Peter Rathlev wrote:
On Wed, 2009-02-11 at 12:31 -0500, Paul Stewart wrote:
I just checked some peers at random here on 6500 and 7600's - they all
defaulted to 1440 in an Ethernet environment...
And they're not directly connected eBGP peers? Those default to using
the interface MTU. Internal neighbors should default to 536 (or 516 or
thereabout) AFAIK. At least that's how it's been for our 6500s running
SXF for years.
Strange though if the OP example of an eBGP peer is directly connected,
yet still uses 536 as default MTU.
My external peers are 1440. Almost all of my internal peers are 516 or
536. There are a few exceptions though. I have one pair of 7600s that
are 1440, me ME3750 that's also 1440, and one ME6524 that's 9132 (the
other ME6524 is 516 for some reason). Perplexing.
So how do one force iBGP sessions to choose a larger datagram size? Do
you have to do it in the BGP config with 'neighbor a.b.c.d transport
path-mtu-discovery' or is there a more automatic solution?
Justin
_______________________________________________
cisco-nsp mailing list [email protected]
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/
