Skip to content

Latest commit

 

History

History
94 lines (77 loc) · 3.24 KB

README.md

File metadata and controls

94 lines (77 loc) · 3.24 KB
Raw

Apache Airflow Pretend ETL - Fully Dockerized

This repo contains a pretend ETL running with Apache Airflow

The repo provides the following functionality:

  1. The top level of this repo will create a docker application of two containers using Docker-compose.
    The first runs Apache Airflow, and the second runs PostGres

  2. The ETLs folder contains a self contained Python ETL application that performs an ETL on some sample data. It contains a DockerFile for building this application as a Docker image

  3. When running Airflow in the docker application mentioned above, Airflow will run the ETL application as a Docker Image using the DockerOperator. The ETL will continue to run on the schedule set in the ieee.py dag
    You can monitor it on the Airflow UI at localhost:8080.
    Be sure to either turn the schedule on, or run it manually

Requirements:

  1. Install Docker if not already installed: https://www.docker.com/

Installation:

  1. Pull down this repo with git clone
  2. In the ETLs directory create a folder called data, and place the sample data in it. I didn't include it because it's too large
  3. In the top level directory create another folder called postgres-data, this is where post-gres will maintain state when the container is not running
  4. Build the swarm containing Airflow and Postgres: 
    ./util.sh airflow-build
  5. Build the ETL application: 
    ./util.sh etl-build-image
  6. Start the airflow and postgres application: 
    ./util.sh airflow-start
  7. Create the tables in postgres using a pretend migration tool: 
    ./util.sh run-migration
  8. View the Airflow UI at: http://localhost:8080/

You can also test the jobs manually using just the ETL application container with: './util.sh etl-run'

Note: The airflow container mounts to your docker socket at /var/run/docker.sock.

Stopping the application

Important To stop the application run

./util-sh airflow-stop

This runs docker-compose down. It's important to stop the application with this command as it will remove the created containers. If you do not do this when stopping the containers then the airflow process may not be properly terminated.
When you running docker-compose up again it will simply restart the existing container, which my cause issues with already running processes.

Development

Using Pycharm PE

The best way to do local development on this repo is with PyCharm Professional Edition using the Docker image is the interpreter.

  1. Run build-image if not already done: ./util.sh etl-build-image
  2. Point Pycharm PE interpreter at the image:
Pycharm -> Preferences -> Project -> Project Interpreter -> Click Drown Down -> Show All -> "+" 
-> Docker -> Etl-application

Testing Airflow DAGS and Tasks

To test Airflow Dags and Tasks you can enter into a container with app already running

# Get the container ID
docker ps

# Connect to it:
docker exec -it [container id] bash

Run various tests

# Test for errors in DAG
python dags/tutorial.py 

# Test run a specific task
airflow test tutorial print_date 2015-06-01

# Run a DAG for a range of dates
airflow backfill tutorial -s 2015-06-01 -e 2015-06-07