Skip to content
openentropy logo openentropy

Telemetry Model

telemetry_v1 is a best-effort environment snapshot model used to contextualize entropy measurements.

Design contract:

  • One model id/version: telemetry_v1 / 1
  • One structured report shape for diagnostics: TelemetryWindowReport (start, end, deltas)
  • One point-in-time shape for ad hoc display: TelemetrySnapshot

Purpose

Section titled “Purpose”
  • Capture host conditions during benchmark/analysis runs.
  • Provide reproducible context for comparing runs across machines and time.
  • Expose measured values instead of embedding hidden assumptions.

Output Shape

Section titled “Output Shape”

TelemetrySnapshot includes:

  • model id/version
  • timestamp (collected_unix_ms)
  • host identity (os, arch, cpu_count)
  • load average (1m, 5m, 15m) when available
  • metric list (domain, name, value, unit, source)

TelemetryWindowReport includes:

  • start snapshot
  • end snapshot
  • elapsed_ms
  • aligned metric deltas (start_value, end_value, delta_value)

Metric Domains

Section titled “Metric Domains”

Depending on host support and permissions:

  • system (e.g. uptime)
  • memory (total/free/available and related counters)
  • scheduling (CPU time counters, runnable/process counters, tasks/threads)
  • pressure (Linux PSI and macOS memory pressure level when available)
  • frequency (timebase/cpu frequency proxies)
  • network (aggregate byte/packet/error counters)
  • disk (aggregate I/O counters and timings)
  • entropy (Linux kernel entropy pool availability)
  • vm (macOS vm_stat counters)
  • thermal (sensor temperatures, typically Linux hwmon)
  • voltage / current / power (typically Linux hwmon)
  • cooling (fan RPM where exposed)

Unavailable domains are omitted rather than synthesized.

Where It Appears

Section titled “Where It Appears”
  • openentropy scan --telemetry
  • openentropy monitor --telemetry
  • openentropy bench --telemetry
  • openentropy analyze --telemetry
  • openentropy analyze --report --telemetry
  • openentropy sessions ... --telemetry
  • openentropy record --telemetry (stored in session.json)
  • openentropy server --telemetry (startup snapshot)
  • HTTP diagnostics:
    • /sources?telemetry=true
    • /pool/status?telemetry=true
    • returns telemetry_v1 as a TelemetryWindowReport

CLI JSON outputs that include telemetry store it at:

  • telemetry_v1 (TelemetryWindowReport)

Session recordings store telemetry at:

  • session.json -> telemetry_v1 (TelemetryWindowReport)
  • Backward compatibility: reader accepts legacy session.json -> telemetry.

scan --telemetry, monitor --telemetry, and server --telemetry print a startup snapshot to stdout for operator context.

Interpretation

Section titled “Interpretation”
  • Treat telemetry as context, not a direct entropy score.
  • Use telemetry deltas to explain run-to-run changes in entropy metrics.
  • Do not infer unavailable physical channels (e.g., rail voltage) on platforms where they are not exposed.