Hi all,

I encounter a problem in getting ConcurrentModificationException when there is more than 1 sessions opened concurrently:.



I am coding against JPA API that EntityManager is referred to manipulate objects. Corresponding entityManager is stored in thread local that its session as well as the persistence context should be threadsafe.

However, there are something weird that I found some debug messages from hibernate on child collection initialization about the parent objects retrieved from thread A appeared in thread B makes me believe that corresponding collection initialization was done under thread B which is not expected because thread A and thread B is working on independent hierarchy of objects. And that's why causing the ConcurrentModificationException when thread B is trying to flush its own.

Does anyone offer me some helps on it as I have being working on it for days but still no clue?

Hi all,

env info:

Hibernate: 3.6.9.Final
OS: Redhat 5
JDK: 1.6.0

Hi all,

Eventually, I spot out that the unexpected modification in thread A is come from the forceInitialization on the PersistentCollection in thread B in my own program before I pass the object from server to client. According to my debug logging, I think that the PersistentCollection is not threadsafe because the associated session and persistence context which the PersistentCollection associated to, is belongs to the session of thread A!

I have written a testing program to simulate the problem. I coded to retrieve a parent object with child objects in oneToMany relationship and in LAZY fetch. After that, my intialization program will initialize corresponding child collection traversely if it is not initialized (in type of PersistentCollection) by calling forceInitialization. At this moment, I debug out the uuid of the session and persistence context the PersistentCollection is associating to.

I run the testing program by creating 5 threads running the same code fragment concurrently and get the odd debug logs as below proving my thought of the non-thread safe PersistentCollection.

14:55:44.751 [Retrieval Thread [4]] DEBUG MyInitializer - Force initiailize in thread 15595312 persistent collection 15976046 in session 15740343 in context 4809601
14:55:47.282 [Retrieval Thread [3]] DEBUG MyInitializer - Force initiailize in thread 24791433 persistent collection 4835957 in session 15740343 in context 4809601
14:55:47.329 [Retrieval Thread [3]] DEBUG MyInitializer - Force initiailize in thread 24791433 persistent collection 19185231 in session 15740343 in context 4809601
14:55:47.923 [Retrieval Thread [2]] DEBUG MyInitializer - Force initiaiize in thread 13986615 persistent collection 14869148 in session 15740343 in context 4809601
14:55:48.891 [Retrieval Thread [4]] DEBUG MyInitializer - Force initiaiize in thread 15595312 persistent collection 4835957 in session 15740343 in context 4809601
14:55:48.891 [Retrieval Thread [4]] DEBUG MyInitializer - Force initiaiize in thread 15595312 persistent collection 19185231 in session 15740343 in context 4809601
14:55:48.891 [Retrieval Thread [4]] DEBUG MyInitializer - Force initiaiize in thread 15595312 persistent collection 23809882 in session 15740343 in context 4809601

I am sure that entityManager who holds the session is well managed in my program and threadsafe. But how come the PersistentCollection refer to the wrong session? Is it no longer for us to work on PersistentCollection by our own? However, it is the same way to initialize the object by calling Hibernate.initialize(). Do any of you have idea on it?