enable DCGM_FI_DEV_GPU_UTIL && update docker file

[zhu733756]

Hello, friends.
Thanks to this repo, we can install DCGM exporter using Helm and fetch metrics using Prometheus.
However, I have a puzzle right now.
Why do we close the metrics label DCGM_FI_DEV_GPU_UTIL ?
Hope that someone can help to answer this question.
Thank you in advance.

BTW, this pr aims to enable metrics label DCGM_FI_DEV_GPU_UTIL and fix a build error in Dockerfile.
The error like below:

E: Failed to fetch https://developer.download.nvidia.cn/compute/cuda/repos/ubuntu2004/x86_64/by-hash/SHA256/751939d95516afc289908a19e447f0acc1506367f72ed356431a2b1a469cc8ca  404  Not Found [IP: 45.43.32.211 443]
E: Some index files failed to download. They have been ignored, or old ones used instead.

Thank you for your reply.

[nikkon-dev]

@zhu733756,

Thank you for the PR.
The explanation behind removing DCGM_FI_DEV_GPU_UTIL from the metrics enabled by default could be found here: NVIDIA/gpu-monitoring-tools#143 (comment)
In general - that metric is deprecated because its value is misleading and expensive to acquire. There are better metrics that are meant to replace the deprecated ones - DCGM_FI_PROF_* family of metrics.

WBR,
Nik

[zhu733756]

Great, thanks. @nikkon-dev