Hi!
I’m trying to use the SNMP to monitor the Routerboards in my network, but I’m having problems with CPU Load when I turn on the SNMP in the RB, like in the picture.
with this, PRTG 's CPU load sensors trigger precisely because of protocol used for monitoring.
Furthermore, in the process Profiles more loads the CPU is the " Management "
They can tell why this happens ?
Thank you!
When you use SNMP you must make sure it is available only to your internal monitoring system, by setting the “addresses” field of the community to your internal monitoring system or trusted subnet where it is, and/or by using the IP Firewall function to block all access to UDP port 161 from untrusted networks like outside internet.
When you don’t do that, the bad guys will abuse your SNMP service to DDoS other people, and your line and router will be loaded as a result. Plus you will get abuse messages.
Snmp is quite cpu problematic even when no bad guys are abusing it. I have the same experience when only one dude server is polling few oids.
Hello pe1chl and Jarda!
I am aware of vulnerabilities using SNMP, so I’m using SNMP V3, with encryption and authentication. The community is also configured to accept solicitations only my monitoring server.
I have created a firewall rule to Action “Acept” to probe the port 161 UDP and I’m not getting packages which are not originated on my server.
Yet the problem persists.
The SNMP protocol uses any more port I need to probe?
Thank you again!
That’s enough. I am using snmp v3 too. Maybe the encryption could be the performance problem. I hadn’t time to test it.