If a node fails, currently its cleanup method does not execute. Configure (Pipeline? Target? Node?) to specify whether any cleanup should still be performed, even after an error. Noted by Kostas.
