[c-nsp] PPPoE Session Limit?

[Tim Franklin]

> Holy cats - thanks Eugene and Joe.  That was it!
> 
> So, what's the relations between having the address on the 
> loopback and the
> '5' limit?

I've seen a very similar thing on (badly-designed) ISDN backup with an
address and what's though to be an appropriate netmask on the virtual
template at the site receiving the calls.

The first four remote sites to dial in work fine, all subsequent sites get a
connection, but no routing - although in this case the remotes all have
hard-wired addresses (still with the covering netmask for the whole pool),
so the central site isn't trying to assign anything.

What *looks* to be happening is that you end up with four virtual-access
interfaces, each with the same address and netmask on, and CEF load-balances
across them - then you add "too many" interfaces with the same address /
netmask and it ignores the new interfaces.  Designing the thing properly
up-front to use 'ip unnumbered' fixed it, so I didn't dig a lot further into
what was happening underneath.

I think there's a tendancy for people to try and treat (and hence configure)
virtual-templates like point-to-multipoint interfaces, rather than
remembering that lots of identical point-to-point interfaces are going to be
created.  Maybe a "design hint" somewhere to make it clear that they're not
would help?

Regards,
Tim.

-- 
____________   Tim Franklin                 e: tim at colt.net 
\C/\O/\L/\T/   Product Engineering Manager  w: www.colt.net 
 V  V  V  V    Managed Data Services        t: +44 20 7863 5714 
Data | Voice | Managed Services             f: +44 20 7863 5876