feat: fan overhaul (sensors, curve types, GPU fan)#97
Merged
Conversation
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
…columns) Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
…Service - Add imports: TempSourceRegistry, HwmonTempSource, GpuTempSource, DiskTempSource, MixTempSource, evaluate_curve, time as _time - Add _registry, _last_pwm_by_fan, _last_tick_ts to __init__ - Add _rebuild_registry() + helpers: _make_hwmon_reader, _make_gpu_reader, _make_disk_reader, _list_smart_devices, _register_composites_from_db, _load_sensor_labels - Call _rebuild_registry() in start() after _initialize_backend(), before _load_fan_configs() - Remove silent CPU-sensor auto-correction elif branch from _load_fan_configs (Step 3b): user-chosen sensors (including composite mix: IDs) now survive service restarts unchanged - Replace _monitor_and_control_fans body: temperature lookup now routes through self._registry.get_temp(), curve evaluation uses evaluate_curve(), hysteresis applied only for graph curve_type - Update TestDbAutoCorrection tests to verify the new no-correction behavior - Add test_fan_sensor_assignment.py: 5 regression tests locking in that user- chosen sensors (non-CPU hwmon, mix:, gpu:, disk:) survive _load_fan_configs Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
- Extend TempSensorInfo schema with custom_label, kind, gpu_vendor fields
- Add TempSensorLabelUpdate, CompositeSensor{Create,Update,Info,ListResponse} schemas
- Replace list_temp_sensors body: reads from registry (all source kinds) instead
of backend-only hwmon sensors
- Add PUT/DELETE /api/fans/sensors/{sensor_id}/label endpoints
- Add GET/POST/PUT/DELETE /api/fans/composite-sensors endpoints with 5-sensor cap
- Module-level MAX_COMPOSITES_PER_SYSTEM = 5 constant in routes/fans.py
- Delete composite unlinks FanConfig rows pointing at the composite
- Add sqlalchemy.select and func imports to routes/fans.py
- 18 new tests in test_fan_sensor_label_api.py and test_fan_composite_api.py
using service-layer integration pattern (db_session fixture)
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
- Add CurveTypeSelector (graph/flat/target/mix/sync tab strip) - Add CurveEditorFlat (single PWM% slider) - Add CurveEditorTarget (target temp + PWM sliders) - Add CurveEditorMix (two profile selects + max/sum function toggle) - Add CurveEditorSync (fan-to-fan sync selector) - Integrate all editors into FanDetails with conditional rendering - Add allFans prop to FanDetails; pass status.fans from FanControl - Extend FanInfo and UpdateFanConfigRequest types with curve-type fields - Export all 5 new components from index.ts Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
…ensor label - Add AdvancedFanSettings component (start_pwm, stop_below_temp, response_time, pwm_steps) - Add GpuManualModeToggle component (AMD-only, local state, POST /gpu-manual-mode) - FanCard: GPU badge, last_write_error chip, sensor label next to temperature, no-sensor amber notice - FanDetails: wire AdvancedFanSettings + GpuManualModeToggle (AMD gate), handleAdvancedChange - SensorsPanel: convert from self-fetching to prop-driven (sensors, composites, onReload) - FanControl: lift sensors/composites fetch into page, pass sensors down to FanCard + SensorsPanel - api/fan-control: add setGpuManualMode(), extend UpdateFanConfigRequest with Task 16 fields - index.ts: export AdvancedFanSettings and GpuManualModeToggle Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
6 tasks
Xveyn
added a commit
that referenced
this pull request
May 24, 2026
## Summary Fixes four issues with the fan SensorsPanel that surfaced in production after the fan overhaul (#97): - **GPU sources showed `—`.** `_make_gpu_reader` reads `data[\"gpu\"]` from `telemetry.json`, but the snapshot writer never published that key. The GPU collector was persisting samples to the DB only. - **Disk sources were missing entirely.** The per-worker SMART cache (`_SMART_CACHE_DATA`) is empty in every web worker unless that worker has personally hit `/api/system/smart/status`. With 4 Uvicorn workers, this was non-deterministic and usually empty. `_rebuild_registry` only ran at startup — when the cache is always empty. - **amdgpu hwmon entries duplicated `gpu:*` sources.** The same physical sensor appeared twice in the panel with different IDs. - **0°C dead sensors cluttered the panel.** Disconnected motherboard inputs (PCH_CHIP_TEMP, PCH_CPU_TEMP, …) always read 0.0°C on this board. ## Approach | Commit | What | |---|---| | `dbd54c8b` | `_write_telemetry_snapshot` includes `data[\"gpu\"]` from `orchestrator.get_gpu_current()`. fan_control reader unchanged. | | `4d222fb1` | Monitoring worker publishes `smart_summary.json` every 60s (gated by the existing 120s smartctl cache). fan_control reads disk temps from there. New `_refresh_disk_sources()` reconciles disk:* registry entries on every `/api/fans/sensors` call. Adds `TempSourceRegistry.unregister()`. | | `212d949e` | `_rebuild_registry` skips hwmon sources with `device_name in {\"amdgpu\", \"nouveau\"}`. Documented trade-off for hypothetical iGPU+dGPU systems. | | `276a10d4` | `SensorsPanel`: 0°C sensors collapse into an \"Inaktiv (N)\" toggle (hidden by default, 60% opacity when shown). Renamed sensors show their original kernel label in the subtitle (\"edge · gpu:edge\") so user-chosen names like \"Composite\" don't blend with actual mix sensors. | ## Tests - 4 new test files, 18 new tests: - `test_worker_service_telemetry_gpu.py` — GPU SHM publish behavior - `test_worker_service_smart_summary.py` — SMART summary SHM publish behavior - `test_fan_disk_sources_from_shm.py` — disk reader/list/refresh - `test_fan_gpu_hwmon_dedup.py` — amdgpu/nouveau hwmon suppression - 300 existing fan + monitoring tests still pass ## Test plan - [ ] After deploy, open `/fans` on prod - [ ] Verify `gpu:edge` / `gpu:junction` / `gpu:mem` show real temperatures (not `—`) - [ ] Verify `hwmon:hwmon1_temp*` (amdgpu duplicates) are gone - [ ] Wait up to ~2 minutes, verify `disk:sda` / `disk:nvme0n1` / etc. appear with temperatures - [ ] Verify PCH 0°C sensors are no longer in the main grid, but still reachable via \"Inaktiv (4)\" toggle - [ ] Verify renamed sensors show their original kernel label below the chosen name 🤖 Generated with [Claude Code](https://claude.com/claude-code)
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Brings the Fan Editor close to the FanControl (Rem0o) feature set, adds GPU fan support, and fixes the silent CPU-sensor override that made case fans always display CPU temperature.
Backend
TempSourceRegistryresolving namespaced sensor IDs (hwmon:,gpu:,disk:,mix:)fan_curve_eval:graph,flat,target,mix,syncstart_pwm_percent,stop_below_temp_celsius,response_time_seconds(EMA),pwm_stepslast_write_errorpower_dpm_force_performance_level=manual,pwm1_enable=1) with restore on disable_load_fan_configs— user-chosen sensors now survive restartsFrontend
SensorsPanelon/fans: inline rename, kind badges (CPU/GPU/Disk/Mix), composite-sensor modalCurveTypeSelector+ 5 typed editors (Flat,Target,Mix,Sync, plus existingGraph)AdvancedFanSettingscollapsible (start PWM, stop-below-temp, response time, PWM steps)GpuManualModeToggle(AMD GPU only) with warning bannerFanCard: GPU badge,PWM-Fehlerchip on write errors, sensor name shown next to temperature, "Kein Sensor zugewiesen" call-to-action when nullSpec & plan
docs/superpowers/specs/2026-05-24-fan-overhaul-design.mddocs/superpowers/plans/2026-05-24-fan-overhaul.mdMigration
One Alembic migration: new
temp_sensor_labels+composite_temp_sensorstables; 12 nullable/defaulted columns added tofan_configs. Existingcurve_typedefaults to"graph"— no breaking change.Test plan
pytest -k fan— 154 passedpytest tests/{api,services,database,middleware}— 1150 passed (4 unrelated failures intest_plugins_marketplace_routes.py, pre-existing)npx tsc --noEmit— cleannpm run build— built in 6.92s/fans, rename sensor, create mix, switch curve types, toggle advanced🤖 Generated with Claude Code