Now, i'm being told that this will terminate the session (or is it all sessions?) in the 15th minute of use, regardless their activity.
This is wrong. It will just kill the session when the associated client (webbrowser) has not accessed the website for more than 15 minutes. The activity certainly counts, exactly as you initially expected, seeing your attempt to solve this.
The HttpSession#setMaxInactiveInterval() doesn't change much here by the way. It does exactly the same as <session-timeout> in web.xml, with the only difference that you can change/set it programmatically during runtime. The change by the way only affects the current session instance, not globally (else it would have been a static method).
To play around and experience this yourself, try to set <session-timeout> to 1 minute and create a HttpSessionListener like follows:
@WebListener
public class HttpSessionChecker implements HttpSessionListener {
public void sessionCreated(HttpSessionEvent event) {
System.out.printf("Session ID %s created at %s%n", event.getSession().getId(), new Date());
}
public void sessionDestroyed(HttpSessionEvent event) {
System.out.printf("Session ID %s destroyed at %s%n", event.getSession().getId(), new Date());
}
}
(if you're not on Servlet 3.0 yet and thus can't use @WebListener, then register in web.xml as follows):
<listener>
<listener-class>com.example.HttpSessionChecker</listener-class>
</listener>
Note that the servletcontainer won't immediately destroy sessions after exactly the timeout value. It's a background job which runs at certain intervals (e.g. 5~15 minutes depending on load and the servletcontainer make/type). So don't be surprised when you don't see destroyed line in the console immediately after exactly one minute of inactivity. However, when you fire a HTTP request on a timed-out-but-not-destroyed-yet session, it will be destroyed immediately.