This content has been marked as final. Show 3 replies
I don't even know what Liferay is. For what that's worth.
Almost certainly two threads are trying to poke the connection at the same time. Clearly one is trying to close the connection. That's the stack trace in the original post. There is probably another thread that is doing something else with the connection, like executing a query or some such. The stack trace looks like some kind of post-error cleanup code. In that case, maybe the connection is dead and the other thread is hung waiting for TCP timeout. That's typically 15 minutes by default. It's better to call abort to clean up possibly dead connections. abort won't hang but close can for various reasons. close is polite and abort isn't. Use close in normal circumstances. Use abort only if there is evidence that the connection is broken. Calling abort needlessly will clutter the server with dangling resources. Be polite. Call close.