When connection leak tracing is ON and the pool is used by the application, in a case when connection is leaked by the application and flush-connection-pool is done to reset the connections in the pool, expected behavior is not seen.
As per the flush connection pool description :
"The flush-connection-pool subcommand resets a JDBC connection
pool or a
connector connection pool to its initial state. Any existing
connections are destroyed, which means that the transactions
these connections are lost. The subcommand then recreates the
connections for the pool, and restores the pool to its steady
pool size. "
the active live connections in the connection pool must be destroyed.
This should be fixed.