cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
3160
Views
0
Helpful
25
Replies

CISCO C3750E CPU utilization 100%

nishhhant
Level 1
Level 1

Dear Team,

I am experiencing an issue with my core switch C3750E (L3 switch) with the cpu utilization going 100% and in cpu processes I see the ARP input is using more than 61% of the cpu.

CPU utilization for five seconds: 99%/25%; one minute: 99%; five minutes: 99%

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

  12  1313789569    66834997      19657 60.10% 61.13% 61.48%   0 ARP Input

  85    75586512    18520293       4081  1.76%  1.46%  1.44%   0 RedEarth Tx Mana

  84    27740856    23022207       1204  1.44%  1.00%  0.94%   0 RedEarth I2C dri

214    51056595   110992559        460  1.28%  1.65%  1.64%   0 IP Input

129    26284746     4806428       5468  0.96%  0.83%  0.82%   0 hpm counter proc

Please suggest what I can provide from my end from show run and help me out resolving this issue.

Warm Regards,

Nishant.


25 Replies 25

Do you want me to downgrade the IOS version from 12.2(58)SE2 to 12.2(55)SE8 ?

My recommendation is for you to downgrade to 12.2(55)SE8. 

Do you have any other suggestion of resolving this issue ...

Do you have any other suggestion of resolving this issue

Aside from downgrading the IOS?  Sure, create a Cisco TAC Case.

Hi Leo,

Lol I could have raised a Cisco TAC but my switch is out of warranty..

The entire 12.2(58)SE train is very buggy.  I've used this version before and downgraded all of my 3560/3750 fleet to 12.2(55)SE8.

Hi Nishant,

Are the hosts on your network not under the management of you IT services group? I would locate the offending machine and talk with the owner about why it is trying to contact so many hosts on the network. If it can't be explained then disconnect it from the network until they can.

This is the sort of behavior you might expect from a machine riddled with viruses!

cheers,

Seb.

Hi Seb,

The hosts not in management vlan but they have full access to internet through proxy.

There are around 325 users in this vlan that is 142 where most of all the hosts trying to reach the few hosts which are not in the network. Due to which ARP entries are incomplete.

What is the 'ip route' config on this switch?

If you are specifiying the next-hop as an interface and not an IP then the switch will generate an ARP request for every packet which it does not have a more specfic route.

Hi Seb,

Please find the show ip route output below :-

D*EX  0.0.0.0/0 [170/28416] via 192.168.24.137, 1d15h, GigabitEthernet1/0/24

      10.0.0.0/8 is variably subnetted, 7 subnets, 4 masks

S        10.70.4.0/23 [1/0] via 192.168.24.158

S        10.70.6.0/24 [1/0] via 192.168.24.158

S        10.70.56.0/21 [1/0] via 192.168.24.158

S        10.70.66.0/23 [1/0] via 192.168.24.162

C        10.70.68.0/23 is directly connected, Vlan207

L        10.70.68.2/32 is directly connected, Vlan207

S        10.70.70.0/23 [1/0] via 192.168.24.174

      192.168.24.0/24 is variably subnetted, 11 subnets, 2 masks

C        192.168.24.136/30 is directly connected, GigabitEthernet1/0/24

L        192.168.24.138/32 is directly connected, GigabitEthernet1/0/24

D        192.168.24.140/30 [90/28416] via 192.168.150.3, 6d12h, Vlan150

                           [90/28416] via 192.168.149.3, 6d12h, Vlan149

                           [90/28416] via 192.168.24.137, 6d12h, GigabitEthernet1/0/24

                           [90/28416] via 10.70.68.3, 6d12h, Vlan207

C        192.168.24.156/30 is directly connected, GigabitEthernet1/0/21

L        192.168.24.157/32 is directly connected, GigabitEthernet1/0/21

C        192.168.24.160/30 is directly connected, GigabitEthernet1/0/18

L        192.168.24.161/32 is directly connected, GigabitEthernet1/0/18

C        192.168.24.168/30 is directly connected, GigabitEthernet1/0/1

L        192.168.24.169/32 is directly connected, GigabitEthernet1/0/1

C        192.168.24.172/30 is directly connected, GigabitEthernet1/0/16

L        192.168.24.173/32 is directly connected, GigabitEthernet1/0/16

S     192.168.135.0/24 [1/0] via 192.168.24.158

S     192.168.136.0/22 [1/0] via 192.168.24.158

S     192.168.140.0/23 [1/0] via 192.168.24.162

C     192.168.142.0/23 is directly connected, Vlan142

      192.168.142.0/32 is subnetted, 1 subnets

L        192.168.142.2 is directly connected, Vlan142

S     192.168.144.0/23 [1/0] via 192.168.24.174

      192.168.146.0/24 is variably subnetted, 2 subnets, 2 masks

C        192.168.146.0/24 is directly connected, Vlan146

L        192.168.146.1/32 is directly connected, Vlan146

      192.168.147.0/24 is variably subnetted, 2 subnets, 2 masks

C        192.168.147.0/24 is directly connected, Vlan147

L        192.168.147.1/32 is directly connected, Vlan147

      192.168.148.0/24 is variably subnetted, 2 subnets, 2 masks

C        192.168.148.0/24 is directly connected, Vlan148

L        192.168.148.1/32 is directly connected, Vlan148

      192.168.149.0/24 is variably subnetted, 2 subnets, 2 masks

C        192.168.149.0/24 is directly connected, Vlan149

L        192.168.149.2/32 is directly connected, Vlan149

      192.168.150.0/24 is variably subnetted, 2 subnets, 2 masks

C        192.168.150.0/24 is directly connected, Vlan150

L        192.168.150.2/32 is directly connected, Vlan150

      192.168.151.0/24 is variably subnetted, 2 subnets, 2 masks

C        192.168.151.0/24 is directly connected, Vlan151

L        192.168.151.2/32 is directly connected, Vlan151

      192.168.152.0/24 is variably subnetted, 2 subnets, 2 masks

C        192.168.152.0/24 is directly connected, Vlan152

L        192.168.152.1/32 is directly connected, Vlan152

Hi Seb,

When you specify IP for a route you have to specify for all segments on the switch...which in turn will increase the processes on the switch and the table length will increase the cpu of switch..so we are working on how to restrict unwanted arp request..If you can provide a resolution to restrict please let me know

Cheers,

Nishant.

For the time being I have managed to minimize the utilization.

Now its as below "-

show proc cpu sorted | in ARP

  12  1461177163    88623776      16487 14.07% 16.87% 19.47%   0 ARP Input

  13      232156     4977278         46  0.00%  0.00%  0.00%   0 ARP Background

  41     2791625     4801887        581  0.00%  0.00%  0.00%   0 Dynamic ARP Insp

  40     8083154     6113999       1322  0.00%  0.00%  0.00%   0 ARP Snoop

212      449891      127484       3528  0.00%  0.00%  0.00%   0 IP ARP Adjacency

213      731410   109891119          6  0.00%  0.00%  0.00%   0 IP ARP Retry Age

216      164817     9172622         17  0.00%  0.00%  0.00%   0 IP ARP Track

277       15835      503671         31  0.00%  0.00%  0.00%   0 Cluster RARP

285           0          12          0  0.00%  0.00%  0.00%   0 RARP Input

show arp | in Incomplete

Internet  192.168.142.147         0   Incomplete      ARPA

Internet  192.168.142.193         0   Incomplete      ARPA

Internet  192.168.149.23          0   Incomplete      ARPA

show proc cpu | in CPU

CPU utilization for five seconds: 42%/14%; one minute: 41%; five minutes: 46%