[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

23456 without AS4_PATH?

These prefixes all appeared with this problem late last December: 35320 196629 23456 35320 196629 23456 35748 35320 196629 23456 35748

The ill side effects of the AS_CONFED_SEQUENCE in an AS4_PATH and analysis
on what is going on were covered in excellent detail by Andy Davidson,
Jonathan Oddy, and Rob Shakir:
- NANOG thread: http://www.merit.edu/mail.archives/nanog/msg14345.html
- NANOG45 presentation: http://www.nanog.org/meetings/nanog45/presentations/Monday/Davidson_asn4_breaks_light_N45.pdf
- AS4 Wiki: http://as4.cluepon.net/index.php/Operational_Issues#AS_CONFED_SEQUENCE_in_AS4_PATH

Numerous attempts to contact AS 35320's NOC and peering folks about the
problem by several people have pretty much been ignored. looks like it just showed up with a broken AS path in the
past couple of days.  We'll probably see it a lot more regular invalid uses
of 23456 in the future... I mean, how often does someone leak a private
ASN :-)?  Perhaps it is a good idea for router and routing software vendors
to add 23456 to "neighbor remove-private-as".

Incidentally, while RFC 4893bis will include better error handling
for 32-bit ASNs, a new I-D to suggest better error handling for
all optional transitive attributes was just released yesterday


Greg Hankins <ghankins at mindspring.com>
+1 404 542 5530