|
|
|
c-tree ISAM clients maintain a tight affinity to their connected server for rich context information. When failing over to a replica server, they need to re-establish this context information, which requires an entirely new connection to the failover instance. It is not always possible for a client application to know when its server has failed over. For example, when using a virtual IP address mapped to different server after failover, a client may hang as it is no longer a valid network client socket connection. A coordinated "notification" is needed informing a client application it must reconnect to continue processing data.
A new back-channel client listening thread is now available to provide this notification. A client can optionally enable this background thread before establishing its server connection. Once enabled, the client application can receive a specific notification when it is required to reconnect. This background thread listens on a configurable network secure UDP port. A specific failover notification message informs all listening clients on the network with a network broadcast message.
Three new client API calls make up this new client failover feature.
