Typically, while many applications may only need to communicate internally within a pod, or even across pods, you may need to expose your application externally so that clients outside of the Kubernetes cluster can interface with the application. You can do this by creating a service definition for the deployment.

To expose a deployment using a service object, you must define the service type that should be used. If you are not using a cloud-based load balancing service, you can set the service type to NodePort. The NodePort service exposes the application running within the cluster on a dedicated port on the public IP address on all of the nodes within the cluster. Use the kubectl expose deployment to create a new service:

$ kubectl expose deployment hello-world --port 80 --type=LoadBalancer
service/hello-world exposed

Use kubectl get services to list the different services that the cluster is running, and to obtain the port information required to access the service:

$ kubectl get services
NAME          TYPE           CLUSTER-IP      EXTERNAL-IP   PORT(S)        AGE
hello-world   LoadBalancer   10.102.42.160   <pending>     80:31847/TCP   3s
kubernetes    ClusterIP      10.96.0.1       <none>        443/TCP        5h13m

In this example output, you can see that traffic to port 80 inside the cluster is mapped to the NodePort 31847. The external IP that can be used to access the service is listed as <pending>, meaning that if you connect to the external IP address for any of the nodes within the cluster on the port 31847, you are able access the service.

For the sake of the example in this guide, you can open a web browser to point at any of the nodes in the cluster, such as http://worker1.example.com:31847/, and it should display the NGINX demonstration application.