Same here, version 15.0(2)SE7 on my 3750X stacks.

For me, only seem to be seeing this on my 48 port switch stacks....  I have multiple 24 port 3750X switch stacks at the same IOS version, and not seeing this issue.  Cisco, any help with this one?

Our issue was resolved by VLAN pruning.  We had a VLAN that was bleeding over the trunk that wasn't being used on that switch.  If the VLAN wasn't in use we removed it.  The culprit was a scanning tool on the network that shouldn't have been scanning that far into our network.

Thanks stansellc!  We are full layer 3 in our network closets, so we don't have trunks between the closets and our distribution switches...  However, we do have a VLAN that is built on the switch that isn't in use, and is currently down due to no interfaces being configured on that VLAN.  I'll try removing that VLAN and SVI from the stack to see if that helps.  I'll report back with results.

No luck in my case...  I removed the one SVI that we had on the switch for a VLAN that we weren't using.  Still seeing the HULC LED process using 20% of the CPU steadily...

We found out a security device probing the network and was actually causing the problem.  When we shut down the VLAN and to solve the problem at the time.  We put it VLAN back on the trunk (months later) the issue hasn't returned.

Pruning won't help here, I only have 3 VLANs, and they are all in use.  There isn't any scanners running.  Besides, even if there was a scanner running, why would it cause the CPU to jump to 20%?

Even at night when there is very little traffic, the CPU sits at 20%.

I found that if I unplug everything, the CPU slowly goes down.  It's as if this process takes .5% of the CPU per port that is plugged in.  If you completely fill the switch, you'll have about 25% CPU usage.

Yep, this problem still persists for us as well...  Still only on the 48 Port Stacks, not the 24 Port ones for some reason...  I'm still running 15.0(2)SE7 on mine.  Would be great if Cisco could figure this one out.  Unfortunately the locations in which I have these 48 port 3750X stacks are critical areas in our hospital...  CPU runs consistently at around 50-55%, with spikes up into the 90s from time to time...

** Update ** Having gone through config's we realised that the switches experiencing spikes were configured with sshv1 and 2. When we changed this to SSHv2 only the spikes disappeared!

So from what I can see, this platform has issues with SSH when both version 1 and 2 are enabled..

i.e. sh ip ssh  and see if version 1.99 is in effect.

I've added a couple of screenshots showing show cpu history for both switches. On the 72 hour readout we applied version 2 24hrs previous and you can see that although the cpu is still on average 40percent (which is normal), the spikes disappeared.

Can't believe Cisco haven't picked that up??

I took a look at this on mine, and I only have ssh v2 running...  Bummer!!  Still no luck for me, HULC_LED process still consistently using ~20% of the CPU...

Yeah HULC_LED will still run at 20 percent but I am not getting the 'spikes' that I was before. I don't think it's every going to be any lower actually.

Cool, gotcha...  Ya, I have never seen too much spiking, just the big chunk of CPU for the LED process...  Mine stays pretty steady at around 20%...  Very surprised that Cisco hasn't figured this one out yet, especially seeing that this is not seen on the same version 24 port version of these switches...  Would be nice to get this fixed...