README.md

Rush

A pipeline execution system named after Rush.

Defining a pipeline

Pipelines are a combination of sub-tasks referred to as "jobs". Each job (or sub-task) may have settings or parameters as well.

Pipelines are defined in JSON. Here is an example:

{
  "jobs": [
    {
      "name": "A readable name",
      "class_name": "fully.classified.class_name",
      "timeout_minutes": 10,
      "retry_gap_minutes": 5,
      "retry_limit": 2
    },
    {
      "name": "BlahJob",
      "class_name": "jobs.dummyJob.DummyJob",
      "timeout_minutes": 10,
      "retry_gap_minutes": 5,
      "retry_limit": 2
    }
  ]
}

Writing a job

A job (i.e. a sub-task of the pipeline) is expected to be a class and needs to have a callable run() method. It can inherit from jobs.job.Job (could that BE any more redundant?) to take advantage of configured job settings being passed in and the file logger.

Setting up to run the engine

The code has been written and tested using python3. Make sure the packages are installed.

Create a database with the tables from the rush-db directory. Only the master_token table needs seed data (just one row, actually).

Running the engine

Configure the PostgreSQL information and run the main.py file, with an optional "worker name" command line argument.

python main.py neil_peart