[j-nsp] bgp.l3vpn.2 and target VRF?

[Dave Diller]

On May 20, 2008, at 4:46 PM, Dave Diller wrote:

> Normally, the system uses the vrf-target to import routes from
> bgp.l3vpn.0 to the correct VRF.inet.0.  This is working fine for
> unicast routes.  I'm trying to do the same now to get routes from
> bgp.l3vpn.2 to VRF.inet.2
>
> Routes are received correctly in bgp.l3vpn.2 from the peer, are
> active, have the correct route-distinguisher, and the correct vrf-
> target, but never make it into the destination table.
>
> Am I missing something additional needed here?  It all "looks" ok to
> me...

It all *was* OK, too, that's the annoying bit of it.

I copied the VRF over to a new VRF, changed the target and  
distinguisher, committed both ends and bounced the BGP for good  
measure.  It all worked fine.  VRF.inet.0 and VRF.inet.2 populated as  
expected.

So then I deactivated the original problem VRF on both routers,  
committed, rolled back, committed, and cleared BGP again. Now IT works  
too.  Everything where it should be.

Something got stuck.  Grr.

I wonder if a 'commit full' would have helped.  Didn't think of that  
until now ;)

-dd