Handling single node failures using agents in computer clusters | IEEE Conference Publication | IEEE Xplore