PPP/ARP routing problem: using a PPP remote address inside an Ethernet subnet

PPP/ARP routing problem: using a PPP remote address inside an Ethernet subnet

Post by Andreas Meil » Sat, 08 Apr 2000 04:00:00



Dear Linux users

I have a small LAN with various Linux boxes. At the moment, they use
192.168.*.* addresses according to RFC 1597/1918. Current topology: Ethernet
LAN with 192.168.0.0/255.255.255.240 subnet. One of these Linux box runs as
default gateway with NAT (IP masqeurading), on another one, I set up a PPP
server which allow me to temporarily connect a guest PC (often Windows 98
boxes of friends) without Ethernet NIC to my LAN.

Currently, I assign the address 192.168.0.17 to the guest machine (=remote
IP for Linux's "pppd"). On the gateway Linux box, I added an appropriate
static route, so especially another Windows box inside my LAN is able to
"ping" to this guest box, because it uses the default gateway.

Meanwhile, I'm changing this Intranet into a part the Internet, that means
that I get a *16* address subnet from my ISP as well as a *preconfigurated*
router, so first, I need this *outside* of the regular subnet placed guest
PC IP address change *inside* the Ethernet subnet to save IP addresses (I
have 8 machines => only 11 of 16 addresses used). The challenging problem
for a TCP/IP guru of you is to do an "ipconfig add eth0" command in such a
way that this Linux box, where the PPP guest machine is connected, listens
(=answer the ARP query) to the guest PC's IP address too, but does *not*
process data itself, instead it forwards every packet to the guest PC.

In my company, I have a RAS account, which assigns me 10.11.3.98 as local IP
and 10.11.3.99 as remote IP but in the company's LAN, there's a
10.11.0.0/255.255.0.0 subnet defined, so this RAS offering Windows NT server
listens both to 10.11.3.98 as well as to 10.11.3.99 on the Ethernet NIC so
no static route is necessary for all other equipment in the LAN => a "ping
10.11.3.98" from *any* machine inside the company works correctly.

How can I do the same with Linux? Any solutions and hints are appreciated.
:-)

               Greetings from Switzerland
                           Andreas

 
 
 

PPP/ARP routing problem: using a PPP remote address inside an Ethernet subnet

Post by Piete » Sat, 08 Apr 2000 04:00:00



Quote:> Dear Linux users

> I have a small LAN with various Linux boxes. At the moment, they use
> 192.168.*.* addresses according to RFC 1597/1918. Current topology:
Ethernet
> LAN with 192.168.0.0/255.255.255.240 subnet. One of these Linux box runs
as
> default gateway with NAT (IP masqeurading), on another one, I set up a PPP
> server which allow me to temporarily connect a guest PC (often Windows 98
> boxes of friends) without Ethernet NIC to my LAN.

> Currently, I assign the address 192.168.0.17 to the guest machine (=remote
> IP for Linux's "pppd"). On the gateway Linux box, I added an appropriate
> static route, so especially another Windows box inside my LAN is able to
> "ping" to this guest box, because it uses the default gateway.

> Meanwhile, I'm changing this Intranet into a part the Internet, that means
> that I get a *16* address subnet from my ISP as well as a
*preconfigurated*
> router, so first, I need this *outside* of the regular subnet placed guest
> PC IP address change *inside* the Ethernet subnet to save IP addresses (I
> have 8 machines => only 11 of 16 addresses used). The challenging problem
> for a TCP/IP guru of you is to do an "ipconfig add eth0" command in such a
> way that this Linux box, where the PPP guest machine is connected, listens
> (=answer the ARP query) to the guest PC's IP address too, but does *not*
> process data itself, instead it forwards every packet to the guest PC.

You should enable bridging in your kernel, set the ethernet adapter of the
linux box to 'promiscue mode' (listen to all packets), and tell it to answer
ARP queries.
Anyway, read some bridging howto's. (I've not tried such a configuration
myself yet, but I think similar configurations are described there)

Good luck,

Pieter

 
 
 

1. another ethernet to ppp routing problem (different subnets)

Speaking about FreeBSD-2.2 June snapshot, Irix-5.3.

I tried. I searched through the mailing archive and through the
helpbook. And through the FAQ. No.

No, I'd like to avoid SOCKS{4|5} if possible (what happened to port in
-current, BTW?).
Yes, I do have the ip.forwarding set to 1 (on FreeBSD).
Yes, routed is running without flags (tried with -s too) (on FreeBSD).
Yes, the FreeBSD machine can see both networks -- entire world
through the ppp-link (199.232.254.68 -- stable IP address) and the
LAN through the ethernet card (222.222.222.22, yes, I will change
this later to smth reserved for local later).

Yes, all the machines on the lan can see the 222.222.222.22 interface
and have it as a default gateway.

My primary test machine is an IRIX box, which will not allow the
default to be set to anything no directly reachable.

`traceroute -n 199.232.254.68' shows 199.232.254.68 in the first line.
`traceroute -n my.ppp.peer.address' on SGI shows and 222.222.222.22
right away, then -- a bunch of `* * *'...

Anything else to set on the FreeBSD box? Other clients are NTs --
even less cooperative then Irix. Ideas? Thanks!

        -mi
--
        "Windows for dummies"

2. part. repeated during df -k

3. ppp: Cannot determine ethernet address for proxy ARP

4. utp crossover cable - 4 or 8 wires needed?

5. Routing from a local ethernet to a remote network via a PPP link-HOW ???

6. HP Laserjet II under HP-UX - accounting info required

7. PPP and PPP and PPP and PPP and PPP and HEADACHE

8. Closing TCP connection

9. Hard Drive Access will using PPP & PPP routing questions

10. ARP <---> PPP routing problems

11. PPP ping problem; PPP local address?

12. Peer is not authorized to use remote address - PPP problem

13. routing tabels loosing default route using PPP