routin loop

Stephen Stuart stuart@pa.dec.com
Wed, 18 Dec 96 08:14:26 -0800


> Stephen,
> 
> Those two loops that I originally posted are fixed. But there is
> another one:
> 
> traceroute -ip6 5f00:3100:8106:3300:0:c0:3302:5a
> [...]
>
> Looking at NIST's stats on the 6bone home page it seems that
> very few 6bone nodes are reachable from NIST.

I pointed my NIST route at NRL, and got there in 3:

% traceroute6 5f00:3100:8106:3300:0:c0:3302:5a 
traceroute to 5f00:3100:8106:3300:0:c0:3302:5a (5F00:3100:8106:3300::C0:3302:5A), 30 hops max, 24 byte packets
 1  gw.ipv6.pa-x.dec.com (5F00:2100:CC7B::12:0:F842:142C)  3.908 ms  1.954 ms  1.954 ms
 2  buzzcut.ipv6.nrl.navy.mil (5F00:3000:84FA:5A00::5)  113.251 ms  108.336 ms  109.312 ms
 3  ipng9.ipng.nist.gov (5F00:3100:8106:3300::C0:3302:5A)  125.904 ms  *  126.88 ms

> I beleive that the problem stems from the fact that we now
> have a combination of rip and static routing on 6bone.  This
> is a sure recipe for loops.  A correct model would be to
> form a backbone of routers that use only rip to pass reachability
> information among themselves and use static routes only to
> point to their leaf clients.

Something like that, to be sure; much of IETF week is a blur to me,
but I have a vague recollection that the opposite was proposed? A
static routed core, with RIP to leaf nodes?

Stephen