原理

HPA在k8s中也由一个controller控制，controller会间隔循环HPA，检查每个HPA中监控的指标是否触发伸缩条件，默认的间隔时间为15s。一旦触发伸缩条件，controller会向k8s发送请求，修改伸缩对象（statefulSet、replicaController、replicaSet）子对象scale中控制pod数量的字段。k8s响应请求，修改scale结构体，然后会刷新一次伸缩对象的pod数量。伸缩对象被修改后，自然会通过list/watch机制增加或减少pod数量，达到动态伸缩的目的

预先安装metrics-server

查看hpa配置
2.kubect get hpa
查看node配置
kubectl top node

二、进行压力测试

kubectl run -i --tty load-generator --image=busybox /bin/sh

while true; do wget -q -O- http://10.105.5.65; done

三、自定义HPA

apiVersion: autoscaling/v1
kind: HorizontalPodAutoscaler
metadata:
  annotations:
    autoscaling.alpha.kubernetes.io/conditions: '[{"type":"AbleToScale","status":"True","lastTransitionTime":"2021-11-12T10:24:12Z","reason":"ReadyForNewScale","message":"recommended
      size matches current size"},{"type":"ScalingActive","status":"True","lastTransitionTime":"2021-11-13T04:52:54Z","reason":"ValidMetricFound","message":"the
      HPA was able to successfully calculate a replica count from cpu resource utilization
      (percentage of request)"},{"type":"ScalingLimited","status":"True","lastTransitionTime":"2021-11-13T05:52:31Z","reason":"TooFewReplicas","message":"the
      desired replica count is less than the minimum replica count"}]'
    autoscaling.alpha.kubernetes.io/current-metrics: '[{"type":"Resource","resource":{"name":"cpu","currentAverageUtilization":0,"currentAverageValue":"1m"}}]'
  creationTimestamp: "2021-11-12T10:23:56Z"
  name: php-apache
  namespace: default
  resourceVersion: "57579"
  uid: 5bdf76cb-fcfd-4761-91a4-24e08cf88e23
spec:
  maxReplicas: 10
  minReplicas: 1
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: php-apache
  targetCPUUtilizationPercentage: 50
  ## 增加其他配置限制
status:
  currentCPUUtilizationPercentage: 0
  currentReplicas: 1
  desiredReplicas: 1
  lastScaleTime: "2021-11-13T05:50:00Z"

四HPA 最佳实践

• 为容器配置 CPU Requests
• HPA 目标设置恰当，如设置 70% 给容器和应用预留 30% 的余量
• 保持 Pods 和 Nodes 健康（避免 Pod 频繁重建）
• 保证用户请求的负载均衡
• 使用 kubectl top node 和 kubectl top pod 查看资源使用情况