> On 03/06/11 12:04, [log in to unmask] wrote:
> 
> >> ...but I would not normally expect them to advertise the Local route.
> >> Certainly our 6500s (12.2(33)SXI5) are not.
> >
> > Certainly, and it doesn't happen in IPv4.  We could filter out the local
> > /128 prefixes on that peering, but it just seems wrong.
> 
> For the avoidance of doubt: you shouldn't need to filter it. It 
> shouldn't be there, period. It's not in our routers (maybe an IOS bug?)

I've checked the release notes and I can't see anything that seems to
match, though I can't get at the bugs themselves at the moment (Cisco
account management - grrr).  CSCsy32000 and CSCsz61156 are the only ones
that seem even vaguely relevant.  I'll get a colleague to look at them
next week.  I'll try putting filters in anyway to see if that's a usable
workaround.

> >> Can you send:
> >>
> >> sh run part router bgp
> >> sh run int<the p2p>
> >
> > I'll do that under separate cover.
> 
> I can't see anything obvious in the config, and I can't reproduce this 
> on the bench with a pair of SXI5 6500s. The config I'm trying is not 
> exactly the same, but it's not wildly dis-similar.
> 
> One thing I do see is that "next-hop-self" doesn't seem to be having any 
> effect; the routes my "test" router receives have, as I would expect in 
> iBGP, next hops from the IGP.
> 
> I'm wondering if something is different about the way IOS handles this 
> config between IPv4 and IPv6, and that's what's causing it.
> 
> You might try asking somewhere like cisco-nsp or of course whoever 
> provides your support.

This isn't the kind of issue where asking our support people to help
generally gets us very far - reaching people with enough expertise far
enough into Cisco takes an awful lot of tenacity and it's usually easier
to ask someone else.  I'll maybe hang around some of the Cisco forums
for a bit.

Thanks for your time,

Sam

-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.