Device Switchover upon Failure

When a failure occurs in the active device, a switchover occurs to the redundant device making it the new active device. Whether a switchover is later done back to the repaired failed device, depends on whether you have enabled the Preempt mode:

Enabled: The Preempt mode specifies one of the device's as the "preferred" device. This is done by assigning different priority levels (1 to 10, where 1 is the lowest) to the two devices. Typically, you would configure the active device with a higher priority level (number) than the redundant device. The only factor that influences the configuration is which device has the greater number; the actual number is not important. For example, configuring the active with 5 and redundant with 4, or active with 9 and redundant with 2 both assign highest priority to the active device. Whenever the device with higher priority recovers from a failure, it first becomes the redundant device but then initiates a switchover to become the active device once again; otherwise, after recovery, it becomes the redundant device and remains as redundant. If you change the priority level of the redundant device to one that is higher than the active device and then restart the redundant device, a switchover occurs to the redundant device making it the active device and the "preferred" device. If both devices are configured with the same priority level, Preempt mode is disabled. Please see note below when using priority level 10.
Disabled: A switchover is done only upon failure of the currently active device.

Failure detection by the devices is done by the constant keep-alive messages they send between themselves to verify connectivity. Upon detection of a failure in one of the devices, the following occurs:

Failure in active device: The redundant device initiates a switchover. The failed device restarts and the previously redundant device becomes the active device in stand-alone mode. If at a later stage this newly active device detects that the failed device has been repaired, the system returns to HA mode. If Preempt mode is enabled and the originally active device was configured with a higher priority, a switchover occurs to this device; otherwise, if it was configured with a lower priority (or Preempt mode was disabled), the repaired device is initialized as the redundant device.
Failure in redundant device: The active device moves itself into stand-alone mode until the redundant device is returned to operation. If the failure in the redundant device is repaired after restart, it's initialized as the redundant device once again and the system returns to HA mode.

Connectivity failure triggering a switchover can include, for example, one of the following:

Loss of physical (link) connectivity: If one or more physical network groups (i.e., Ethernet port pair) used for one or more network interfaces of the active device disconnects (i.e., no link) and these physical network groups are connected OK on the redundant device, a switchover occurs to the redundant device.
Loss of network (logical) connectivity: No network connectivity, verified by keep-alive packets between the devices. This applies only to the Maintenance interface.
Switchover triggered by loss of physical connectivity in one or more Ethernet Group is not done if the active device has been configured to a Preempt mode level of 10. In such a scenario, the device remains active.
After HA switchover, the active device updates other hosts in the network about the new mapping of its Layer-2 hardware address to the global IP address, by sending a broadcast gratuitous Address Resolution Protocol (ARP) message.