| 摘要 |
A cluster system for executing jobs with high processing load is allowed to continue its service while software modification is applied to the system. When input with modification data (3), a modification application scheduled node decision means (1b) generates a modification application scheduled node list (1c). A modification applicable node selection means (1e) successively extracts the node IDs of nodes which are not executing a job, from the modification application scheduled node list to set the extracted node IDs as modification applicable node IDs until the value of a modification-in-progress node counter (1f) indicating the number of nodes to which software modification is being applied reaches a predetermined upper limit value. A service management means (1g) stops the service of nodes corresponding to the modification applicable node IDs. In accordance with the modification data, a modification means (1h) modifies target software installed on the nodes whose service has been stopped.
|
