| 摘要 |
PROBLEM TO BE SOLVED: To avoid an event, in which failure of a server having a queue is not detectable by failure of a server detecting a failure of server in a cluster-structured queue system, and takeover of messages left in a queue cannot be performed. SOLUTION: The cluster-structured queue system in which a plurality of servers share queues, a message management program spontaneously determines, upon detection of a failure of another server, whether to take over messages managed by the failure server or not. The message management program which determines the takeover takes over the messages managed by the failure server by updating "processed" messages of messages within the queue managed by the failure server to its own ID and copying copies of the messages to its message cache area. COPYRIGHT: (C)2007,JPO&INPIT
|
