Plethora of UUnet outages and instabilities

From: Peter Cole <>
To:, Josh Beck <>
Subject: RE: Plethora of UUnet outages and instabilities
Date: Tue, 1 Jul 1997 11:34:42 -0500


Our customer support started getting complaints about customers not
being able to reach us(AS7201) from MSN dialup in Houston. The network was in my BGP tables but not my IP route table. I called
Cisco and after a little work they concluded that the reason the routes
were being removed was more specific routes &
inside the CIDR were removing the I didn't understand
the logic but a simple distribute list ignoring announcements for the
two class Cs seemed to solve the problem. Later I sent a note to
Sprint, MCI, MSN, & UUNET. Spint and MCI responded quickly stating that
Cisco was mistaken. Cisco wants to look into it further. I have herd
nothing from MSN or UUNET. I have to update the software on the router
for other reasons. I thought I would wait to see if the problem is
still present after my upgrade. I have no route Damping setup and can
think of no other explanation for this bizarre behavior.

My point is that maybe there are other reasons that so many ASs are
having problems with UUNET routes.

You can only blame UUNET for announcing 2 subnets of their Class B via BGP,
especially when there is no valid reason for them to be doing so... It will
not be long before most ISPs will be installing filters to deny subnets
of Class B address space (nothing less than /16).

You probably need "ip classless" command implemented on your Cisco 7206. The
Cisco "show ip route x.x.x.x" probably assumes classfull routing, and there
are bugs in it:

  > sh ip route
  Routing entry for, 2 known subnets

  B [210/1] via, 13:51:42
  B [210/1] via, 13:51:42

Note that is not a /24 (Cisco bug).

If you dump the full routing table via "sh ip route" you should also
find the aggregate classless route:

  B [210/1] via, 06:02:56

This should be used if you have installed "ip classless" on your router.