[c-nsp] frame-relay bug on C7609?

[Zitibake]

I was migrating some functions from a GSR (12.0(28)S1) to C7609 (12.2(18)SXD5).
 One application being moved was an OC12 Internet access link with frame encaps
(you can probably guess where that is going).  

With the OC12 moved over to the 7609, I could not ping the far-end of my
frame-relay static map; also, I could ping out the OC12 successfully provided
that the ping reply came back to my network via another path (so I know that I
could transmit OK out the frame-relay interface).  

No input packets were initially counted on the interface.  But (here's where it
gets weird), when BGP was enabled over the interface, eBGP adjacency came up
just fine.  I couldn't ping my BGP neighbor, but BGP was up.  BGP packets came
in OK, and incremented the input counters.  They seemed to be the only input
packets that were recognized on the interface.

So:
Transmited packets: OK
Received packets: 
-into-me ICMP echo-replies... no-go
-into-me BGP packets... OK
-switched/forwarded packets... no go

Reconfiguring the interface as PPP encaps worked fine.  I never liked using
frame-encaps there anyways, so I'll leave it at PPP.  I wonder if there is a
bug with forwarding/exception-path under frame encaps on
OC12/7609/12.2(18)SXD5?

Zitibake


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com