06-14-2010 06:47 AM - edited 03-06-2019 11:34 AM
Ciscoworks is giving me a HighUtilization on the PSR component of one of my 6500's with a sup 720. When I look at cpu utilization using 'show process cpu sorted' it doesnt show any high utilization. Any ideas?
Solved! Go to Solution.
06-14-2010 08:14 AM
What does "show proc cpu hist" show you?
06-14-2010 08:14 AM
What does "show proc cpu hist" show you?
06-14-2010 08:45 AM
Spikes of high utilization:
7774444477777444449999944444777774444466666444441111133333
100
90
80
70
60
50
40
30
20
10 *** ***** ***** ***** ***** *****
0....5....1....1....2....2....3....3....4....4....5....5....
0 5 0 5 0 5 0 5 0 5
CPU% per second (last 60 seconds)
15 1 55 2521 41 7512 144211519 2237115212311 612114111
1698080198271195498513183406312809798515943783399506213283
100
90 *
80 * * *
70 * * * *
60 * * * * * * *
50 * ** * * ** * * * * *
40 * ** * * ** ** * * ** * * * *
30 * ** * * ** *** * * **** * ** * * *
20 * ** *** * ** * *** **# **** *** ** # * * *
10 ##########################################################
0....5....1....1....2....2....3....3....4....4....5....5....
0 5 0 5 0 5 0 5 0 5
CPU% per minute (last 60 minutes)
* = maximum CPU% # = average CPU%
9785964585664955577366568954458585448555595558557544555655959556585788
0323456168398275646991278506805234883844704883023276413871404654719398
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)
* = maximum CPU% # = average CPU%
06-14-2010 12:32 PM
Any suggestions on how to find the traffic causing this?
06-14-2010 10:11 PM
Hello Randy,
because sh proc cpu history does not provide details about processes or interrupts using cpu you can use:
sh proc cpu sorted in case there are processes loading the cpu
if the high spikes are caused by peaks of traffic that is process switched = sent to main cpu you can follow the following link:
http://www.cisco.com/en/US/products/hw/switches/ps708/products_tech_note09186a00804916e0.shtml
refer to the section about tools for finding traffic sent to main cpu
and
Hope to help
Giuseppe
06-15-2010 12:05 AM
What show proc cpu hist does show you is an over view of the utilisation from the last 60 seconds, 60 minutes and 72 hours.
As you can see there are high peaks (peak noted by the * sign), but nothing sustained (average noted by the # sign). Your polling/alerting must be set as such to catch this high peak event.
I would at first glace say that this is nothing too much to worry about as the CPU is not underload for very long, but it may well be worth getting to the bottom of it if it is a frequent and/or regular occurance.
The problem you have is that the show proc cpu sorted command only gives you at max the last 5 mins of processes utilising the cpu. You either need to logon immediatley when you recieve the alarm to see what process spiked (this depends on whether you monitoring alerts straight away or not) or set up some other method in understanding what is hitting the CPU at that time. In the past I have set SPAN sessions up to monitor the control plane to see what is being sent to the cpu at specific times.
It may just be that there is a network event, the requirement for STP or other mechanisms to re-compute at these times so it would be worth looking around the network to see if anything else has a similar spike or event that may have caused some topology change. If not, then it is something specific to this 6500. A little assumption here that you are using this 6500 as the root bridge or some central point of control.
SNMP is a fairly common cuase of these kind of spikes. Usually a good place to start to ensure these strings are secured.
HTH
Andy
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide