Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I have two 3750-X switches operating in a stack. The SDM template is set to desktop routing, as we need a lot of next-hop unicast routes.

The tcam usage is not looking quite right...every value is maxed out. Has anyone seen this before (or have any explanation to it), or should I open a TAC case right away?

CAM Utilization for ASIC# 0                      Max            Used
                                             Masks/Values    Masks/values

Unicast mac addresses:                       3292/3292        627/627
IPv4 IGMP groups + multicast routes:         1120/1120          1/1
IPv4 unicast directly-connected routes:      3072/3072         92/92
IPv4 unicast indirectly-connected routes:    8144/8144       3730/3730
IPv4 policy based routing aces:               498/498          13/13
IPv4 qos aces:                                474/474          21/21
IPv4 security aces:                           972/972          33/33

Edit:

Here is an example from Cisco, wich looks quite different: http://www.cisco.com/en/US/products/hw/switches/ps5023/products_tech_note09186a00807213f5.shtml#SDM

Edit2:

I might want to add the reason why I'm investigating this: We're experiencing very random latency when we ping this switch, no matter what VRF or VLAN we use. The CPU utilization looks a bit weird as well:

    6644646454444584445446554355447464554555554444464645444445457455556545
    6907344850716408995114547934140092790235780624077383760641040825126164
100
 90
 80               *
 70 **            *               * *              *            *     *
 60 **  * * *     *   *  **       * * **   ***     * *          *  *  *
 50 ** ** *** * *******  **** **  * * ** ***** *   ******* * * ***********
 40 **********************************************************************
 30 **********************************************************************
 20 ######################################################################
 10 ######################################################################
   0....5....1....1....2....2....3....3....4....4....5....5....6....6....7.
             0    5    0    5    0    5    0    5    0    5    0    5    0
                   CPU% per hour (last 72 hours)

Current usage:

CPU utilization for five seconds: 25%/1%; one minute: 24%; five minutes: 24%
 PID Runtime(ms)   Invoked      uSecs   5Sec   1Min   5Min TTY Process
  81   105562156    425102     248330 13.41% 11.52% 11.62%   0 Adjust Regions
  75    38660112   6968319       5547  3.03%  2.94%  2.84%   0 RedEarth Tx Mana
  74    24316130  11091855       2192  1.59%  1.59%  1.59%   0 RedEarth I2C dri
   4    15330586    677415      22631  0.00%  1.14%  1.14%   0 Check heaps
share|improve this question
    
I wouldn't consider that CPU usage to be too high. It's fairly stead, with only an average of 24% or so. The asterisks you see are the max value recorded in the hour, which is normal. –  Hyppy Mar 22 '11 at 13:05
    
I've noticied the new 3750-X and 2950-S switches to respond oddly to ping requests and sometimes ssh connections. I've never actually had performance problems with the port to port traffic though. I'm thinking there is some prioritizing going on, but I haven't actually seen it mentioned in any docs yet. –  Bad Dos Jun 3 '12 at 5:20
    
@BadDos Confirmed, they have low prio on stuff like icmp/ssh/telnet. –  pauska Jun 6 '12 at 14:29

1 Answer 1

up vote 4 down vote accepted

You're not maxed out. The second set of numbers is the current utilization, and the first column is the max.

Edit 1: On page 17 of this link is a decent explanation of it.

Edit 2: For your underlying issue that you are noting, take a look at show controller utilization to get a good feel for the traffic you're seeing. Also, here's a decent guide from Cisco on troubleshooting high CPU usage in 3750s (linked here)

share|improve this answer
    
Ok. I've edited the question to show an example from Cisco, wich looks quite different. Is it still normal output? –  pauska Mar 22 '11 at 12:46
    
I edited the answer to include a pdf from cisco with a little more info. –  Hyppy Mar 22 '11 at 12:56
    
Ok, that was very helpful. Thank you. Guess I have to dig deeper to find the problem :( –  pauska Mar 22 '11 at 12:59
    
2nd edit added to address your 2nd edit :-) –  Hyppy Mar 22 '11 at 13:02
    
sh controllers utilization is zeroed out on every value.. –  pauska Mar 22 '11 at 13:06

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.