Multi-partition computer system, failure handling method and program therefor
The present invention determines an incorrect packet from a faulty partition quickly and reliably and prevents the packet from flowing into normal partitions through simple control actions. The multi-partition computer system is a multi-partitioned computer system in which a plurality of nodes are logically divided into a plurality of partition, and each node contained in the partitions includes a packet identification unit which, upon receiving a packet, compares the partition identification information uniquely assigned to own partition against the partition identification information contained in the receive packet, and if these pieces of information do not match each other, judges and discards the receive packet as an incorrect packet.
1. A multi-partitioned computer system in which a plurality of nodes are logically divided into a plurality of partitions, wherein a node contained in one of said partitions includes
a packet identification unit which, upon receiving a packet, compares the partition identification information uniquely assigned to its own partition against the partition identification information contained in the receive packet, and if these pieces of information do not match each other, judges and discards said receive packet as an incorrect packet,
wherein a crossbar which connects between said nodes is provided; and
said crossbar has an error detection unit which detects an error in a packet received,
a packet identification unit which extracts said partition identification information in the receive packet for which the error has been detected, and
an error notification unit which notifies the error to the partition having said partition identification information.
2. The multi-partition computer system according to claim 1 , wherein said node has a packet transmission unit which generates and transmits a packet containing partition identification information for identifying own partition.
3. The multi-partition computer system according to claim 1 , wherein said node has a partition identification information registration unit which stores partition identification information uniquely assigned to its own partition; and
said packet identification unit extracts said partition identification information contained in said receive packet, and
compares said partition identification information extracted against said partition identification information registered in said partition identification information registration unit.
4. The multi-partition computer system according to claim 3 , wherein said packet transmission unit generates said transmit packet by adding said partition identification information registered in said partition identification information registration unit.
5. The multi-partition computer system according to claim 1 , wherein said node has an error notification unit which, if said packet identification unit has discarded said receive packet, notifies an error to another partition having said partition identification information contained in said receive packet.
6. A failure handling method for a multi-partitioned computer system in which a plurality of nodes are logically divided into a plurality of partitions, comprising:
a node contained in one of said partitions performing steps of:
upon receiving a packet, comparing the partition identification information uniquely assigned to its own partition against the partition identification information contained in the receive packet; and
if these pieces of information do not match each other, judging and discarding said receive packet as an incorrect packet,
wherein a crossbar which connects between said nodes performs the processes of:
detecting an error in a packet received;
extracting said partition identification information in the receive packet for which the error has been detected; and
notifying the error to the partition having said partition identification information extracted.
7. The failure handling method according to claim 6 , wherein said node generates and transmits a packet which contains partition identification information for identifying own partition.
8. The failure handling method according to claim 6 , wherein said node
extracts said partition identification information contained in said receive packet, and
compares said partition identification information extracted against said partition identification information registered in the partition identification information registration unit which stores a partition identification information uniquely assigned to its own partition.
9. The failure handling method according to claim 8 , wherein said transmit packet is generated by adding said partition identification information registered in said partition identification information registration unit.
10. The failure handling method according to claim 6 , wherein said node, if said node has discarded said receive packet, notifies an error to another partition having said partition identification information contained in said packet.
11. A non-transitory computer readable medium storing a failure handling program executed on said node and a crossbar which connects between said nodes in a multi-partitioned computer system in which a plurality of nodes are logically divided into a plurality of partitions, wherein said failure handling program causes a node contained in one of said partitions, upon receiving a packet, to execute a processing of comparing the partition identification information uniquely assigned to its own partition against the partition identification information contained in the receive packet, and if these pieces of information do not match each other, judging and discarding said receive packet as an incorrect packet,
wherein said failure handling program causes said crossbar to execute a processing of:
detecting an error in a packet received;
extracting said partition identification information in the receive packet for which the error has been detected; and
notifying the error to the partition having said partition identification information extracted.
12. The non-transitory computer readable medium according to claim 11 , wherein said failure handling program causes said node to execute a processing of generating and transmitting a packet which contains partition identification information for identifying own partition.
13. The non-transitory computer readable medium according to claim 11 , wherein said failure handling program causes said node to execute a processing of:
extracting said partition identification information contained in said receive packet; and
comparing said partition identification information extracted against said partition identification information registered in said partition identification information registration unit which stores a partition identification information uniquely assigned to its own partition.
14. The non-transitory computer readable medium according to claim 13 , wherein said failure handling program causes said node to execute a processing of generating said transmit packet by adding said partition identification information registered in said partition identification information registration unit.
15. The non-transitory computer readable medium according to claim 11 , wherein said failure handling program causes said node to execute a processing of, if said node has discarded said receive packet, notifying an error to another partition having said partition identification information contained in said packet.