> They've been sending accounting "stop" packets after an Access-Reject. Uh... no. If the user was rejected, there's no session, and no need for a "stop" packet. Just a FYI, I can't reproduce this in the beta code that I am testing - I presume it has been fixed at some point.