scala - Concurrent tasks on a Spark executor -
what determines how many tasks can run concurrently on spark executor? maybe kind of thread pool , shared memory resources?
what parameters control behavior?
does mean code used in executors should written thread-safe?
what determines how many tasks can run concurrently on spark executor?
spark maps number tasks on particular executor number of cores allocated it. default, spark assigns 1 core task controlled spark.task.cpus
parameter defaults 1.
does mean code used in executors should written thread-safe?
no. working rdds or dataframe/set aimed work locally inside transform, without sharing global resources. should think thread-safety when have global resource execute in parallel inside single executor process, can happen when multiple tasks executed on same executor.
Comments
Post a Comment