cpu utilization high

nagalingammk · ‎12-23-2009

hi,

I am using 3750 stack switch of 4 where we are facing a issue with the CPU utilization. the utilization is going high to 50 - 70 %.

when i use the command "sh processes cpu history sor" the cpu utilization for hrpc<- response is more.

please let me know what does hrpc - response means and for what purpose its used for.

thanks

Naga

Reza Sharifi · ‎12-23-2009

Hello Naga,

There is list of items that can contribute to high CPU utilization. If your network has been running smoothly, I would focus on what config changes you have made recently.

Take a look at this link and see if any of them apply to you.

http://www.cisco.com/en/US/products/hw/switches/ps5023/products_tech_note09186a00807213f5.shtml

HTH

Reza

sachinraja · ‎12-23-2009

Hi Naga

Take the output of show proc cpu, and paste it in the output interpreter tool.. this can give you more info on what issues you could have... you can find output interpreter in Cisco URL:

https://www.cisco.com/cgi-bin/Support/OutputInterpreter/home.pl

or give us the "show proc cpu" command output and we would look for you.. As Reza pointed out, check for recent changes on your network .. check for abnormalities like spurious bandwidth, drops, logs etc.. "show log" on the switches and see if you find anything suspicious..

Hope this helps.. all the best..

Raj

nagalingammk · ‎12-23-2009

thanks Reza/Sachin,

please find the cpu output on the attachment.

sachinraja · ‎12-23-2009

Hi Naga

the output interpretor didnt give exact solutions because it saw all processes under 10 % (1 min interval).. are all the switches running 50-70 % CPU on the stack ? on the file you attached, we see that the SNMP process is upto 14 % .. this is quite common when the switches are polled for SNMP and on a 5 sec interval the CPU might go high... since SNMP is a low priority process, it will not affect other production traffic ! can you look at the "show log" on the switches, and let us know if you see something.. this looks a bit normal to me ( since the 5 min CPU is just 16 %), are you seeing any packet drops or performance degradation on the network ?

Raj

nagalingammk · ‎12-23-2009

hi raj,

you can find the total cpu utilization is going around 59%. for five seconds.

we are facing a problem that there is an delay in getting the snmp informaiton means its taking around 10 min if we need an snmp Uptime information from the switch

this is happening only when the cpu utilization is high.

sachinraja · ‎12-23-2009

Hi Naga

5 sec cpu utils, can just be spikes... when we see issues with components, the cpu utils are normally high for 1 min - 5 min interval (which looks good in your case).. as i see snmp does take some cpu cycles.. as said before snmp is by default considsered a low priority traffic and is processed only after other traffic is forwarded ... normally we dont force snmp higher than normal data traffic, since this is only used for management purpose (unless it is highly critical).. anyways you can look at the release notes of the software IOS you are running and see if there are any bugs related to snmp/cpu util.. if there are open caveats, then you might have to upgrade your IOS..

Hope this helps.. all the best..

Raj

nagalingammk · ‎12-23-2009

thanks raj, i will look into it.

Reza Sharifi · ‎12-23-2009

Hi Naga,

The numbers don't look too bad. The highest one is:

258 4607595461197331596 384 17.09% 7.20% 2.08% 0 SNMP ENGINE

There are several bugs associated with SNMP causing high CPU on the router.

Have a look at this link to see if you are hitting any of them

http://www.cisco.com/en/US/tech/tk648/tk362/technologies_tech_note09186a00800948e6.shtml#intro

HTH

Reza

nagalingammk · ‎12-23-2009

thanks reza,

i will go thru the document and get back 2 u