The following figure shows how RPCs use ports.

Figure 7-1 How RPCs use ports

1. Each OpenTP1 process (system process or user process) secures a receive port during process initialization.
2. To issue an RPC, the client UAP process secures a send port, establishes a connection with the receive port of the scheduler daemon, and sends the RPC. If a connection is already established, it is used. Each send port is assigned with a different number for each destination process (scheduler daemon of each node).
3. The message queue facility of the operating system is used to contact the server UAP.
4. To return a response to the RPC, the server UAP process secures a send port, establishes a connection with the receive port of the client UAP, and sends the response. If a connection is already established, it is used. Each send port is assigned with a different number for each destination process (client UAP process).
   For the subsequent transmission during commitment, the connection that is already established between the client UAP and the server UAP is used.