Performance and scalability: Heap size maxing out - weblogic clustered environment

  1. Hi, I our prod environment we have two linux boxes and weblogic cluster setup . Number of java server processes are same on both servers. Some how on one box one java server is maxing out heaping size(1GB). But it is not happening on other server. This problem started only few days ago. Earlier it was fine. On other box I have set max heap size as 512 MB and it perfectly fine. I need to keep bouncing the server whenever the server is hitting max heap size. If there is any memory leak it should create issue on both servers right? I tried to get Thread dump using kill -3 . But it is not working. Some how I am not able to connect jconsole also. I am not sure how to debug this problem. Do you guys think of any possible reasons? Any help appreciated.
  2. What version of Java are you using? For the more modern versions of Java using jstack allows you to get a stack dump. The thread dump for kill -3 ends up in stdout. Please check where stdout is redirected to. Kees Jan
  3. Another idea might be to set the command line flag that causes the JVM to dump its heap on OOME. That way, you just wait for the JVM to die, restart it and then copy the heap dump to your local machine for analysis. http://java-monitor.com/forum/showthread.php?t=273 You need to check your load balancing strategy. Are you sure that both servers see the sam traffic? Are there any configuration differences between the two machines? Kees Jan