6509 High CPU utilization

[Ramkumar1988]

Hi All,

        Am facing a critical issue with 6509 core switch.The cpu utilization of teh switch is reached 99% often. Am checked the logs and reaveled thta debug commend is enabled automatically. after issuing the " U all " utilization reduced to normal . Please find some details below and screen captures below :

Model : 6509 E

sup engine : S32

IOS ver:

s3223-ipbasek9-mz.122-33.SXH8a.bin

so i suspect there is an bug in ios.. any suggestions..

Regards,

Prabakar

[nkarpysh]

Hello,

First of all - what do you mean by debug enabled automatically? What was the debug enabled? Do you have any script which may enable debug on some event? Did you consider that someone might login to box before issuing some debugs?

The CPU was High due to interrupts - mainly traffic. So something triggered traffic to be soft switched. That can be debug but it woould depened on the debug itself and other condition on this switch.

Please provide the requested above info to clarify it further.

Nik

[Ramkumar1988]

Hi Nikolay,

               thanks for your Revert..

   It was a regular working hour.In that am noticed that the device is not pinging.so am trying to login.after logged in the perfomence is very slow.so i issued the commend show process cpu and reaveled that some logging events(like output of the debug commend . is going on.so i issued the commend "u all ". suddenly utilization reduced to normal..so it is clear some  events(such as debuging ,traffic  is going in.). Since am the the only guy having the credential to access the device.so suspect the ios which means bug..i dont have any script to enable debug..

Regards,

Prabakar

[nkarpysh]

Thanks, Got you.

It is very strange that debugs started automatically - if you have AAA set up - I would recomend to check the TACACs server for commands run on this switch - may be it was something enabled before. The other this is that switch might be just shwoing some err-rlogs without debug run due to heavy traffic coming it. As you see CPU value is shown like 100%/90%. Second value after slash means the CPU driven by traffic processed by cpu.

So it looks like some traffic was hitting the switch causing CPU to go up high - it can be some user triggering it or some other event. I would say almost not possible to TS now without symptom itself. What you can check are the following commands:

show int | i line|drop  ----- to see if any interface has high number of input drops

show span det | i exec|of top|from    ---- to see if STP flapped recently and where last TCN came from

Next time try configure putty to capture all output to a file - then we can observe it later.

Nik

[Ramkumar1988]

Hi Nik,

        Thanks .. Since STP in not running in the device and i will check the other feasiblities ...

Regards,

Prabaakar