Hello everyone, Spark has a dynamic resource allocation scheme, where, when available Spark manager will automatically add executors to the application resource.
Spark's default configuration is for executors to allocate the entire worker node they are running on, but this is configurable, my question is, if an executor is set to use half of the worker node. Is it possible that Spark will spawn two executors which belong to the same application on the same worker node? Thanks, Anton.