This Knowledgebase article provides information about the 'Failover started' event.
The Primary and Secondary servers in a Neverfail server pair are continuously connected via the Neverfail Channel.
In the absence of replication data, the passive server issues heartbeats at regular intervals of time to ensure the active server is online and available. The active server can become unresponsive due to a hardware crash, lost network connection, or a power outage. When the passive server finds that the heartbeat process has failed, it will initiate a failover.
The number of heartbeats issued, as well as the time interval between two heartbeats, can be changed using the Neverfail Heartbeat Management Client. The default values are for a heartbeat timeout of 20-seconds, and a failover will be initiated after three such timed out heartbeats. These values should be carefully changed as they can affect the server pair’s functionality.