COMMUNICATION FAILURE SOURCE ISOLATION IN A DISTRIBUTED COMPUTING SYSTEM

摘要

In accordance with one aspect of the present description, an indication that a communication failure reported in a predetermined time interval is more likely the result of a software failure than a hardware failure may be made if the number of communication links reporting a communication failure in the predetermined time interval exceeds a communication link failure threshold, and the number of communication link devices such as nodes or communication paths which have been implicated as causing a communication failure, exceeds an implicated device threshold. Other features and aspects may be realized, depending upon the particular application.

主权项

1. A method, comprising:
receiving within a predetermined time interval from at least one communication link at least one report of at least one communication failure in a distributed computing system having a plurality of communication links, each communication link comprising a pair of nodes and a communication path linking the pair of nodes for communication between the nodes of the pair; determining how many communication links have reported communication failure within the predetermined time interval; determining how many devices of the communication links reporting communication failure within the predetermined time interval are implicated as causing a communication failure within the predetermined time interval; and indicating that communication failure reported in the predetermined time interval is more likely the result of a software failure than a hardware failure if the number of communication links reporting a communication failure in the predetermined time interval exceeds a communication link failure threshold and the number of devices implicated as causing a communication failure exceeds an implicated device threshold.