Re: as4bytes - 4byte speaker receiving new* attributes?

[Jeffrey Haas <[e-mail]>]

On Tue, May 08, 2001 at 10:09:54AM -0700, Enke Chen wrote:
> > In the current as4bytes draft (draft-ietf-idr-as4bytes-02.txt):
> > :    The new attributes, NEW_AS_PATH and NEW_AGGREGATOR should not be
> > :    carried in the updates between NEW BGP peers. A NEW BGP speaker that
> > :    receives an UPDATE message from a NEW BGP speaker, with the
> > :    NEW_AS_PATH attribute carried in the UPDATE message must ignore the
> > :    attribute. The same applies to the NEW_AGGREGATOR attribute.

> There is no need to send a notification in this case as the receiver can deal with
> the message and function properly. It is consistent with the famous protocol
> rule: "Be conservative with what you send, and be liberal with what you accept".

I'm picturing two cases here:
1. The receiver gets a route containing both an as_path and a new_aspath.
   Since both speakers are 4byte speakers, AS_PATH should contain
   the same things as NEW_AS_PATH.  What if they don't?
2. If we simply ignore it, its still transitive.  When this finally
   gets to a 2byte boundary (possibly several AS's over), should
   this speaker discard the NEW_AS_PATH prior to generating a new one?

It is my general impression that the NEW_* path attributes aren't
really meant to exist within a 4byte domain and are only added
on egress into a 2byte domain.  Perhaps I'm mistaken.

> -- Enke

-- 
Jeff Haas 
NextHop Technologies