I have an EMR cluster with 1 master node (16GB RAM, 4 vCPUs), 1 core node (16GB RAM, 4 vCPUs) and 4 task nodes (64GB RAM, 16 vCPUs). And I have this issue when running a ste