feat: Add servicemonitor switch for metrics acquisition

[CoderTH]

The servicemonitor switch is added to collect indicators

[archlitchi]

could your provide a document regarding this feature?

[CoderTH]

could your provide a document regarding this feature?

OK, I will add some document content and open some other parameter settings for servicemonitor.

[chaunceyjiang]

Good work, but I think introducing serviceMonitor will force the installation of corresponding CRD resources.

Therefore, I highly recommend using the following method to expose metrics.
Expose the corresponding metrics on annotations.

    prometheus.io/scrape: "true"
    prometheus.io/port: {{ .Values.device-plugin.prometheus.port | quote }}

[CoderTH]

干得好，但我认为引入serviceMonitor会强制安装相应的 CRD 资源。

因此，我强烈建议使用以下方法来公开指标。 在 上公开相应的指标annotations。

    prometheus.io/scrape: "true"
    prometheus.io/port: {{ .Values.device-plugin.prometheus.port | quote }}

The advantage of using servicemonitor is that users can use servicemonitor to do some processing on metrics, such as replacing label, etc., but I'm not sure if I have this ability in this way of annotation.

[chaunceyjiang]

such as replacing label,

Got you.

https://github.com/cilium/cilium/blob/main/install/kubernetes/cilium/values.yaml#L1024-L1054
Perhaps we can refer to some graduate projects to see how they are done.