CSCtn42790 - 3560X/3750X Elevated CPU usage due to Hulc LED process

[limmingchun]

We found out that there is an exception for this bug.

Our environment have 2 type of 3750X PoE stacked switches, all the stacks are almost having the same amount of ip phone and ap connected:

A) 1x24ports, 4x48ports

B) 5x48ports

We have more than 6 of type A stack, the cpu utilization for Hulc LED process of all stacks just around 1%

Besides we have more than 7 of type B stack, all of the stack are showing high cpu utilization around 20% for Hulc LED process.

If Cisco claimed that the high cpu of Hulc LED because of does the following tasks:
- Check Link status on every port
- If the switch supports POE, it checks to see if there is a Power Device (PD)
detected
- Check the status of the transceiver
- Update Fan status
- Set Main LED and ports LEDs
- Update both Power Supplies and RPS
- Check on system temperature status

But why the stack with 24 ports switch did not faced this issue? 

[andrei]

I have the same behaviour on the WS-C3750X-48T-S running under 15.0(2)SE6 version.

Switch Ports Model              SW Version            SW Image
------ ----- -----              ----------            ----------
*    1 54    WS-C3750X-48       15.0(2)SE6            C3750E-UNIVERSALK9-M
     2 54    WS-C3750X-48       15.0(2)SE6            C3750E-UNIVERSALK9-M

Switch#show processes cpu sorted | exclude 0\.
CPU utilization for five seconds: 33%/0%; one minute: 35%; five minutes: 34%
 PID Runtime(ms)     Invoked      uSecs   5Sec   1Min   5Min TTY Process
 179  1597050939   213538343       7478 19.51% 19.85% 19.66%   0 Hulc LED Process
  90   254625571    41668722       6110  2.87%  2.80%  2.86%   0 RedEarth Tx Mana
 134   132649041     8566912      15483  2.39%  2.36%  2.38%   0 hpm counter proc

[Massimo Baschieri]

I've seen that behaviour on different platforms with a lot of different ios versions.

As someone explained erlier, 33% of cpu utilization is not that alarming, specially considering that the process under examination is a low priority one on ios.

In my experience the only annoying issue about that is when you raise over 40/50% cpu utilization and you have management software that starts sending alarms.

[Joel Girard]

Same here. 6 stack of 3750X's with a 20% Hulu LED process.

Switch Ports Model              SW Version            SW Image
------ ----- -----              ----------            ----------
*    1 54    WS-C3750X-48       15.0(1)SE2            C3750E-UNIVERSALK9-M
     2 54    WS-C3750X-48       15.0(1)SE2            C3750E-UNIVERSALK9-M
     3 54    WS-C3750X-48       15.0(1)SE2            C3750E-UNIVERSALK9-M
     4 54    WS-C3750X-48       15.0(1)SE2            C3750E-UNIVERSALK9-M
     5 54    WS-C3750X-48       15.0(1)SE2            C3750E-UNIVERSALK9-M
     6 54    WS-C3750X-48       15.0(1)SE2            C3750E-UNIVERSALK9-M

CPU utilization for five seconds: 52%/4%; one minute: 50%; five minutes: 45%
 PID Runtime(ms)     Invoked      uSecs   5Sec   1Min   5Min TTY Process
 176  2215782113   249642443       8875 20.48% 19.81% 20.11%   0 Hulc LED Process
 404   156180063   139911469       1116  6.40%  6.67%  3.36%   0 SNMP ENGINE
  95    44752857   134945153        331  3.68%  2.21%  0.73%   0 hrpc <- response
 132   218355531    20972257      10411  2.24%  2.13%  2.07%   0 hpm counter proc

Causes us grief because we're getting alarms when the overall CPU goes higher than 80% and we hit it often.