Zabbix template for multiple Nvidia GPUs

in #cryptocurrency7 years ago

A couple of months ago, myself and a friend decided to resurrect our old mining hardware as cryptocurrency prices were getting higher and higher.

After the initial setup, some hardware upgrades (new GPUs mostly), etc., I decided it was time I used my Zabbix server to monitor the miners. Unfortunately, I found only one Zabbix template for monitoring Nvidia GPUs and it works with one GPU only. Therefore, I used it conceptually to create a new template, featuring:

  • low-level discovery of all the graphics cards
  • item prototypes for:
    • fan speed
    • total, free and used memory
    • power draw in decaWatts (tens of Watts, so that it can fit in the graphs nicely)
    • temperature
    • utilization
  • a graph prototype having the fan speed, power draw and temperature in one graph
  • trigger prototypes set at different GPU temperatures
  • a BASH script for the low-level discovery

Events caused by the triggers:
Triggers:

And the graphs:
Graphs

Here's the template at github: https://github.com/plambe/zabbix-nvidia-smi-multi-gpu

Enjoy!