Your data processing job is experiencing data spill, where intermediate data exceeds the memory capacity of the nodes in the cluster. What steps would you take to handle this situation and optimize resource management?