High cpu load of 6509 with MLS

[thomaswu74]

Hi all,

I used have one GE link connected between two 6509 by layer3 (in one vlan).

After I enable one more GE in different vlan(total two GE). The cpu of one 6509 raise to almost 100%. Only one 6509 have this problem. Anyone knows how can I solve this loading problem?

Both are running MLS by gobal and logical interface MLS configuration (mls rp ip). Following are some outputs shows high cpu with Fib process in the switch.

(enable) sh proc cpu

CPU utilization for five seconds: 99.08%

one minute: 98.16%

five minutes: 100.00%

28 53577804 25399935 501000 97.14% 97.92% 97.00% 0 Fib

enable) sh mls

Total packets switched = 189027735148

Total bytes switched = 88132869441442

Total routes = 139598

Long-duration flows aging time = 1920 seconds

[Georg Pauwen]

Hello,

can you post the config of your 6509 ?

High CPU in combination with a high FIB Process value could mean some sort of a DoS attack, check out this document:

Configuring Denial of Service Protection

http://www.cisco.com/en/US/products/hw/switches/ps708/products_configuration_guide_chapter09186a008019c6e7.html#1086290

Regards,

GP

[foxpreacher]

maybe you can use the packet-caputure tool for analyse the packet getting through your MLS,and give the full picture of what had happenen in your MLS.

[thomaswu74]

I guess that it doesn't related DoS.. What I found is that the cef entry in MSFC can't be installed into the switch. See as follow for adding new route.

MSFC(config)# ip route 1.1.1.0 255.255.255.0 Null0

MSFC#sh ip cef 1.1.1.0

1.1.1.0/24, version 12105295, attached

0 packets, 0 bytes

via Null0, 0 dependencies

valid null adjacency

cat6509> sh mls entry cef ip 1.1.1.0/24

Mod FIB-Type Destination-IP Destination-Mask NextHop-IP Weight

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

I know there's bug (CSCdy75968) related to this problem, but my catOS is Sw : 6.3(7) that shouldn't be affected.