routing loops

[Tony Li <[e-mail]>]

   I've just reread the BGP-3 specification.  The general looping problem
   I am talking about is addressed in a more specific form in section 10,
   Detection of Inter-AS Policy Contradictions.  The last paragraph of
   that section says:"... the protocol itself does not provide any
   mechanism for suppressing the route oscillation that may result from
   these unsatisfiable policies".

   I don't mean to pick on BGP.  As I've already said, the architecture
   document discusses general distance vector algorithms for policy
   routing.  It's just that I don't see how you can say that this looping
   problem is not a problem in BGP, unless the implementation differs
   from the protocol specification.

Ding.  Congratulations, you win the prize.  Recall the
_implementation_ sections that I quoted you?  Now, from section 10
look at:

   In some cases, this may allow a BGP speaker to detect the fact that
   its policies, taken together with the policies of some other AS,
   cannot simultaneously be satisfied.  For example, consider the
   following situation involving AS A and its neighbor AS B.  B
   advertises a route with a path of the form <B,...>, where A is not
   present in the path.  A then decides to use this path, and advertises
   <A,B,...> to all its neighbors.  B later advertises <B,...,A,...>
   back to A, without ever declaring its previous path <B,...> to be
   unreachable. 

The sections that I quoted insure that this last action is not
possible.  If B advertises <B,....,A,....> back to A, that _implies_
that the previous path is no longer reachable.  A correct
implementation at A, then would have to remove <A,B,...> from its path
table.  No loop.  I will gladly stipulate that RFC 1267 is not
internally self-consistent in this case.

Of course, it may still oscillate.  Is this what you real complaint is?

Tony