METHOD AND A SYSTEM FOR RELIABLY RESOLVING A FAILURE IN A CLUSTER, PARTICULARLY FOR SUITABLY OVERCOMING A FAILURE BY CHECKING A STARTING POINT OF THE FAILURE BY DETECTING AND SEPARATING THE FAILURE

摘要

PURPOSE: A method and system for reliably resolving a failure in a cluster are provided to reliably and effectively detect a failure from a highly used integrated architecture and solve the failure. CONSTITUTION: A heartbeat message is transmitted to a peer node to monitor detection of a failure(202). It is determined whether there is a loss of a heartbeat in one network interface(204). If there is a loss of a heartbeat, a node which has detected the loss of the heartbeat issues an ICMP(Internet Control Message Protocol) echo(206). It is checked whether at least one echo return has been received with respect to one network interface(208). If at least one echo return has been received, it means that a network path is being properly functioned(210). Echo responses received from a target node set with respect to network interfaces are compared to determine a network path with an optimum connectivity according to one network interface in a cluster(212). It is checked whether connectivity of a different network path has been improved(214). If connectivity of the different network path has been improved, a network path failure is overcome(216).