That routes detail command doesn't really give me that much extra info over the other one. Could it be that my import at RIPE for my customer AS is set to action pref=100 instead of 120 is causing this? Regards, Rens -----Original Message----- From: Richard A Steenbergen [mailto:ras@e-gerbil.net] Sent: jeudi 17 septembre 2009 8:33 To: Rens Cc: nanog@nanog.org Subject: Re: BGP question On Thu, Sep 17, 2009 at 06:49:37AM +0200, Rens wrote:
*>i customer_range Local IX 110 0 XXXX YYYY i * customer_range upstream provider 100 0 AAAA BBBB XXXX YYYY i * customer_range my customer 0 120 0 YYYY YYYY i
Last update to IP routing table: 0h23m34s, 1 path(s) installed: Route is not advertised to any peers
...
But it seems to choose the set the one as local IX as best.
When I disable local IX it sets the one with my upstream provider as best.
Shouldn't it set the customer_range I receive from my customer as best since it's highest local pref and advertise it to my upstream provider peer?
Yes, which means there must be something else wrong with that route to keep it from getting installed as best path. That CLI output smells like Foundry, and I don't remember exactly how it would show up, but on Cisco for example "show ip bgp" not only shows your rib but also the adj rib in for neighbors which have soft-reconfig enabled. This would show up as (received) but not say "used", though again I'm not sure if Foundry does this or how it would show up. If you aren't doing something silly like outright rejecting the route in the route-map or prefix-list on the neighbor, make sure the next-hop is valid and reachable. IIRC Foundry doesn't do bgp next-hop recursion by default, you have to manually enable it. If I remember my Foundry BGP correctly (despite many years and a lot of therapy trying to repress those memories) you can see more details with "show ip bgp routes detail x.x.x.x". -- Richard A Steenbergen <ras@e-gerbil.net> http://www.e-gerbil.net/ras GPG Key ID: 0xF8B12CBC (7535 7F59 8204 ED1F CC1C 53AF 4C41 5ECA F8B1 2CBC)