Issue

• We are facing issues in our production environment with the server hitting very high peaks in memory and going out of memory. (GC overhead)
• A heap dump shows most of the memory taken up in oracle.jdbc.driver.BufferCaches referenced from oracle T4CConnections' oracle.jdbc.driver.PhysicalConnection$BufferCacheStores.
• A heap dump shows the following retention under a single instance of org.jboss.jca.core.connectionmanager.pool.mcp.SemaphoreArrayListManagedConnectionPool:

Class Name                                                                                 | Shallow Heap | Retained Heap | Percentage
--------------------------------------------------------------------------------------------------------------------------------------
org.jboss.jca.core.connectionmanager.pool.mcp.SemaphoreArrayListManagedConnectionPool @ ...|           72 |   730,013,848 |     66.37%
|- java.util.ArrayList @ 0xa80022f0                                                        |           24 |   730,008,864 |     66.37%
|  '- java.lang.Object[300] @ 0xa8002308                                                   |        1,216 |   730,008,840 |     66.37%
|     |- org.jboss.jca.core.connectionmanager.listener.NoTxConnectionListener @ ...        |           72 |    10,806,344 |      0.98%
|     |  |- org.jboss.jca.adapters.jdbc.local.LocalManagedConnection @ 0xaf882c78          |           72 |    10,806,168 |      0.98%
|     |  |  |- oracle.jdbc.driver.T4CConnection @ 0xaf882cc0                               |          840 |    10,805,152 |      0.98%
|     |  |  |  |- oracle.jdbc.driver.PhysicalConnection$BufferCacheStore @ 0xaf885618      |           24 |    10,675,088 |      0.97%
|     |  |  |  |  |- oracle.jdbc.driver.BufferCache @ 0xaf885d10                           |           32 |    10,578,624 |      0.96%
...
--------------------------------------------------------------------------------------------------------------------------------------