How does Spark running on YARN account for Python memory usage?

Question

After reading through the documentation I do not understand how does Spark running on YARN account for Python memory consumption.

Does it count towards spark.executor.memory, spark.executor.memoryOverhead or where?

In particular I have a PySpark application with spark.executor.memory=25G, spark.executor.cores=4 and I encounter frequent Container killed by YARN for exceeding memory limits. errors when running a map on an RDD. It operates on a fairly large amount of complex Python objects so it is expected to take up some non-trivial amount of memory but not 25GB. How should I configure the different memory variables for use with heavy Python code?

Answer 1

I would suggest you to try to increase memory to spark.python.worker.memory default (512m) because of heavy Python code and this property value does not count in spark.executor.memory.

The amount of memory to use per python worker process during aggregation, in the same format as JVM memory strings (e.g. 512m, 2g). If in any case the memory used during aggregation goes above this amount, it is going to spill the data into disks. Check this link.

ExecutorMemoryOverhead calculation in Spark:

MEMORY_OVERHEAD_FRACTION = 0.10 

MEMORY_OVERHEAD_MINIMUM = 384 

val executorMemoryOverhead = 

  max(MEMORY_OVERHEAD_FRACTION * ${spark.executor.memory}, MEMORY_OVERHEAD_MINIMUM))

The property is spark.{yarn|mesos}.executor.memoryOverhead for YARN and Mesos.

YARN automatically kills the processes which takes more memory than they requested which is the sum of executorMemoryOverhead and executorMemory.

In the image given below python processes in worker uses spark.python.worker.memory, then spark.yarn.executor.memoryOverhead + spark.executor.memory is specific JVM.