I'd like that, during a distributed IrrMap computation or during distributed render, if one or more servers fail because they are doing some other job (for example rendering something else), as they are able again they would be added to job... it would be very cool!!!!!!