IMO, bad negototiation messages are a bit more indicitave of a malfunctioning router that a bad prefix is, as it's unquestioningly something that was originated by the router in question, where a bad prefix could easily have originated elsewhere. Receipt of a malformed negotiation message should definitely be grounds for terminating the BGP session. Whether or not a BGP peer shuts down the peering session upon receipt of a bad prefix, it should definitely refuse to propagate the invalid data. The fact that Brand "C" routers propagated the bad prefix was the primary cause of what happened in October. -C On Fri, Jan 18, 2002 at 06:46:42PM -0800, Jake Khuon wrote:
### On Fri, 18 Jan 2002 21:39:10 -0500, Susan Hares <skh@nexthop.com> ### casually decided to expound upon Vijay Gill <vijay@umbc.edu> the ### following thoughts about "Re: Persistent BGP peer flapping - do you ### care? ":
SH> What else causes repeative peer bounces other than the broken prefix?
Well... I remember when bad capability negotiation messages would cause the session to drop. Although this is before any update messages were sent. However it still caused repeating session bouncing.
-- /*===================[ Jake Khuon <khuon@NEEBU.Net> ]======================+ | Packet Plumber, Network Engineers /| / [~ [~ |) | | --------------- | | for Effective Bandwidth Utilisation / |/ [_ [_ |) |_| N E T W O R K S | +=========================================================================*/
-- --------------------------- Christopher A. Woodfield rekoil@semihuman.com PGP Public Key: http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xB887618B