In automatic failover mode, the Watchdog process initiates a failover process if all of the following conditions are met:
- The secondary Sequencer connects to all systems in the Unity cluster before losing the primary Sequencer. This condition verifies that basic network connectivity is valid.
- The Watchdog confirms that the other Sequencer has not communicated with any of the systems in the Unity cluster.
- The Watchdog communicates with all managed Teradata systems and verifies the Dispatcher on the other Unity server has not sent updates.
- All these conditions must be true for the entire wait period defined in the HAAutoFailoverWait configuration value.
During an automatic failover, the Watchdog process waits before initiating a Sequencer failover. The Watchdog confirms the active Sequencer is not running.
The Sequencer process emits a heartbeat every five seconds (default). The heartbeat frequency is defined in the ProcessPingInterval configuration parameter. The ProcessTimeout configuration parameter defines how long the Watchdog process waits for a heartbeat before it allows automatic failover to the Standby Sequencer process.
The Watchdog process might not receive heartbeats if either of the following occur:
- The Sequencer, network, or server fails
- The Sequencer process is unresponsive