This content has been marked as final. Show 1 reply
The best values for DB_REP_HEARTBEAT_MONITOR and DB_REP_HEARTBEAT_SEND are highly dependent on your particular environment. You need to have some idea of how long it could potentially take for a message to travel from the master to a client and make sure your MONITOR value is significantly longer than that.
It is certainly safe for the MONITOR value to be double the SEND value.
If the heartbeat monitor expires on a replication client, the client concludes that it has lost its connection to the master and calls an election to find a new master. If the connection disruption is brief and the connection to the master is restored in time, the election stops when the client again finds the original master. If the disruption takes a longer time, the client's election could progress enough to elect another master and the resulting duplicate master situation could take a while to resolve.
It is not a good idea to risk frequent heartbeat monitor timeouts by setting the MONITOR value small enough that timeouts are easily triggered by the expected communications speed of your environment.