Skip to main content
WebsiteGitHub last commitGitHub commit activityGitHub IssuesDocker PullsDiscordLocalized

Scrutiny: Monitor Your Hard Drives Before They Fail

· 3 min read
BankaiTech
BankaiTech
Homelab Enthusiast & Self-Hosting Advocate

Hard drives fail. It's not a matter of if, but when. Scrutiny helps you predict failures before they happen by monitoring SMART data across all your drives. Here's how to set it up.

What is SMART?​

Self-Monitoring, Analysis, and Reporting Technology (SMART) is built into most storage devices. It tracks:

  • Reallocated Sectors - Bad blocks moved to spare area
  • Spin Retry Count - Failed spin-up attempts
  • Current Pending Sectors - Unstable sectors waiting to be remapped
  • Temperature - Operating temperature
  • Power On Hours - Total running time
  • And many more...

When these metrics go bad, your drive is telling you it's dying.

Why Scrutiny?​

Scrutiny provides:

  • Web dashboard - Visual overview of all drives
  • Historical data - Track trends over time (via InfluxDB)
  • Alerting - Get notified before failures
  • Multi-server - Monitor drives across multiple machines

Docker Setup​

Here's a complete stack with InfluxDB for history:

docker-compose.yml
services:
  scrutiny:
    image: ghcr.io/analogj/scrutiny:master-omnibus
    container_name: scrutiny
    cap_add:
      - SYS_RAWIO
    ports:
      - "8080:8080"
      - "8086:8086"  # InfluxDB
    volumes:
      - scrutiny_config:/opt/scrutiny/config
      - scrutiny_influxdb:/opt/scrutiny/influxdb
      - /run/udev:/run/udev:ro
    devices:
      - /dev/sda
      - /dev/sdb
      - /dev/sdc
      # Add all your drives
    restart: unless-stopped

volumes:
  scrutiny_config:
  scrutiny_influxdb:

Finding Your Drives​

# List all block devices
lsblk

# Get detailed drive info
sudo fdisk -l

Add each drive to the devices section.

Distributed Setup (Multiple Servers)​

For monitoring drives across multiple machines:

Hub (Central Server)​

docker-compose.yml (hub)
services:
  scrutiny-web:
    image: ghcr.io/analogj/scrutiny:master-web
    container_name: scrutiny-web
    ports:
      - "8080:8080"
    volumes:
      - scrutiny_config:/opt/scrutiny/config
    restart: unless-stopped

  influxdb:
    image: influxdb:2.1
    container_name: scrutiny-influxdb
    ports:
      - "8086:8086"
    volumes:
      - influxdb_data:/var/lib/influxdb2
    restart: unless-stopped

volumes:
  scrutiny_config:
  influxdb_data:

Collector (Each Server with Drives)​

docker-compose.yml (collector)
services:
  scrutiny-collector:
    image: ghcr.io/analogj/scrutiny:master-collector
    container_name: scrutiny-collector
    cap_add:
      - SYS_RAWIO
    environment:
      - COLLECTOR_API_ENDPOINT=http://hub-ip:8080
    volumes:
      - /run/udev:/run/udev:ro
    devices:
      - /dev/sda
      - /dev/sdb
    restart: unless-stopped

Understanding the Dashboard​

Drive Health Status​

  • 🟢 Passed - All metrics healthy
  • 🟡 Warning - Some metrics outside normal range
  • 🔴 Failed - Critical issues detected

Key Metrics to Watch​

MetricWarning Signs
Reallocated SectorsAny non-zero value
Pending SectorsAny non-zero value
Uncorrectable ErrorsAny increase
Spin Retry CountValues > 0
TemperatureAbove 50°C sustained
Power On HoursReference for age

Attribute Thresholds​

Scrutiny uses thresholds from various sources:

  • Backblaze failure data
  • Manufacturer specifications
  • Community research

Alerting Setup​

Configure notifications in /opt/scrutiny/config/scrutiny.yaml:

notify:
  urls:
    - discord://webhook_id/token
    - smtp://user:[email protected]:587/?[email protected][email protected]
    - pushover://user_key:api_token

Supported services:

  • Discord, Slack, Teams
  • Email (SMTP)
  • Pushover, Gotify, Ntfy
  • Webhook (generic)

Backup Strategy Based on SMART​

Use SMART data to prioritize backups:

  1. Green drives - Normal backup schedule
  2. Yellow drives - Increase backup frequency
  3. Red drives - Immediate backup, plan replacement

Real World Example​

Last year, Scrutiny warned me about rising reallocated sectors on a 4-year-old drive. Over 2 weeks:

  • Week 1: 2 reallocated sectors (warning)
  • Week 2: 47 reallocated sectors (critical)

I replaced the drive before any data loss. Without monitoring, I would have lost data.

Drives That Fail Silently​

Some failure modes aren't caught by SMART:

  • Firmware bugs
  • Controller failures
  • Cable/connection issues

Always maintain backups regardless of SMART status!

Learn More​

What storage monitoring tools do you use? Share on Discord!