[c-nsp] BGP "keep alives throttled do to tcp", MTU mismatch?

[Scott Granados]

Absolutely dead on, thank you so much for the nudge it pointed me in the 
right direction.

Here's what worked

using the show ip bgp nei a.b.c.d | I data I was presented with an MSS of 
1482
1482+40 = 1522.  Using the DF bit set and extended ping commands I was 
unable to pass data larger than 1504.  I assume this is because the mtu 1522 
command set on the interfaces was set and the medium between us can't pass 
larger packets.  I had increased the MTU to allow for VLAN tags based on an 
example I had googled.

I set ip mtu on the subinterface of 1500, reestablished the sessions and 
viewed the MSS again which showed 1460.  1460+40 = 1500 or MTU joy!

Sessions have been up for an hour now and traffic is passing.

Thank you and the others on and off list for pointing me in the right 
direction!

Scott



-----Original Message----- 
From: Olga
Sent: Friday, July 22, 2011 4:04 AM
To: cisco-nsp at puck.nether.net
Subject: Re: [c-nsp] BGP "keep alives throttled do to tcp", MTU mismatch?



> Sounds reasonable I can try to bring the link up with the loop unloaded to
> see if that helps but I still sort of feel possible MTU issues but can't 
> put
> my finger on it.  Some more googling shows that there can be issues if
> path-mtu-discovery and timestamps are enabled at the same time but I don't
> have a tcp timestamps entry in the global config.

This is 99% MTU issue. Do

show ip bgp nei x.x.x.x | i Data

on both sides.

You will get something line this "Datagrams (max data segment is 1460 
bytes):"

This is the MSS neighbors negotiated while opening BGP session.

Add 40 byte to it and try to ping  your neighbor with this packet size (with 
df bit to be completely sure). If it fails, lower mtu and reset session.

_______________________________________________
cisco-nsp mailing list  cisco-nsp at puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/