How do I serve webapp A only from port A and webapp B only from port B?

The most efficient way to do this is with two org.mortbay.jetty.Server instances. There is also a second, less efficient alternative.

Instance A will have a connector listening on port A with webapp A defined, and instance B will have webapp B and connector listening on port B defined.

Let's clarify that with an example.

Suppose we have webapp A that we want served from port 8080, and webapp B that we want served from an SSL connector on port 8443. We would set up 2 different Server instances, each in it's own jetty xml file like so:

Now we need to set up 2 context files, one to describe the webapp we wish to deploy on Server A and the other for the webapp to deploy on Server B. We then put these files into $JETTY_HOME/contextsA or $JETTY_HOME/contextsB respectively.

You run both Server instances in the same JVM by providing them both on the runline:

java -jar start.jar jettyA.xml jettyB.xml

Of course, you could also start two separate jetty instances, one with jettyA.xml and the other with jettyB.xml, if you wanted to. However, it is usually more efficient to run both Servers in the same JVM.

Alternative

There is also a alternative way to achieve the same result as above, however it is less efficient. It involves setting the list of connectors on a webapp from which it will accept requests. This is a less efficient solution than the one described above because the request is presented to each webapp, which then must decide if it will accept it or not. In the first solution, only one webapp will ever be passed the request.

Here's how to configure the alternative solution: