Proc CPU hit to 96%

Unanswered Question
Apr 3rd, 2008

Hi NetPro,

does anyone who encountered this problem before ? show proc cpu hit to 96% .

what is the most common problem caused this ?

CPU utilization for five seconds: 82.56%

one minute: 69.16%

five minutes: 65.30%

PID Runtime(ms) Invoked uSecs 5Sec 1Min 5Min TTY Process

--- ----------- ---------- -------- ------- ------- ------- --- ---------------

1 3119734053 0 0 17.44% 30.84% 34.70% -2 Kernel and Idle

23 287829042 45111228 55000 4.45% 4.00% 4.00% -2 SptBpduRx

24 79253724 7987102 39000 1.53% 1.92% 1.98% -2 SptBpduTx

28 565262007 169065436 10000 7.97% 9.76% 9.95% -2 McastRx

38 44394 3705 42000 3.77% 3.00% 3.00% -2 Acct Send Bkg

147 33486 442 41000 5.23% 6.84% 6.96% 1084656276 telnet147

193 580308540 989270583 12783 7.43% 7.66% 7.47% 0 Packet forwardi

194 1601268361 400927856 8696 28.78% 26.46% 25.23% 0 Switching overh

195 497364026 527775876 24650 6.09% 6.18% 6.17% 0 Admin overhead

your reply will be highly appreciated.

regards,

jack

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 2.3 (3 ratings)
Loading.
ney25 Thu, 04/03/2008 - 06:43

hi dandy,

the switch model is 4506 and running on Sup2

what do you think of the consequences causing switch hit to 96% ?

thanks.

regards,

jack

Danilo Dy Thu, 04/03/2008 - 09:14

Well from the document it mentioned the MAC address.

1. It can be done by flushing the MAC address table. But if the CPU is high all the time, I don't think someone is too free to keep doing that 24x7.

2. The switch may have flapping connection to another switch which in turn connected to a lot of hosts - which makes the MAC Addresses widthrawn and re-learned.

If its a L3 switch, look at routing loop. Routing loop can happen both in Static and Dynamic routing. One way to avoid this is to create a NULL interface and route shorter prefix of your entire network to it.

You can check on attacks also by performing sniffing (Wireshark) in your trunk.

Regards,

Dandy

Actions

This Discussion