Hello there! I wonder why Spark Streaming keeps all processed batches in memory? It leads to getting out of memory on executors but I really don't need them after processing. Can it be configured somewhere so that batches are not kept in memory after processing?
Respectfully, Artem Moskvin
