For the metric cpu.total_usage on the containerd plugin (returned by POST /api/infrastructure-monitoring/metrics), what is the precise unit and range?
Specifically: is 1.0 equivalent to one fully-utilized CPU core (i.e. the value can exceed 1.0 on multi-core containers, capped by the container's CPU limit or host core count), or is it a normalized 0–1 fraction? Or something else altogether? The metric metadata declares formatter: PERCENTAGE but we observe values up to ~2.0 on a container with cpu_limit = 2000m, which suggests the unit is "cores", not “fraction”.
Does anyone know anything about this? I have been search in the docs without success