The invention provides a method of synchronous detection and failure isolation for a double control highly available system. The system is provided with two controllers which use special channels integrated on a PCI mainboard, serial ports or network card cables to monitor the HADaemon process of the counterpart. When a failure appears, a primary controller and a secondary controller conduct the following control steps of: 1), the primary controller releases virtual IPs of a server and the secondary controller takes over the IPs through ARP deception; 2), the resources of the main controller are hung and send state information to the primary controller, and a related service is started to realize seamless switching; 3), during the taking over of IP and the switching of servers, an alarm is given in an email log; 4), the primary controller after restoration choose takeover service or no takeover as a backup of synchronous detection; 5), in order to improve the availability of heartbeat transmission, network cards and serial ports are used together to avoid single point failure of a link; 6), a failure occuring in a network for external services of the primary controller can also be detected by the HADaemon which then triggers the primary controller to take over the virtual IP and switch the resources.