From: <tom...@jb...> - 2006-01-31 03:53:34
|
The reason for sending this byte for the socket invoker is to keep the socket connection open between the client and server. After each request on the server side, it will send an ACK byte to the client. After each client call, the socket will be pooled and if this socket is pulled out of the pool before the timeout value, it will send the an ACK back to the server (in the checkConnection()) and then make the request. The whole point is to keep the socket connection open if going to be reusing it within a short amount of time (as the ACK is much faster than creating a whole new socket from scratch). I can see where if going to make continous calls back to back on the same connection, then this would add overhead. Is this for a real use case, or for perforamnce testing where want to see max throughput? Guess we can add a configuration so the ACK is turned off. If you think this is needed, please open a jira issue for it. Also, we probably need to discuss what remoting jira issues you need and when. The next release in jira is for 1.6.0, but I doubt that will be till this summer. If you need some of these issuses resolved before then, we may need to add a 1.4.1 release with an earlier release date. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=3920565#3920565 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=3920565 |