[c-nsp] Multipoint tunnels, routing protocols, default route

Bob Tinkelman bob at tink.com
Fri Dec 29 06:32:14 EST 2006


I'm trying to do something for the first time and could use
some help, probably just a pointer to a sample config or the
appropriate features and documentation.

A customer with T1s to two upstreams, and with the experience
of telco problems breaking both of them at the same time,
wanted to be able to use a cable-modem connection as a
backup, with some sort of tunnel to one of our routers.

I figured I'd try to set it up in a way that could support
similar future requests, and started experimenting with a
test router connected directly to a Time Warner cable modem
at my place.

I've included below the portion of the configuration related
to the tunnel. 

The test router is configured to run bgp over the tunnel,
announce a single network, and accept only the 0.0.0.0/0
route.  
 
All this works.  However, the configuration has (hard-coded
into it) the ip address of the default-gateway on the cable
company connection.  This is reasonable when the cable
company is providing a static ip address (as is the case for
my customer) but it's not reasonable with dynamic ip address
assignment (e.g., on my home test system).

  | ! cisco 2514 test system
  | version 12.2
  ...
  | interface Tunnel202
  |  description Dynamic multi-point tunnel
  |  bandwidth 1000
  |  ip address 69.48.189.10 255.255.255.0
  |  no ip redirects
  |  ip mtu 1416
  |  ip nhrp authentication xxxxxxxx
  |  ip nhrp map multicast 165.254.97.2
  |  ip nhrp map 69.48.189.1 165.254.97.2
  |  ip nhrp network-id xxxxxxxxx
  |  ip nhrp holdtime 300
  |  ip nhrp nhs 69.48.189.1
  |  ip nhrp server-only
  |  no ip route-cache
  |  no ip mroute-cache
  |  delay 1000
  |  tunnel source Ethernet1	! Cable-modem port
  |  tunnel mode gre multipoint
  |  tunnel key xxxxxxxxx
  ...
  | interface Ethernet1
  |  description Cable-modem via point-to-point Ethernet
  |  ip address dhcp client-id Ethernet0
  |  no ip mroute-cache
  ...
  | ! **** Temporary work-around ****
  | ip route 165.254.97.2 255.255.255.255 Ethernet1 24.193.28.1


When I first coded the configuration, I didn't have the
static route.  The router got a dhcp assignment on Ethernet1
with a gateway address, and the tunnel came up.  However,
when I turned up bgp, the test router received a 0.0.0.0/0
announcement, and that broke the tunnel because (I think) it
now thought that the route to 165.254.97.2 was through the
tunnel.  

The quick-and-dirty work-around was the insertion of the
static route.  And, for a site with a static ip address
cable-modem connection, this seems a workable solution.

This sort of "recursive" routing must be a standard issue
that people need to handle all the time.  I'm hoping that
there is something simple I missed.

I assume that the general idea of using a tunnel as a backup
connection is pretty common.

Did I just choose the wrong tool for the job?

Thanks in advance.
--
Bob Tinkelman   <bob at tink.com>   +1 718 464-4747


More information about the cisco-nsp mailing list