[jboss-remoting-issues] [JBoss JIRA] Commented: (JBREM-982) Performance testing

[ http://jira.jboss.com/jira/browse/JBREM-982?page=comments#action_12414709 ] Ron Sigal commented on JBREM-982: --------------------------------- Added new parameter org.jboss.remoting.transport.socket.MicroSocketClientInvoker.USE_ONEWAY_CONNECTION_TIMEOUT (actual value "useOnewayConnectionTimeout"), which determines the behavior of MicroSocketClientInvoker and org.jboss.remoting.transport.socket.ServerThread when handling client side oneway invocations. When set to "true", the default value, ServerThread will send, and MicroSocketClientInvoker will expect, a version byte, which indicates that the server side processing of the invocation is complete and that it's safe to reuse the connection. If the read() for the version byte times out, the connection is discarded. This is the behavior implemented in reaction to JBREM-706 "In socket transport, prevent client side oneway invocations from artificially reducing concurrency". When "useOnewayConnectionTimeout" is set to "false", there is no transmission of a version byte, which was the original behavior. When the socket and bisocket client side oneway invocation tests are run with "useOnewayConnectionTimeout" is set to "false", the performance in Remoting 2.4.0.GA is now consistent with the performance in 2.2.2.SP7. The overall ratios for those transports are now: bisocket: 0.99 socket: 0.98 The conclusion is that the appropriate setting of the "useOnewayConnectionTImeout" parameter will vary according to circumstances. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira