not rewriting next-hop, pointing default, ...

Avi Freedman freedman at
Fri Sep 12 01:15:49 UTC 1997

> >> no neighbor
> > they should not care if you peer with them or not, they can have
> > the upstream provider to give them your routes, then:
> > !
> >  set nexthop

The danger with this approach is, obviously, that the router that you
try to do this to can go away.  In which case you shoot yourself in the
foot.  Some day, someone will send me a valid use of 'set ip next-hop'
but I haven't seen a good one yet.  

> The problem is not agreements.  The problem is, as Scott says, detecting
> violation thereof is not easy.  But traceroute -g is your friend.
> I also think it may be time we refuse to peer with anyone who inhibits LSR,
> as it seems that validation is now mandatory.  I think we should be sending
> out a "LSR is mandatory" notice to our peers.  Comments?

We have LSR turned off.

We periodically run XP mappers that insert temporary /32 routes to map
who sends traffic to whom, and to flag asymmetry so a human can look at
it.  A few routing engineers are on a mini-mailing-list that I have to
get that info, esp. wrt who sends next-hop to them and who defaults to
them.  (eunet defaulting into uunet is OK, anyone else isn't)

It's a perl/expect combo - no trace -g required.

> Smaller peers.  A non-trivial few are doing seriously disgusting things
> which are going to cost you a lot of pain and some cash.  There is little
> incentive for larger folk to peer with smaller ones.  Hence, larger peers
> will simply cut the smaller masses off rather than spending the resource to
> differentiate the bad apples from the good.  Maybe you want to do something
> about it.  And soon.

It's a danger, so the smaller peers should warn the larger peers and each
other when they see extreme funniness happening.

> randy


More information about the NANOG mailing list