Hi all,

I guess I found a major bug, but I request your opinion before report it in Jira. By the way any workaround are welcome.

So, here is the bug.

Test Case:
I use a TableHiloGenerator to generate uid for an object (e.g Event).
The max_lo value is set to a small value (e.g 1024).

The max connection number of my pool is set to 10 per example.

Now if I persist more than 10x1024 new Events, a dead lock occurs.

Explanation:
The TableHiloGenerator use the Isolater class to increment the counter in a dedicated transaction. To do so a connection is retrieved from the pool, but never released!
After a while, the pool is empty and the max connection number is reached. So the next session or Isolater execution will wait indefinitely for a free connection.

This issue is not limited to our test case. This dead lock may occur after a while if a large number of objects are created.

Code & Fix:
Check the inner class org.hibernate.engine.transaction.Isolater.JdbcDelegate, line 191.
A curly bracket is at the wrong place.
lines 191 and 192 must be switched round.

Impacts:
Major, Connection Management is a key point of any application.

Hibernate version: 3.1.3

Thread Dump

"Thread-64" daemon prio=5 tid=0x00bca0e8 nid=0x7a in Object.wait() [acdfe000..acdffc28]
at java.lang.Object.wait(Native Method)
- waiting on <0xcbf98020> (a com.mchange.v2.resourcepool.BasicResourcePool)
at com.mchange.v2.resourcepool.BasicResourcePool.awaitAcquire(BasicResourcePool.java:966)
at com.mchange.v2.resourcepool.BasicResourcePool.checkoutResource(BasicResourcePool.java:208)
- locked <0xcbf98020> (a com.mchange.v2.resourcepool.BasicResourcePool)
at com.mchange.v2.c3p0.impl.C3P0PooledConnectionPool.checkoutPooledConnection(C3P0PooledConnectionPool.java:232)
at com.mchange.v2.c3p0.PoolBackedDataSource.getConnection(PoolBackedDataSource.java:94)
at org.hibernate.connection.C3P0ConnectionProvider.getConnection(C3P0ConnectionProvider.java:35)
at org.hibernate.jdbc.AbstractBatcher.openConnection(AbstractBatcher.java:495)
at org.hibernate.engine.transaction.Isolater$JdbcDelegate.delegateWork(Isolater.java:147)
at org.hibernate.engine.transaction.Isolater.doIsolatedWork(Isolater.java:38)
at org.hibernate.engine.TransactionHelper.doWorkInNewTransaction(TransactionHelper.java:51)
at org.hibernate.id.MultipleHiLoPerTableGenerator.generate(MultipleHiLoPerTableGenerator.java:183)
- locked <0xcc026d68> (a org.hibernate.id.MultipleHiLoPerTableGenerator)
at org.hibernate.event.def.AbstractSaveEventListener.saveWithGeneratedId(AbstractSaveEventListener.java:91)
at org.hibernate.event.def.DefaultSaveOrUpdateEventListener.saveWithGeneratedOrRequestedId(DefaultSaveOrUpdateEventListener.java:186)
at org.hibernate.event.def.DefaultSaveOrUpdateEventListener.entityIsTransient(DefaultSaveOrUpdateEventListener.java:175)
at org.hibernate.event.def.DefaultSaveOrUpdateEventListener.performSaveOrUpdate(DefaultSaveOrUpdateEventListener.java:98)
at org.hibernate.event.def.DefaultSaveOrUpdateEventListener.onSaveOrUpdate(DefaultSaveOrUpdateEventListener.java:70)
at org.hibernate.impl.SessionImpl.fireSaveOrUpdate(SessionImpl.java:502)
at org.hibernate.impl.SessionImpl.saveOrUpdate(SessionImpl.java:494)

Hi alexism - I think that's a known issue: http://opensource.atlassian.com/project ... se/HB-1246

Thanks for the link R1ch,

but I think my issue is not exactly the same.
I point out a non-release connection issue not a "out of connection" pb.

I can confirm this is a real "not release connection" bug.
I set up c3p0 with a timeout of 5mn, I create a lot of a new objects with a small max_lo value.
37 connections has been check out from the pool, but 12 hours later they are still not released!

But it is already reported in jira: HHH-1669