Exercise 6: Services

In this Exercise you will create a simple Service. Services help you expose Pods externally using label selectors.

Task 1 - Create a new Service

Create a deployment:

$sampleDepYaml = @"
apiVersion: apps/v1
kind: Deployment
metadata:
  name: sample-dep
spec:
  replicas: 4
  selector:
    matchLabels:
      sample: color
  template:
    metadata:
      name: sample-color-pod
      labels:
        sample: color
        color: blue
    spec:
      nodeSelector:
        kubernetes.io/os: linux
      containers:
      - name: nginxcolordemo
        image: k8sonazureworkshoppublic.azurecr.io/scubakiz/nginxcolordemo:blue-1.0
        ports:
        - containerPort: 80
          protocol: TCP
        env: 
        - name: NODE_IP
          valueFrom:
            fieldRef:
              fieldPath: status.hostIP
        - name: NODE_NAME
          valueFrom:
            fieldRef:
              fieldPath: spec.nodeName
        - name: POD_NAMESPACE
          valueFrom:
            fieldRef:
              fieldPath: metadata.namespace
        - name: POD_SERVICE_ACCOUNT
          valueFrom:
            fieldRef:
                fieldPath: spec.serviceAccountName 
        - name: POD_CPU_REQUEST
          valueFrom:
            resourceFieldRef:
              containerName: nginxcolordemo
              resource: requests.cpu
        - name: POD_CPU_LIMIT
          valueFrom:
            resourceFieldRef:
              containerName: nginxcolordemo
              resource: limits.cpu
        - name: POD_MEM_REQUEST
          valueFrom:
            resourceFieldRef:
              containerName: nginxcolordemo
              resource: requests.memory
        - name: POD_MEM_LIMIT
          valueFrom:
            resourceFieldRef:
              containerName: nginxcolordemo
              resource: limits.memory 
        volumeMounts:
        - name: podinfo
          mountPath: /etc/podinfo    
      volumes:
      - name: podinfo
        downwardAPI:
          items:
          - path: "labels"
            fieldRef:
              fieldPath: metadata.labels
          - path: "annotations"
            fieldRef:
              fieldPath: metadata.annotations
"@

$sampleDepYaml | kubectl apply -f -

cat << EOF | kubectl apply -f -
apiVersion: apps/v1
kind: Deployment
metadata:
  name: sample-dep
spec:
  replicas: 4
  selector:
    matchLabels:
      sample: color
  template:
    metadata:
      name: sample-color-pod
      labels:
        sample: color
        color: blue
    spec:
      nodeSelector:
        kubernetes.io/os: linux
      containers:
      - name: nginxcolordemo
        image: k8sonazureworkshoppublic.azurecr.io/scubakiz/nginxcolordemo:blue-1.0
        ports:
        - containerPort: 80
          protocol: TCP
        env: 
        - name: NODE_IP
          valueFrom:
            fieldRef:
              fieldPath: status.hostIP
        - name: NODE_NAME
          valueFrom:
            fieldRef:
              fieldPath: spec.nodeName
        - name: POD_NAMESPACE
          valueFrom:
            fieldRef:
              fieldPath: metadata.namespace
        - name: POD_SERVICE_ACCOUNT
          valueFrom:
            fieldRef:
                fieldPath: spec.serviceAccountName 
        - name: POD_CPU_REQUEST
          valueFrom:
            resourceFieldRef:
              containerName: nginxcolordemo
              resource: requests.cpu
        - name: POD_CPU_LIMIT
          valueFrom:
            resourceFieldRef:
              containerName: nginxcolordemo
              resource: limits.cpu
        - name: POD_MEM_REQUEST
          valueFrom:
            resourceFieldRef:
              containerName: nginxcolordemo
              resource: requests.memory
        - name: POD_MEM_LIMIT
          valueFrom:
            resourceFieldRef:
              containerName: nginxcolordemo
              resource: limits.memory 
        volumeMounts:
        - name: podinfo
          mountPath: /etc/podinfo    
      volumes:
      - name: podinfo
        downwardAPI:
          items:
          - path: "labels"
            fieldRef:
              fieldPath: metadata.labels
          - path: "annotations"
            fieldRef:
              fieldPath: metadata.annotations
EOF

Review running Pods and their labels:
kubectl get pods --show-labels
kubectl get pods --show-labels
Note
Notice the label sample=color that is associated with the Pods.

Create a Service manifest with a selector that matches the sample=color label. This Service will track all Pods that have this label and load balance traffic between them:

$sampleSvcYaml = @"
apiVersion: v1
kind: Service
metadata:
  name: sample-svc
spec:
  ports:
  - port: 80
  selector:
    sample: color
  type: LoadBalancer
"@

$sampleSvcYaml | kubectl apply -f -

cat << EOF | kubectl apply -f -
apiVersion: v1
kind: Service
metadata:
  name: sample-svc
spec:
  ports:
  - port: 80
  selector:
    sample: color
  type: LoadBalancer
EOF

Check the newly created service:
kubectl get svc -o wide
kubectl get svc -o wide
Tip
The command above will display the details of all available services along with their label selectors. You should see the sample-svc Service with SELECTOR sample=color. This service should get an external IP that you can use in the next step. It may take several minutes for this IP to appear, it will show <pending> in the meantime.

Task 2 - Access the sample-svc Service

Open a browser and navigate to the IP address shown in the output of the previous command.
The website displays the Node IP/Pod IP address of the pod currently receiving the traffic through the service’s load balancer. The page refreshes every 3 seconds and each request may be directed to a different pod, with a different IP address. This is the service’s internal load balancer at work.
Tip
You may not see the Pod IP address change if you are only accessing the webpage from a single tab in your browser. If this is the case, try opening the same URL in a private browsing window, or a different browser / different device - you should see a different Pod IP on these other devices.

Task 3 - Clean Up

Leave the current deployment and service running for the next exercise. You will clean them up at the end of this lab.