Research Article
A Novel Low-Overhead Recovery Approach for Distributed Systems
Algorithm 1
Nonblocking Algorithm.
At each process | βif receives | β ββββtakes checkpoint ; | β βsends its and to the initiator process ; | β ββ β// all such vectors from each are used by to determine the lost messages | β ββ ββββsent by the processes during () in the event of a failure continues its normal operation; | βββelse if receives a piggybacked application message m, $ && has not yet received | βββ β βfor the current execution of the checkpointing algorithm | β ββββββtakes checkpoint without waiting for ; | β ββββ βsends its and to the initiator process ; | β ββ β// all such vectors from each are used by to determine the lost messages | β ββ ββββsent by the processes during () in the event of a failure continues its normal operation; | β ββ β// processes the received message and ignores , when received later |
|