| 摘要 |
<P>PROBLEM TO BE SOLVED: To properly decide an operation state of each node that is a monitoring target in a cluster to prevent performance deterioration of the cluster. <P>SOLUTION: A failure monitored node (200) executes a process for transmitting a periodic signal to a failure monitoring node (100) at a prescribed time interval. The failure monitoring node decides whether receiving the periodic signal from the failure monitored node or not, and once cuts off it from a group by stopping input of requirement from an external device to the failure monitored node when not receiving the periodic signal. After the cut-off, detailed information showing an operation situation of the failure monitored node is acquired from the node, and it is decided whether the operation situation of the failure monitored node is overload generation or failure generation on the basis of the detailed information. As a result, when deciding that the operation situation is the overload generation, the input stop to the failure monitored node is released, so that the failure monitored node is returned to the group. <P>COPYRIGHT: (C)2006,JPO&NCIPI |
