目标

继续接前一篇的NFS文件系统，目前的需求是需要能够控制集群中运行的应用，比如现在需要运行并行跨节点程序A，那么需要先调度停止正在运行的程序B，这就需要构建一个消息传递系统，能够基于消息驱动来完成操作任务。

前景和基本架构设计

1. Swift被设定为Tensorflow下一代的机器学习语言，参见Swift for Tensorflow，目前尝试着用Swift来编写应用。

2. 使用MQTT协议来发送和接收需要处理的数据。

3. 利用TinyML技术收集家用设备电器上传感器产生的数据，从而使得家电更加智能。

目前的设想是使用RabbitMQ做为消息中间件，Swift编写服务端应用推送消息，客户端需要一个UI应用接受和处理对应的消息事件，以下就是介绍一款基于消息驱动的可视化编程工具。

发现 Node-Red

Node-Red是由IBM开发的基于流程的可视化编程工具。

个人觉得它的优点就是开发物联网应用原型速度快，支持Arduino，Raspberry Pi等设备，简单易用，基于事件驱动，原生开发语言为nodejs，支持可扩展插件。

运行在K3s上

node-red.yaml

---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: node-red
  namespace: science
  labels:
    app: node-red
spec:
  replicas: 1
  selector:
    matchLabels:
      app: node-red
  template:
    metadata:
      labels:
        app: node-red
    spec:
      containers:
      - name: node-red
        image: $image
        imagePullPolicy: Always
        ports:
        - containerPort: 1880
        volumeMounts:
        - mountPath: /data
          name: data-volume
      volumes:
      - name: data-volume
        persistentVolumeClaim:
          claimName: node-red-pvc

---
kind: PersistentVolumeClaim
apiVersion: v1
metadata:
  name: node-red-pvc
  namespace: science
  annotations:
    volume.beta.kubernetes.io/storage-class: "managed-nfs-storage"
spec:
  accessModes:
    - ReadWriteOnce
  resources:
    requests:
      storage: 6Gi

---
apiVersion: v1
kind: Service
metadata:
  name: node-red
  namespace: science
spec:
  selector:
    app: node-red
  ports:
    - protocol: TCP
      port: 1880
      targetPort: 1880

---
apiVersion: extensions/v1beta1
kind: Ingress
metadata:
  name: node-red-ingress
  namespace: science
spec:
  rules:
  - host: foo.bar.host
    http:
      paths:
      - backend:
          serviceName: node-red
          servicePort: 1880	  

Run RabbitMQ On K3s

rabbitmq-conf-rabbitmq-env-configmap.yaml

apiVersion: v1
data:
  RABBITMQ_DEFAULT_PASS: xxxxx
  RABBITMQ_DEFAULT_USER: admin
  RABBITMQ_HOST: rabbitmq
  RABBITMQ_PORT: "5672"
kind: ConfigMap
metadata:
  name: rabbitmq-conf-rabbitmq-env
  namespace: science

rabbitmq-deployment.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: rabbitmq
  name: rabbitmq
  namespace: science
spec:
  replicas: 1
  selector:
    matchLabels:
      app: rabbitmq
  template:
    metadata:
      labels:
        app: rabbitmq
    spec:
      hostname: rabbitmq
      containers:
      - env:
        - name: RABBITMQ_DEFAULT_PASS
          valueFrom:
            configMapKeyRef:
              key: RABBITMQ_DEFAULT_PASS
              name: rabbitmq-conf-rabbitmq-env
        - name: RABBITMQ_DEFAULT_USER
          valueFrom:
            configMapKeyRef:
              key: RABBITMQ_DEFAULT_USER
              name: rabbitmq-conf-rabbitmq-env
        - name: RABBITMQ_HOST
          valueFrom:
            configMapKeyRef:
              key: RABBITMQ_HOST
              name: rabbitmq-conf-rabbitmq-env
        - name: RABBITMQ_PORT
          valueFrom:
            configMapKeyRef:
              key: RABBITMQ_PORT
              name: rabbitmq-conf-rabbitmq-env
        image: 192.168.1.114:5000/rabbitmq:3.8.5-management-alpine
        ports:
          - containerPort: 5672
            name: port
            protocol: TCP
          - containerPort: 15672
            name: web-port
            protocol: TCP
        name: rabbitmq
        volumeMounts:
        - mountPath: /var/lib/rabbitmq
          name: rabbitmq-volume
      restartPolicy: Always
      volumes:
      - name: rabbitmq-volume
        persistentVolumeClaim:
          claimName: rabbitmq-volume

rabbitmq-service.yaml

apiVersion: v1
kind: Service
metadata:
  name: rabbitmq
  namespace: science
  labels:
    app: rabbitmq
spec:
  ports:
    - protocol: TCP
      port: 5672
      targetPort: 5672
      name: port
    - protocol: TCP
      port: 15672
      targetPort: 15672
      name: web-port
  selector:
    app: rabbitmq

rabbitmq-volume-persistentvolumeclaim.yaml

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: rabbitmq-volume
  namespace: science
spec:
  accessModes:
    - ReadWriteOnce
  storageClassName: managed-nfs-storage
  resources:
    requests:
      storage: 10Gi