The spec says that the Agent should use OpenTelemetry semantic conventions for reporting its process metrics (e.g. CPU, memory usage).
Do we also need to define OpenTelemetry semantic conventions for reporting typical collection Agent-specific metrics (e.g. input/processing/output data rates, throughput, latency, etc)?
The spec says that the Agent should use OpenTelemetry semantic conventions for reporting its process metrics (e.g. CPU, memory usage).
Do we also need to define OpenTelemetry semantic conventions for reporting typical collection Agent-specific metrics (e.g. input/processing/output data rates, throughput, latency, etc)?