Re: RADIUS accounting bug

John W. Temples (john@kuwait.net)
Tue, 19 Mar 1996 13:00:52 +0300 (GMT)

On Mon, 18 Mar 1996, Leo Savage wrote:

> It now appears that it is possible to receive a "start" record and then
> never receive a corresponding "stop" record. Instead, we receive a new
> "start" record when the port is reused.

I've done a lot of hacking on radiusd, and I can attest to the
existance of this problem. This complicates implementation of
duplicate login prevention with radiusd -- since you cannot reliably
know when a user has logged off, it's not a good idea to deny them a
login because you think they're still logged on. My approach to
handling a duplicate login is to disconnect the existing session (even
though it might not be there), remove the user from the list of logged
in users (even if we don't get a stop record for him) and let the new
login proceed.

A related problem is getting a stop record before a start record for a
given session, but this is more easily dealt with.

--
John W. Temples, III       ||       Providing the only public access Internet
Gulfnet Kuwait             ||            site in the Arabian Gulf region