Spark context is created every time a batch job is executed in yarn

Question

I am wondering, is there any way where i create the spark-context once in the YARN cluster, then the incoming jobs will re-use that context. The context creation takes 20s or sometimes more in my cluster. I am using pyspark for scripting and livy to submit jobs.

Answer 1

No, you can't just have a standing SparkContext running in Yarn. Maybe another idea is to run in client mode, where the client has it's own SparkContext (this is the method used by tools like Apache Zeppelin and the spark-shell).