Creating a Loadbalancer

In the previous section we created a pod running an nginx webserver. Now we need a way to make the webserver accessible from the Internet. In the Nectar Cloud, we can do this by creating a load balancer.

A load balancer has a public (floating) IP. Client accessing via this public IP are redirected to one or more private addresses within the cluster.

(Note that a Magnum cluster is configured with 2 loadbalancers for cluster management. The loadbalancer we will be creating here is for a different purpose.)

  1. Create a yaml file to describe the configuration for your load balancer. Save the following as nginxservice.yaml.

    apiVersion: v1
    kind: Service
      name: nginxservice
      - port: 80
        targetPort: 80
        protocol: TCP
        run: webserver
      type: LoadBalancer
  2. Create the loadbalancer using the yaml file:

    $ kubectl apply -f nginxservice.yaml
    service/nginxservice created
  3. Run the following to see the details of the loadbalancer, and wait until the output shows EXTERNAL-IP as populated.

    $ kubectl get services
    NAME           TYPE           CLUSTER-IP       EXTERNAL-IP   PORT(S)        AGE
    nginxservice   LoadBalancer   80:31086/TCP   5m

    Tip: If you use the -w flag with any kubectl get command, it will watch for changes from the server and print out status updates as they happen.

  4. Verify that the webserver is accessible. From your browser, visit the IP in the EXTERNAL-IP field. You should see the default nginx page.

More information

This is what we just did:

  1. We started an external LoadBalancer service in Kubernetes, and directed it at the webserver pod that we created previously.

  2. Kubernetes created the loadbalancer (externally) by calling out to the OpenStack loadbalancer provider plugin.

  3. The plugin sent Openstack requests to create the resources to make it all work, namely: a floating ip address, a security group, and a load balancer, pool listener and members.

  4. We observed what had been set up:

    • You can see the loadbalancer from the Kubernetes perspective by running kubectl get services

    • You can also see the resources from the Openstack perspective using the openstack command line; e.g. openstack floating ip list, openstack loadbalancer list and so on.

  5. Finally, we checked that the loadbalancer was working using a web browser.