[c-nsp] GSR CPU Process is very HIGH 95%

[Oliver Boehmer (oboehmer)]

Lasher, Donn <mailto:DLasher at newedgenetworks.com> wrote on Wednesday,
October 07, 2009 17:08:

> -----Original Message-----
> From: cisco-nsp-bounces at puck.nether.net
> [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Oliver Boehmer
> (oboehmer)
> Sent: Wednesday, October 07, 2009 6:22 AM
> To: bharath kondi
> Cc: BHARATH KONDI; bharath at vtelecoms.com.my; cisco-nsp at puck.nether.net
> Subject: Re: [c-nsp] GSR CPU Process is very HIGH 95%
> <snip>
>> hmm, your high interrupt load (the 71 in "CPU utilization for five
>> seconds: 95%/71%") suggests that the RP's CPU switches traffic, which
>> should not happen on a GSR as all traffic is switched by hardware..
>> It requires a closer look at the config and linecards to see what
>> could cause this. I'd contact TAC..
> 
> To clarify, this depends on both the card type (engine 0/1/2/3/4/5)
> and traffic type (mpls, DSCP marking, etc).  You can be doing
everything
> right, and still have 50% CPU with the wrong combination of those
> two.. (For example, MPLS Labeling and Engine0 GIG-E card at 100M of
> traffic) 

sorry, you are obviously right. I should have said "all traffic is
switched by the linecards". If the linecard forwarding hardware can't
switch the packet, it's punted to the linecard CPU, but should never be
switched by the RP's CPU. The only exception are packets switched out
the RP's Mangement-Ethernet ports (so this should be avoided).

	oli