Facing a memory leak issue due to internal weblogic objects

I have an app running on Weblogic 10 app server.

I am facing a memory leak issue due to internal weblogic objects. The following class objects are occupying a lot of memory and not clearing it off. Can anyone help me figure out what these classes are meant for and what possible could be causing the leak? Googling didn't help much.

weblogic.store.xa.internal.PersistentStoreXAImpl 

T开发者_JAVA技巧hanks, Sid

---

This is the internal WL class for an XA-enabled persistent store. If you are using JDBC then check if your DB is full. If you are using the default store, or a file store, check your filesystem for errors or to see if its full.

There should not be very many of these. Are you using JMS in your app? Perhaps you also have a configuration issue. Try opening a ticket with Oracle support, or if you can share the code and configuration with me you can email me at Jeffrey.west@oracle.com. If you have a support contract you should open a case as opposed to emailing me. If you open a case, please email me the SR# and I can look into it. If you don't have a support contract then email me with the details of your problem...

---

To add to Jeff's answer, This is the persistent store used by Weblogic for various subsytems such as JMS, JTA and even Web services and EJB timers

If you have not configured this in your instance, it usually should default to a Filestore located at

bea_home\user_projects\domains\domain-name\servers\server-name\data\store\default

So you could check for file system disk space issues there. If you can afford to shut down your Weblogic, can you clear out the file system above - or check how old the objects persisted there are?

---

"XA" usually means "two-phase commit" to me. I think this is a JDBC driver for two-phase commit to a persistent store.

I can't understand why these wouldn't be cleaned up. How are you looking at the heap. What tool is telling you that this is the source of the leak? Could it be that your classes aren't cleaning up persistent resources?

---

We had seen a similar problem once, when JDBC profiler was enabled. Do you have any profiling enabled either on application or WebLogic level?