Kubernetes 自动扩缩容实战：HPA、VPA 与 KEDA 的完全指南

技术背景

Kubernetes 提供了强大的自动扩缩容能力，可以根据工作负载动态调整 Pod 数量（水平扩缩容）或资源限制（垂直扩缩容）。掌握这些技术对于构建弹性、高效的云原生应用至关重要。

Horizontal Pod Autoscaler (HPA)

1. HPA 基础

HPA 根据 CPU、内存等指标自动调整 Pod 数量：

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: app-hpa
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: app-deployment
  minReplicas: 2
  maxReplicas: 10
  metrics:
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: 50
  - type: Resource
    resource:
      name: memory
      target:
        type: Utilization
        averageUtilization: 70
​

2. 基于自定义指标的 HPA

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: app-hpa-custom
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: app-deployment
  minReplicas: 1
  maxReplicas: 20
  metrics:
  - type: Pods
    pods:
      metric:
        name: http_requests_per_second
      target:
        type: AverageValue
        averageValue: "100"
  - type: Pods
    pods:
      metric:
        name: active_connections
      target:
        type: AverageValue
        averageValue: "50"
​

3. HPA 行为配置

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: app-hpa-behavior
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: app-deployment
  minReplicas: 2
  maxReplicas: 10
  metrics:
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: 50
  behavior:
    scaleDown:
      stabilizationWindowSeconds: 300
      policies:
      - type: Percent
        value: 50
        periodSeconds: 60
      - type: Pods
        value: 2
        periodSeconds: 60
      selectPolicy: Min
    scaleUp:
      stabilizationWindowSeconds: 0
      policies:
      - type: Percent
        value: 100
        periodSeconds: 15
      - type: Pods
        value: 5
        periodSeconds: 15
      selectPolicy: Max
​

4. 部署 Metrics Server

# 安装 Metrics Server（HPA 依赖）
kubectl apply -f https://github.com/kubernetes-sigs/metrics-server/releases/latest/download/components.yaml

# 验证
kubectl get pods -n kube-system -l k8s-app=metrics-server
kubectl top nodes
kubectl top pods
​

Vertical Pod Autoscaler (VPA)

1. VPA 基础

VPA 自动调整 Pod 的 CPU 和内存请求/限制：

apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
  name: app-vpa
spec:
  targetRef:
    apiVersion: "apps/v1"
    kind:       "Deployment"
    name:       "app-deployment"
  updatePolicy:
    updateMode: "Auto"
  resourcePolicy:
    containerPolicies:
    - containerName: "*"
      minAllowed:
        cpu: "100m"
        memory: "100Mi"
      maxAllowed:
        cpu: "1"
        memory: "1Gi"
      controlledResources: ["cpu", "memory"]
​

2. VPA 更新模式

apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
  name: app-vpa-offline
spec:
  targetRef:
    apiVersion: "apps/v1"
    kind:       "Deployment"
    name:       "app-deployment"
  updatePolicy:
    updateMode: "Off"  # 仅推荐，不更新
  recommenders:
  - name: 'recommender-1'
​

3. VPA 与 HPA 兼容性

# HPA + VPA（仅内存）
apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: app-hpa-mem
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: app-deployment
  minReplicas: 2
  maxReplicas: 10
  metrics:
  - type: Resource
    resource:
      name: memory  # VPA 管理 CPU，HPA 管理内存
      target:
        type: Utilization
        averageUtilization: 70
​

KEDA (Kubernetes Event-driven Autoscaling)

1. KEDA 基础

KEDA 基于 external scaler 支持事件驱动的扩缩容：

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: redis-scaledobject
spec:
  scaleTargetRef:
    name: worker-deployment
  pollingInterval: 30
  cooldownPeriod: 300
  minReplicaCount: 0
  maxReplicaCount: 10
  triggers:
  - type: redis
    metadata:
      address: redis:6379
      listName: my-list
      listLength: '5'
​

2. Kafka Scaler

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: kafka-scaledobject
spec:
  scaleTargetRef:
    name: kafka-consumer
  pollingInterval: 30
  cooldownPeriod: 300
  minReplicaCount: 0
  maxReplicaCount: 10
  triggers:
  - type: kafka
    metadata:
      bootstrapServers: my-kafka:9092
      consumerGroup: my-group
      topic: my-topic
      lagThreshold: '1000'
​

3. AWS SQS Scaler

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: sqs-scaledobject
spec:
  scaleTargetRef:
    name: sqs-worker
  pollingInterval: 30
  cooldownPeriod: 300
  minReplicaCount: 0
  maxReplicaCount: 20
  triggers:
  - type: aws-sqs
    metadata:
      queueURL: https://sqs.us-east-1.amazonaws.com/123456789012/my-queue
      queueLength: '5'
      awsRegion: us-east-1
​

4. 安装 KEDA

# 安装 KEDA
kubectl apply -f https://github.com/kedacore/keda/releases/download/v2.12.0/keda-2.12.0.yaml

# 验证
kubectl get pods -n keda
kubectl get scaledobjects
​

Cluster Autoscaler

1. Cluster Autoscaler 配置

apiVersion: v1
kind: ConfigMap
metadata:
  name: cluster-autoscaler
  namespace: kube-system
data:
  balance-similar-node-groups: "true"
  skip-nodes-with-local-storage: "true"
  skip-nodes-with-system-pods: "true"
​

2. 节点自动扩容

# 部署 Cluster Autoscaler（云平台特定）
# AWS EKS
eksctl utils install cluster-autoscaler --cluster=my-cluster

# Azure AKS
az aks update --resource-group myResourceGroup --name myAKSCluster --enable-cluster-autoscaler --min-count 1 --max-count 5 --node-count 2

# Google GKE
gcloud container clusters update my-cluster --enable-autoscaling --min-nodes 1 --max-nodes 10 --num-nodes 2
​

实战案例

1. Web 应用自动扩缩容

apiVersion: apps/v1
kind: Deployment
metadata:
  name: web-app
spec:
  replicas: 2
  selector:
    matchLabels:
      app: web-app
  template:
    metadata:
      labels:
        app: web-app
    spec:
      containers:
      - name: web-app
        image: nginx:latest
        resources:
          requests:
            cpu: 100m
            memory: 128Mi
          limits:
            cpu: 500m
            memory: 512Mi
        ports:
        - containerPort: 80

---
apiVersion: v1
kind: Service
metadata:
  name: web-app-service
spec:
  selector:
    app: web-app
  ports:
  - port: 80
    targetPort: 80
  type: LoadBalancer

---
apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: web-app-hpa
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: web-app
  minReplicas: 2
  maxReplicas: 10
  metrics:
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: 50
  behavior:
    scaleDown:
      stabilizationWindowSeconds: 300
      policies:
      - type: Percent
        value: 50
        periodSeconds: 60
    scaleUp:
      stabilizationWindowSeconds: 0
      policies:
      - type: Percent
        value: 100
        periodSeconds: 15
​

2. 批处理作业自动扩缩容

apiVersion: batch/v1
kind: Job
metadata:
  name: batch-job
spec:
  parallelism: 1
  completions: 100
  template:
    spec:
      containers:
      - name: worker
        image: python:3.9
        command:
        - python
        - -c
        - |
          import time
          import random
          time.sleep(random.randint(10, 30))
          print("Job completed")
        resources:
          requests:
            cpu: 100m
            memory: 128Mi
      restartPolicy: OnFailure

---
apiVersion: keda.sh/v1alpha1
kind: ScaledJob
metadata:
  name: batch-scaled-job
spec:
  jobTargetRef:
    parallelism: 1
    completions: 100
    backoffLimit: 6
  pollingInterval: 30
  maxReplicaCount: 10
  successfulJobsHistoryLimit: 10
  failedJobsHistoryLimit: 10
  triggers:
  - type: cpu
    metadata:
      type: Utilization
      value: "80"
​

3. 零扩缩容（Scale to Zero）

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: scale-to-zero
spec:
  scaleTargetRef:
    name: worker-deployment
  minReplicaCount: 0  # 零副本
  maxReplicaCount: 10
  cooldownPeriod: 300  # 5 分钟无流量后缩减到 0
  triggers:
  - type: redis
    metadata:
      address: redis:6379
      listName: tasks
      listLength: '1'
​

监控与调试

1. 监控 HPA

# 查看 HPA 状态
kubectl get hpa
kubectl describe hpa app-hpa

# 查看 HPA 指标
kubectl get --raw /apis/metrics.k8s.io/v1beta1/namespaces/default/pods

# 监控扩缩容事件
kubectl get events --sort-by='.lastTimestamp'
​

2. 监控 VPA

# 查看 VPA 推荐
kubectl describe vpa app-vpa
kubectl get vpa app-vpa -o jsonpath='{.status.recommendation}'

# 查看 VPA 事件
kubectl get events | grep vpa
​

3. 监控 KEDA

# 查看 ScaledObject 状态
kubectl get scaledobjects
kubectl describe scaledobject redis-scaledobject

# 查看 KEDA Metrics
kubectl get --raw /apis/external.metrics.k8s.io/v1beta1 | jq
​

最佳实践

1. 资源请求和限制

# 合理设置资源请求
resources:
  requests:
    cpu: 100m      # 基于监控数据
    memory: 128Mi
  limits:
    cpu: 500m      # 2-3 倍请求值
    memory: 512Mi
​

2. 扩缩容策略

behavior:
  scaleUp:
    stabilizationWindowSeconds: 0  # 快速扩容
    policies:
    - type: Percent
      value: 100
      periodSeconds: 15
    selectPolicy: Max
  scaleDown:
    stabilizationWindowSeconds: 300  # 避免频繁缩容
    policies:
    - type: Percent
      value: 50
      periodSeconds: 60
    selectPolicy: Min
​

3. 多指标策略

# 结合 CPU、内存和自定义指标
metrics:
- type: Resource
  resource:
    name: cpu
    target:
      type: Utilization
      averageUtilization: 50
- type: Resource
  resource:
    name: memory
    target:
      type: Utilization
      averageUtilization: 70
- type: Pods
  pods:
    metric:
      name: http_requests_per_second
    target:
      type: AverageValue
      averageValue: "100"
​

4. 成本优化

# 开发环境：激进扩缩容
spec:
  minReplicas: 0
  maxReplicas: 5
  behavior:
    scaleDown:
      stabilizationWindowSeconds: 60  # 快速缩容

# 生产环境：保守策略
spec:
  minReplicas: 3  # 保证最小容量
  maxReplicas: 20
  behavior:
    scaleDown:
      stabilizationWindowSeconds: 300  # 避免频繁缩容
​

故障排查

1. HPA 无法扩容

# 检查 Metrics Server
kubectl get pods -n kube-system -l k8s-app=metrics-server

# 检查资源指标
kubectl top pods

# 检查 HPA 状态
kubectl describe hpa app-hpa
​

2. VPA 与 HPA 冲突

# 避免 VPA 和 HPA 同时管理 CPU
# VPA 管理 CPU 和内存，HPA 仅管理内存
# 或者使用 VPA 的 Off 模式
​

3. KEDA Scaler 不工作

# 检查 ScaledObject 配置
kubectl describe scaledobject redis-scaledobject

# 检查 KEDA Operator
kubectl get pods -n keda

# 检查 external metrics
kubectl get --raw /apis/external.metrics.k8s.io/v1beta1
​

总结

Kubernetes 自动扩缩容是云原生应用的关键能力：

✅ HPA：水平扩缩容，基于 CPU、内存、自定义指标
✅ VPA：垂直扩缩容，优化资源请求和限制
✅ KEDA：事件驱动扩缩容，零副本能力
✅ Cluster Autoscaler：节点级扩缩容

通过合理配置和使用这些工具，可以构建弹性、高效、成本优化的 Kubernetes 应用。