| 摘要 |
In a cluster computer which connects a plurality of nodes that include memory and processors through an interconnection network, shutting down of a node due to an irrecoverable error that occurs in a common communication area is prevented, and the availability of the cluster computer is increased.A system control apparatus in each node sends a system error stop notification to the memory access origin when an irrecoverable error occurred during a memory access request generated in one node is sent to the node proper memory located on the same node step (S 17). However, a common communication area error notification is sent to the memory access origin (step S 10 and step S 18), when an irrecoverable error occurs when a memory access request generated in one node is sent to the common communication area of the memory located on another node and when a memory access request generated on one node is sent to the common communication area of the memory of the same node, and the node shut down is prevented.
|
