router working can't piing internet gw

[mbx_9_99]

Hi,
RH ES 2.1. 2 Ethernet NIC's.

I've got a rh box sitting between a 10.10.11.0/24 network and 10.10.12.0/24 network. I've echo'd 1 into ip_forward & in general the rh box is routing nicely. I have one problem with the internet connection - from the 10.10.12.0 network I can't ping the internet gateway - some details...;

The 10.10.11.0/24 network has an internet gateway (cisco pix) at 10.10.11.97/24 - all the machines on this network are fine surfing the net using this as their default gateway.

The rh box has two ethernet interfaces (10.10.11.88/24- eth0) and (10.10.12.97/24 - eth1).

I've added a route to machines on both networks pointing to the appropriate sides of the rh router to allow them to see the other network (the pix can't route internally and until I get things stable I don't want to make the rh box the site default router - although that is a longer term plan).

We're running various applications from the 10.10.11.0/24 to the 10.10.12.0/24 network - they all run ok (http/https/terminal svcs/ms-file & print/etc).

*My Problem* - from machines within 10.10.12.0/24 I can't ping the PIX on 10.10.11.97/24 and therefore can't get out to the internet! But they can ping any other address on 10.10.11.0/24

I've got to a point where I've confused myself about what the default gateway for the interfaces of the rh box should be. The 10.10.11.88/24 interface is pointing to the PIX as it's def gw (on 10.10.11.97/24) & it can ping the PIX. I'm confused what the gw for the 10.10.12.97/24 interface should be - should it be the PIX or the 10.10.11.88/24 address which is the other interface of the same machine?

All the machines within 10.10.12.0/24 have a default gateway of 10.10.12.97 and a route defined for 10.10.11.0/24 to be 10.10.12.97. I didn't seem to get any connectivity until I defined the routes on the machines..although I would have thought the def gw would have done it.

Ok, that was a lot of info. To summarise;

- If the def gw for a 10.10.12.0/24 machine is the 10.10.12.97/24 interface of the rh box then should it really need me to manually add a route to reach 10.10.11.0/24? I sort of think no.

- Why can't machines in 10.10.12.0/24 ping 10.10.11.97/24 when they have no problem ping'ing any other address within 10.10.11.0/24?

I've seen something about proxy arp needing to be enabled??

Anyone got any views??

Ta

[JohnT]

If the def gw for a 10.10.12.0/24 machine is the 10.10.12.97/24 interface of the rh box then should it really need me to manually add a route to reach 10.10.11.0/24? I sort of think no. Have you even tried it?

[mbx_9_99]

When I put the router in at first I wasn't getting icmp through it so i put the routes on machines at both ends and it started working.

I've just added a bunch of new 10.10.12.0/24 machines so will try on those machines again just in case.

Would you agree though that it's not needed?

Still don't get why even having the additional route would stop traffic getting to the pix which is just one addres within an otherwise ok 10.10.11.0 network.

Unless - does the PIX need a route to be able to send traffic back to 10.10.12.0/24?? I would have thought the source address in the icmp packet would have been enough to send the echo reply back to the right address (which would be the 10.10.11.0/24 side of the rh box which would in turn pass the packet to the original source within 10.10.12.0/24).

Ta