摘要 |
<p><P>PROBLEM TO BE SOLVED: To provide a method and a system to find the location of a fault, and to resolve it in a cluster environment. <P>SOLUTION: The cluster (100) consists of at least one multi-homed node (110), at least one gateway (140) for each of network interfaces (112, 114). Heartbeat messages are sent between a peer-node and the gateway at a predetermined periodic interval. If there is a loss of the heartbeat messages by any node or gateway occurs, an ICMP echo is issued to each node and gateway in the cluster on a network interface basis. When the ICMP echo is responded to and a node loss or a network loss is not identified, an application-level ping is issued, to determine whether the fault related to the heartbeat message missing is a transient error condition or an application software fault. <P>COPYRIGHT: (C)2005,JPO&NCIPI</p> |