We have client with multiple threads. To reuse TCP connection, each thread on the client opens a new RcfProtoChannel and uses that to create a service stub. After that large number of synchronous requests are executed from each of the client thread. We have observed that many time "The server has reached its incoming connection limit" is thrown even though the total requests fired on the server is less than the maximum configured. How will the client knows that the previous call has completed?