OpenLMIS HAPI FHIR service

This repository contains openlmis-hapifhir service.


  • Docker 1.11+
  • Docker Compose 1.6+

All other dependencies, such as Java, are delivered automatically via the Docker image. It is unnecessary to install them locally to run the service, though often helpful to do so for the sake of development. See the Tech section of openlmis/dev for a list of these optional dependencies.

Quick Start

  1. Fork/clone this repository from GitHub.
git clone <openlmis-your-service-name>
  1. Add an environment file called .env to the root folder of the project, with the required project settings and credentials. For a starter environment file, you can use this one. e.g.
cd openlmis-hapifhir
curl -o .env -L
  1. Develop w/ Docker by running docker-compose run --service-ports <your-service-name>. See Developing w/ Docker.
  2. You should now be in an interactive shell inside the newly created development environment, start the Service with: gradle bootRun
  3. Go to http://<yourDockerIPAddress>:8080/ to see the service name and version. Note that you can determine yourDockerIPAddress by running docker-machine ip.
  4. Go to http://<yourDockerIPAddress>:8080/api/ to see the APIs.

Building & Testing

Gradle is our usual build tool. This template includes common tasks that most Services will find useful:

  • clean to remove build artifacts
  • build to build all source. build, after building sources, also runs unit tests. Build will be successful only if all tests pass.
  • generateMigration -PmigrationName=<yourMigrationName> to create a “blank” database migration file. The file will be generated under src/main/resources/db/migration. Put your migration SQL into it.
  • test to run unit tests
  • integrationTest to run integration tests
  • sonarqube to execute the SonarQube analysis.

The test results are shown in the console.

While Gradle is our usual build tool, OpenLMIS v3+ is a collection of Independent Services where each Gradle build produces 1 Service. To help work with these Services, we use Docker to develop, build and publish these.

See Developing with Docker.

Developing with Docker

OpenLMIS utilizes Docker to help with development, building, publishing and deployment of OpenLMIS Services. This helps keep development to deployment environments clean, consistent and reproducible and therefore using Docker is recommended for all OpenLMIS projects.

To enable development in Docker, OpenLMIS publishes a couple Docker Images:

  • openlmis/dev - for Service development. Includes the JDK & Gradle plus common build tools.
  • openlmis/postgres - for quickly standing up a shared PostgreSQL DB

In addition to these Images, each Service includes Docker Compose instructions to:

  • standup a development environment (run Gradle)
  • build a lean image of itself suitable for deployment
  • publish its deployment image to a Docker Repository

Development Environment

Launches into shell with Gradle & JDK available suitable for building Service. PostgreSQL connected suitable for testing. If you run the Service, it should be available on port 8080.

Before starting the development environment, make sure you have a .env file as outlined in the Quick Start instructions.

> docker-compose run --service-ports <your-service-name>
$ gradle clean build
$ gradle bootRun

Build Deployment Image

The specialized docker-compose.builder.yml is geared toward CI and build servers for automated building, testing and docker image generation of the service.

Before building the deployment image, make sure you have a .env file as outlined in the Quick Start instructions.

> docker-compose -f docker-compose.builder.yml run builder
> docker-compose -f docker-compose.builder.yml build image

Publish to Docker Repository


Docker’s file details

A brief overview of the purpose behind each docker related file

  • Dockerfile: build a deployment ready image of this service suitable for publishing.
  • docker-compose.yml: base docker-compose file. Defines the basic composition from the perspective of working on this singular vertical service. These aren’t expected to be used in the composition of the Reference Distribution.
  • docker-compose.override.yml: extends the docker-compose.yml base definition to provide for the normal usage of docker-compose inside of a single Service: building a development environment. Wires this Service together with a DB for testing, a gradle cache volume and maps tomcat’s port directly to the host. More on how this file works:
  • docker-compose.builder.yml: an alternative docker-compose file suitable for CI type of environments to test & build this Service and generate a publishable/deployment ready Image of the service.
  • Docker-compose file suitable for production. Contains nginx-proxy image and virtual host configuration of each service.

Running complete application with nginx proxy

  1. Enter desired VIRTUAL_HOST for each service in the file.
  2. Start up containers
> docker-compose -f docker-compose.yml -f up
  1. The application should be available at port 80.


Logging is implemented using SLF4J in the code, Logback in Spring Boot, and routed to an external Syslog server. There is a default configuration XML (logback.xml) in the resources folder. To configure the log level for the development environment, simply modify the logback.xml to suit your needs.

Configuring log level for a production environment is a bit more complex, as the code has already been packaged into a Spring Boot jar file. However, the default log configuration XML can be overridden by setting the Spring Boot logging.config property to an external logback.xml when the jar is executed. The container needs to be run with a JAVA_OPTS environment variable set to a logback.xml location, and with a volume with the logback.xml mounted to that location. Some docker compose instructions have been provided to demonstrate this.

  1. Build the deployment image. (See Build Deployment Image)
  2. Get a logback.xml file and modify it to suit your log level configuration.
  3. Modify docker-compose.builder.yml to point to your logback.xml location. a. Under volumes, where it shows two logback.xml locations separated by a colon, change the location before the colon.
  4. Run the command below.
> docker-compose -f docker-compose.builder.yml run --service-ports hapifhir

Internationalization (i18n)

Internationalization is implemented by the definition of two beans found in the Application class, localeResolver and messageSource. (Alternatively, they could be defined in an application context XML file.) The localeResolver determines the locale, using a cookie named lang in the request, with en (for English) as the default. The messageSource determines where to find the message files.

Note there is a custom message source interface, ExposedMessageSource, with a corresponding class ExposedMessageSourceImpl. These provide a method to get all the messages in a locale-specific message file.

See the MessageController class for examples on how to get messages.

Additionally, Transifex has been integrated into the development and build process. In order to sync with the project’s resources in Transifex, you must provide values for the following keys: TRANSIFEX_USER, TRANSIFEX_PASSWORD.

For the development environment in Docker, you can sync with Transifex by running the script. This will upload your source messages file to the Transifex project and download translated messages files.

The build process has syncing with Transifex seamlessly built-in.


To debug the Spring Boot application, use the --debug-jvm option.

$ gradle bootRun --debug-jvm

This will enable debugging for the application, listening on port 5005, which the container has exposed. Note that the process starts suspended, so the application will not start up until the debugger has connected.

Production by Spring Profile

By default when this service is started, it will clean its schema in the database before migrating it. This is meant for use during the normal development cycle. For production data, this obviously is not desired as it would remove all of the production data. To change the default clean & migrate behavior to just be a migrate behavior (which is still desired for production use), we use a Spring Profile named production. To use this profile, it must be marked as Active. The easiest way to do so is to add to the .env file:


This will set the similarly named environment variable and limit the profile in use. The expected use-case for this is when this service is deployed through the Reference Distribution.

Demo Data

A basic set of demo data is included with this service, defined under ./demo-data/. This data may be optionally loaded by using the demo-data Spring Profile. Setting this profile may be done by setting the environment variable.

When building locally from the development environment, you may run:

$ export spring_profiles_active=demo-data
$ gradle bootRun

To see how to set environment variables through Docker Compose, see the Reference Distribution

Environment variables

Environment variables common to all services are listed here:

  • LARGEST_GEO_LEVEL_NUMBER - Define the level number of the largest area in the OpenLMIS eco-system. The default value is 1.
  • FACILITY_TYPE_ID - Define the id of the default facility type which will be used when a new facility resource will be created. The default value is ae9715b4-2a72-4769-8121-e3894aec5b70.