AlgoManager

The API used to perform the underlying AlgoRun container management of AlgoPiper

Prerequisites

• Docker

Download and Install

• Download this repository.
• Navigate to the downloaded folder.
• Run the script run.sh
• Go to http://localhost:8080 and make sure it is working

Configuring Production Environment

If you want to set AlgoManager on a shared server, edit algomanager/settings.py file. Change SERVER_PATH = 'http://localhost' to SERVER_PATH = 'http://server_IP'

Add Available Algorithms

Now, let AlgoManager be aware of what algorithms (AlgoRun containers) are available on your machine (or server).

• Run docker exec -it algomanager bash
• Run python manage.py createsuperuser. This will prompt you to create an admin user to manage the available algorithms on this algomanager instance.
• Now exit this bash using exit. Go to http://localhost:8080/admin/. Enter your newly created username and password. After you login, click on +Add, right beside Available Algorithms. Enter the name of the algorithm and its AlgoRun container. For example: Name=REACT and Docker Image=algorun/react:latest. Don't forget to docker pull those images from Docker Hub, before you make them available.

Congratulations! AlgoManager is fully functional on your machine/server.

API

If you are setting up AlgoManager to be used from AlgoPiper, no further steps are needed at this point.

If you plan to use AlgoManager for a different tool, you can read the API endpoints definitions below.

HTTP GET /api/v1/list

Returns a JSON object showing available images on the server.

    {"images":[{"name":"algo_name1","docker":"algo_docker_image_1"},
               {"name":"algo_name2","docker":"algo_docker_image_2"}]}

HTTP POST /api/v1/deploy

Parameters

• docker_image: the name of the docker image to run
• node_id: the node id of node-red that is requesting the deploy
• cpu_share: (optional) an integer value as defined by the official docker run command option: --cpu-shares
• memory_limit: (optional) a string value to limit the memory available for that container. Examples: '128m', '2g', '512k'

Results

If the docker run is successfull, it returns JSON object in the following format:

{
    'status': 'success',
    'endpoint': <hostname:portnumber>
}

If the docker run is unsuccessfull, it returns JSON object in the following format:

{
    'status': 'fail',
    'error_message': <the-stderr-returned-by-docker-daemon>
}