It is an object of the present invention to allow easy upgrading of the version of the OS in a failover cluster system. When the user mounts a storage medium 4 on the node 1, and gives instructions to install a program, the installation program 12 requests the failover program 11 for the stopping of heartbeat communications. The node 1 voluntarily stops heartbeat communications in a state in which the system of this node is operating normally. When the failover program 21 of the node 2 detects the stopping of heartbeat communications, this program starts failover, and takes over the business service of the node 1. During the execution of failover by the node 2, the program stored in the storage medium 4 is installed in the node 1. When this installation is completed, the node 1 transmits a failback request to the node 2, and restarts the business service.