[c-nsp] CEF problems when adding IPv6 to SVIs on 6500/sup720/SXI3

[Phil Mayers]

All,

We've seen this several times now - just wondering if anyone can give me 
some pointers to troubleshoot it.

In brief: we have a 6500/sup720 running SXI3. It runs MPLS and L3VPNs, 
with hundreds of SVIs, most of them in one VRF or another.

We're rolling out IPv6 one SVI at a time, and after a change we're 
seeing CEF problems on *unrelated* SVIs - sometimes in a VRF, sometimes 
non-VRFed.

For example this morning we did:

int VlanXX
  vrf forwarding YYY
  ! ipv4 is already configured; now add:
  ipv6 nd prefix 2001:db8:1::/64  900 600
  ipv6 nd router-preference High
  standby 1 ipv6 autoconfig
  standby 1 preempt delay reload 180
  standby 1 authentication XXX
  standby 1 track 100 decrement 4
  standby 1 track 101 decrement 4

...and we then started to get reports of problems on unrelated, 
unchanged SVIs in the default VRF e.g.

interface Vlan1934
  ip address 193.61.68.41 255.255.255.248
  ip verify unicast source reachable-via rx
  no ip proxy-arp
  ip flow ingress
  arp timeout 1200

The symptoms were basically very high traffic loss. I suspect what was 
actually happening was punt to CPU, with the traffic loss caused by our 
CoPP policy.

A "shut / no shut" of the unchanged SVIs fixed the problem. I had to do 
this on 6 of them.

Any ideas? There are probably a couple of unused backup links on the box 
which I didn't "fix" via shut/no-shut. How can I tell if their 
forwarding entries are broken?