I have a.b.c.1 through a.b.c.6 as the ether addresses of the six PMs,
and the assigned addresses at a.b.c.33 and so on.
Everything works ok with routing off (presumably with ARP) and no
netmasks in the PMs. Everything still works ok with routing on, though
my Cisco sees a host route for each dialup user.
My first problem is with my static IP users. I have a block of
addresses allocated to them in a.b.d.160. With routing off, they can't get
past the PM they're connected to. With routing on, things seem to work
ok, but my Cisco sees something I don't understand:
R a.b.d.0 [120/1] via a.b.c.5, 00:00:16, Ethernet0
[120/1] via a.b.c.1, 00:00:03, Ethernet0
This is with two static users connected, one on PM1, one on PM5. I
don't understand why the PMs choose to generate a network route for
these users instead of a host route, while they're generating host
routes for the dynamic IP users that should have network routes.
Traceroutes from the Cisco to one of the static users show paths
randomly passing through either or both of PM1 and PM5. Traceroutes
from hosts on the LAN give only "*", even though said hosts can reach
the users and vice versa.
If I add a 255.255.255.224 netmask to PM1 and PM5, the Cisco now sees
the same thing, only with a.b.d.160 instead. And routing stops
working.
How can I convince the PMs to generate host routes for my static IP
users, and network routes for my dynamic IP users?
-- John W. Temples, III || Providing the first public access Internet Gulfnet Kuwait || site in the Arabian Gulf region