Workflow controller

Architecture

This is a mini framework to manage workflows asynchronous.

A workflow is a sequence of one or more ordered tasks. It consist of four components:

• A controller for workflow management
• Server message queue
• Client message queue agents
• Hotfolders for workflow generation

Infrastructure

The entire setup consists of:

• ActiveMQ server
• Java ActiveMQ client (Camel)
• Nodejs for the workflow controller
• MongoDB for storing temporary workflow events

Quick start

Run the docker compose file:

docker-compose up

Define workflows

A workflow specifies its tasks and hotfolders and are defined config.json file.

For example:

workflows": {
   "helloworld": {
     "mail_archivist_on_success": "archivist@localhost",
     "tasks": [
       {
         "description": "Say 'hello world 1'",
         "queue": "hello_world_1"
       },
       {
         "description": "Say 'hello world 2'",
         "queue": "hello_world_2"
       },
      "events": [
        "/data/helloworld"
      ]

This workflow is called 'helloworld'.

It has two tasks, each associated with a message queue or topic: 'hello_world_1' and 'hello_world_2'

And monitors a hotfolder: '/data/helloworld'

Agent configuration

For each queue in the config.json@workflows.name.tasks.queue element, there is a corresponding bash queues/queuename/startup.sh script.

Agent on github

Run the controller

First startup the surrounding stack with docker-compose:

docker-compose up

Then run the idea:

nodejs bin/wwwworkflow/config/controller/config.json

with the environmental variable pointing to the configuration file:

CONFIG_FILE=../workflow/config/controller/config.json