Dikan Xing wrote:
When two (almost) concurrent requests to the SAME servlet occurs (for example,I type the same url in two individual browser window), I find that Tomcat seems not to work on the second request until the first is done.
Are the two browser windows truly individual, or could they be part of the same browser process? It seems that browsers may constrain themselves in several ways when accessing a single resource (server, URL).
However, in my situtaion, my servlet should wait for some other resources to respond. Those resources are quick-responded, but transimssion of requests to & responses from those resources cost much time. Therefore, if the servlet could be run (service method be called) immediately instead of the previous is done, the total response time could be reduced.
Then you should, as Christopher mentioned, check that the contention is not caused by those other resources (either via some pooling/synchronization mechanism within your app only letting one request at a time to touch the external resource, or somehow the external resource itself being able to only service a single request at a given time).
Is there any option in Tomcat so that I can configurate it to call the same servlet service method (doGet, doPost) simultanously instead of one by one?
No options needed -- not even if you're using SingleThreadModel in your servlet. I just wrote a small test and ran it in Tomcat 6.0.14 (not current, but that's what I happened to have at hand); I wrote two servlets, with request body essentially "Thread.sleep(5000)", followed by a printout identifying the servlet (class and instance). One of the servlets was a regular servlet, another was a servlet implementing SingleThreadModel. Then I wrote a test program; a Java command-line application to spawn five threads for making requests to the regular servlet (MTServlet, as in multi-threaded), and another five threads for making requests to the servlet implementing SingleThreadModel (STServlet). Below is output from a run of the test program against a server hosting these two servlets: Start: Tue Apr 22 21:56:28 EEST 2008 Hello from STServlet instance 3 Hello from MTServlet instance 1 Hello from MTServlet instance 1 Hello from STServlet instance 2 Hello from MTServlet instance 1 Hello from MTServlet instance 1 Hello from MTServlet instance 1 Hello from STServlet instance 5 Hello from STServlet instance 4 Hello from STServlet instance 6 Stop: Tue Apr 22 21:56:34 EEST 2008 So, all servlets finished within six seconds of starting the first one -- even those implementing the SingleThreadModel. It should also be noted that all MTServlet calls are handled by a single MTServlet instance (in parallel), but with STServlet the parallelism comes from Tomcat creating several instances of the servlet (as allowed by the Servlet Specification for servlets implementing the SingleThreadModel). I've seen something like you describe, and it was when someone had either declared the doGet() method (or another method called within doGet()) as synchronized, or had written a synchronized code block referring to some single object for all requests. -- ..Juha --------------------------------------------------------------------- To start a new topic, e-mail: users@tomcat.apache.org To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
