While the Active Broker is unavailable, the following limitations have been observed on components in a failover deployment.
The Failover Manager polls each Domain Manager and Broker at fixed intervals to determine their status. Consequently, a few polling cycles pass before the Failover Manager determines that an Active Broker has failed. During this time frame, the Active Domain Managers may not be able to communicate with each other until the Broker for them is switched by the Failover Manager to point to another running Broker. This behavior occurs only with topology-synchronization among Domain Managers.
To overcome the impact due to this limitation, configure separate polling intervals for Brokers and other Domain Managers. For example, if the Failover Manager polls IP Manager, SAM, and other Domain Managers at an interval of 4 minutes, then configure the polling interval for Brokers at 1 minute.