment/deploy_manually.md • administration/management/FE_configuration.md

observer node is not supported by operator right now.