Skip to content
A simple customer database for a fictitious company. Kubernetes + Java backed by MongoDB. Makes heavy use of Microsoft managed services to keep things easy.
Java Smarty Makefile
Branch: master
Clone or download
Pull request Compare This branch is 2 commits ahead of nomisbeme:master.
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.


A simple customer database for a fictitious company. Kubernetes + Java backed by MongoDB. Makes heavy use of Microsoft managed services to keep things easy. Traffic to/from the application is secured using TLS.

This is a demonstration application that optimizes for simplicity and learning. Not intended for production use.



Provisioning the initial resources can take a little time so I did these steps upfront.

  1. Download and setup the command-line tools you'll need on your computer. The examples below assume Mac but should work on other platforms.
  1. Create an Azure group and an empty Kubernetes cluster in that group
az group create --name bayazure --location eastus
az aks create --name k8s --resource-group bayazure
az aks get-credentials --name k8s --resource-group bayazure
  1. Add some needed components to Kubernetes
$ helm init
$ helm install stable/nginx-ingress
$ helm install --name kube-lego stable/kube-lego --set --set config.LEGO_URL=

We use nginx-ingress to route external traffic within our cluster. kube-lego is used to automatically obtain a TLS certificate from Let's Encrypt. Adjust the example values above for your environment and domain.

  1. Next we'll create a CosmosDB MongoDB database to store our application data in.
$ az cosmosdb create --kind MongoDB --name customerdatabase --resource-group bayazure

Building and deploying the application

  1. First create a Kubernetes secret so that the application can connect to MongoDB:
kubectl create secret generic customerdatabasesecret \
    az cosmosdb list-connection-strings -n customerdatabase -g bayazure --query "connectionStrings[0].connectionString" \
      --out tsv | tr -d '\n')"
  1. Next we'll build our Spring Boot application code:
mvn clean package

This creates an file target/customerapp-0.1.2.war that contains our application.

  1. Then package the application in a Container image and push it to the Docker Hub:
docker build -t nomisbeme/customerapp:0.1.2 .
docker push nomisbeme/customerapp:0.1.2
  1. Use the Kubernetes package manager, helm, to deploy the resulting application to Kubernetes.
helm install helm/customerapp/ -n capp

Adjust the to point to the correct domain name if needed.

  1. Wait for the application to deploy then access it at your domain name.

Updating the application

Kubernetes and Helm make it easy to deploy new versions of an application, so we'll modify the application and deploy the result.

  1. Modify the application. Hint: Uncomment list.jsp

  2. Rebuild the application and push a new version to the Docker Hub:

mvn clean package
docker build -t nomisbeme/customerapp:fixed .
docker push nomisbeme/customerapp:fixed
  1. Now update the existing deployment to reference the new code.
helm upgrade --set image.tag=fixed capp helm/customerapp


  • Use Open Service Catalog for Azure to expose Cosmos within the Kubernetes cluster
  • Use the Spotify maven plugins to simplify container image creation
  • Use a private registry e.g. Azure Container Registry to avoid making application code publicly available.
  • Replace kube-lego with the more modern replacement cert-manager
You can’t perform that action at this time.