ESMP values in Cisco 4500 series switch

[katberts]

Hello there,

We have a 4500 series switch that is running around 39% cpu utilization but seems to be dragging a bit.   I did a show platform health and GalChassisVp-review has a target of 3% but an actual of 17%.  All other actual values were at or below target values.  I then went to show platform cpu packet statistics and in the "Packets received by Packet Queue" all queue packet values at 5 sec, 1 min, 5 min and 1 hour avg were either 0 or 1 with the exception of Esmp - that averaged 351/377/317 and 313 respectively.  I then checked show platform hardware acl input entries static and the entry type InputEsmpToCPU had a whopping 10121370 hit count.  All other entry types were well under that number.  I am not able to provide screen grabs/files but does anyone have any idea why this one area would be so high?  And what, exactly, is ESMP?  I read where it said it was "packets for line card ASICs or other component mgmt" but that explanation left a landing strip on my head.  Any help would be appreciated.

Thanks,

Kiley

[Arumugam Muthaiah · Accepted solution]

Hi,

Normally the 'GalChassisVp-review' in output of "show platform health" could be expected and tend to increase with increase in POE line cards on the chassis (for example, WS-X4548-RJ45V+ cards).

Those cards require a software polling model for their management and their hardware access S2w mechanism which uses CPU bit more, because of which you could always see a higher CPU utilization when using these linecards. The more the number of such linecards in the chassis, the higher will be the CPU utilization, and this could be expected.

This can occur on below condition usually:

• In a redundant chassis, we might observe this bug when the chassis has lots of K2 POE linecards and a switchover activity was performed in the system, and cpu utilization shots up to around 80% on the new Active supervisor.
  
• In a non-redundant chassis, we might see this issue if the chassis has lots of K2 POE linecards in the system. (WS-X4548-GB-RJ45V, WS-X4548-RJ45V+). However the cpu utilization should not be as high as 80%.
  
• In the fixed config boxes this issue is not applicable.
  

Suggestion:

The SW fix has been provided for this issue. SW fix is available in release 15.0(00.99)XO22,  12.2(53)SG03,  15.0(01)XO.

In Catalyst 4500 switch, The Supervisor packet Engine has 16 queues in order to handle various types of packets or events. The 16 queues allow the Catalyst 4500 to queue the packets on the basis of the packet type or priority

Even Simple Management Protocol (ESMP)
This belongs to Queue 0 and ESMP packets (internal management packets) for the line card ASICs or other component management like Those cards require a software polling model for their management and their hardware access S2w mechanism which uses CPU

Refer:

High CPU Utilization on Cisco IOS Software-Based Catalyst 4500 Switches

http://www.cisco.com/en/US/products/hw/switches/ps663/products_tech_note09186a00804cef15.shtml

Troubleshooting High CPU on the Catalyst 4500-E Series Switch

http://www.cisco.com/en/US/docs/switches/lan/catalyst4500/troubleshooting/cpu_util.html

Regards,

Aru

*** Please rate if the post is useful ***

[katberts]

Aru,

Thank you very much for the information.  We do have quite a number of PoE linecards in the switch and it is running an older IOS.  Will provide this extremely helpful info to my management.

Kiley