When both failure transparency and load balancing are enabled in a HotStandby (HSB) configuration, the applied failover policy depends on the failure type:
▪ Primary server failure: All the load is directed to the new primary server, which is in the PRIMARY ALONE state.
▪ Secondary server failure: All the load is directed to the primary server, which is in the PRIMARY ALONE state.
▪ Connection break between the servers (the servers are in the PRIMARY ALONE and SECONDARY ALONE states): If there is an ongoing read-only transaction executing in the secondary server, it is successfully committed in the secondary server. All the subsequent transactions are directed to the primary server (in PRIMARY ALONE).
When the normal HSB operation is resumed (with servers being in PRIMARY ACTIVE and SECONDARY ACTIVE states), the load is rebalanced between the primary and the secondary servers.
Note Even when failure transparency is not enabled, some rudimentary failover capability is available: failover from secondary server to primary server when the secondary server fails. All other failures result in a communication link failure. Thus, in most failure cases where failure transparency is not enabled, the application must reconnect with the same information. To avoid reconnection, enable failure transparency when load balancing is used.