During a rolling update, kube-controller-manager performs zero filling on pods whose monitoring data cannot be collected. This may add excess pods. You can prevent this issue by configuring an annotation for workloads.

Solution

tip

To use this feature, you need to upgrade the metrics-server add-on of clusters to v0.7.1-vke.2. For details, see Upgrading add-ons.

You can add the HPARollingUpdateSkipped configuration when configuring workloads to skip HPA during rolling updates to prevent extra pods from being created. Sample configurations:

apiVersion: apps/v1​
kind: Deployment​
metadata:​
  name: demo-workload​
  annotations: ​
    HPARollingUpdateSkipped: "true"# Skip HPA during rolling updates​
spec:​
  replicas: 1​
  selector:​
    matchLabels:​
      app: demo-workload​
  template:​
    metadata:​
      labels:​
        app: demo-workload​
    spec:​
      containers:​
      - name: workload​
        image: doc-cn-beijing.cr.volces.com/vke/nginx-demo:v1.0​
        resources:​
          requests: ​
            cpu: "500m" ​
            memory: "1024Mi"​
          limits:​
            cpu: "1" ​
            memory: "2Gi"​

Last updated: 2026.03.31 10:58:10

Public security network record in Beijing: No. 11010802032137