Takeover method of process, cluster construction program and cluster construction apparatus

1. A method for taking over a process in a processing system, the processing system including a plurality of node groups, each node group including one or more of nodes, the method comprising:

determining, by a processor of a computer configured to belong to a first group among the plurality of node groups, whether a communication failure with any other of the plurality of node groups is detected;

starting a time count in response to a determination result indicating that the communication failure is detected; and

starting a takeover process in response to detecting that a handicap time has elapsed in the time count, the takeover process including performing a fencing process to the nodes included in the any other of the plurality of node groups, the handicap time having a time length adjusted in accordance with a total number of nodes included in the first group.

2. The method according to claim 1 ,

wherein the handicap time of the first group is less than the handicap time of the any other of the plurality of node groups when the first group includes nodes exceeding one half of a total number of nodes included in the processing system.

3. The method according to claim 1 ,

wherein the handicap time of the first group is less than the handicap time of a second group when the first group includes an operation node, the second group being a node group that includes nodes that does not exceed one half of a total number of nodes included in the processing system and that does not include an operation node.

4. The method according to claim 3 , wherein

the handicap time of a second group is the minimum one of handicap times allocated to nodes in the second group, the nodes in the second groups being nodes to each of which the handicap time longer than the handicap time allocated to the operation node is allocated in accordance with a performance.

5. The method according to claim 1 , wherein

the operation node is configured to update a database;

a standby node from among the nodes includes a duplicate of the database, the standby node being configured to receive an update difference of the database from the operation node to update an own database, the standby node being a node other than the operation node; and

the handicap time is adjusted based on a writing performance into a file for temporarily accumulating the update difference.

6. An apparatus for taking over a process in a processing system, the processing system including a plurality of node groups, each node group including one or more of nodes, the apparatus being configured to belong to a first group among the plurality of node groups, the apparatus comprising:

a memory; and

a processor coupled to the memory and configured to

determine whether a communication failure with any other of the plurality of node groups is detected,

start a time count in response to a determination result indicating that the communication failure is detected, and

start a takeover process in response to detecting that a handicap time has elapsed in the time count, the takeover process including performing a fencing process to the nodes included in the any other of the plurality of node groups, the handicap time having a time length adjusted in accordance with a total number of nodes included in the first group.

7. The apparatus according to claim 6 ,

wherein the handicap time of the first group is less than the handicap time of the any other of the plurality of node groups when the first group includes nodes exceeding one half of a total number of nodes included in the processing system.

8. The apparatus according to claim 6 ,

wherein the handicap time of the first group is less than the handicap time of a second group when the first group includes an operation node, the second group being a node group that includes nodes that does not exceed one half of a total number of nodes included in the processing system and that does not include an operation node.

9. The apparatus according to claim 8 , wherein

the handicap time of a second group is the minimum one of handicap times allocated to nodes in the second group, the nodes in the second groups being nodes to each of which the handicap time longer than the handicap time allocated to the operation node is allocated in accordance with a performance.

10. The apparatus according to claim 6 , wherein

the operation node is configured to update a database;

a standby node from among the nodes includes a duplicate of the database, the standby node being configured to receive an update difference of the database from the operation node to update an own database, the standby node being a node other than the operation node; and

the handicap time is adjusted based on a writing performance into a file for temporarily accumulating the update difference.

11. A non-transitory computer-readable storage medium for storing a program which causes a processor to perform processing for taking over a process in a processing system, the processing system including a plurality of node groups, each node group including one or more of nodes, the processor being included in a computer configured to belong to a first group among the plurality of node groups, the processing comprising:

determining whether a communication failure with any other of the plurality of node groups is detected;

starting a time count in response to a determination result indicating that the communication failure is detected;

starting a takeover process in response to detecting that a handicap time has elapsed in the time count, the takeover process including performing a fencing process to the nodes included in the any other of the plurality of node groups, the handicap time having a time length adjusted in accordance with a total number of nodes included in the first group.

12. The non-transitory computer-readable storage medium according to claim 11 ,

wherein the handicap time of the first group is less than the handicap time of the any other of the plurality of node groups when the first group includes nodes exceeding one half of a total number of nodes included in the processing system.

13. The non-transitory computer-readable storage medium according to claim 11 ,

wherein the handicap time of the first group is less than the handicap time of a second group when the first group includes an operation node, the second group being a node group that includes nodes that does not exceed one half of a total number of nodes included in the processing system and that does not include an operation node.

14. The non-transitory computer-readable storage medium according to claim 13 , wherein

the handicap time of a second group is the minimum one of handicap times allocated to nodes in the second group, the nodes in the second groups being nodes to each of which the handicap time longer than the handicap time allocated to the operation node is allocated in accordance with a performance.

15. The non-transitory computer-readable storage medium according to claim 11 , wherein

the operation node is configured to update a database;

a standby node from among the nodes includes a duplicate of the database, the standby node being configured to receive an update difference of the database from the operation node to update an own database, the standby node being a node other than the operation node; and

the handicap time is adjusted based on a writing performance into a file for temporarily accumulating the update difference.