10-12-2023 06:05 AM
Hello everybody,
out customer has a SD-WAN router:
cisco C1111X-8P (1RU) processor with 1971490K/6147K bytes of memory
running IOS Version 17.6.3a.
Our monitoning generated an alalm for high CPU load.
The command 'sh process cpu sorted' shows small values:
AT-IN-RT-0001# sh process cpu sort
CPU utilization for five seconds: 1%/0%; one minute: 1%; five minutes: 1%
PID Runtime(ms) Invoked uSecs 5Sec 1Min 5Min TTY Process
659 276 383 720 0.31% 0.05% 0.01% 1 SSH Process
92 38010512 94348841 402 0.23% 0.20% 0.21% 0 IOSD ipc task
141 3891156 1182038916 3 0.07% 0.01% 0.00% 0 L2 LISP Punt Pro
86 90351 9963028 9 0.07% 0.00% 0.00% 0 SASRcvWQWrk1
4 0 1 0 0.00% 0.00% 0.00% 0 Retransmission o
3 0 1 0 0.00% 0.00% 0.00% 0 PKI Trustpool
5 0 1 0 0.00% 0.00% 0.00% 0 IPC ISSU Dispatc
...
But in the 'sh tech' (attached) I found high values for the command
'show process cpu platform sorted':
------------------ show process cpu platform sorted ------------------
CPU utilization for five seconds: 95%, one minute: 92%, five minutes: 89%
Core 0: CPU utilization for five seconds: 99%, one minute: 92%, five minutes: 83%
Core 1: CPU utilization for five seconds: 99%, one minute: 90%, five minutes: 83%
Core 2: CPU utilization for five seconds: 90%, one minute: 92%, five minutes: 94%
Core 3: CPU utilization for five seconds: 97%, one minute: 97%, five minutes: 98%
Pid PPid 5Sec 1Min 5Min Status Size Name
--------------------------------------------------------------------------------
14198 14181 173% 174% 176% S 253440 qfp-ucode-tsn-l
6902 4999 65% 68% 71% R 981048 snort
11670 11661 27% 3% 2% R 15372 hman
10483 10474 14% 2% 0% S 73296 smand
5158 4999 10% 11% 11% R 271024 climgr
9834 9826 9% 10% 9% S 1356764 linux_iosd-imag
13129 13104 5% 0% 0% S 13848 btman
16385 16361 3% 4% 4% S 8664 ngiolite
13892 13851 3% 3% 3% S 147464 fman_fp_image
34 2 3% 4% 5% S 0 ksmd
3635 3628 1% 1% 1% S 61120 ftmd
32206 32190 0% 0% 0% S 7676 ngiolite
...
What could cause this high CPU-related load values and how this can be changed?
Thanks a lot for every hint!
Bye
R.
10-12-2023 06:28 AM
you have errors :
%IOSXE_QFP-2-LOAD_EXCEED: Slot: 0, QFP:0, Load 83% exceeds the setting threshold 80%
Do you see performance issue . try to increase this to 90%
platform qfp utilisation monitor load 90
check utilisation :
# show platform hardware qfp active datapath infra sw-cio
#show platform hardware qfp active datapath utilization
10-12-2023 07:08 AM
Hi Balaji,
thanks for the fast reply!
I will try to set the platform qfp utilisation monitor load to higher values but
how can I store the change permanently?
AT-IN-RT-0001(config)# platform qfp utilization monitor load ?
Possible completions:
<0..65535, 50 .. 99>[80]
AT-IN-RT-0001(config)# platform qfp utilization monitor load 90
AT-IN-RT-0001(config)# wr mem
-----------------------^
syntax error: unknown command
AT-IN-RT-0001(config)# copy ?
^
% Invalid input detected at '^' marker.
Thanks a lot!
Bye
R.
10-12-2023 08:19 AM
AT-IN-RT-0001(config)#
once you are in config mode try end or exit
then issue write mem
10-12-2023 09:43 AM - edited 10-13-2023 04:03 AM
Snort and traffic forwarding are cause. You need to monitor qfp and cpu utilisation. What about licence perf or hsec? On controlled mode without license 1111x can forward 250 mbit security traffic for all platform (in +out)
10-12-2023 04:30 PM
The OS, IOS-XE, was designed specifically for multi-CPU operation. And any platform running on IOS-XE is guaranteed to have multiple CPU. The command everyone used to know and love, "sh process cpu sorted", gives only an average look of the data-plane. It averages the total number of CPU in a nice form.
Use a different command to look at the control-plane: sh process cpu platform sort location r0
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