On Mar 26, 2009, at 3:41 PM, Robert Raszuk wrote:
John,
> The fact is that some RFC 1966 (and pre RFC 1966) RR implementations
> *did* reflect routes back to the client, and relied on the client to
> suppress them, because of this:
>
> A BGP speaker SHALL NOT install a route with itself as the next hop.
Isn't the below quote you cite talks about EBGP cases ?
No, it's not conditional. (Section 5.1.3 if you want to check.)
Also, regardless of what the spec says, it would seem pretty clear
that it never makes sense to install a route with yourself as the next
hop, whether EBGP or IBGP. (Modulo the issues discussed in the
ACCEPT_OWN spec -- in that case the caveat is that the notion of
"yourself" is a little tricky since multiple forwarding contexts are
in use.)
For IBGP AFAIK no one was at that time doing next hop self on ASBRs
hence just per the below sentence reflected routes would end up on
the ASBRs.
In fact 1771 says just below your quote:
" When a BGP speaker advertises the route to a BGP speaker located in
its own autonomous system, the advertising speaker shall not modify
the NEXT_HOP attribute associated with the route."
Sure, but I don't see how that's material to the case under discussion.
--John
_______________________________________________
GROW mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/grow
