Time, in seconds, before a heartbeat on a secondary server declares the coordinator offline if the heartbeat fails to reconnect to the coordinator after the first disconnect. This option also determines how long the coordinator keeps a global transaction in a suspended state.
0 to 604800 (1 week) in seconds
3600 seconds (1 hour)
This option affects all multiplex nodes and has no node-specific or connection-specific value. Option can be set at the database (PUBLIC) level only.
Requires the SET ANY SYSTEM OPTION system privilege to set this option. If you change the value of MPX_LIVENESS_TIMEOUT on a running server, the new value takes effect immediately for connections that might suspend in the future. The changed value also immediately affects the remaining timeout period for all current suspended transactions.
If a writer fails to resume a suspended transaction within the MPX_LIVENESS_TIMEOUT period, the transaction can no longer commit, and the user should roll back the transaction. The coordinator keeps a global transaction in a suspended state for a period of 2 * MPX_LIVENESS_TIMEOUT. If the corresponding writer fails to resume the transaction before the 2 * MPX_LIVENESS_TIMEOUT period, the coordinator rolls back the suspended transaction.
Always specify an MPX_LIVENESS_TIMEOUT value that is a multiple of the current MPX_HEARTBEAT_FREQUENCY value, which controls the aliveness check period. The coordinator internally doubles the value of MPX_LIVENESS_TIMEOUT.