Skip to main content

Microservice

Intent

OpenFn is used by numerous health and humanitarian organizations around the world to scale their programs through real-time interoperability, systems integration, and workflow automation. OpenFn/microservice makes use of OpenFn's open-core technology—namely OpenFn/core, OpenFn/engine, and the various OpenFn adaptors—to create standalone microservices which can be deployed on any hardware.

This microservice approach helps to ensure that governments and NGOs are never locked-in to OpenFn's SaaS offering, and can port their existing jobs, triggers, and credentials from OpenFn.org to their own infrastructure easily.

Introduction

Similar to platform, OpenFn/microservice runs on project.yaml files. This means that when organizations or governments have an open-source license requirement, all their jobs, credentials, and project configurations can be exported from OpenFn's iPaaS and used to create a microservice deployment.

While this approach doesn't provide the OpenFn platform front-end with its various project management and configuration features, it's perfect for groups with DevOps experience and 100% compatible with the platform. You can even build and test entire projects on platform and then export the project.yaml file to run on your own servers using microservice.

This microservice approach provides flexibility to governments and NGOs, so they are never locked-in to OpenFn's SaaS platform offering. At any time, an organization can port their existing jobs, triggers, and credentials from OpenFn.org to run with our FOSS integration toolkit, using their own infrastructure.

Prerequisites

Familiarity with other elements of OpenFn's open source integration toolkit is helpful when considering the microservice approach.

Docker up and running

Assuming you've got an .env and a sample project at ./sample-project directory with a project.yaml spec:

docker-compose up

You can configure either the compose file or the .env, or run the container using docker run:

docker run -v <path-to-your-project-folder>:/home/microservice/<path-to-your-project-folder> \
--env-file <path-to-your-env-file> \
--network host \
openfn/microservice:v0.3.2

Development up and running guide

  • Clone this repo with git clone git@github.com:OpenFn/microservice.git
  • Enter the directory with cd microservice
  • Install dependencies with mix setup
  • Run the tests with mix test
  • Make a project directory to hold your project artifacts with mkdir sample-project
  • Create a new project specification with cp project.yaml.example ./sample-project/project.yaml
  • Create a .env file with cp .env.example .env
  • Install necessary adaptors via npm install @openfn/language-http --prefix priv/openfn/runtime/node_modules --no-save --no-package-lock --global-style
  • Start your microservice server with env $(cat .env | grep -v "#" | xargs ) iex -S mix phx.server

Up and running inside Docker

  • Build a docker image with docker build -t openfn/microservice:v0.3.0 .
  • Run with the docker run command

Project configuration

You can configure the jobs, triggers, credentials and language packs used in your microservice in the project.yaml config file.

First setup using the sample config

The sample project configuration file describes an example project setup to help you get acquainted with this structure.

By default microservice is configured with 4 sample jobs:

  1. job-1 is triggered when a matching message arrives to the inbox (see trigger-1).
  2. recurring-job is a timed job scheduled to run every minute and is linked to the every-minute cron trigger.
  3. flow-job and catch-job run after the success and failure of job-1, respectively.

All of the jobs are configured with the language pack openfn/language-common.

In the default sample configuration a new message posted to localhost:4000/inbox that matches trigger-1 (i.e. the message contains "number":2) is greeted with an asynchronous acknowledgement receipt (HTTP 202 Data accepted and processing has begun) and will trigger job-1 to run.

You can try this out with the following snippet:

curl -X POST -H "Content-Type: application/json" \
-d '{
"number":2,
"surveyId": 37479
}' \
http://localhost:4000/inbox

Posting a message not matching any of the triggers (e.g. “number”:3) equally prompts an acknowledgement but doesn’t trigger any jobs.

Example message post for this non-match scenario:

curl -X POST -H "Content-Type: application/json" \
-d '{
"number":3,
"surveyId": 37479
}' \
http://localhost:4000/inbox

HTTP post requests made to localhost:4000/inbox will be processed by the Receiver, according to the credential, expression, and adaptor defined in the project configuration YAML file.

Time-based jobs will be run by Engine according to the credential, expression, and adaptor defined in your project.yaml file.

Setup from your existing OpenFn platform project

If you have a project configured on OpenFn, you have two ways for exporting your config on the Project Settings page and running your project in microservice.

  1. If you export as project.yaml, you can download your settings in yaml format from your platform project Download page or from a link in the auto-generated email sent to your address. You can plug this file into your environment as set up using the Development Up and Running Guide.

  2. If you export as microservice.zip, you'll get your microservice folder ready to run with docker, containing

  • a docker-compose.yaml config file
  • a project folder containing project.yaml
  • .env file with the default environment variables for docker
  • a Readme file

cd into the folder and run the project with docker-compose up. If you don't have the docker image, it will be auto-pulled from hub.docker.com.

Export Microservice Zip