Ransomware guard, heartbeat, Prometheus metrics, one-click repair, skip next run, disk power

[firsttris]

This PR adds the six features snapraid-daemon 2.0 has and SnapRAID UI lacked.

Features

1. Sync guard for changed files (ransomware)
   • A scheduled sync is skipped when diff reports more updated files than Max. changed files. The default is 100 for new schedules, the same as the daemon's sync_threshold_updates.
   • Why: ransomware encrypts files in place, and the next sync would overwrite their parity, which is the copy you could restore them from.
   • The schedule row shows the limit. The skip reason and the notification name the risk.
   • Existing schedules have no limit until you edit them.
2. Heartbeat
   • After every successful scheduled run, the UI sends a GET to a healthchecks.io, Uptime Kuma or Dead Man's Snitch URL.
   • If the pings stop (server down, runs failing or skipped), the monitoring service raises the alarm.
   • It sits as a card under Notifications, with a Send ping button.
3. Prometheus metrics at /api/metrics
   • Switched on under Automation; a token is generated, and the card shows a ready scrape_configs entry.
   • Covers array health, last sync and scrub, bad blocks, disk usage, SMART temperatures and sectors, and schedules.
   • Values come only from cached data (logs, last status read, histories), so a scrape never wakes a disk.
   • Not behind the login; it checks its own bearer token instead.
4. Repair and verify
   • One button runs fix -e and then scrub -p bad (POST /api/snapraid/heal).
   • The scrub only runs after a successful fix and only when no other job started in between.
   • Verify only stays available as a second button.
5. Skip next run
   • A button per schedule skips the next timed run once, and a second click takes it back.
   • The skipped run is recorded with "Skipped once, as asked." and sends no notification. Run now ignores the flag.
6. Disk power and temperature
   • A ⋯ menu per disk and Spin up/down in the header for the whole array (snapraid up/down, POST /api/snapraid/power).
   • SnapRAID's error message is shown when a disk can't be controlled.
   • The status column shows the temperature from SMART with a sparkline of the last 30 reads.

Other changes

• executeSnapraidCommand also returns the exit code.
• The status route and the scheduler's disk check remember the last status, for the metrics.
• docs/architecture.md now also lists POST /api/schedules/:id/run from End-to-end tests against real SnapRAID, snapraid-daemon and the Docker image #26, which was missing.

Tests

• Backend: 135 passed. New tests:
  • update guard and skip-next in the scheduler
  • heartbeat logic, and the ping against a local server
  • metrics rendering
  • the heal sequence
  • disk power arguments and errors
• Frontend: Biome, typecheck, Vitest 27 (new: temperature trend).
• E2E: 15/15, locally and against the Docker image. New E2E cases:
  • one-click repair of bit rot
  • ransomware-like mass changes stopping the sync
  • skip next and taking it back
  • heartbeat on Send ping and after a scheduled run
  • metrics refused without the token and correct with it
  • spin up/down plumbing
• Docs: mkdocs build --strict passes. Updated pages:
  • scheduling.md, notifications.md (Heartbeat), automation.md (Prometheus metrics with metric list and alert examples)
  • usage.md, architecture.md (API), development.md
  • README feature list

🤖 Generated with Claude Code

https://claude.ai/code/session_01JuXnquCPdbG5sBbDyin7yF