Have you figured anything out to fix this? We are having a very similary situation. As soon as the connection pool runs into a situation where all the connections are busy, it waits for the blocking timeout milis and then throws the exception complaining that it can't open a managed connection.