Resolution:
A SmartSockets connection actually involves different layers.  A warm connection is one in which the structures (such as buffers), and information which is internal to the process (this is the conn-layer) is maintained or created, but the actual TCP-level socket (which is managed by the OS and is known as the socket-layer) is not. Whenever an RTclient gets disconnected from RTserver, the client maintains a warm connection on its side so that messages in the output buffer waiting to be flushed are not lost, and so that callbacks and other message processing functionality is maintained.  The only time this would not be true would be if the developer had intentionally included a call to TipcSrvDestroy(T_IPC_SRV_CONN_NONE) in their code.
Whether or not the RTserver mainatins a warm connection on its side for a disconnected RTclient is dependent on the Server_Disconnect_Mode that was set in the RTclient prior to connecting to RTserver. It is very common for applications that are using GMD to set the Server_Discaonnect_Mode to warm, so that the RTserver will continue buffering the GMD messages for the client during any tansient disconnects.  If the RTclient reconnects in a timely fashion, (i.e., before the Client_Reconnect_Timeout is reached in the RTserver) then all messags buffered for the client will be delivered as soon as the RTclient reconnects. If an RTclient with a Server_Disconnect_Mode of warm disconnects from the RTserver but does not reconnect in a timely fashion then when the Client_Reconnect_Timeout is reached the Rtserver converts the connection from WARM to NONE.

What is the difference between a Warm vs. Full connection