Docker and Docker Compose

Summary

Using Docker and Docker Compose is highly recommended in development, and is also a good solution in production.

Description
Prerequisites
Differences when using a microservices architecture
Building a Docker image of your application
Working with databases
Elasticsearch
Sonar
Common commands

Description

Please note: this Docker configuration is used to run your generated application(s) inside a container image. It’s completely different from the Docker setup that JHipster also provides, which is for running the JHipster generator inside a container

JHipster provides a complete Docker support, in order to:

Facilitate development, as you can start a full infrastructure very easily, even when using a complex microservices architecture
For people using Docker Swarm, deploying to production directly, as it uses the same Docker Compose configuration

One great feature of using Docker Compose is that you can easily scale your containers, using the docker-compose scale command. This is very interesting if you use JHipster with a a microservices architecture.

When generating your application, JHipster generates for you:

A Dockerfile for building a Docker image and running your application inside a container
Several Docker Compose configurations to help you run your application with third-party services, for example a database

Those files are located inside folder src/main/docker/.

Prerequisites

You have to install Docker and Docker Compose:

Tip: On Windows and Mac OS X, Kitematic is an easy-to-use graphical interface provided with the Docker Toolbox, which will makes using Docker a lot easier.

Warning: If you are using Docker Machine on Mac or Windows, your Docker daemon has only limited access to your OS X or Windows file system. Docker Machine tries to auto-share your /Users (OS X) or C:\Users\<username> (Windows) directory. So you have to create the project folder under this directory to avoid any issues especially if you are using the JHipster Console for monitoring.

Differences when using a microservices architecture

if you are working on a monolithic application, you can skip this section

If you have selected to generate a microservices architecture, each application (gateway, microservice) has a Dockerfile and a Docker Compose configurations, like with a normal monolithic application.

But you can use the specific docker-compose sub-generator, which will generate a global Docker Compose configuration for all your gateway(s) and microservices. This will allow you to deploy and scale your complete architecture with one command.

You need to have all your gateway(s) and microservices in the same directory.
Create another directory, for example mkdir docker-compose.
Go into that directory: cd docker-compose.
Run the sub-generator: yo jhipster:docker-compose.
The sub-generator will ask you which application you want to have in your architecture, and if you want to have monitoring with ELK included.

This will generate a global Docker Compose configuration, type docker-compose up to run it, and have all your services running at once.

This configuration will have a pre-configured JHipster Registry, that will configure your services automatically:

Those services will wait until the JHipster Registry is running until they can start (this can be configured in your bootstrap-prod.yml file using the spring.cloud.config.fail-fast and spring.cloud.config.retry keys).
The registry will configure your applications, for example it will share the JWT secret token between all services.
Scaling each service is done using Docker Compose, for example type docker-compose scale test-app=4 to have 4 instances of application “test” running. Those instances will be automatically load-balanced by the gateway(s), and will automatically join the same Hazelcast cluster (if Hazelcast is your Hibernate 2nd-level cache).

Building and running a Docker image of your application

To create a Docker image of your application, and push it into your Docker registry:

With Maven, type: ./mvnw package -Pprod docker:build
With Gradle, type: ./gradlew bootRepackage -Pprod buildDocker

This will package your application with the prod profile, and install the image.

To run this image, use the Docker Compose configuration located in the src/main/docker folder of your application:

docker-compose -f src/main/docker/app.yml up

This command will start up your application and the services it relies on (database, search engine, JHipster Registry…).

Working with databases

MySQL, MariaDB, PostgreSQL, MongoDB or Cassandra

Running docker-compose -f src/main/docker/app.yml up already starts up your database automatically.

If you just want to start your database, and not the other services, use the Docker Compose configuration of your database:

With MySQL: docker-compose -f src/main/docker/mysql.yml up
With MariaDB: docker-compose -f src/main/docker/mariadb.yml up
With PostgreSQL: docker-compose -f src/main/docker/postgresql.yml up
With MongoDB: docker-compose -f src/main/docker/mongodb.yml up
With Cassandra: docker-compose -f src/main/docker/cassandra.yml up

MongoDB Cluster Mode

If you want to use MongoDB with a replica set or shards and a shared configuration between them, you need to build and set up manually Mongo images.
Follow these steps to do so:

Build the image: docker-compose -f src/main/docker/mongodb-cluster.yml build
Run the database: docker-compose -f src/main/docker/mongodb-cluster.yml up -d
Scale the mongodb node service (you have to choose an odd number of nodes): docker-compose -f src/main/docker/mongodb-cluster.yml scale <name_of_your_app>-mongodb-node=X
Init the replica set (param is the number of node, folder is the folder where the YML file is located, it’s docker by default): docker exec -it <yml_folder_name>_<name_of_your_app>-mongodb-node_1 mongo --eval 'var param=X, folder="<yml_folder_name>"' init_replicaset.js
Init the shard: docker exec -it <name_of_your_app>-mongodb mongo --eval 'sh.addShard("rs1/<yml_folder_name>_<name_of_your_app>-mongodb-node_1:27017")'
Start your application: docker-compose -f src/main/docker/app.yml up -d <name_of_your_app>-app

If you want to add or remove some MongoDB nodes, just repeat step 3 and 4.

Cassandra

Unlike the other databases, where the schema migrations are executed by the application itself, Cassandra schema migrations are executed by a dedicated Docker container.

Cassandra in development

To start a Cassandra cluster to run your application locally, you can use the docker_compose file for development use:
docker-compose -f src/main/docker/cassandra.yml up -d

Docker-compose will start 2 services:

<name_of_your_app>-cassandra: a container with the Cassandra node contact point
<name_of_your_app>-cassandra-migration: a container to automatically apply all CQL migrations scripts (create the Keyspace, create the tables, all data migrations, …)

See the Cassandra page for more information on how to add new CQL scripts without restarting the local cluster.

Cassandra in production:

The app.yml docker-compose file uses cassandra-cluster.yml to configure the cluster.
The application starts after few seconds (see JHIPSTER_SLEEP variable) to gives the time to the cluster to start and the migrations to be executed.

One big difference between Cassandra and the other databases, is that you can scale your cluster with Docker Compose. To have X+1 nodes in your cluster, run:

docker-compose -f src/main/docker/cassandra-cluster.yml scale <name_of_your_app>-cassandra-node=X

Elasticsearch

Running docker-compose -f src/main/docker/app.yml up already starts up your search engine automatically.

If you just want to start your Elasticsearch node, and not the other services, use its specific Docker Compose configuration:

docker-compose -f src/main/docker/elasticsearch.yml up

Sonar

A Docker Compose configuration is generated for running Sonar:

docker-compose -f src/main/docker/sonar.yml up

To analyze your code, run Sonar on your project:

With Maven: ./mvnw sonar:sonar
With Gradle: ./gradlew sonar

The Sonar reports will be available at: http://localhost:9000

Common commands

List the containers

You can use docker ps -a to list all the containers

$ docker ps -a
CONTAINER ID        IMAGE               COMMAND                  CREATED             STATUS              PORTS                    NAMES
fc35e1090021        mysql               "/entrypoint.sh mysql"   4 seconds ago       Up 4 seconds        0.0.0.0:3306->3306/tcp   sampleApplication-mysql

Scale a container

Run docker-compose scale test-app=4 to have 4 instances of application “test” running.

Stop containers

docker-compose -f src/main/docker/app.yml stop

You can also use directly Docker:

docker stop <container_id>

When you stop a container, the data is not deleted, unless you delete the container.

Delete a container

Be careful! All data will be deleted:

docker rm <container_id>