We have got a weblogic server installation deployed on six servers HP-UX. Our installation has 21 weblogic servers and we have got a big problem on one of the clusters; this cluster has got two servers that have to be restarted after working for several days because the heap memory consumption does not have limits. For that reason I did a memory dump using JCONSOLE and after analyzing it using the Eclipse Memory Analyzer; I could detect that the main leak suspect is: weblogic.servlet.internal.session.ReplicatedSessionData. I did this analysis several times and the main suspect has been the same class.
Problem Suspect 1
One instance of "weblogic.cluster.replication.ReplicationManager" loaded by "sun.misc.Launcher$AppClassLoader @ 0x9ffffffeca2bf030" occupies 120,817,128 (11.38%) bytes. The memory is accumulated in one instance of "weblogic.utils.collections.ConcurrentHashMap$Entry" loaded by "sun.misc.Launcher$AppClassLoader @ 0x9ffffffeca2bf030".
Problem Suspect 2
7,921 instances of "*weblogic.servlet.internal.session.ReplicatedSessionData", loaded by "sun.misc.Launcher$AppClassLoader @ 0x9ffffffeca2bf030" occupy 634,519,688 (59.77%) bytes*. These instances are referenced from one instance of "java.util.Hashtable$Entry", loaded by "<system class loader>"
Problem Suspect 3
67 instances of "weblogic.utils.classloaders.ChangeAwareClassLoader", loaded by "sun.misc.Launcher$AppClassLoader @ 0x9ffffffeca2bf030" occupy 107,663,504 (10.14%) bytes
Cluster´s JVM configuration: We have two nodes
-d64 -Xms3584m -Xmx3584m -XX:MaxPermSize=1024m -XX:+UseParNewGC
This is the information about my product:
Product: Oracle Weblogic Server
Product Version: 10.3.1
Operating System: HP-UX Itanium
OS Version: 11.31
S: HPUX 11.31
Processpr: 64 bit Itanium
Java HotSpot(TM) 64-Bit Server VM version 1.6.0.02 jinteg:09.27.08-08:51 IA64W
Vendor: Hewlett-Packard Company