finally sorted after 5 hours then in a remarkable coincidence the other cluster member has now developed the same fault. Is it likely that the new firmware I put on is less tolerant of memory erros? At least my HW engineer stayed onsite he is a good bloke kiwi