External Reference Entity

To prepare for failover situations, configure an External Reference Entity (ERE) for the primary and secondary master nodes.

When the communication link between the primary and secondary master nodes fails, both database servers might mistakenly assume the other one is down. As a result, a dual primary situation can arise and we might lose transactions when databases are later synchronized. To avoid this situation, we can use a network reference device, such as a network router, as an ERE to check the health of the network.

If we configure a secondary master, we must also configure an ERE for the primary and secondary masters. If the primary master loses its connection to the secondary master, it can contact the ERE to determine whether there is a network fault or the secondary master is down.

In a distributed configuration, we can separate the primary and secondary masters into one data center and the tertiary and quaternary masters into another data center. If the data center link fails, the primary and tertiary masters operate in parallel and service requests in their local networks. When the data center link is restored, the tertiary master becomes inactive and reconciles its updates with the primary master. The above mentioned data reconciliation applies only to the distributed session cache. It does not apply to other databases such as the configuration database and geolocation database.

Parent topic: High availability of cluster services