(usagi-users 01208) Re: Is this a bug?

[Pekka Savola <pekkas@xxxxxxxxxx>]

On Thu, 24 Jan 2002, Nathan Lutchansky wrote:
> > How badly would you have to much up your routes to get into this 
> > situation?  How many *real* (non-erroneous) usage scenarios are there?
> 
> This should apply to any p-t-p link; I see no reason to assign global 
> addresses on p-t-p interfaces if the nodes on each side have global 
> addresses assigned on other interfaces.  -Nathan

Note that this question was about the global destination address: I don't
think it's necessarily the best practise to select a link-local source for
that.

Issues:
 - if you try to reach the either end of p-t-p, link-local source is fine
 - if you try to reach any other address, link-local is not fine (would be
silently discarded when forwarding unless explicit route is added to e.g.
loopback address in the sending node).

-- 
Pekka Savola                 "Tell me of difficulties surmounted,
Netcore Oy                   not those you stumble over and fall"
Systems. Networks. Security.  -- Robert Jordan: A Crown of Swords