Hi!
Solarwinds can monitor Nvidia's gpu? I can't take de metrics..
I've done some work before on this, but don't have access to any Nvidia currently.
Have you had a look at nvidia-smi?
There's a genuine difference if talking about the linux-flavoured-hostOS vmware ESXs are running, and if it's a random linux VMSimilarly in the host scenario there's additional metrics exposed using one licensing type with NvidiaAlso if it's a host (spot who had more hosts with GPUs to monitor) you can sometimes get by using their idrac (health) or vmware (is not broken)In either flavour, but easier in your OS - If you can use it as a server you can run a script, so linux script to run nvidia-SMI and parse the output into components
I'm being asked about this now too. Can Observability monitor Nvidia GPUs in servers whether in the cloud or on prem? I started looking at the API: https://github.com/NVIDIA/go-nvml and am glad you listed https://docs.nvidia.com/deploy/nvidia-smi/index.html tool as well. It seems with all the AI going on now that we'd have some support in Orion? Or could a SAM monitor take the output from nvidia-smi and display it on a node detail page? Which SAM template would you use?
Bill
Just added a feature request: Observability monitor Nvidia GPU utilization in linux or windows servers through the Orion agent as a part of node details vital stats
