2018-05-06

Kubernetes: Using MetalLB to expose a service

Intro

Once we have our Kubernetes cluster, it’s time to deploy something in it. I’m going to follow a simple deploying a Ratpack application I did sometime ago and that I packed as a docker image at Bintray.

Installing POD

Pod declaration

In the POD declaration I’m defining the Docker image url. And the port where the application is exposed.

apiVersion: v1
kind: Pod
metadata:
  name: pnlp-pod
spec:
  containers:
    - name: pnlp-pod
      image: pacotheai-docker-public.bintray.io/pnlp:0.1.0
      ports:
        containerPort: 5050

How to deploy my POD

kubectl create -f pnlp-pod.yml

How to see deployed POD details

If we want to see things like which private IP is the app exposed, we can use:

kubectl describe pods/pnlp-pod.yml

How to destroy my POD

In case we would like to terminate the POD:

kubectl delete pods pnlp-pod.yml

Installing service

We’re using a Load Balancer to expose our service to a public IP.

MetalLB configuration

MetalLB is nowadays the only load balancer available for a barebone Kubernetes solution. It will expose our application in the available public IPs. In order to provide MetalLB with a range of available IPs we can deploy a MetalLB configuration to Kubernetes like the following:

I’m using my Vagrant Kubernetes image which already has MetalLB installed and configured.

apiVersion: v1
kind: ConfigMap
metadata:
  namespace: metallb-system
  name: config
data:
  config: |
    address-pools:
    - name: my-ip-space
      protocol: layer2
      addresses:
      - 192.168.250.112/29

Service definition

Ok so we’ve already deployed our application but it’s not publicly available, it would be nice to publish it to the outside world. In order to do that we need to define a service of type LoadBalancer.

apiVersion: v1
kind: Service
metadata:
  name: pnlp-pod
  labels:
    name: pnlp-pod
spec:
  type: LoadBalancer
  ports:
    - port: 80
  selector:
    name: pnlp-pod

How to deploy service

kubectl create -f pnlp-service.yml

Check where the service’s been deployed ?

kubectl get services