MDStudio workflow

MDStudio workflow is a Python based workflow engine developed for the MDStudio microservice environment. MDtudio workfow makes it easy to construct complex workflows as DAGs (Directed Acyclic Graphs) combining calls to MDStudio microservices with Python scripts or even commandline tools. Examples on how to construct and use these workflows is provided in the MDStudio_examples repository on GitHub.

Installation Quickstart

MDStudio workflow can be used in the MDStudio environment as (Docker) service or as stand-alone Python library.

Install option 1 Pre-compiled Docker container

MDStudio propka can be installed quickly from a pre-compiled docker image hosted on DockerHub by:

docker pull mdstudio/mdstudio_workflow
docker run (-d) mdstudio/mdstudio_workflow

In this mode you will first need to launch the MDStudio environment itself in order for the MDStudio workflow service to connect to it. You can unify this behaviour by adding the MDStudio workflow service to the MDStudio service environment as:

MDStudio/docker-compose.yml:
    
    services:
       mdstudio_workflow:
          image: mdstudio/mdstudio_workflow
          links:
            - crossbar
          environment:
            - CROSSBAR_HOST=crossbar
          volumes:
            - ${WORKDIR}/mdstudio_workflow:/tmp/mdstudio/mdstudio_workflow

And optionally add mdstudio_workflow to MDStudio/core/auth/settings.dev.yml for automatic authentication and authorization at startup.

Install option 2 custom build Docker container

You can custom build the MDStudio workflow Docker container by cloning the MDStudio_workflow GitHub repository and run:

docker build MDStudio_workflow/ -t mdstudio/mdstudio_workflow

After successful build of the container follow the steps starting from docker run in install option 1.

Install option 3 standalone deployment of the service

If you prefer a custom installation over a (pre-)build docker container you can clone the MDStudio_workflow GitHub repository and install mdstudio_workflow locally as:

pip install (-e) mdstudio_workflow/

Followed by:

./entry_point_mdstudio_workflow.sh

or

export MD_CONFIG_ENVIRONMENTS=dev,docker
python -u -m mdstudio_workflow

Install option 4 use mdstudio_workflow as Python library

Install the project as standalone library as described in option 3. Build workflows as Python scripts explained in the MDStudio_examples repository on GitHub.