Description of problem: Presently there is no mechanism by which nodes which are being scaled-down / deleted can run any automated clean up; they simply get shutoff and removed from the pool. Examples of house-keeping tasks that may be desired include: Unsubscribing themselves from subscription manager, thus freeing resources. Cleaning up service entries for compute nodes. There is a desire for these functions and as such it would be helpful for work to be done to allow such hooks.