[f-nsp] XMR Software Upgrade Path from 4.0.1a

[Abello, Vinny]

You'd see this if your IP MTU's are dissimilar across your network anywhere.
This bit us because our links can vary in MTU size depending on the provider
of the circuit or link type. On our Cisco devices, they by default try to
use path-mtu-discovery to optimize the iBGP communication. This resulted in
MTU blackholes and stalled/flapping iBGP sessions on IPv6 between Cisco
routers where an XMR was in the path. It could affect many things depending
on your topology, configuration and applications, or you might not see it at
all.

Vinny Abello
Network Engineer
Dell | Physician Services
office +1 973 940 6125
mobile +1 973 868 0610
vinny_abello at dell.com


-----Original Message-----
From: igor at ergens.org [mailto:igor at ergens.org] On Behalf Of Igor Ybema
Sent: Wednesday, January 19, 2011 9:37 AM
To: Abello, Vinny
Cc: foundry-nsp at puck.nether.net
Subject: Re: [f-nsp] XMR Software Upgrade Path from 4.0.1a

Hi,

> We specifically have issues with path-mtu-discovery being broken in IPv6
due
> to the XMR not responding with icmp unreachables with IPv6. This is a
known
> issue according to Brocade and is fixed in later releases.

Didn't knew this. We do use IPv6 but haven't stumbled on a problem yet
with it. In which case should I see a problem with the
path-mtu-discovery not working?

regards, Igor
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 4999 bytes
Desc: not available
URL: <https://puck.nether.net/pipermail/foundry-nsp/attachments/20110119/81911acd/attachment.p7s>