From the client side, enabling high availability means the application receives the error code CS_RET_HAFAILOVER during a network failure, and the client library automatically reconnects to a server. In high availability, the application again attempts to connect first to the original server, and then tries the secondary server (as set in the interfaces file entry for the primary server) if the connection fails. If you use the Open Client libraries shipped with the Cluster Edition, the instance sends a failover target list to the client. The application uses this list to determine where to connect when the instance fails.
Whether you use high availability or the Cluster Edition, failover can occur at any point.
