Memory leak in JBossAS

[Naman Dhaliwal]

Memory leak associated with the JBossAS:-

JBossAS server version:- 4.0.5.GA
Container:- EJB Container ( with MDB)

While processing message within MDB, whenever a runtime exception (Wrapped with the EJBException) occurs , the instance of the MDB is not garbage collected. We can see a new instance of MDB generated but all the remnants of the previous MDB still in the heap. While examining the JBOSS with eclipse TPTP, we could see the reference of the bad MDB instance still in a HashMap, managed by the JBoss.

Has anyone encountered this issue. Appreciate any help .

Thanks
Nam

[Jaikiran Pai]

4.0.5 is a bit too old. So even if it was a bug, it probably would have been fixed in the later versions. Have you tried with the latest 4.2.3 version in 4.x series?

[Naman Dhaliwal]

We currently have this working instance in prod which no body want to migrate, untill there is a show stopper.
I actually wanted to check if this is a known issue and community is aware of it.


Thanks
Nam

[Vikram Saxena]

Hey ..

I had encountered a memory leak issue as well. And I stated it exactly as you have stated .-> Memory leak in JBoss.

But later I realised that is not JBoss which is responsible, it is bad coding which was retaining a hashmap in memory and was referenced forever and hence the problem.

This is how I got to the issue :
1. I analysed the heap dumps [taken at relative times in Production] and then i analysed the objects/worker threads that are grwoing in size.
2. Now when you open these worker threads you will see Hash Maps which are holding some information.
3. Check in the code where these objects are being made and where you have done lose coding by still referncing them.


However, this is easier said than done, it took me a month to get to the root of it .