User Process Checkpoint/Restart. Revision 1.1

摘要

This document describes the design of the Tera user process checkpoint facility.Checkpoint is a means for saving the state of an executing process or group of processes and restarting them later on demand. The motivation for providing checkpoint on the Tera is to allow long running computationally intensive jobs to be restarted after scheduled system down-time or unanticipated system faults.