We have a fault on one of our XMRs which is showing slighly odd FIB behaviour. For some reason the mgmt card is not properly installing all active routes on a particular XMR16. It has the full BGP table and should be sending traffic via LSPs directly to the edge routers.
I have another router connected directly to this one advertising a default into the IGP. When traffic hits the XMR and the line card doesn't know what to do with it, it's following the default route. The next-hop router then has LSPs to all edge routers and it's working perfectly fine. Some of those LSPs actually go back through the affected router but at that point it's only looking at the transport label. So I'm guessing depending on your topology you might still be able to forward traffic even when your fib is full, as long as that default route is there. Thanks Darren http://www.mellowd.co.uk/ccie Date: Wed, 5 Mar 2014 15:33:56 -0600 From: [email protected] To: [email protected] CC: [email protected] Subject: Re: [f-nsp] BigIron RX FIB Capacity Nick, We experienced the exact same problem that Josh mentioned in 2007 on an RX-16; the traffic to those destinations without a route got dropped. Jason On Wed, Mar 5, 2014 at 2:53 PM, Nick Hilliard <[email protected]> wrote: On 05/03/2014 20:17, Josh Farrelly wrote: > If you exhaust the max ip-route count the router will simply reject any > additional entries to the route table. so does traffic flowing through the box to those destinations: 1. get dropped with host unreachable 2. get punted to the cpu Nick _______________________________________________ foundry-nsp mailing list [email protected] http://puck.nether.net/mailman/listinfo/foundry-nsp _______________________________________________ foundry-nsp mailing list [email protected] http://puck.nether.net/mailman/listinfo/foundry-nsp
_______________________________________________ foundry-nsp mailing list [email protected] http://puck.nether.net/mailman/listinfo/foundry-nsp
