[c-nsp] PFC3 and VPN CAM

[Saku Ytti]

After TAC and Account Team have been unable to get answer to my question
using cisco internal channels. I'd like to check if someone in community
is experiencing same symptoms, who could perhaps add pressure, or
someone from cisco could pick this up and find whats going on.
 The TAC case is/was 603584313 and was opened 1.5months ago.

To simply the issue lets just look at two boxes that have exact
same hardware, same IOS and very similiar configuration.
When issuing 'show platform hardware capacity pfc' I can see that
other one of them is populating VPN CAM and other one is not.

TAC suggested several reason why this could happen, we investigated
them all, and could not:
  a) get VPN CAM unpopulated in the other box
  b) get VPN CAM populated in the other box

Currently there is no business impact, but I expect that if I had
more considerable amount of traffic, I should notice and perhaps
have serious problems due to unnecessary recirculation.

Could someone running PFC3 and VRF's on more than one box 
check the above command to verify if VPN CAM is being
consistently either always populated or unpopulated?

And then second one for the architecture inclined people.
Could it not be possible to avoid recirculation of MPLS
explicit-null packets that are single label deep if VPN-CAM
would have entry for 'Internet VRF', that is special case
where single deep explicit-null would be considered VRF label
for 'Internet VRF'. With my limited knownledge of PFC3 
it would appear possible, but perhaps it would also mean
loosing the EXP value?

Thanks,
-- 
  ++ytti