> Well, there are systems out there that return ICMP destination unreachable
>messages.
> aviator:3 $ traceroute -n 204.128.145.3
>traceroute to 204.128.145.3 (204.128.145.3), 30 hops max, 40 byte packets
> 1 128.227.224.100 3 ms 3 ms 2 ms
> 2 128.227.162.1 3 ms 3 ms 3 ms
> 3 128.227.254.2 4 ms 5 ms 4 ms
> 4 128.227.252.10 10 ms 14 ms 31 ms
> 5 192.221.6.13 76 ms 56 ms 84 ms
> 6 192.221.48.1 179 ms 147 ms 130 ms
> 7 * 192.221.42.100 136 ms 157 ms
> 8 * * *
> 9 * * 192.221.42.100 168 ms !H
>10 * 192.221.42.100 87 ms !H *
>11 * 192.221.42.100 153 ms !H *
> And, yes, the packets do not bounce around FOREVER. Nor are they limited
>to 15 hops, or none of the providers in Gainesville (each of which is 16-22
>hops away from the University of Florida, also in Gainesville) would be able
>to reach the U. Traceroute often gives up around 30 hops.
Name: cpe1-fddi1-0.atlanta.mci.net
Address: 192.221.42.100
Ok, 192.221.42.100 looks like a backbone router for mci. In other
words, it doesn't have a route to the destination network at all. In
this situation, a Host Unreachable is perfectly acceptable and expected.
This is not the situation that was being asked about though. The
situation that was being asked about involved having the routes in the
backbone routers, so they would not give a host unreachable message.
If the destination network is running a routing protocol which is
capable of advertising that the eventual destination doesn't exist (ie
BGP4) then, yeah, you're going to get a Host Unreachable as well...if
things are statically routed though....its unlikely that you will get a
host unreachable (unless someone went in and blew away the static routes
when they found out the destination was unreachable...which is unlikely)
and you'll get the ping-pong.
-- Jeff McAdams | A feature is a bug IgLou Internet Services | with seniority. e-mail: jeffm@iglou.com | -- unknown