In a previous article, we covered how easy it is to create Spring Boot containers with Rockcraft. So the next logical step is to deploy and operate your application in a production environment. The Juju ecosystem is the key to making this process straightforward.

In this article we walk through the steps required to deploy a Spring Boot application to production using Juju and Kubernetes. The goal is to showcase the integration of the application with essential services like PostgreSQL for database management and Traefik for ingress control.

Ordinarily, deploying and operating Spring Boot containers is easier said than done. Any non-trivial application will require integrations with other software components like databases, message brokers, and authentication services. Some of these components are complex to operate. What’s more, your application will need to be exposed to the outside world. And for Day 2 operations, you’ll also need to perform routine maintenance and properly monitor your application.

Fortunately, Juju provides a strong ecosystem composed of a multitude of high quality software operators that enables deployment, integration and lifecycle management at any scale, on any infrastructure.

With the release of the Spring Boot framework extensions for Rockcraft and Charmcraft, if your application follows the 12-factor methodology best practices, you can benefit from the Juju ecosystem out of the box.

If you are developing a Spring Boot application we recommend you to try out the devpack support for Spring and then follow this tutorial to write your Spring Boot charm. 

But without further ado, let’s see how it all works by charming spring-petclinic. 

Set up a developer machine with Canonical Kubernetes and Juju

We will start with a virtual machine created from scratch using Multipass and prepare our development machine with the help of the Concierge tool.

Launch the new virtual machine and shell into it:

multipass launch --cpus 4 --disk 30G --memory 4G --name blogpost 24.04

multipass shell blogpost

We will provision the testing machine with Concierge, which will install and configure Rockcraft, Charmcraft, Canonical Kubernetes and Juju. Besides that, we will update the load balancer to serve under the machine’s main IP and install an OCI image registry:

sudo snap install --classic concierge

sudo concierge prepare -p k8s --charmcraft-channel latest/edge

PREFSRC=$(ip -4 -j route get 2.2.2.2 | jq -r '.[] | .prefsrc')

sudo k8s set load-balancer.cidrs=$PREFSRC/32

curl https://raw.githubusercontent.com/canonical/spring-petclinic/refs/heads/resources/registry.yaml | kubectl apply -f - 

Get spring-petclinic application

spring-petclinic is the app we’ll be charming. It’s a sample application created by the Spring team to demonstrate a real-world web app using the Spring Framework. 

Clone the spring-petclinic repository. 

git clone https://github.com/canonical/spring-petclinic.git

cd spring-petclinic

Following the spring-petclinic repository instructions, you can build a jar and run it with the command line. It is quite easy, but clearly not production ready, even the database in-memory.

Create the rock

Let’s prepare a rock – a compliant OCI image – for the application. We could use either Maven or Gradle to build the application, but for this example we will use Maven. From the spring-petclinic directory, remove the Gradle artifacts and initialize the rockcraft project:

rm -rf build.gradle gradlew gradle

rockcraft init --profile spring-boot-framework

Pack the rock and upload it to the OCI registry :

ROCKCRAFT_ENABLE_EXPERIMENTAL_EXTENSIONS=True rockcraft pack

rockcraft.skopeo copy --insecure-policy --dest-tls-verify=false oci-archive:spring-petclinic_0.1_amd64.rock docker://localhost:32000/spring-petclinic:0.1

At this point we have a fully compliant OCI image. But to fully integrate with the Juju ecosystem we need a charm, a software operator. 

Create the charm

Let’s build it:

mkdir charm

cd charm

charmcraft init --profile spring-boot-framework --name spring-petclinic

As we want to use PostgreSQL instead of an in-memory database, add the following lines to the file charmcraft.yaml:

requires:
  postgresql:
    interface: postgresql_client
    optional: false
    limit: 1

And we are ready to create a charm. Pack it:

CHARMCRAFT_ENABLE_EXPERIMENTAL_EXTENSIONS=True charmcraft pack

Let’s test it

At this point we have a charm for spring-petclinic that will integrate seamlessly with PostgreSQL. 

Let’s test it with a new Juju model in a Kubernetes cloud:

juju add-model spring-petclinic

juju deploy ./spring-petclinic_amd64.charm spring-petclinic --resource app-image=localhost:32000/spring-petclinic:0.1 --config app-profiles=postgres

Deploy PostgreSQL and Traefik and integrate with spring-petclinic:

juju deploy postgresql-k8s --trust

juju integrate spring-petclinic postgresql-k8s

juju deploy traefik-k8s --trust

juju integrate spring-petclinic traefik-k8s

After a while, your production ready application will be ready for use! You can check it with the command juju status:

You can get the URL of the application with:

juju run traefik-k8s/0 show-proxied-endpoints

And just like that, you can use it!

The application is running in a Kubernetes cluster, with Trafik acting as an ingress. Your spring-petclinic application is now using PostgreSQL to store its data. The Charmed PostgreSQL K8s has provided credentials for the database to the spring-petclinic application, and is ready to be operated in a production-grade environment. 

Not bad for a start.

Get in touch

Do you have a project or setup where you want to employ this solution? If you have any questions about the 12-factor tooling for Rockcraft and Charmcraft, reach out in our Matrix Channel.

What’s next?

• Follow the Spring Boot tutorial
• Publish your charm in CharmHub