High CPU utilization on Router 2821 with 12.4(24)T1

Unanswered Question
Jan 20th, 2010
User Badges:

Hi all,


I have a Router 2821 with c2800nm-advipservicesk9-mz.124-24.T1.bin and with a Cisco Integrated Service Engine(s) Cisco Unity Express 7.1.1 in slot 1 , this router shows a high cpu utilization.

I think is a bug with the IOS , because there is no voicemail configured on the CUE, there is no traffic to the CUE , just traffic from:

- voice calls

- ospf


The ospf  is not the cause of the high cpu , because the high cpu utilization appears before I put the ospf proccess on the router.


Thanks for any help about this.

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
Giuseppe Larosa Wed, 01/20/2010 - 12:55
User Badges:
  • Super Silver, 17500 points or more
  • Hall of Fame,

    Founding Member

Hello Jal,

post in txt attachment file

sh proc cpu sorted 1min


with

and without OSPF running


Hope to help

Giuseppe

Giuseppe Larosa Thu, 01/21/2010 - 11:54
User Badges:
  • Super Silver, 17500 points or more
  • Hall of Fame,

    Founding Member

Hello Jal,


>> CPU utilization for five seconds: 99%/96%; one minute: 98%; five minutes: 62%


the second figure 96% is very close to the first 99%, there are a lot of interrupts that causes the cpu to be high, not single processes.


see following docs:


http://www.cisco.com/en/US/products/hw/routers/ps359/products_tech_note09186a00801c2af0.shtml


http://www.cisco.com/en/US/products/hw/routers/ps133/products_tech_note09186a00800a70f2.shtml#symptons


this high interrupts can be caused by a SW bug as you think or by some misconfiguration in the VoIP part.


About the Unity Express service module: you can easily check what happens if you power off the router and you remove the module.

if after restart cpu is low your suspect is correct.

save the few lines of the module config on a txt file for safety.


if this is not a solution you can try to unconfigure all the VOIP part and see what happens



Hope to help

Giuseppe

edwardlassotovitch Thu, 05/24/2012 - 13:37
User Badges:

Just curious. Has anyone ever seen a real, live case where OSPF was all in area 0 and was taking too much CPU. Then when you segmented the network into areas it actually helped the CPU?

Actions

This Discussion