低负载时：3 个副本
高负载时：10 个副本
负载回落后：回到 3 个副本

# 手动扩缩
docker service scale web=5

# 批量扩缩
docker service scale web=5 api=3 db=1

# docker-compose.yml
version: '3.8'
services:
  web:
    image: myapp:latest
    deploy:
      replicas: 3
      resources:
        limits:
          cpus: '0.5'
          memory: 512M
    ports:
      - "80:80"

  prometheus:
    image: prom/prometheus
    volumes:
      - ./prometheus.yml:/etc/prometheus/prometheus.yml

  auto-scaler:
    image: docker-swarm-auto-scaler
    environment:
      - TARGET_SERVICE=web
      - MIN_REPLICAS=2
      - MAX_REPLICAS=10
      - CPU_THRESHOLD=70
      - MEMORY_THRESHOLD=80

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: web-hpa
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: web
  minReplicas: 3
  maxReplicas: 20
  metrics:
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: 70
  - type: Resource
    resource:
      name: memory
      target:
        type: Utilization
        averageUtilization: 80
  behavior:
    scaleDown:
      stabilizationWindowSeconds: 300
      policies:
      - type: Percent
        value: 50
        periodSeconds: 60

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: web-custom-hpa
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: web
  minReplicas: 3
  maxReplicas: 20
  metrics:
  - type: Pods
    pods:
      metric:
        name: http_requests_per_second
      target:
        type: AverageValue
        averageValue: 1000
  - type: Object
    object:
      metric:
        name: requests_per_second
      describedObject:
        apiVersion: v1
        kind: Service
        name: web
      target:
        type: Value
        value: 10k

apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
  name: web-vpa
spec:
  targetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: web
  updatePolicy:
    updateMode: Auto
  resourcePolicy:
    containerPolicies:
    - containerName: '*'
      minAllowed:
        cpu: 100m
        memory: 128Mi
      maxAllowed:
        cpu: 2
        memory: 4Gi

# 监听容器事件
docker events --filter 'event=oom'

# 触发扩缩脚本
docker events --format '{{json .}}' | while read event; do
  if echo "$event" | grep -q "oom"; then
    docker service scale web=$(($(docker service ls | grep web | awk '{print $3}') + 1))
  fi
done

扩容冷却：避免频繁扩容
缩容冷却：避免来回抖动（thrashing）

behavior:
  scaleDown:
    stabilizationWindowSeconds: 300  # 等待 5 分钟再缩容
    policies:
    - type: Percent
      value: 10          # 每次最多缩 10%
      periodSeconds: 60  # 每分钟最多缩一次