Expose classification metrics from LowNodeUtilization plugin

[tiraboschi]

Enhance the plugin interface introducing an additional Metrics interface to be easily implemented by plugins
that intend to register and expose metrics.

Expose two new metrics from the LowNodeUtilization plugin:

• descheduler_low_node_utilization_thresholds
• descheduler_low_node_utilization_classification

These metrics allow users to properly monitor plugin behavior,
which was previously only visible in descheduler logs.

[k8s-ci-robot]

Hi @tiraboschi. Thanks for your PR.

I'm waiting for a kubernetes-sigs member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.

Once the patch is verified, the new status will be reflected by the ok-to-test label.

I understand the commands that are listed here.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository.

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by:
Once this PR has been reviewed and has the lgtm label, please assign damemi for approval. For more information see the Code Review Process.

The full list of commands accepted by this bot can be found here.

Needs approval from an approver in each of these files:

• OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[ingvagabund]

This will not scale well for external plugins. Maybe a combination of init() and metrics registration within each plugin? Or, have you tried to invoke RegisterMetrics within the plugin instead?

[tiraboschi]

Done, I tried adding an optional Metrics interface to complement Handle without breaking backward compatibility

[ingvagabund]

There could be a new handler for setting metrics. E.g. l.handle.Metric(metricsName).WithLabelValues or similar that would be a no-op in case metrics are not enabled.

[tiraboschi]

Done, I implemented a new metric handler wrapping all the metric setters embedding there the skip metric setting.

[ingvagabund]

There's probably no way to disallow users to invoke this in the NewXXX functions. Also, the descheduler registers the metrics independent of whether metrics server is enabled or not. I wonder what would be the overhead of just registering the metrics directly. Allowing to skip ShouldExport logic.

[ingvagabund]

Also, a single plugin can be initialized in each profile. Thus, a metric entry needs to be labeled with a profile name as well.

[tiraboschi]

I added a bit of logic to register metrics by profile name and by plugin name.
A metric named Classification registered by LowNodeUtilization plugin required by CustomProfile profile will result in descheduler_custom_profile_low_node_utilization_classification.