You could give a look @ the different options on the JVM u r using !
I would suggest U to add the following paramters to your JBoss start up script :
It sets the initial (to 32MB) & maximum (to 256 MB) size of the heap. These values should be worked out when optimizing the envrioment...
This should confine a little bit the Memory hungry system. It forces regular GC & usually make the application run more smoothly :-)
To know what flag to set, just type :
"java" at the prompt. In most case you have the
"java -X" command propose for the unsupported jvm-options. chack for appropriate syntax.
p.s. : My Linux install has very similar option the Win2000 VM does have. I posted you directly the result of the
java -X command
You are right . Under linux , both ps and top don't distinguish between process mem and thread mem. Thus if you have 100 threads in one process and the process is using 100MB , ps will report 100 java entries each using 100MB. You will notice the total memory usage NOT be 100*100MB .
I've set the JVM heap size as stated (-Xms32MB and -Xmx256MB), but, from ps output, I still see jboss's VmSize as big as 1.7GB, and my system (with ~700MB physical memory and >1GB swap space) just grinds down to halt. All I did was making about 50 invocations on a SLSB that invoke another 3 EBs with CMP20, and the abnormal menory growth happens within 5 minutes of the invocation. I'm using JBOSS 3.0 RC1 on Linux 2.4.7 with JVM of Blackdown 1.3.1 FCS. Any clue? I saw a bunch of memory leak postings on the forum. Does the latest CVS code fix the problem?
Do at some point a :
Runtime.getRuntime ().freeMemory ();
Runtime.getRuntime ().totalMemory() ;
You will see that the value return are not consistent with the values shown in "ps".
Look for other trheads with similar posts.