Quick post mostly for my own reference since I always need to re-learn how to do this.

This used to be more difficult in older versions of Spark, but when using Spark 2.4 or later, all you have to do is:

wget https://repo1.maven.org/maven2/com/amazonaws/aws-java-sdk/1.7.4/aws-java-sdk-1.7.4.jar -P $SPARK_HOME/jars/
wget https://repo1.maven.org/maven2/org/apache/hadoop/hadoop-aws/2.7.3/hadoop-aws-2.7.3.jar -P $SPARK_HOME/jars/

That’s all there is to it. The s3a:// prefix should work now for reading and writing data using Spark.