> On an LOCAL IRX-111, I have set up three frame-relay PVC's so far...for the
> sake of argument assume all remote networks have subnet .248 (30-hosts).
> All of these sites are remote POPs with one subnet going to the Ethernet
> interface so that office/school machines may connect to the PM's ethernet
> port, and another subnet is used by the modem bank (PM2ER-30).
>
> Now, one of these remote locations needs an *additional* 30 IP addresses
> routed to the Ethernet interface! Because I anticipated this, I had left a
> contiguous subnet to the one he's using open (Had I *REALLY* known that
> they were definately going to use the additional 30 IP's I would have put
> them on a .192 subnet, but that's a different story). I *thought* that
> even though the PM's don't do VLSM that I would be able to route this
> second subnet to it via static routing. In fact, a traceroute of an IP
> address in the second subnet DOES get to the remote PM, but the customer is
> unable to see it or get any machines in the second subnet to see the PM
> (who's ethernet address is in the first subnet). Is there anything I can
> do to get the second subnet recognized by computers on the remote PM's
> ethernet port???
>
> GORY DETAILS:
>
> On the IRX (local):
> add netmask XXX.YYY.ZZZ.0 255.255.255.248
> route add XXX.YYY.ZZZ.64 XXX.YYY.WWW.2 1 (ethernet subnet...this works)
> route add XXX.YYY.ZZZ.32 XXX.YYY.WWW.2 1 (modem subnet..also works)
>
> route add XXX.YYY.ZZZ.96 XXX.YYY.WWW.2 1 (SECOND SUBNET of 30 IP's)
> Traceroutes to XXX.YYY.ZZZ.97-126 DO make it to XXX.YYY.WWW.2!!
>
> I have also tried
> route add XXX.YYY.ZZZ.96 XXX.YYY.ZZZ.65 (the IP address of the remote PM
> ethernet interface)
>
>
> I entered the same netmask table and routing commands on the remote PM as
> shown above...no dice. I cannot ping/traceroute to any IP on the second
> subnet. Do I have to renumber the original 30 IP's to a .192 network???
> Just say NO! :-) Is this an error in PC configuration or a limitation
> of RIPv1??
Its a limitation of RIPv1 and the fact that even with static routes the
Livingston and more importantly the hosts on the other end are probably
treating the two ethernets as completely separate networks (which they
are) and don't support/understand variable subnets. It is analogous to
trying to carry 2 Class C's on the same wire, both networks will function
fine but won't see each other. Why? Because when one machine doesn't know
(or forgets) the ethernet MAC address of another machine what does it do?
It sends out an ARP broadcast message to the net (in this case subnet)
using its own network's broadcast address which the other network's
machines aren't listening to. The remote Livingston has static routes in
its tables to route both subnets to its ethernet port I assume. It may or
may not route in this case, since all traffic for one subnet would come in
and the have to go out the same interface as destination. If this doesn't
work you might consider setting up a linux or whatever gateway machine
that has IP forwarding turned on, IP aliasing for both subnets, gated,
and/or proxy arping for each subnet. A little more of a pain, but you
have finer grain control over what is going on with routing.
You were clear that your packets were getting to the remote portmaster,
you weren't clear in whether the remote portmaster knew what to do with
the packets once it got them. Does it have the proper static route
tables?
*******************************************************************
Stephen Zedalis Exis Net Inc.
System Administrator Hampton Roads' Premier Internet Provider
tintype@Exis.Net Web Page: http://www.Exis.Net
Email: support@Exis.net
*******************************************************************