python

Spark Python template

The Spark Python template image serves as a base image to build your own Python application to run on a Spark cluster. See big-data-europe/docker-spark README for a description how to setup a Spark cluster.

Package your application using pip

You can build and launch your Python application on a Spark cluster by extending this image with your sources. The template uses pip to manage the dependencies of your project, so make sure you have a requirements.txt file in the root of your application specifying all the dependencies.

Extending the Spark Python template with your application

Steps to extend the Spark Python template

1. Create a Dockerfile in the root folder of your project (which also contains a requirements.txt)
2. Extend the Spark Python template Docker image
3. Configure the following environment variables (unless the default value satisfies):

• SPARK_MASTER_NAME (default: spark-master)
• SPARK_MASTER_PORT (default: 7077)
• SPARK_APPLICATION_PYTHON_LOCATION (default: /app/app.py)
• SPARK_APPLICATION_ARGS

4. Build and run the image

docker build --rm -t bde/spark-app .
docker run --name my-spark-app -e ENABLE_INIT_DAEMON=false --link spark-master:spark-master --net dockerspark_default -d bde/spark-app

The sources in the project folder will be automatically added to /app if you directly extend the Spark Python template image. Otherwise you will have to add the sources by yourself in your Dockerfile with the command:

COPY . /app

If you overwrite the template's CMD in your Dockerfile, make sure to execute the /template.sh script at the end.

Example Dockerfile

FROM bde2020/spark-python-template:3.3.0-hadoop3.3

MAINTAINER You <you@example.org>

ENV SPARK_APPLICATION_PYTHON_LOCATION /app/entrypoint.py
ENV SPARK_APPLICATION_ARGS "foo bar baz"

Example application

See big-data-europe/docker-spark/examples/python.