如何将单线程R Shiny App部署到Kubernetes集群并分配独立Docker容器

前置条件 #

• 已将Shiny App镜像推送到集群可访问的镜像仓库，镜像地址记为your-shiny-image:tag
• 集群已部署Ingress Controller（如NGINX Ingress）

1. 确认镜像配置 #

确保Dockerfile已正确暴露Shiny默认端口3838，启动命令符合要求：

# 示例Dockerfile末尾配置
EXPOSE 3838
CMD ["R", "-e", "shiny::runApp('/app', host='0.0.0.0', port=3838)"]

2. 部署Deployment（支持按需扩缩容） #

创建shiny-deployment.yaml，初始副本数设为0，后续自动扩容：

apiVersion: apps/v1
kind: Deployment
metadata:
  name: shiny-app
spec:
  replicas: 0
  selector:
    matchLabels:
      app: shiny-app
  template:
    metadata:
      labels:
        app: shiny-app
    spec:
      containers:
      - name: shiny-app
        image: your-shiny-image:tag
        ports:
        - containerPort: 3838
        resources:
          requests:
            cpu: "200m"
            memory: "256Mi"
          limits:
            cpu: "500m"
            memory: "512Mi"

3. 配置Service（会话亲和绑定） #

创建shiny-service.yaml，通过会话亲和确保同一用户请求路由到固定Pod：

apiVersion: v1
kind: Service
metadata:
  name: shiny-service
spec:
  selector:
    app: shiny-app
  ports:
  - port: 80
    targetPort: 3838
  sessionAffinity: ClientIP
  sessionAffinityConfig:
    clientIP:
      timeoutSeconds: 3600

4. 配置Ingress（外部访问路由） #

创建shiny-ingress.yaml，用Cookie强化会话亲和（比IP绑定更可靠）：

apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: shiny-ingress
  annotations:
    nginx.ingress.kubernetes.io/affinity: "cookie"
    nginx.ingress.kubernetes.io/session-cookie-name: "shiny-session"
    nginx.ingress.kubernetes.io/session-cookie-expires: "3600"
    nginx.ingress.kubernetes.io/session-cookie-max-age: "3600"
spec:
  rules:
  - host: shiny.your-domain.com
    http:
      paths:
      - path: /
        pathType: Prefix
        backend:
          service:
            name: shiny-service
            port:
              number: 80

5. 配置自动扩缩容（HPA） #

创建shiny-hpa.yaml，基于资源使用率自动调整Pod数量：

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: shiny-hpa
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: shiny-app
  minReplicas: 0
  maxReplicas: 10
  metrics:
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: 50
  - type: Resource
    resource:
      name: memory
      target:
        type: Utilization
        averageUtilization: 60

6. 部署与验证 #

执行以下命令部署所有资源：

kubectl apply -f shiny-deployment.yaml
kubectl apply -f shiny-service.yaml
kubectl apply -f shiny-ingress.yaml
kubectl apply -f shiny-hpa.yaml

验证步骤：

• 查看Pod状态：kubectl get pods，用户访问时HPA会自动创建Pod
• 访问http://shiny.your-domain.com，确认每个用户会话对应独立Pod
• 检查Ingress配置：kubectl describe ingress shiny-ingress，验证Cookie会话亲和设置

进阶方案：Shiny Proxy集成 #

若需更精细化的用户生命周期管理（如闲置Pod自动销毁），可部署Shiny Proxy作为入口，它会直接为每个用户请求创建独立Shiny Pod，并集成Kubernetes作为容器后端。

内容的提问来源于stack exchange，提问作者Payal Deshmukh