b15 Did you find a solution to this. I am facing exactly the same issue while running a spark job in GCP dataproc.