Pete, I did open a case with the TAC on this. They reaffirm what Nate said, that it's normal. I still don't like it though. I rely on my Load Averages to gage performance of my system. With it like this, I can no longer do that.

Quote from the TAC:


        You make a good point about troubleshooting.

        You can tell if another process other than kclick is taking CPU
        time by running the top command at the CAS cli and looking at the
        %CPU column. If something other than Kclick is taking the CPU
        time, you will see what it is.

        The click router is open source and the behavior is documented
        here:

        https://amsterdam.lcs.mit.edu/pipermail/click/2003-November/002345.html

        It may not be this way in every configuration all the time because
        of the nature of Click itself. A scheduler is using an algorithm
        to determine how much processor time to actually use, how much
        time to reserve, and how much to allocate to other processes like
        Java or tcpdump or whatever.

        Again, as a point of reassuring you, the TAC has been taking this
        question since we have supported the product. Other processes will
        take processor priority as needed.

-Mike


On Sat, 5 Dec 2009, Pete Boynton wrote:

Just did the upgrade to 4.7.1 and my CPU is being pegged by kclick now:

top - 11:46:18 up 43 min,  1 user,  load average: 1.00, 1.00, 0.91
Tasks:  72 total,   3 running,  69 sleeping,   0 stopped,   0 zombie
Cpu(s):  0.3%us, 59.0%sy,  0.0%ni,  0.0%id,  0.0%wa,  0.0%hi, 40.7%si,  0.0%st
Mem:    903796k total,   217512k used,   686284k free,    12436k buffers
Swap:  4192956k total,        0k used,  4192956k free,   120880k cached

 PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
2048 root      25   0     0    0    0 R 99.7  0.0  42:48.74 kclick


Not too sure what this means .....





-----Original Message-----
From: Cisco Clean Access Users and Administrators 
[mailto:[email protected]] On Behalf Of Mike Diggins
Sent: Thursday, December 03, 2009 9:45 PM
To: [email protected]
Subject: Re: NAC 4.7.1 Upgrade - High CPU on CAS


This is my production CAS (3140) running 4.1.6 lightly loaded:

[r...@cas ~]# top
top - 23:12:17 up 464 days,  3:18,  1 user,  load average: 0.17, 0.21, 0.18
Tasks:  63 total,   2 running,  61 sleeping,   0 stopped,   0 zombie
Cpu(s):  1.7% us,  6.7% sy,  0.0% ni, 69.7% id,  2.3% wa, 10.0% hi,  9.7% si
Mem:   1035588k total,   962836k used,    72752k free,    42060k buffers
Swap:  2040244k total,     2876k used,  2037368k free,   588684k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
 2835 root      15   0     0    0    0 S  8.7  0.0  25757:49 kclick

This is my test CAS (3140) running 4.7.1 with NO load:

[r...@test-cas ~]# top
top - 21:41:25 up 12:02,  1 user,  load average: 1.00, 1.00, 1.00
Tasks:  71 total,   3 running,  68 sleeping,   0 stopped,   0 zombie
Cpu(s):  0.3%us, 49.0%sy,  0.0%ni,  0.0%id,  0.0%wa,  0.0%hi, 50.7%si,
0.0%st
Mem:    904184k total,   462560k used,   441624k free,    90544k buffers
Swap:  4192956k total,        0k used,  4192956k free,   261208k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
 2726 root      25   0     0    0    0 R 99.5  0.0 718:37.89 kclick


Is anyone running 4.7.1 on one of the newer appliances like the 3310 or
3350? If so, is the same thing happening?

-Mike



On Thu, 3 Dec 2009, Mike King wrote:

I haven't looked at the command line of a CAS in a year or so, but if I 
remember correctly, on the CAS it was always normal for the kclick to replace 
the CPU idler.

On Thu, Dec 3, 2009 at 11:24 AM, Kyle Torkelson 
<[email protected]> wrote:
      Mike
       We are exact same as you...Virtual Gateway In Band on two 3140's.  
Whether I have 0, 50 or 200 users online, kclick is always using 95-99%.  TAC 
told me it
      was nothing out of the ordinary...

      I'm still skeptical though...

      Kyle Torkelson
      University of Sioux Falls




      -----Original Message-----
      From: Cisco Clean Access Users and Administrators 
[mailto:[email protected]] On Behalf Of Mike Diggins
      Sent: Thursday, December 03, 2009 10:02 AM
      To: [email protected][root@test-cas1 ~]# top
Subject: NAC 4.7.1 Upgrade - High CPU on CAS

Just upgraded my test NAC system (1x3140 CAM, 1x3140 CAS) from 4.6.1 to
4.7.1. The "kclick" process on the CAS is taking 99% of my CPU, and there
is nobody on the system! I recall someone else posting with this same
issue. I'm in Virtual Gateway Inband mode. Did anyone find a cause?

Reply via email to