[c-nsp] 6509 broken netflow

[Jon Lewis]

On 2 6509s with very similar configs and traffic, one is giving me 
reasonable netflow data while the other gives what looks like 0.5% sampled 
netflow.  Back when I was first configuring netflow export on this pair of 
6509s, I had a feeling at one point after making changes to the flow 
aging, that netflow had become "sampled" on both units, but for what I was 
using the data for (abuse tracking) it generally didn't matter and I 
didn't pursue the issue.  After one of them rebooted recently, whatever 
was broken got fixed by the reboot.  With no change in config, suddenly it 
was producing about 200x as much netflow data.

Are there known issues in 12.2SXD that can cause this behavior?  Is there 
anything I can do, short of a reboot, that might fix the sampled one?

On the one that's working:
#show mls netflow ip count
Displaying Netflow entries in Supervisor Earl

  Number of shortcuts = 70629 (bounces around 60000-80000)

The other:
#show mls netflow ip count
Displaying Netflow entries in Supervisor Earl

  Number of shortcuts = 74 (stays in 50-100)

Each is configured with:

mls aging fast time 8 threshold 3
mls aging long 480
mls aging normal 32
mls flow ip interface-full
mls flow ipx destination
mls nde sender version 5
mls rate-limit unicast cef receive 10000
mls qos
ip flow-export source Loopback0
ip flow-export version 5 peer-as
ip flow-export destination <IP> <PORT>

----------------------------------------------------------------------
  Jon Lewis                   |  I route
  Senior Network Engineer     |  therefore you are
  Atlantic Net                |
_________ http://www.lewis.org/~jlewis/pgp for PGP public key_________