[c-nsp] possible CEF issue?

[Oliver Boehmer (oboehmer)]

Hi,

> I have two 7206VXR Routers w/NPE-400s connected via FastEthernet. 
> The configuration is as follows: 

> [Router2]
> 
> interface Loopback0
>  ip address x.x.149.251 255.255.255.255
> !
> interface FastEthernet1/1
>  description Connection to Router1 FastEthernet1/0
>  ip address x.x.149.30 255.255.255.252
> 
> Both routers run IOS 12.2(17a), CEF, EIGRP and an iBGP peer.
> From each router, I can successfully ping the attached interface of
the other.
>> From Router1 however, I can not ping the Loopback interface of
>> Router2. From Router2, I can successfully ping the Loopback of
>> Router1.  PINGS 
> from Router1 to the Loopback interface of Router2 receive a response
> from an IP address attached to a different interface
> (FastEthernet0/0.500) on Router1.

> For example:

> Router1#ping x.x.149.251
> Type escape sequence to abort.
> Sending 5, 100-byte ICMP Echos to x.x.149.251, timeout is 2 seconds:
> Reply to request 0 from host.domain-name.com (x.x.142.10), 1 ms
> Reply to request 1 from host.domain-name.com (x.x.142.10), 1 ms

This looks like the output of a broadcast ping, so router1 thinks it is
sending a broadcast. Strange..  

> Originally, the iBGP peer between these two routers used
> these Loopback
> interfaces as the update-source.  When I took these routers on, it was
> not working.  I added a new Loopback with a new IP address on Router2
> and changed the update-source on the two routers to band-aid the
> problem. 

does the "sh ip route" output looks any different for these two
addresses from router1's perspective?

> In my initial troubleshooting, the Loopback IP Addresses were known
> correctly via the routing protocol from both router's perspectives.
> However, from Router1's perspective, the CEF entry was not correct in
> fact, there was not much output.  It looked as if you were to
> do a "show
> ip cef 0.0.0.0"  Unfortunately, a "show ip cef x.x.149.251" appears
> normal now so I can't provide an example.

> Router1#show ip cef x.x.149.251
> x.x.149.251/32, version 45468681, per-destination sharing
> 925231 packets, 81667986 bytes
>   Flow: AS 0, mask 32
>   via x.x.149.30, FastEthernet1/0, 0 dependencies
>     next hop x.x.149.30, FastEthernet1/0
>     valid adjacency
>   925231 packets, 81667986 bytes switched through the prefix
>   tmstats: external 0 packets, 0 bytes
>            internal 925231 packets, 81667986 bytes
>   30 second output rate 0 Kbits/sec
> Router1#

looks good, what makes you think this looks like "sh ip cef 0.0.0.0"?

> Router2#show ip cef x.x.149.252
>x.x.149.252/32, version 19140878, per-destination sharing 
> 0 packets, 0 bytes
>   Flow: AS 0, mask 32
>   via x.x.149.29, FastEthernet1/1, 0 dependencies
>     next hop x.x.149.29, FastEthernet1/1
>     valid adjacency
>   0 packets, 0 bytes switched through the prefix  <--this doesn't
> look correct 

actually we only update the counters if cef accounting is enabled, so
this might be harmless.

Does it work when you disable cef? I'd be interested to see the full,
unsanitized configs as well as "show ip route eigrp" and "show ip cef",
you can unicast them directly to me..

	oli