Allow adding custom labels to the "gpu-operator" ServiceMonitor

[peihsuant]

feature description

There is a ServiceMonitor named gpu-operator that was automatically created and is owned by the ClusterPolicy. I would like to add custom labels to it so the Prometheus instance with a serviceMonitorSelector can scrape it.

Thanks.

[cdesiniotis]

@peihsuant this is already possible today with the dcgmExporter.serviceMonitor.additionalLabels field in clusterpolicy.

[peihsuant]

Hi @cdesiniotis, what I need to modify is the servicemonitor of gpu-operator not dcgmExporter, thanks.

[cdesiniotis]

@peihsuant apologies, I misread the description. I see we do not have any fields for configuring the gpu-operator ServiceMonitor, unlike for dcgmExporter.

If you are interested in working on this, PRs against our gitlab repository are always welcome. https://gitlab.com/nvidia/kubernetes/gpu-operator

[csauoss]

@cdesiniotis I would like to give this a try if its still needed.

[cdesiniotis]

@csauoss yes this is open. PRs are welcome here https://github.com/NVIDIA/gpu-operator/pulls

[csauoss]

@cdesiniotis thank you, but should I create a MR in gitlab(since documentation mentions it) or a PR here is fine as well?

[csauoss]

@cdesiniotis I created MR 1099 in gitlab for this issue for now. Please review it when you get a chance.

[cdesiniotis]

@csauoss we have recently migrated to GitHub and now perform development here. Can you open your PR against https://github.com/NVIDIA/gpu-operator?

(since documentation mentions it)

Can you point me to where our documentation states this?

[csauoss]

Thanks @cdesiniotis! sounds good, will PR here in that case. Here in CONTRIBUTING.md file is where I saw references to Gitlab.

[csauoss]

Thanks @cdesiniotis! sounds good, will PR here in that case. Here in CONTRIBUTING.md file is where I saw references to Gitlab.

Actually the pr template in github too

[cdesiniotis]

Thank you for pointing this out. I have filed #851 to update our docs.

[csauoss]

@cdesiniotis I created pr #850 in github to address this issue. Please review it when you get a chance. Thanks!