Luxor Documentation Hub Logo

Critical Temperature Sensor Behavior

How LuxOS reacts to working/failed critical temperature sensors per board and miner-wide

Critical Temperature Sensor Behavior Overview

This page documents how LuxOS decides whether to keep hashing, reboot a board, or shut the miner down based on the working temperature sensors available on each hashboard. The behavior depends on the operator-configured min_per_board floor, the miner's hardware (cooling type and ASIC die-temperature coverage), and which sensors are marked as critical.

At a glance

  • Hydro water-sensor failure always reboots the affected board, regardless of min_per_board or die-temp coverage.
  • min_per_board > 0 enforces an operator-configured floor — a board drops below it ⇒ board reboot.
  • ASIC on-die temps and water sensors are fallback protections — either one alone is enough to keep the miner hashing when critical board sensors are gone.
  • The miner only shuts itself down when min_per_board == 0 and no critical board sensors, no die-temp coverage, and no working water sensors remain anywhere.

Decision Table Column definitions

  • Water Temps (Hydro): N/A (air-cooled), OK (≥1 working), FAIL (any failed). Water is only ever critical on hydro.
  • ASIC Die Temp Coverage: on-die overheat coverage — Yes for BM1368 / BT1978, No otherwise. See HasDieTempOverheatCoverage in config.
  • Critical Board Temps: working critical (hot-side, non-water) board sensors on this board; min = min_per_board. See IsCritical in temps.
  • min_per_board: configured floor; 0 disables per-board minimum enforcement and enables the miner-level shutdown check.
  • Any critical board temps left miner-wide: ≥1 if any board still has a working critical board sensor, die coverage, or working water sensor; 0 if none anywhere.

Decision table

#Water Temps (Hydro)ASIC Die Temp CoverageCritical Board Tempsmin_per_boardAny critical board temps left miner-wideBoard actionMiner action
1FAILanyanyany—Reboot boardRunning
2anyany< min> 0—Reboot boardRunning
3N/A or OKNo0 (all hot-side failed)any—Reboot board †Running
4N/A or OKYes0 (all hot-side failed)≥ min (i.e. min=0)—Keep hashing (die coverage)Running
5N/ANo0 on every board00—Miner shutdown
6OK anywhereNo0 on every board0≥ 1 (water OK somewhere)Keep hashingRunning
7anyYes on any board0 on every board0≥ 1 (die coverage)Keep hashingRunning
8N/A or OKany≥ min and ≥ 1 hot-side OKany—Keep hashingRunning

† Row 3 has a pre-existing exception for hashboards with a PIC (has_pic_on_hashboard == true): the PIC provides an independent on-chip overtemp backstop, so the critical-failure check tolerates "all hot-side failed" on those boards and the row-3 reboot does not fire. Affected classes: S19j, S19jpro, S19k Pro, older PIC S19 variants, S19 Pro+ Hydro, S19 XP Hydro, S21 Hydro.

How each cooling type and ASIC chip combination maps to the Decision Table

ClassExample modelsWater Temps (Hydro)ASIC Die Temp CoverageReachable rows
Air, no chip die temp coverageS19j, S19jpro, S19k Pro, older T19/S19 PIC boardsN/ANo2, 3 †, 5, 8
Air, die temp coverage (BM1368)S19 XP+, S21, T21N/AYes2, 4, 7, 8
Air, no die temp coverage (BM1370)S21 Pro, S21 XP, S21+N/ANo2, 3, 5, 8
Air, die temp coverage (BT1978)Whatsminer M50 / M50S / M50S+ / M60 / M60sN/AYes2, 4, 7, 8
Hydro, die temp coverage (BM1368)S19 XP+ Hydro, S21 HydroOK/FAILYes1, 2, 4, 7, 8
Hydro, no die temp coverageS19 Pro+ Hydro, S19 XP HydroOK/FAILNo1, 2, 3 †, 6, 8
Hydro, no die temp coverage (BM1370)S21+ Hydro, S21 XP Hydro, S21e Hydro, U3 S21e XP HydroOK/FAILNo1, 2, 3, 6, 8

Key invariants

  • Row 1 fires before any other check on hydro: water-sensor loss always reboots the board, regardless of min_per_board or die coverage.
  • Row 2 is the only path that can reboot a board for "below the operator's configured minimum"; it is disabled when min_per_board == 0.
  • Row 5 is the only path that can shut the miner down from this subsystem; it requires min_per_board == 0 and zero working critical board sensors miner-wide and no die coverage and no working water sensors anywhere.
  • Rows 4, 6, 7 are the fallback protections: ASIC die temps and water temps, each of which individually keeps the miner hashing when critical board temps are gone.

On this page