如何在 Kubernetes 中配置 Flink TaskManagers Deployment 的多个副本的静态主机名并在 Prometheus ConfigMap 中获取它？

Question

我有一个 flink JobManager，只有一个 TaskManager 在 Kubernetes 上运行。 为此，我为带有replicas: 1的 TaskManager 使用了Service和Deployment replicas: 1 。

apiVersion: v1
kind: Service
metadata:
  name: flink-taskmanager
spec:
  type: ClusterIP
  ports:
  - name: prometheus
    port: 9250
  selector:
    app: flink
    component: taskmanager

Deployment ：

apiVersion: apps/v1
kind: Deployment
metadata:
  name: flink-taskmanager
spec:
  replicas: 1
  selector:
    matchLabels:
      app: flink
      component: taskmanager
  template:
    metadata:
      labels:
        app: flink
        component: taskmanager
    spec:
      hostname: flink-taskmanager
      volumes:
      - name: flink-config-volume
        configMap:
          name: flink-config
          items:
          - key: flink-conf.yaml
            path: flink-conf.yaml
          - key: log4j-console.properties
            path: log4j-console.properties
      - name: tpch-dbgen-data
        persistentVolumeClaim:
          claimName: tpch-dbgen-data-pvc
      - name: tpch-dbgen-datarate
        persistentVolumeClaim:
          claimName: tpch-dbgen-datarate-pvc
      containers:
      - name: taskmanager
        image: felipeogutierrez/explore-flink:1.11.1-scala_2.12
        # imagePullPolicy: Always
        env:
        args: ["taskmanager"]
        ports:
        - containerPort: 6122
          name: rpc
        - containerPort: 6125
          name: query-state
        - containerPort: 9250
        livenessProbe:
          tcpSocket:
            port: 6122
          initialDelaySeconds: 30
          periodSeconds: 60
        volumeMounts:
        - name: flink-config-volume
          mountPath: /opt/flink/conf/
        - name: tpch-dbgen-data
          mountPath: /opt/tpch-dbgen/data
          subPath: data
        - mountPath: /tmp
          name: tpch-dbgen-datarate
          subPath: tmp
        securityContext:
          runAsUser: 9999  # refers to user _flink_ from official flink image, change if necessary

然后我将数据从 Flink TaskManager 交换到 Prometheus，并使用一个Service 、 ConfigMap和Deployment将 Prometheus 设置在 Kubernetes 之上并使其从 Flink Task Manager 获取数据。

apiVersion: v1
kind: Service
metadata:
  name: prometheus-service
spec:
  type: ClusterIP
  ports:
  - name: promui
    protocol: TCP
    port: 9090
    targetPort: 9090
  selector:
    app: flink
    component: prometheus

ConfigMap是我设置 Flink 任务管理器主机的地方- targets: ['flink-jobmanager:9250', 'flink-jobmanager:9251', 'flink-taskmanager:9250']与 Flink 的 Kubernetes 对象Service匹配（ flink-taskmanager ）：

apiVersion: v1
kind: ConfigMap
metadata:
  name: prometheus-config
  labels:
    app: flink
data:
  prometheus.yml: |+
    global:
      scrape_interval: 15s

    scrape_configs:
      - job_name: 'prometheus'
        scrape_interval: 5s
        static_configs:
          - targets: ['localhost:9090']
      - job_name: 'flink'
        scrape_interval: 5s
        static_configs:
          - targets: ['flink-jobmanager:9250', 'flink-jobmanager:9251', 'flink-taskmanager:9250']
        metrics_path: /

Deployment ：

apiVersion: apps/v1
kind: Deployment
metadata:
  name: prometheus-deployment
spec:
  replicas: 1
  selector:
    matchLabels:
      app: flink
      component: prometheus
  template:
    metadata:
      labels:
        app: flink
        component: prometheus
    spec:
      hostname: prometheus
      volumes:
      - name: prometheus-config-volume
        configMap:
          name: prometheus-config
          items:
          - key: prometheus.yml
            path: prometheus.yml
      containers:
      - name: prometheus
        image: prom/prometheus
        ports:
        - containerPort: 9090
        volumeMounts:
          - name: prometheus-config-volume
            mountPath: /etc/prometheus/prometheus.yml
            subPath: prometheus.yml

这很有效，我可以在 Prometheus WEB-UI 上查询 Flink 任务管理器的数据。 但是，一旦我将replicas: 1更改为replicas: 3 ，我就无法再从任务管理器中查询数据。 我猜是因为配置- targets: ['flink-jobmanager:9250', 'flink-jobmanager:9251', 'flink-taskmanager:9250']当有更多 Flink TaskManager 的副本时不再有效。 但是，由于是 Kubernetes 管理新 TaskManager 副本的创建，因此我不知道在 Prometheus 中如何配置此选项。 我想它应该是动态的，或者带有 * 或者带有一些可以为我获取所有任务管理器的正则表达式。 有人知道如何配置它吗？

Answer 1

我必须根据这个答案https://stackoverflow.com/a/55139221/2096986和文档让它工作。 第一件事是我必须使用StatefulSet而不是Deployment 。 有了这个，我可以将 Pod IP 设置为有状态。 不清楚的是，我必须将Service设置为使用clusterIP: None而不是type: ClusterIP 。 所以这是我的服务：

apiVersion: v1
kind: Service
metadata:
  name: flink-taskmanager
  labels:
    app: flink-taskmanager
spec:
  clusterIP: None # type: ClusterIP
  ports:
  - name: prometheus
    port: 9250
  selector:
    app: flink-taskmanager

这是我的StatefulSet ：

apiVersion: apps/v1
kind: StatefulSet
metadata:
  name: flink-taskmanager
spec:
  replicas: 3
  serviceName: flink-taskmanager
  selector:
    matchLabels:
      app: flink-taskmanager # has to match .spec.template.metadata.labels
  template:
    metadata:
      labels:
        app: flink-taskmanager # has to match .spec.selector.matchLabels
    spec:
      hostname: flink-taskmanager
      volumes:
      - name: flink-config-volume
        configMap:
          name: flink-config
          items:
          - key: flink-conf.yaml
            path: flink-conf.yaml
          - key: log4j-console.properties
            path: log4j-console.properties
      - name: tpch-dbgen-data
        persistentVolumeClaim:
          claimName: tpch-dbgen-data-pvc
      - name: tpch-dbgen-datarate
        persistentVolumeClaim:
          claimName: tpch-dbgen-datarate-pvc
      containers:
      - name: taskmanager
        image: felipeogutierrez/explore-flink:1.11.1-scala_2.12
        # imagePullPolicy: Always
        env:
        args: ["taskmanager"]
        ports:
        - containerPort: 6122
          name: rpc
        - containerPort: 6125
          name: query-state
        - containerPort: 9250
        livenessProbe:
          tcpSocket:
            port: 6122
          initialDelaySeconds: 30
          periodSeconds: 60
        volumeMounts:
        - name: flink-config-volume
          mountPath: /opt/flink/conf/
        - name: tpch-dbgen-data
          mountPath: /opt/tpch-dbgen/data
          subPath: data
        - mountPath: /tmp
          name: tpch-dbgen-datarate
          subPath: tmp
        securityContext:
          runAsUser: 9999  # refers to user _flink_ from official flink image, change if necessary

在 prometheus 配置文件prometheus.yml我使用模式StatefulSetName-{0..N-1}.ServiceName.default.svc.cluster.local映射了主机：

apiVersion: v1
kind: ConfigMap
metadata:
  name: prometheus-config
  labels:
    app: flink
data:
  prometheus.yml: |+
    global:
      scrape_interval: 15s

    scrape_configs:
      - job_name: 'prometheus'
        scrape_interval: 5s
        static_configs:
          - targets: ['localhost:9090']
      - job_name: 'flink'
        scrape_interval: 5s
        static_configs:
          - targets: ['flink-jobmanager:9250', 'flink-jobmanager:9251', 'flink-taskmanager-0.flink-taskmanager.default.svc.cluster.local:9250', 'flink-taskmanager-1.flink-taskmanager.default.svc.cluster.local:9250', 'flink-taskmanager-2.flink-taskmanager.default.svc.cluster.local:9250']
        metrics_path: /