[c-nsp] Route reflection - need second set(s) of eyes please

[Pete Templin]

All,

I'm trying to figure out some unexpected behavior relating to route 
reflection.  First, a diagram:

RAS/A1---Dist/A1---Core/A1---Core/B1---Dist/B1---RAS/B1

(City A and City B, one device of each type represented)

All routers shown above are in the same ASN.  As currently configured, 
the cores are fully-meshed.  The dists are neighbored with the cores, 
and the dists are route reflector clients of the cores (I guess from a 
configuration perspective, the cores see the dists as route reflector 
clients).  The dists are neighbored with the RASen, but the RASen are 
not currently considered route reflector clients.

To the best of my knowledge, since the dists aren't configured to treat 
the RASen as route reflector clients, they shouldn't be passing the RAS 
routes along to the core (who would then reflect them to the other 
city's core, who would then reflect them to that city's dists). 
However, it appears that the RAS routes are making it across the network.

Should I be confused?  Should I be configuring the dists to treat the 
RASen as RRC, to avoid future surprises?

Thanks,

Pete