Failure recovery system and failure recovery method in loosely coupled multi-computer system, and medium for storing failure recovery program

摘要

In a loosely coupled multi-computer system, even when a failure happens to occur in any one of the loosely coupled computers, users can continue their works without having adverse influences caused by this failure. When a process operation for transaction used to update the shared data is commenced, a first computer registers such a message that it is using this shared data into an exclusive access device. After updating process information corresponding to the shared data before being updated has been outputted to the updated record holder, the first computer updates this shared data, and deletes the updating process information of the updated record holder when the transaction is ended. Either a second computer or a third computer, which is notified by the failure notifier that a failure has occurred in the first computer before the transaction is ended, registers such a message that the recovering computer uses this shared data into the exclusive access device. Then, the second or third computer returns the updating process information about "no recovery is required" to its original updating process information, among the updating process information stored in the updated record holder, thereby recovering this shared data. As a consequence, this updating process information is brought into such updating process information about "no recovery is required".