truenas-burnin/claude-sandbox
echoparkbaby 3e0000528f TrueNAS Burn-In Dashboard v0.9.0 — Live mode, thermal monitoring, adaptive concurrency
Go live against real TrueNAS SCALE 25.10:
- Remove mock-truenas dependency; mount SSH key as Docker secret
- Filter expired disk records from /api/v2.0/disk (expiretime field)
- Route all SMART operations through SSH (SCALE 25.10 removed REST smart/test endpoint)
- Poll drive temperatures via POST /api/v2.0/disk/temperatures (SCALE-specific)
- Store raw smartctl output in smart_tests.raw_output for proof of test execution
- Fix percent-remaining=0 false jump to 100% on test start
- Fix terminal WebSocket: add mounted key file fallback (/run/secrets/ssh_key)
- Fix WebSocket support: uvicorn → uvicorn[standard] (installs websockets)

HBA/system sensor temps on dashboard:
- SSH to TrueNAS and run sensors -j each poll cycle
- Parse coretemp (CPU package) and pch_* (PCH/chipset — storage I/O proxy)
- Render as compact chips in stats bar, color-coded green/yellow/red
- Live updates via new SSE system-sensors event every 12s

Adaptive concurrency signal:
- Thermal pressure indicator in stats bar: hidden when OK, WARM/HOT when running
  burn-in drives hit temp_warn_c / temp_crit_c thresholds
- Thermal gate in burn-in queue: jobs wait up to 3 min before acquiring semaphore
  slot if running drives are already at warning temp; times out and proceeds

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-02-27 06:33:36 -05:00
..
truenas-burnin TrueNAS Burn-In Dashboard v0.9.0 — Live mode, thermal monitoring, adaptive concurrency 2026-02-27 06:33:36 -05:00