wlongxiang/pyspark_docker

Run pyspark cluster with docker on your local laptop

Pyspark docker environment

If you simply want to test a single node spark cluster, just run docker run -it wlongxiang/pyspark-2.4.4 pyspark, this will bring you yo spark-shell.

Or you can also create a muti-worker cluster with a simple compose file, the example will mount the data and code local dir to the cluster worker and master such that you can easily change your code and data locally and test it within docker cluster.

• Checkout the master branch
• Run docker-compose up to spin up 2 workers and 1 master (or you can also set COMPOSE_FILE env to a diff. compose file)
• Go to http://localhost:8080 to see the spark master UI
• Run docker exec -it pyspark_docker_master_1 bash to shell into the spark container
• word count: spark-submit /code/wordcount.py /data/logs.txt

Supported spark versions: 2.4.4.

Some examples to test out, see code directory.

How to publish a new image (manually)

• first you do docker login with your credentials to docker-hub
• then docker build -t wlongxiang/pyspark-2.4.4:<version_tag> ., this command will build your email with name pyspark and tag with 2.4.4
• verify your image is built successfully by docker images
• finally docker push wlongxiang/pyspark-2.4.4:<version_tag>, in the end this will be available in your docker repo
• now, everything should be able to run your image or use it in the docker-compose file, such as docker run -it pyspark-2.4.4:<version_tag> bash