We are trying to move from using spark-submit to a programmatic solution using SparkLauncher (Spark 2.4 running on Cloudera Hadoop), and have encountered this problem:
