search

open-telemetry / semantic-conventions

Defines standards for generating consistent, accessible telemetry across a variety of domains
Apache License 2.0
231 stars 146 forks source link

Should process.cpu.utilization and system.cpu.utilization be opt-in? #1130

Open ChrsMark opened 1 month ago

ChrsMark commented 1 month ago

Area(s)

area:system

Is your change request related to a problem? Please describe.

Following up from https://github.com/open-telemetry/semantic-conventions/issues/647#issuecomment-2152625724 we should decide on the requirement level for the process.cpu.utilization and system.cpu.utilization. At the moment these are recommended metrics but maybe we can consider if those can be opt-in since both of them can be calculated from their respective *.cpu.time metric.

JVM metrics define the jvm.cpu.recent_utilization as recommended however reading its description I'm not sure if that one can be derived from the jvm.cpu.time, so that might be a different case.

/cc @open-telemetry/semconv-system-approvers @braydonk @open-telemetry/semconv-jvm-approvers

Describe the solution you'd like

Having an alignment (if possible) on the requirement level for the *.cpu.utilization metrics across SemConv.

Describe alternatives you've considered

No response

Additional context

No response

mx-psi commented 1 month ago

I support making them opt-in

trask commented 1 month ago

there was a semi-related discussion a while back:

https://github.com/open-telemetry/opentelemetry-specification/pull/2392#discussion_r833018035

We had the same debate for system.cpu.utilization and the result of the discussion is that system.cpu.utilization was simpler for some backends than calculating from two metrics.