reminder of IPv6 Ops Topics meeting THURSDAY, 11:45-12:45, Regency Room prior to IPng

[Kunihiro Ishiguro]

Hi, this is Kunihiro Ishiguro.

>6bone cleanup/hardening
> Rob Rockell - HARDEN and routes (along with Ivano)

I know there are several ugly routes in current 6bone.  Definitely it
is an operational issue (announce routes which shouldn't be
announced).

Adding to that I'm thinking there could be an BGP-4+ implementation
issue.  Even though originator stopped the announcement, it seems that
some routes never disappear from the 6bone.

Let me show an example.  Todays BGP-4+ logging information at Merit
 shows that:

http://www.merit.net/mail.archives/html/6bone-routing-report/msg00596.html
==========================================================================
Poorly Aggregated Prefixes (>24 in 3ffe:0000::/17 or >28 in 3ffe:8000::/17):
Format: Prefix path AS-Path (Origin-AS -- Availability)
--------------------------------
  WIDE (3ffe:500::/24) had 18 route(s)
    3ffe:508:0:3::/64 path 1225 2547 559 5408 1752 3185 786 1849 2839 237 7680 ( -- 0%)
    3ffe:508:0:2::/64 path 1225 2547 559 5408 1752 3185 786 1849 2839 237 7680 ( -- 0%)
    3ffe:508:1::/64 path 1225 2547 559 5408 1752 3185 786 1849 2839 237 7680 ( -- 0%)
    3ffe:508::/64 path 1225 2547 559 5408 1752 3185 786 1849 2839 237 7680 ( -- 0%)
    3ffe:503:1050::/48 path 1225 2547 559 5408 1752 3185 786 1849 2839 237 7680 ( -- 0%)
    3ffe:508:5::/48 path 1225 2547 559 5408 1752 3185 786 1849 2839 237 7680 ( -- 0%)

I'm sure that originator already stopped the announcement.  So the bad
guy is not the originator at this moment.  I'm thinking there may be
an router which never forget BGP-4+ routes.

And worse, those routes are flapping!.  The logging shows it is 0%
life time.  The routes are withdrawn right after the announcement.  So
it is hard to find the problem.  When you type `show ipv6 bgp' like
command from your terminal interface, it will not be shown.  Let me
call those routes as `zombie routes'.

If we have time, I and Mr. Ikuo Nakagawa talk about this problem at
IPv6 Ops Topics meeting.
-- 
Kunihiro Ishiguro