Re: [RadiusNT] Concurrency Control
Dale E. Reed Jr. ( (no email) )
Fri, 25 Jun 1999 12:15:26 -0700
Jeff Woods wrote:
>
> I believe the concurrency checks that RadNT 2.5 performs are invalid. By
> relying solely on the contents of the Calls table, you create
> problems. For example, if a NAS spontaneously reboots (as is happening
> frequently with the latest PM-3 code), the calls table isn't cleared, and
> Stop records never come. Anyone who was online at the time of the reboot
> CANNOT log on again until someone manually clears the calls table. Since
> these reboots are so common with the PM-3 these days, this is a Bad Thing
> (tm). Wouldn't using an SNMP check (or in the case of multiple NAS's, a
> series of checks) to determine if the user is online, and how many times,
> be better?
SNMP Concurrency checking has been in RadiusNT for quite a while. Its
in the documentation and there is a more detailed FAQ on the 2.5.162 and
later updates to it at /radiusnt/faqs on our website.
There should be code in RadiusNT that resets a server if its sends
the "00000000" AcctSessionID that a PM used to do. I'm not sure if the
later ones do this anymore, though. I'd be interested is anyone could
verify this by watching RadiusNT accounting when they reboot their PM3.
If you could send me the accounting request, I would appreciate it as
well.
If your NAS is rebooting that frequently, you need to call Lucent and
find out why. If Lucent can't fix it, find another vendor. NASes
should
NOT frequently reboot.
Dale