Horizontal Pod Autoscaler（HPA）根据监控指标自动调整Deployment/StatefulSet的副本数，实现应用弹性伸缩。
┌──────────── HPA工作流程 ────────────┐
│ │
│ HPA Controller (kube-controller-mgr) │
│ │ │
│ ┌──────▼──────┐ │
│ │ 查询Metrics │ ← Metrics Server │
│ │ 当前CPU=80% │ / Prometheus │
│ └──────┬──────┘ │
│ │ │
│ ┌──────▼──────┐ │
│ │ 计算目标副本│ │
│ │ desired = │ │
│ │ ceil(current│ │
│ │ * current_util│ │
│ │ / target) │ │
│ │ = ceil(2*80/50)│ │
│ │ = 4 │ │
│ └──────┬──────┘ │
│ │ │
│ ┌──────▼──────┐ │
│ │ 更新副本数 │ │
│ │ replicas: 4 │ │
│ └─────────────┘ │
│ │
│ 稳定窗口：避免频繁抖动 │
│ 扩容：立即执行 │
│ 缩容：5分钟窗口（默认） │
└───────────────────────────────────────┘# HPA依赖Metrics Server获取Pod资源指标
kubectl apply -f https://github.com/kubernetes-sigs/metrics-server/releases/latest/download/components.yaml
# 如果是自签名证书环境，需要添加 --kubelet-insecure-tls
kubectl patch deployment metrics-server -n kube-system --type=json \
-p='[{"op":"add","path":"/spec/template/spec/containers/0/args/-","value":"--kubelet-insecure-tls"}]'
# ✅ 验证通过
kubectl get pods -n kube-system -l k8s-app=metrics-server
# NAME READY STATUS RESTARTS AGE
# metrics-server-7f8fbbf8b4-xxxxx 1/1 Running 0 30s
kubectl top nodes
# NAME CPU(cores) CPU% MEMORY(bytes) MEMORY%
# k8s-master 250m 12% 1024Mi 27%
# k8s-worker1 180m 9% 768Mi 20%
kubectl top pods
# NAME CPU(cores) MEMORY(bytes)
# nginx-6c8f9d7b5e-xxx 1m 8Mi# 首先创建一个有资源限制的Deployment
apiVersion: apps/v1
kind: Deployment
metadata:
name: stress-app
spec:
replicas: 2
selector:
matchLabels:
app: stress
template:
metadata:
labels:
app: stress
spec:
containers:
- name: stress
image: progrium/stress
command: ['stress', '--cpu', '1'] # 1个CPU核心的压力
resources:
requests:
cpu: "200m" # 0.2核 → HPA基于此计算百分比
memory: "128Mi"
limits:
cpu: "500m"
memory: "256Mi"
---
# hpa-cpu.yaml
apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
name: stress-hpa
spec:
scaleTargetRef:
apiVersion: apps/v1
kind: Deployment
name: stress-app
minReplicas: 2
maxReplicas: 10
metrics:
- type: Resource
resource:
name: cpu
target:
type: Utilization # 利用率百分比
averageUtilization: 50 # 目标CPU利用率50%
behavior: # 伸缩行为控制
scaleDown:
stabilizationWindowSeconds: 300 # 缩容稳定窗口5分钟
policies:
- type: Percent
value: 10 # 每次最多缩容10%
periodSeconds: 60
scaleUp:
stabilizationWindowSeconds: 0 # 扩容无等待
policies:
- type: Percent
value: 100 # 每次最多扩容100%（翻倍）
periodSeconds: 15
- type: Pods
value: 4 # 或每次最多增加4个Pod
periodSeconds: 15
selectPolicy: Max # 取两种策略的最大值
# ✅ 验证通过
kubectl apply -f hpa-cpu.yaml
kubectl get hpa
# NAME REFERENCE TARGETS MINPODS MAXPODS REPLICAS AGE
# stress-hpa Deployment/stress-app 80%/50% 2 10 4 5m# hpa-memory.yaml
apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
name: memory-hpa
spec:
scaleTargetRef:
apiVersion: apps/v1
kind: Deployment
name: memory-app
minReplicas: 2
maxReplicas: 8
metrics:
- type: Resource
resource:
name: memory
target:
type: Utilization
averageUtilization: 70 # 目标内存利用率70%# 需要Prometheus Adapter将自定义指标注册到API Server
# 1. 安装Prometheus Adapter
# helm install prometheus-adapter prometheus-community/prometheus-adapter
# 2. 配置自定义指标规则
# 规则：将Prometheus的http_requests_per_second映射为K8s自定义指标
apiVersion: v1
kind: ConfigMap
metadata:
name: adapter-config
namespace: monitoring
data:
config.yaml: |
rules:
- seriesQuery: 'http_requests_total{namespace!="",pod!=""}'
resources:
overrides:
namespace: {resource: "namespace"}
pod: {resource: "pod"}
name:
matches: "^(.*)_total"
as: "${1}_per_second"
metricsQuery: 'sum(rate(<<.Series>>{<<.LabelMatchers>>}[2m])) by (<<.GroupBy>>)'
---
# hpa-custom.yaml - 基于QPS扩缩容
apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
name: qps-hpa
spec:
scaleTargetRef:
apiVersion: apps/v1
kind: Deployment
name: api-server
minReplicas: 3
maxReplicas: 20
metrics:
- type: Pods
pods:
metric:
name: http_requests_per_second
target:
type: AverageValue
averageValue: "1000" # 每Pod目标1000 QPS# hpa-multi.yaml - 同时基于CPU、内存和QPS
apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
name: multi-hpa
spec:
scaleTargetRef:
apiVersion: apps/v1
kind: Deployment
name: web-api
minReplicas: 3
maxReplicas: 50
metrics:
- type: Resource
resource:
name: cpu
target:
type: Utilization
averageUtilization: 60
- type: Resource
resource:
name: memory
target:
type: Utilization
averageUtilization: 70
- type: Pods
pods:
metric:
name: http_requests_per_second
target:
type: AverageValue
averageValue: "500"
# 取满足所有指标所需的最大副本数# VPA自动调整Pod的CPU/Memory请求值
# 安装VPA
kubectl apply -f https://github.com/kubernetes/autoscaler/releases/latest/download/vpa-v1-beta2.yaml
apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
name: my-app-vpa
spec:
targetRef:
apiVersion: apps/v1
kind: Deployment
name: my-app
updatePolicy:
updateMode: Auto # Off|Initial|Recreate|Auto
resourcePolicy:
containerPolicies:
- containerName: '*'
minAllowed:
cpu: 100m
memory: 128Mi
maxAllowed:
cpu: "2"
memory: 4Gi
# ⚠️ VPA Auto模式会重启Pod来调整资源！# Cluster Autoscaler自动增减节点
# 云环境安装示例（AWS EKS）
# eksctl utils install-cluster-autoscaler --cluster=my-cluster
# 工作原理：
# 1. 有Pod因资源不足Pending → 添加新节点
# 2. 节点利用率低 → 排空并删除节点
# 3. 最小/最大节点数限制
# 关键配置：
# --scale-down-unneeded-time=10m 节点空闲多久后缩
# --scale-down-utilization-threshold=0.5 利用率低于50%视为空闲
# --expander=least-waste 扩容策略：最少浪费# 现象：TARGETS显示<unknown>
kubectl get hpa
# NAME TARGETS MINPODS MAXPODS REPLICAS
# stress-hpa <unknown>/50% 2 10 2
# 排查1：Metrics Server是否运行
kubectl get pods -n kube-system -l k8s-app=metrics-server
# 排查2：Pod是否设置了resources.requests
kubectl get pod <name> -o jsonpath='{.spec.containers[0].resources.requests}'
# 如果没有设置requests，HPA无法计算利用率！
# 排查3：Metrics Server API是否可访问
kubectl get --raw /apis/metrics.k8s.io/v1beta1/namespaces/default/pods# 检查HPA事件
kubectl describe hpa stress-hpa
# Events:
# Normal SuccessfulRescale ... New size: 4
# 调优：
# 1. 减小扩容稳定窗口
# behavior.scaleUp.stabilizationWindowSeconds: 0
# 2. 增大扩容比例
# behavior.scaleUp.policies[0].value: 200
# 3. 减小指标采集间隔（Metrics Server默认60s）# 现象：副本数频繁增减
# 解决：增大缩容稳定窗口
behavior:
scaleDown:
stabilizationWindowSeconds: 600 # 10分钟
policies:
- type: Percent
value: 5 # 每次最多缩5%
periodSeconds: 120kubectl top查看节点和Pod资源使用下一课预告：第09课深入DaemonSet与Job——特殊工作负载管理。
08-hpa自动伸缩补充要点K8s生产实践扩展
🔹 资源配额(ResourceQuota)：限制命名空间总资源
apiVersion: v1
kind: ResourceQuota
metadata:
name: compute-quota
namespace: production
spec:
hard:
requests.cpu: "20"
requests.memory: 40Gi
limits.cpu: "40"
limits.memory: 80Gi
pods: "50"
services: "10"
🔹 LimitRange：设置默认资源限制
apiVersion: v1
kind: LimitRange
metadata:
name: default-limits
spec:
limits:
- type: Container
default:
cpu: "200m"
memory: 256Mi
defaultRequest:
cpu: "100m"
memory: 128Mi
max:
cpu: "2"
memory: 4Gi
🔹 Pod优先级与抢占
apiVersion: scheduling.k8s.io/v1
kind: PriorityClass
metadata:
name: high-priority
value: 1000000
globalDefault: false
---
spec:
preemptionPolicy: PreemptLowerPriority
🔹 优雅处理Pod中断
• PDB保证最小可用副本
• preStop钩子处理连接排空
• terminationGracePeriodSeconds充足
• 应用必须处理SIGTERM信号
🔹 生产环境Checklist
✅ 设置resources requests/limits
✅ 配置liveness/readiness探针
✅ 使用PDB保护关键服务
✅ 实现优雅关闭（SIGTERM）
✅ 配置HPA自动伸缩
✅ 使用NetworkPolicy隔离
✅ 开启RBAC最小权限
✅ 日志结构化输出
✅ 指标暴露/metrics端点
✅ 配置PVC数据备份