I'm using spark-0.9 with YARN. Q: Why spark.cleaner.ttl setting could remove broadcast that still in use? I think cleaner should not remove broadcasts still in the dependences of some RDDs. It makes the value of spark.cleaner.ttl need to be set more carefully.
POINT: cleaner should not crash the task. -- View this message in context: http://apache-spark-user-list.1001560.n3.nabble.com/Suggestion-Strange-behavior-for-broadcast-cleaning-with-spark-0-9-tp5512.html Sent from the Apache Spark User List mailing list archive at Nabble.com.
