ISR 4400 CPU or memory threshold value configuration

rajulpar · ‎09-30-2020

Hi Team,

I am looking for some recommendation around 3750 and 4431 CPU or memory threshold value configuration.

Each month we have a few switches and routers that operate above the configured high tide values, almost every time the Team investigate they find no issue and report that the device is busy but operating OK.

What should be criteria to configure this value or is there any guideline on it ?

For example we have;

3750X-48P in one site that run an average cpu of 60-64% - current alert level is 60%
ISR 4331 routers with average mem utilization of 82% - current alert level is 80%

Is there any guidance or recommendation as to the possibility of lifting the current alert levels to avoid non required alarms?

joelm · ‎09-30-2020

Run this command on the switch module:

switch1#show processes cpu sorted 5sec
CPU utilization for five seconds: 61%/0%; one minute: 62%; five minutes: 62%
 PID Runtime(ms)     Invoked      uSecs   5Sec   1Min   5Min TTY Process 
 171  1706026751   101436650      16818 39.83% 39.92% 39.86%   0 Hulc LED Process

If the top-running process is Hulc LED Process, then you are being bitten by a known bug. It is fixed in later releases of IOS.

As you can see, I'm suffering from it too on my SM-X-ES3D-48-P board (functionally equivalent to a Catalyst 3750), running version 15.2(3)E1 in my ISR4451-X router. You won't see this bug if nothing is plugged into the module because all this process does is control the blinky LEDs for each port. <sigh>

Upgrading to a fixed version of IOS might eliminate your issue and you can leave the alert at 60%.

rajulpar · ‎10-06-2020

Hi Joelm,

Thanks for your response :

yes, I can it is going around 20% in 3750-x

#sh processes cpu sorted
CPU utilization for five seconds: 50%/1%; one minute: 58%; five minutes: 63%
PID Runtime(ms) Invoked uSecs 5Sec 1Min 5Min TTY Process
185 1537677029 172976322 8889 20.00% 20.28% 20.50% 0 Hulc LED Process

running code : version 15.2

Do you think upgrade will help ? can you point to specific bug here ?

Leo Laohoo · ‎09-30-2020

What firmware are the 3750X and ISR 4k running on and what are their uptimes?

NOTE:

For the IOS-XE, refrain from using the traditional command "sh proc cpu h" or "sh proc mem" because the result can be misleading. A better command is "sh platform software status control brief".

For the 3750X, use the command "sh proc cpu sort | ex 0.00" to determine what CPU is cycle is chewing up resources.

rajulpar · ‎10-06-2020

Hi leo,

Thanks for reply, 3750 is running 15.2 and I have following output

#sh processes cpu sorted
CPU utilization for five seconds: 50%/1%; one minute: 58%; five minutes: 63%
PID Runtime(ms) Invoked uSecs 5Sec 1Min 5Min TTY Process
185 1537677029 172976322 8889 20.00% 20.28% 20.50% 0 Hulc LED Process
88 498170816 87636045 5684 6.24% 6.36% 6.46% 0 RedEarth Tx Mana
87 327534337 154289071 2122 3.04% 3.77% 4.08% 0 RedEarth I2C dri
134 155182674 7005935 22150 2.55% 2.17% 2.11% 0 hpm counter proc
473 431419088 386912991 1115 2.24% 3.86% 5.08% 0 SNMP ENGINE

Can I change the CPU threshold value based on the output or need to change in the configuration to reduce it?

Leo Laohoo · ‎10-06-2020

@rajulpar wrote:

3750 is running 15.2

For someone who works for Cisco, this is really NOT a helpful/useful comment.

Try again: What is the exact firmware the switch is running on?