Quarkus in Real World Deployments

This is the Github repository with sample applications for the presentation Quarkus in Real World Deployments.

The purpose of these sample applications is to showcase typical use cases and the experience that the authors had when adopted Quarkus to develop applications.

Pre Requisites

• OpenJDK 11
• GraalVM 19.3.1
• Maven 3.6.2
• Docker
• Docker-compose

Project Contents

The project showcases two services. The Legume Service, used to Create, Read, Update and Delete Legumes. Each time a Legume is created, a new event with that information is produced and published to RabbitMQ. This allows other services, interested in the event to subscribe and consume it to perform additional actions.

The Hero Service consumes the event published by the Legume Service and transform the Legume into a Super Hero Legume.

Both Legume and Hero Service store the data into a Postgres database.

A separate R Service also exists in the project structure. This service showcases the Polyglot capabilities of GraalVM by mixing Java and R code in the same project. The R Service exposes a Calculator with the most common operations, with JAX-RS and the actual Calculator functions are provided by R.

Project Structure

The project is split into several modules:

• legume-service (a CRUD service to manage Legumes. When a new Legume is created it will be posted to RabbitMQ).
• hero-service (Consumes messages from RabbitMQ. It will transform Legumes into Hero Legumes).
• camel-support (A custom Camel ProcessorFactory to fix a cache issue with the RabbitMQ adapter).
• r-service (A Calculator service, with operations implemented in R and exposed with JAX-RS endpoints).

Build

Just use Maven to build the project with the following command:

mvn install

With Docker

You can also build Docker Images to run the apps:

docker-compose build

Run

With Docker

The easiest way to run the the entire system is to use docker-compose. This will run the apps, plus some required infrastructure, like a posgres database and a rabbitmq message broker.

docker-compose up

And then you can use the following command to remove all the containers:

docker-compose down

With Java

You always going to need to use docker-compose to run the required infrastructure, but you can run the apps directly with your JVM so you can play and change them around. Run only the infrastructure:

docker-compose up database rabbitmq

Then you run both legume-service and hero-service by going to each module folder and running the commands (depending on the project):

java -jar target/legume-service-1.0-SNAPSHOT-runner.jar

java -Dquarkus.http.port=8081 -jar target/hero-service-1.0-SNAPSHOT-runner.jar

Test the Applications

There is no UI to test the application. You can use a curl command to send a payload and check the applications:

curl -v -XPOST http://localhost:8080/legumes -H 'Content-Type: application/json' -d '{"name":"Broccoli","description":"Green Plant"}' 

Native Images

Install GraalVM Native Image binary with:

gu install native-image

Make sure that you have an envirobment variable GRAALVM_HOME pointing to your GraalVM installation. With this you should be able to compile both Legume and Hero services as native binaries and run them without a JVM:

mvn package -Pnative

This is going to generate a binary executable. Run with:

./target/legume-service-1.0-SNAPSHOT-runner

./target/hero-service-1.0-SNAPSHOT-runner -Dquarkus.http.port=8081

Polyglot Java - R Service

Build

Make sure that you use GraalVM as your JVM to run the application. You need to point JAVA_HOME to GraalVM.

First your must have the R language installed in your GraalVM:

gu install R

Then, just use Maven to build the project:

mvn install -Pall

Run

Move to the r-service folder and run:

java -jar target/r-service-1.0-SNAPSHOT-runner.jar

Test

Just use the following curl commands to perform operations:

curl -XPOST http://localhost:8090/calculator/subtract -d 'x=1&y=2.1'

curl -XPOST http://localhost:8090/calculator/add -d 'x=1&y=1'

The first call is slow, since it requires to start the R context. After that, subsequent calls are fast.