[Freeswitch-users] Zombie Records in core db

We have FreeSWITCH Version 1.0.4 (exported) running at a high volume traffic. I normally check the concurrent calls by looking at the number of sessions from status command. However, the number of concurrent calls in FS is normally higher than it's supposed to be after we ran traffic for about a week. Thus, I routed the traffic away from the FS and found out from "show calls" that there were so many old calls from previous days. We are running a pass-thru traffic in signaling only. I wonder whether there is a way to have those "zombied" records clean up automatically. Also, what should I do to prevent this problem? Thank you, Dorn B.