Needed tools

• Docker
• Minikube (+ Ingress)
• Maven
• Make

How to run it

After cloning this repository, navigate into the root directory and execute make setup.

This will:

• create a namespace
• build a docker image and tag it
• install the Strimzi operator, deploy a Kafka instance and create a Kafka topic. This is done purely to demonstrate how to deploy "infrastructre"
• deploy the Java application on minikube

Check if it is up and running my executing kubectl get pods -n q8s

After making some changes to your Java application, you can simply use make build deploy to rebuild and redeploy your app.

How does it work?

While trying to find a way to develop with minikube locally, I've ran into the problem that rebuilding and tagging a docker image with latest doesn't work in combination with kubectl apply. Kubernetes won't do anything if you apply the same configuration file without any changes.

I also didn't want to push the image to a Docker registry everytime I make code changes and make my local installation to pull the image everytime. What if I'm working somewhere where the internet connection is bad?

If you have a look at the pom.xml file, you will see that this POC uses the dockerfile-maven plugin for building and tagging our image. You'll also see that there is a maven property called imageTag. This property is used to define how our images should be tagged.

Now check the Makefile. The build recipe does three things:

• It configures the shell to use Minikube as Docker host. By doing so, Minikube does not need to pull the image.
• Pruning unused Docker images. Attention: This is just for testing. Obviously you don't want to delete all your images just becaused they are not running in a container right now.
• Building the app and a Docker image by using Maven and the dockerfile-maven plugin. Mind the argument that is passed: -DimageTag=${IMAGE_TAG}. This will override the property defined in the pom.xml. In this POC, this is a timestamp. This means, after the build succeeds, the image will be tagged with something similar to 2019-04-14---10:03:01.

Have a look at your images by using docker images.

Now check the quarkusk8s.yaml. This is where the deployment configuration for our app is defined. Two things are important:

image: zinnsoldat/quarkusk8s:${IMAGE_TAG}

Instead of using a fixed tag, we're using a placeholder for the tag name.

imagePullPolicy: IfNotPresent

By setting the imagePullPolicy to IfNotPresent, Kubernetes will use local images (if they are present) instead of pulling images from a remote registry.

Obviously, kubectl does not care about our placeholders. Thats why the deploy recipe in the Makefile uses envsubst to replace the placeholder and copies it to the /tmp/ directory. Then the temporaraly created configuration file is applied.