[c-nsp] 7206 cpu utilization

Vikas Sharma vikassharmas at gmail.com
Wed Mar 16 05:01:00 EDT 2011


Hi All,

Another issue i am seeing on 7206

setup looks like Spirent ---- 12k XR --(10 gig link)--- CRS1 ---(10 gig
link)----- (l2 switch) --(1 gig link)---- 7206 npeG1 ---- Spirent

I have shaper on  CRS1 towards 7206 which shape 10gig to 300 mbps

RP/0/RP0/CPU0:crs1.rtr#sh run int Te0/1/2/0.52
Wed Mar 16 01:36:57.751 UTC
interface TenGigE0/1/2/0.52
 description
 service-policy output CR_QOS_10GB_SHAPED_300MB_TO_CR_SAR
 ipv4 mtu 1500
 ipv4 address 10.54.1.8 255.255.255.254
 arp timeout 1200
 load-interval 30
 dot1q vlan 52
!
RP/0/RP0/CPU0:crs1.rtr#sh run policy-map CR_QOS_10GB_SHAPED_300MB_TO_CR_SAR
Wed Mar 16 01:37:08.041 UTC
policy-map CR_QOS_10GB_SHAPED_300MB_TO_CR_SAR
 class class-default
  service-policy CR_QOS_300MB_TO_CR_SAR
  shape average percent 3
 !
 end-policy-map
!
Issue is when I am sending normal traffic i.e. within 300 mbps (even 500
mbps), things works fine and output of policy on crs interface looks fine
also. But as soon as i increase BE traffic (128 byte), cpu utilization on
7205 increase to 99% but on crs1 still I can see all traffic is in the
respective limit.

So my question is when I increase traffic (simulate attack) in BE class with
128 byte, why CPU on 7206 increase BUT on crs still traffic is bounded to
the assigned value in respective class.

any reaoson !!!

Regards,
Vikas


More information about the cisco-nsp mailing list