zkan / data-pipelines-with-airflow

Skooldio: Data Pipelines with Airflow

Data Pipelines with Airflow

Contents

• Prerequisites
  • Installing Docker Desktop
  • Installing Visual Studio Code
• Data Source
• Starting Airflow
• Project Instruction
• Airflow S3 Connection to MinIO
• Running Tests
• References

Data Source

• CCXT - CryptoCurrency eXchange Trading Library
• Hello, CCXT!

Starting Airflow

Before we run Airflow, let's create these folders below first. Please note that if you're using Windows, you can skip this step.

mkdir -p mnt/dags mnt/logs mnt/plugins mnt/tests

On Linux, please make sure to configure the Airflow user for the docker-compose:

echo -e "AIRFLOW_UID=$(id -u)" > .env

With LocalExecutor

docker-compose build
docker-compose up

With CeleryExecutor

docker-compose -f docker-compose-celery.yml build
docker-compose -f docker-compose-celery.yml up

With SequentialExecutor (NOT recommended for production use)

docker-compose -f docker-compose-sequential.yml build
docker-compose -f docker-compose-sequential.yml up

To clean up the project, press Ctrl+C then run:

docker-compose down

Airflow Connection to MinIO

Since MinIO offers S3 compatible object storage, we can set the connection type to "Amazon Web Services". However, we'll need to set an extra option, so that Airflow connects to MinIO instead of S3.

• Connection Name: minio or any name you like
• Connection Type: Amazon Web Services
• AWS Access Key ID: <replace_here_with_your_minio_access_key>
• AWS Secret Access Key: <replace_here_with_your_minio_secret_key>
• Extra: a JSON object with the following properties:

  {
    "host": "http://minio:9000"
  }

See the example below:

Note: If you were using AWS S3 already, you don't need to specify the host in the extra.

Running Tests

First we need to install pytest:

pip install pytest

Run tests:

export PYTHONPATH=/opt/airflow/plugins
pytest

References

• Running Airflow in Docker
• MinIO Docker Quickstart Guide
• Deploy MinIO on Docker Compose