- Travis CI
- CWL conformance tests

cwl-airflow

Python package to extend Apache-Airflow 1.9.0 functionality with CWL v1.0 support.

Try it out

1. Install cwl-airflow

   $ pip3 install cwl-airflow --user --find-links https://michael-kotliar.github.io/cwl-airflow-wheels/
   

2. Init configuration

   $ cwl-airflow init
   

3. Run demo

   $ cwl-airflow demo --auto
   

4. When you see in the console output that Airflow Webserver is started, open the provided link

Installation requirements

• Ubuntu 16.04.4
• python 3.5.2
• pip3

    wget https://bootstrap.pypa.io/get-pip.py
    python3 get-pip.py --user
  

• setuptools

  pip3 install setuptools --user
  

• docker

  sudo apt-get update
  sudo apt-get install apt-transport-https ca-certificates curl software-properties-common
  curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo apt-key add -
  sudo add-apt-repository "deb [arch=amd64] https://download.docker.com/linux/ubuntu $(lsb_release -cs) stable"
  sudo apt-get update
  sudo apt-get install docker-ce
  sudo groupadd docker
  sudo usermod -aG docker $USER
  

  Log out and log back in so that your group membership is re-evaluated.
• python3-dev

  sudo apt-get install python3-dev
  

Configuration

When running cwl-airflow init the following parameters can be specified:

• -l LIMIT, --limit LIMIT sets the number of processed jobs kept in history. Default 10 for each of the category: Running, Success, Failed
• -j JOBS, --jobs JOBS sets the path to the folder where all the new jobs will be added. Default ~/airflow/jobs
• -t DAG_TIMEOUT, --timeout DAG_TIMEOUT sets timeout (in seconds) for importing all the DAGs from the DAG folder. Default 30 seconds
• -r WEB_INTERVAL, --refresh WEB_INTERVAL sets the webserver workers refresh interval (in seconds). Default 30 seconds
• -w WEB_WORKERS, --workers WEB_WORKERS sets the number of webserver workers to be refreshed at the same time. Default 1
• -p THREADS, --threads THREADS sets the number of threads for Airflow Scheduler. Default 2

If core/dags_folder parameters from Airflow configuration file (default location ~/airflow/airflow.cfg) has been updated manualy, make sure to rerun cwl-airflow init

Running

Batch mode

To automatically monitor and process all the job files present in a specific folder

1. Make sure your job files include the following mandatory fields:

   • uid - unique ID, string
   • output_folder - absolute path the the folder to save result, string
   • workflow - absolute path the the workflow to be run, string

   Aditionally, job files may also include the tmp_folder parameter to point to the temporary folder absolute path.

2. Put your JSON/YAML job files into the directory set as jobs in cwl section of airflow.cfg file (by default ~/airflow/cwl/jobs)

3. Run Airflow scheduler:

   $ airflow scheduler
   

Manual mode

To perform a single run of the specific CWL workflow and job files

cwl-airflow run WORKFLOW_FILE JOB_FILE

If uid, output_folder, workflow and tmp_folder fields are not present in the job file, you may set the them with the following arguments:

  -o, --outdir      Output directory, default current directory
  -t, --tmp         Folder to store temporary data, default /tmp
  -u, --uid         Unique ID, default random uuid

Demo mode

1. Get the list of the available demo workflows to run

   $ cwl-airflow demo
   

2. Run demo workflow from the list (if running on macOS, consider adding the directory where you installed cwl-airflow package to the Docker / Preferences / File sharing options)

   $ cwl-airflow demo super-enhancer.cwl
   

3. Optionally, run airflow webserver to check workflow status (default webserver link)

   $ airflow webserver