README

to start processing one needs to:

    redis: this is basically the head node (steers the messaging between headcli and workers). runtime: unlimited         
    headcli: this is what knows the DAG and submits tasks to workers. runtime: length of DAG processing
    workers: these are celery worker that execute the tasks the headcli submits

know for docker swarm one would need to have a 3 step process:

1) start redis and worker containers (i.e. know we have messaging setup and workers waiting for jobs)
    docker-compose scale redis=1 worker=X # note: order is important.. workers want to link to redis, so it needs to exist before
2) start head container (the run cmd here will submit actual jobs
    docker-compose run --rm head recastworkflow-headnode /bla
3) tear down redis, workers after head exited.
    docker-compose stop && docker-compose rm