[c-nsp] Strange RIP behaviour

[Marco Matarazzo]

Hi all,

today I had a problem I cannot explain.
I'm using Cisco 3550-48-SMI with IOS  12.1(22)EA1 as access switches for our
customers. Each customer has a routed port on the 3550. The 3550 announces
the connected and static routes via RIP to a couple of 7206 (my borders).
The 7206 announce via RIP the default to the 3550. Everything works fine,
but today all of sudden, a static route was marked as "possibly down", and
my borders began to throw all traffic destined to that subnet to Null0 (as
designed).

This is the relevant sanitized config:

interface FastEthernet0/35
 description CustomerX PtP
 no switchport
 ip address 1.2.3.4 255.255.255.252
 service-policy input 2mbit-in
 service-policy output 2mbit-out
 no cdp enable

router rip
 version 2
 timers basic 3 10 0 10
 redistribute connected
 redistribute static
 passive-interface default
 no passive-interface FastEthernet0/32
 network 1.0.0.0
 neighbor 1.2.2.3
 neighbor 1.2.2.4
 no auto-summary

ip route 1.6.7.8 255.255.255.240 1.2.3.4

Normaly a show ip rip database shows:

1.6.7.8/28    redistributed
1.2.3.4/30 directly connected, FastEthernet0/35

During the outage (that affected only this customer out of 40 on this 3550),
it showed:

1.6.7.8/28    possibly down
1.2.3.4/30 directly connected, FastEthernet0/35

What strikes me is: how could it be possible to have a static selected as
possibly down, when the connected route is (and has always be) up? I could
ping the point to point fine from the router, no problems. The way to
recover was simply to shut/unshut the relevant interface. Any hints what
could be happened?

Thanks!
]\/[arco