Hi,
I have a CCR1036-8G-2S+ acting as PPPoE server, with at the moment 170 sessions.
I see constantly one CPU having a high load. Randomly this load moves to another CPU. And it gradually increases, and then sometimes drops a bit. This is now happening for a week. See the images: The firmware is Version 6.43.4 (stable), the routerboard version is the same.

Any ideas what might cause this?

Could it we scheduled script. They run in the management process if I remember correctly.

I only have one script, that emails me the configuration once a week. Can't imagine that's the problem. Especially since the load shifts to different CPU's multiple times a day.

Are u doing too much nat or mangle?

Enviado desde mi Moto G (5) mediante Tapatalk

Hi,
Did you verify that the front LCD works correctly? Try however to disable it. Sometimes a failure on the serial interface that manages it generates a CPU overload.

Inviato dal mio HUAWEI MT7-TL10 utilizzando Tapatalk

We experienced this previously - a reboot appeared to clear it. I noticed it happens on our router when a user session is "stuck" - sometimes winbox disconnects but the router still thinks the user is logged in, and continues to think so until it is rebooted. I'm not sure if this is the cause or not, but it is correlated.

The device doesn't do nat'ing or has any magle rules.
I rebooted the device this night, and the CPU load is gone.
I noticed that the memory usage was also gradually increasing since the CPU load appeared.
Fortunatly it's fixed for now, but not to happy i didn't find a specific cause to prevent it in the future.

I am experiencing this on a CCR1036-12G-4S-EM, when I disable SNMP, it goes away so think it is SNMP related.

Have upgraded a couple of versions already since I originally noticed it and currently running 6.43.8 but still the same

I read about the disabling SNMP, tried it too, didn't help for me.

This is by the way what the CPU looks like after the reboot. Busy, but normal:

I read about the disabling SNMP, tried it too, didn't help for me.

This is by the way what the CPU looks like after the reboot. Busy, but normal:
cpu.JPG

Yes this is exactly what happened to us. In our case it seemed to start when a winbox session terminated abnormally that had a large table open (I think it was the ip routes table). The user session showed as permanently active on the router (under system->users->active users) even though it had disconnected. My guess is that the management CPU was high because it was trying to redraw the table every single time there was a change, even though the user was no longer connected.