This is probably a really dumb question, but.....

I've got a RedHat 7.3 machine with two NIC's installed.

eth0 has a fixed ip address of 192.168.1.2
eth1 has a fixed ip address of 192.168.67.1

The two cards are connected to one another with a CAT 5 cross-over network cable.

I'm trying the following ping commands....

ping -I eth0 192.168.67.1 and
ping -I eth1 192.168.1.2

Both fail. Am I missing something basic? Does the machine require additional routing information? I have already turned on IP forwarding.

Any help appreciated!

-Chris

post the output of ifconfig and netstat -ar please

By the way, are you trying to create some sort of endless packet loop?

I tried your command on my setup and it failed also, I think ping -I (a capital letter i) expects an interface address, not an interface name. Try:

ping -I 192.168.1.2 192.168.67.1

and vice-versa.

-Bob

Bob,

Thanks you're right. Specifying the interface as an IP address does the trick.

I've got an ADSL router on the 192.168.1.X network with an IP address of 192.168.1.1 when I ping this from the 192.168.1.2 interface I get replies OK, however if I ping it using the 192.168.67.1 interface I get nothing back. Any thoughts?

I've included the netstat output below.

[root@ph-firewall root]# netstat -ra
Kernel IP routing table
Destination Gateway Genmask Flags MSS Window irtt Iface
192.168.1.0 * 255.255.255.0 U 40 0 0 eth0
192.168.67.0 * 255.255.255.0 U 40 0 0 eth1
127.0.0.0 * 255.0.0.0 U 40 0 0 lo
default 192.168.1.1 0.0.0.0 UG 40 0 0 eth0

Well, your default route may be overriding the interface specification in the ping command. Try deleting your default route while pinging on the "wrong" interface. Or better, put another machine on the .67.x net and ping it.

-Bob

Bob,

If I put a machine on the 192.168.67.X network with an address of 192.168.67.10 and try and ping it from the 192.168.1.2 interface then that's OK.

If I delete the default route in the routing table I'm still not able to ping 192.168.1.1. This is now really starting to anoy me!

In summary then...
From eth0 (192.168.1.2) I can successfully ping the following addresses
192.168.1.1, 192.168.67.1, 192.168.67.10

From eth1 (192.168.67.1) I can successfully ping these addresses
192.168.67.10, 192.168.1.2

What will NOT work is pinging 192.168.1.1 from eth1.

Is it possible that it's some configuration of the router not responding to networks outside of the 192.168.1.0 network?

-Chris

Yes, it's possible that the router is either
a) ignoring a ping originating from a private subnet it is not part of,
or
b) attempting to reply via a different route which does not return to the originator.
But, I have to say, I think your NICs appear to be working fine. Why are you trying so hard to route on the "wrong" subnets?

-Bob

If you look at my other post a bit further down (regarding topology) you'll see that I'm trying to setup a home network.

The problem was that.... It didn't work! Whilst the Linux box connects quite happily to the internet the clients on the 192.168.67.X network were failing. This exercise was an attempt to break things down into manageable chunks in order to find out where the problem was.

I'm convinced that the problem is the router either a) It can't perform NAT for the .67 network or b) it doesn't know how to route the traffic back to the clients.

The topology I went for was based around the need to ensure that the network was firewalled by the Linux machine (firewall is not present at present in order to avoiding complicatiing things!) and I thought in order to do that I had to ensure that every IP packet passed through that machine.

I'm starting to wonder if this could have been achieved by having everything on the 192.168.1.X network and the use of routing tables? Any thoughts?

-Chris

OK, I see your other thread. Maybe I'll chime in over there.

-Bob