Do I Need to Remove a Broadcasted Spark Variable

In Apache Spark, you don’t need to explicitly remove a broadcasted variable. Spark automatically handles the lifecycle of broadcast variables, and they will be cleaned up by the garbage collector when no longer needed.

Spark has a ContextCleaner which runs on periodic interval to delete broadcast variable if it is not being used. It does that using Weak References (link is about weak references in Python, but the same applies to Java). Reference: https://stackoverflow.com/questions/54731255/spark-broadcast-variables-life-time