Hi,

on some Routers I have a high CPU utilization at management and SNMP (see attached picture).
This problem started with the 5.x ROS, but it is also possible since the 4.17 ROS, I´m not realy shure.
I´ve tried Dude 4.0beta2 and 4.0beta3 but it is still the same. I can see constantly SNMP traffic between the
Dude and the Router from 200 up to 500kbit/s.

I have this problem on an Axiomtek Hardware with 386 ROS and also an a RB/750G, bot with ROS 5.5.
The RB/750G has no special configuration and minimal traffic. Other RB/750G with identical configurations
are running with a max. of 2-5% CPU utilization.
The Axiomtek Router is equipped with 7 GigE ports, 5 of them are used. There are about 50 VLANs configured,
OSPF and some VRRP interfaces. Previously with a 4.x ROS this Router has a max. CPU utilization of 10%, now
he is running with min. 50 up to 80%.

I´ve seen a thread with a high CPU utilization on Cisco boxes done by the Dude and also one request in
that thread with a similar problem, but on a RB/1000. But there was no answer on the RB CPU utilization.

Any idea???

Best Regards

Ivan

Yes, same here. Management goes nuts if SNMP is enebled (v5.5)

Same problem here, v5.6 x86 P4 2,8GHz HT. CPU goes in 100% if SNMP gets enabled.
Tested again on another x86 platform CPU P4 3,2GHz, same problem.

If change snmp trap version from 1 to 2?

I wanted to see if anyone had found an answer to this issue. I am running Axiomtek as well with ver 5.7 and see the same thing when SNMP is enabled. I thought this might be due to the full BGP routes that I have on the router but it sounds as if others are seeing the same issue.

As a work around, we looked at using routerOS to monitor the links from the Dude; however, it still pulls in the entire routing table to the Dude server. Anyone know of a way to filter that out so that the Dude does not pole the routes via routeros login or SNMP? It's pretty easy to setup a filter on a Cisco IOS device but I have not found a way to do it in routeros

I have also had issues with high CPU utilisation on some Enterasys switches due to The Dude (and have heard of others with different types of gear).

This seems to be because of the large amount of info that The Dude gathers frome every device that has SNMP probing turned on (generally a good thing - but in certain circumstances, very troublesome).

I have dealt with it in a couple of differenet ways. The easiest and most brutal is to just select "noSNMP" as the SNMP profile on the Settings=>General tab of devices that are having trouble - but then you don't get much info.

A bit gentler method that can be used on devices with SNMP v3 is to set up a view that only includes some of the MIB tree (such as interface stats, CPU, system info - you need to tailor this to cull out the stuff you don't care about but keep the info you want). Doing it this way stops the device from expending CPU cycles on extracting and sending all the data you don't want. Cisco (and probably others too) have config options to achieve similar results with snmp v1 and v2 as well. This is explained in detail in other places in this forum.

You will need to look closely at the documentation for your particular equipment to see what options you have available to limit the data that The Dude extracts using SNMP queries.

Maybe in a future release Mikrotik may allow us to configure what information The Dude queries for

What is your mac mapping refresh interval? Mine is set to 24 hours. I am not having CPU issues. I am only running 1 probe on most Cisco devices, besides the link monitoring. I have 26 Cisco mibs in my files list. The only routers I have had to be careful with are the ones running BGP. I set them to not return the routing table. We are almost entirely Cisco devices.