Urgent! Tomcat4 closed giving Out of Memory Error.

Hi,

I have stunk on this problem for almost 2 weeks. :-( 
In these two weeks, I have tried many methods, but it still doesn't work.

I built a system to run TPCW benchmark. One web and application server, 
with tomcat4.0 running, one database with mysql4.0 running on it and 
several client machines. The Jave version is j2sdk1.4.0_03. The system 
worked perfect under light load. But when the number of clients 
exceeded some number (like 600), the tomcat4 didn't work anymore.
I checked catalina.out. It appeared the error:

SEVERE: Caught exception executing 
org.apache.tomcat.util.net.TcpWorkerThread-NuMP4iMkatM@public.gmane.org, terminating thread
java.lang.OutOfMemoryError: unable to create new native thread
        at java.lang.Thread.start(Native Method)
        at 
org.apache.tomcat.util.threads.ThreadPool$ControlRunnable.<init>(ThreadPool.java:497)
        at 
org.apache.tomcat.util.threads.ThreadPool.openThreads(ThreadPool.java:387)
        at 
org.apache.tomcat.util.threads.ThreadPool.runIt(ThreadPool.java:226)
        at 
org.apache.tomcat.util.net.TcpWorkerThread.runIt(PoolTcpEndpoint.java:503)
        at 
org.apache.tomcat.util.threads.ThreadPool$ControlRunnable.run(ThreadPool.java:530)
        at java.lang.Thread.run(Thread.java:536)


But my tomcat4 is running with option -server -Xmx1536M. (I also tried to 
change -Xss and -Xms, which didn't help.) And the actual memory 
utilization was less than 40%. (I have 2GB memory in this machine.)

I also changed 'ulimit -n 8192' and server.xml :

<!-- Define a non-SSL Coyote HTTP/1.1 Connector on port 8080 -->
    <Connector className="org.apache.coyote.tomcat4.CoyoteConnector"
               port="8080" minProcessors="5" maxProcessors="1750"
               enableLookups="true" redirectPort="8443"
               acceptCount="1000" debug="0" connectionTimeout="-1"
               useURIValidationHack="false" disableUploadTimeout="true" />
    <!-- Note : To disable connection timeouts, set connectionTimeout 
value  to -1 -->

Why this error happened? I was trying to use strace -f -p to trace the
system call, but it only gave me one line "accept(13, " and waited at that
point. Did anyone meet this problem before? Can I get some detailed error 
log besided catalina.out?

I really appreciate your kind help

Qi

-- 
Qi Zhang, Ph.D Student

Department of Computer Science,
College of William and Mary
P.O. Box 8795
Williamsburg, VA 23187-8795