[nsp] Netflow config change : GSR stuck for a few seconds

Antoine Versini vox at t-online.fr
Tue Nov 4 09:11:00 EST 2003


I encountered a rather strange problem when doing some flow-export config
on GSR's running 12.0(25)S1. The gears have two GRP-B's running in the SSO
mode. Here are the configuration statements :

ip flow-export source Loopback0
ip flow-export version 5 origin-as
ip flow-export destination xxx.xxx.xxx.xxx pppp
ip flow-sampling-mode packet-interval 1000
ip flow-aggregation cache as

The routers allready had a minimal flow-export config, I just changed the
flow-export destination, changed sampling from 100 to 1000 and added
flow-aggregation cache as. «ip route-cache flow sampled» was not set on
any interfaces at this time.

After a few second changing the export destination, the CLI was stuck for
a few seconds (say 3 ou 4 seconds) and the router stoped forwarding
traffic. Then it came back to life, resuming normal operation with nothing
in the logging buffer. The only thing I can see is a CPU usage peak.

Then, I have set «ip route-cache flow sampled» without any side effect and
the router started exporting flows to the collector. On older IOS
versions, at this time I used to see the PSA of the involved LC
downloading a new microcode including netflow support (resulting in a very
short traffic disruption), but not this time.

What I don't understand is why only changing some general statements when
no interface was involved in flow-export made the router stop forwarding
traffic, even with Non-stop Forwarding configured and the GRP's operating

I've done the same changes on other GSR's, but I rerouted the traffic on
other parts of the network before, so the routers were not handling any
traffic anymore when I made the flow-export changes : no trouble, CLI not
stuck and routers answering to pings...

Antoine Versini
IP Networks Project Manager / T-Online France - Club-Internet

More information about the cisco-nsp mailing list