opentelemetry-instrumentation-system-metrics: Configuration by EnvVar

[sfozz]

What problem do you want to solve?

Currently when using the system-metrics instrumentation via auto-instrumentation it is not possible to specify the configuration via an alternative mechanism (e.g. an EnvVar).

When running workloads in a orchestrated environment generally there are classes of metrics that are collected at the orchestration level. An example of this would be cpu and memory metrics in a kubernetes environment, where there are opentelemetry-collectors running each of the nodes.

Describe the solution you'd like

A EnvVar that permits excluding metrics using a glob based pattern.

For example:

OTEL_PYTHON_SYSTEM_METRICS_EXCLUDED_METRICS=system.*

This would exclude the collection of all of the metrics that start in the system namespace.

Describe alternatives you've considered

In a slack thread there was a suggestion that custom distribution could be created that check for the loading of the systems-metrics instrumentation and then injecting a custom config dict.

https://cloud-native.slack.com/archives/C01PD4HUVBL/p1763433132238179

Additional Context

No response

Would you like to implement a fix?

None

Tip

_{React with 👍 to help prioritize this issue. Please use comments to provide useful context, avoiding +1 or me too, to help us triage it. Learn more here.}