Deploying RMI Across a Network


As shown so far, the server and the client must be on the same machine -- some distributed system!


Get the RMI registry to dish out the client stubs on demand.


RMI does not provide true location transparency , which means that you must at some point know the network name of the machine the server is running on. The server machine must be running the RMI registry program as well, though there’s no need for the RMI registry to be running on the client side.

Now the RMI registry needs to send the client stubs to the client. The best way to do this is to provide an HTTP URL and ensure that the stub files can be loaded from your web server. This can be done by passing the HTTP URL into the RMI server’s startup by defining it in the system properties:

java -Djava.rmi.server.codebase=http://serverhost/stubsdir/ ServerMain

In this example, serverhost is the TCP/IP network name of the host where the RMI server and registry are running, and stubsdir is some directory relative to the web server from which the stub files can be downloaded.

Be careful to start the RMI registry in its own directory, away from where you are storing (or building!) the RMI stubs. If RMI can find the stubs in its own CLASSPATH, it will assume they are universally available and won’t download them!

The only other thing to do is to change the client’s view of the RMI lookup name to something like rmi://serverhost/foo_bar_name. And for security reasons, the installation ...

Get Java Cookbook now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.