Control plane

Fabric unknown — the byte-rate cannot be labelled, because the adapter has not been scraped yet. Absent reads —; a real zero reads 0. The backend says what computes the physical layer; the fabric says what carries it. A sim or loopback byte-rate is never labelled a radio number. A sample is pushed to the trace only on a successful read, so use the freshness strip above to tell a flat line from a dead feed.

Where this number comes from, and what it is a count of. The top figure is the RAN adapter's own ran_attached_ues series, read once at /api/ran/metrics. The body is checked before a number is taken out of it: it must be a Prometheus exposition carrying ran_backend_info, so a health blob from some other service on the same port cannot become a UE count. If the check fails the panel says Unknown and prints nothing.

It counts UEs. It does not count PDU sessions. One UE can hold several; the UE on this stack holds four. Four is a different quantity, it is not shown here, and the two are never added. Sim load is a demand generator with no radio behind it, and is never added to either.

Research preview, not a spectrum-access authority. On the 28-channel UHF survey of 2026-08-04 (560 records, 10 occupied / 15 vacant, ground truth from measured physics) this checkpoint detected 2 of 10 live multiplexes and called 48 of 300 vacant records occupied, at 0.93 to 0.99 confidence. With the physics detector wired in — same unchanged checkpoint, one process, differing only in whether a calibrated control floor is passed — that becomes 10 of 10 multiplexes and 0 of 15 vacant channels called occupied. Not clean yet: UHF 28 still asserts OCCUPIED on 8 of 20 records, and that threshold was not moved. This caveat used to blame the threshold for calibration on synthetic noise; a measured receiver-noise null falsified that on 2026-08-05 — real receiver noise tops out at comb z = 2.289 against the synthetic 2.19, and 0 of 300 records reach the 3.5 threshold. The firing arrives through the antenna and its cause is now open: a genuine emission below the energy floor, which would mean the ground truth needs revising rather than the detector, or front-end intermodulation. The RX-gain sweep that separates them has not been run. The earlier “40 of 80” figure came from a 2026-07-29 set that was entirely occupied, on which answering OCCUPIED unconditionally also scores 80 of 80 — it measured no discrimination. Rollout is opt-in and this service was not restarted: with no calibrated control floor configured the engine falls back to the legacy CNN verdict and reports occupancy_source: cnn-confidence, so the occupancy shown above is the classifier’s unless a floor is set. Model confidence is not accuracy: high confidence has been measured on wrong answers, so the theta_sense = 0.85 gate does not catch them. PAWS grant state is the authority.

Force locks the controller. Reset returns it to autonomous hysteresis (trace-driven). Degrade at —, recover at —, — transitions so far. Controller state: Awaiting the controller..

Disaster scenario: down drops the regulator; the waiver pre-authorises sensed channels. The state shown is read back from the database, never assumed from the button you pressed.

Everything on this card is simulated demand. None of it registers on a radio, and none of it appears in the real-attach figure in section 02.

no job started

Allowlisted profiles only, confirmation required, and never down -v. The line above is the control API's answer, not an assumption about what happened.

A service reported absent is not in the active profile, which is different from unhealthy. Absence is neither a pass nor a failure.

The capability register has not answered yet.

/api/control/{whoami,users,provision,access,auth,esim}. Ki and OPc are never returned by the API and are never rendered here. Invite codes are shown once.