Hello.
I need to achieve high CPU usage notification system.
What I did for now is a script running from scheduller every 30 seconds which takes probes every second for 28 seconds in a loop and calculates average value and then logs it. If it goes over 20% it’s a warning and MT sends e-mail notification. I’m not really happy with this solution.
Do you know a better way to achieve this?
Any SNMP monitoring tool that supports alerting will do better, like Prometheus. You will need external server to run that or ROS container if you have device that supports containers and enough RAM.
Take a look at: https://www.observium.org and try the community edition one of my favorites.
Thanks guys, I will dig into it. If someone is using something else - feel free to share it.
What I like the most is that it’s not “inside” router.
It’s still my supervisor’s “wish” for gateway when it’s running at 0-2 % usage when I look at it, but you know - it is what it is.
Few times I had to restart MT Access points because their CPU was flopping in a weird way so maybe it’s not a bad idea after all.
Maybe Zabbix has this functionality as well? I got a running server in my network, but I’ve inherited it and didn’t have time for last months to look at it 
Maybe I should in this case?
Never worked with Zabbix, but look at its functionality with SNMP traps - https://www.zabbix.com/documentation/current/en/manual/config/items/itemtypes/snmptrap, you can set up SNMP condition trap and process that with script which can send email.
I have worked with Zabbix (about 10 years ago) and also inherited it. A few years later used Observium, it was like it knew what I wanted before I did, as it did discover most devices by it self (after providing snmp config).