[Enhancement] Detect scale-in and drop CN node from FE

[yandongxiao]

Description

Fixes: #550

Background

When a scale-in operation happens:

For BE with shared-nothing deployment, the cleanup actions include:

1. Decommission operation (time-consuming)
2. Stop pod operation (need to wait for termination of BE pod)
3. Drop BE from SR

For CN with shared-data deployment, the cleanup actions include:

1. Stop pod operation (need to wait for termination of CN pod)
2. Drop CN from SR

Among them, Decommission and Stop pod are time-consuming operations, and the Operator cannot wait for the operation to complete in a blocking manner.

In summary, the cleanup actions Stop pod and Drop node are likely not in the same tuning loop. Therefore, we cannot rely on whether the current operation is a scaling-in operation to execute these logics.

How

At the end of each tuning cycle, the operator performs the following validation steps:

1. Verify Replica Consistency

Compare the replicas field in the StarRocksCN Custom Resource Definition (CRD) with the spec.replicas field of the corresponding CN StatefulSet. These values must be identical.

2. Validate Running Pod Count

Compare the replicas field in the StarRocksCluster CRD with the number of running and ready CN pods. These values must match.

3. Confirm Revision Hash Match

Ensure that the controller-revision-hash label on all running CN pods exactly matches the status.updateRevision field of the CN StatefulSet.

4. Perform DROP COMPUTE NODE

If all three conditions are met, the operator will compare the list of compute nodes registered in the Frontend (FE) cluster against the current running CN pods. Initiate the DROP COMPUTE NODE operation for any nodes that are no longer present in the pod list.

Checklist

For operator, please complete the following checklist:

• run make generate to generate the code.
• run golangci-lint run to check the code style.
• run make test to run UT.
• run make manifests to update the yaml files of CRD.

For helm chart, please complete the following checklist:

• make sure you have updated the values.yaml
  file of starrocks chart.
• In scripts directory, run bash create-parent-chart-values.sh to update the values.yaml file of the parent
  chart( kube-starrocks chart).