[AusNOG] ICMP weirdness inside Internode's network

Lincoln Dale ltd at cisco.com
Fri Oct 28 17:37:28 EST 2011


On 28/10/2011, at 1:52 PM, Julien Goodwin wrote:
> My first guess is just broken ICMP handling inside MPLS tunnels, but without
> knowledge of Internode's network internals I can't make an intelligent suggestion.

not likely anything broken, more likely you ran the traceroute at a time when a lot of other people also happened to be running a trace route or something else which elicited ICMP responses from the router(s) in question.

in short, ICMP responses are ALWAYS rate limited to some sane packet/sec level by routers (pick your router vendor, they all do this).

its sort of in the same bucket as "why its not meaningful to ping a router" to determine round-trip.
a. the router control plane responses to that, not whatever the data plane, b. it often does it when it feels like it, i.e. low priority task,


cheers,

lincoln.


More information about the AusNOG mailing list