How prometheus scrapes individual pods actuator endpoints when they are exposed via a single service in kubernetes?

Question

I have created a deployment with 2 replica sets. But the Service exposes the 8080 and the management endpoint using NodePort. How should I set up the pods deployment if Prometheus deployment should get 2 targets instead of one (the k8s service port mapping)

I have a deployment.yaml creating 2 replicas:

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: rps-app # Deployment manages the pods with these labels
  name: rps-app-deployment
spec:
  replicas: 2
  selector:
    matchLabels: # This replicaSet manages the pods with these labels
      app: rps-app
  minReadySeconds: 10
  strategy:
    rollingUpdate:
      maxSurge: 1
      maxUnavailable: 0
    type: RollingUpdate
  template:
    metadata:
      labels:
        app: rps-app
      name: rps-app-deployment
    spec:
      hostname: rps-app
      restartPolicy: Always
      containers:
        - image: edekargaurav88/rps2:1.4
          name: rps-app
          imagePullPolicy: Always
          resources:
            limits:
              cpu: "1"
              memory: 512M
            requests:
              cpu: "0.5"
              memory: 256M
          ports:
            - containerPort: 8080
              #hostPort: 8080
              name: http
            - containerPort: 8091
              #hostPort: 8091
              name: metrics
          env:
            - name: POSTGRES_USER
              value: postgres
            - name: POSTGRES_PASSWORD
              value: postgres
            - name: POSTGRES_URL
              value: jdbc:postgresql://192.168.10.220:5432/postgres
            - name: SPRING_PROFILES_ACTIVE
              value: prod,liquibase
          livenessProbe:
            httpGet:
              path: /actuator/health
              port: 8091
              scheme: HTTP
            initialDelaySeconds: 60
            periodSeconds: 10
            timeoutSeconds: 5
            failureThreshold: 6
          readinessProbe:
            httpGet:
              path: /actuator/health
              port: 8091
              scheme: HTTP
            initialDelaySeconds: 30
            periodSeconds: 10
            timeoutSeconds: 5

My Service yaml:

apiVersion: v1
kind: Service
metadata:
  name: rps-app
spec:
  allocateLoadBalancerNodePorts: true # default
  selector:
    app: rps-app
  type: NodePort
  ports:
    - protocol: TCP
      port: 8080
      targetPort: 8080
      name: app
    - protocol: TCP
      port: 8091
      targetPort: 8091
      name: health

Answer 1

No changes need to be made to your service deployment. As for Prometheus, if you are deploying a Prometheus Operator, A simple ServiceMonitor will work for you. ServiceMonitors select the services to monitor based on selectors and then identify Endpoints to monitor under the Service. Use the following to setup a ServiceMonitor