I seem to be having a strange problem. I have set up a CIPE connection between two Red Hat Linux 9 machines, one at work and one at home. I would like to access my home network from work and vice verce. My home network is NAT'ed behind a Linksys router. My home linux machine is my DMZ.

I have set up the CIPE connection and I can do everything I want to from work, even access my home computers by name using DNS on my home linux! The problem is accessing the work network from home. I'll handle the routing/forwarding later but I seem to be having a strange problem with a defualot route right now. If I ping work using the comand:

ping -I cipcb0 <ip of work machine>

I do just fine. However, if I add a default route using either

route add -host <work machine> dev cipcb0

or

route add -host <work machine> gw <ip of either end of cipe>

my ping stops working, whether or not I use the -I on the ping. Also, if I start a ping on the work computer it will go until I add the route on the home computer. It then stops until I delete the route.

Any suggestions would be appreciated.

If ping is working without adding this route, what are you adding the route to do?

Normally, ping wouldn't work, and you would add one or more routes to the routing table and it would start working.

If ping is working through whatever routes are there then it is probably that the new route you have added to your work machine is clobbering whatever route it was using before.

If you are trying to add a default route for your home PC through your work PC, or vice versa, then the command you want is probably:

route add default gw <ip of other end of cipe>

Perhaps I didn't explain what I'm trying to do very well. What I want to do is to communicate securely over the internet between subnets. A normal ping does not go through the tunnel, and hence is not secure. If I force the ping through the tunnel by specifying the CIPE interface (the -I cipcb0 on the ping command), the ping works. If I try to set a default route to the host on the other end (or the subnet) everything breaks. I would like to make all traffic between the two subnet go through the tunnel by default.

Have you tried / are you able to configure both ends to route to each other down your secure pipe?

If you set up one end, but not the other I can see how that would cause problems.

e.g. On home PC:

route add -net <work subnet> -netmask <mask> gw <work end of secure pipe>

On work PC:

route add -net <home subnet> -netmask <mask> gw <home end of secure pipe

I solved the problem! FYI you cannot make a route to the remote end through the cipe device. (Makes sense if you think of it - but then most things do in hindsight.)

what you have to do is set up a route to the subnet excluding the one ip address which is the internet connection on the other end. I.E.

route add -host 123.456.789.10 gw <addr of eth0>
route add -net 123.456.0.0/16 gw <addr of cipcb0>

Then everything works!

Thanks for the replies.