GPU (nvml)

[kazem26]

I cannot see the metrics for nvml in grafana for kepler.
What should I do exactly?

[marceloamaral]

@kazem26 which metrics are you looking for?

Few questions:

• Is your environment based on Bare-metal or VMs?
• Can you share the Kepler logs?
• I am assuming that you have access to GPUs in your cluster. Do you see kepler_container_gpu_compute_util_total?

[thisiskazem]

It's a bare-metal environment. there's a kubernetes cluster only with a master-node running and all the pods are running on the master-node. I've uploaded the logs of kepler here.
metrics.pdf

[rootfs]

from the metrics, there is no kepler_node_gpu or kepler_container_gpu metrics, have you enabled gpu in kepler container? i.e. -enable-gpu=true

[thisiskazem]

Yes. It is enabled

[thisiskazem]

I've installed kepler on a bare-metal server. My server has a graphic card installed. How should i check if the enable-gpu=true or not because I cannot see the metrics related to gpu.