Hi Rick, Thanks for the response.
> When you fix your mapping, does the connection leak stop? Also, how are you > confirming that there is a connection leak? Well, yes, it stops. But after further research, there are other places in my code where "normal" operation could lead to a SQLException getting thrown inside OpenJPA -- violation of a constraint, for example. So I still need to track this down. Glassfish logs a leak warning when a connection isn't returned to the pool within a configurable timeout (in my case, 60 seconds). Unless I enable leak reclaiming (which is prohibitively expensive, performance-wise), enough of these warnings and I run out of connections. My transactions are short-lived (a couple of seconds, max) and I only get the leak warning as the result of flushes that throw SQL exceptions (as corroborated by the stack traces included in the warning). Regards, Julian
