John & list, On 02/13/2011 07:58 PM, [email protected] wrote: > Regarding 1: Are you using ip tcp path-mtu-discovery? That can drastically > improve convergence time, as default neighbor MSS is around 500 bytes while > it can be improved to ~ 1440 in most environments.
As I read this thread I was reminded that I have the same problem with the PRP-2. Regarding the MSS, my eBGP peers are fine (they inherit probably the MSS from the interface to the exterior peer), but the iBGP sessions are all on 536 bytes. Configuring ip tcp path-mtu-discovery solves this, of course. Problem: I have a iBGP peer connected with one POS link (MTU 4470) and one Ethernet link (MTU 1500). Normally, the POS link has a better OSPF cost, therefore the iBGP sessions gets a MSS of 4430 (w/ path-mtu-discovery). Question: What if the POS links fails? If a bigger BGP update occurs, is the path-mtu-discovery "fast enough"? Does someone has experience here and can give me a recommendation? Thanks, -tim _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
