Hard Drive Health Dashboard & Monitoring for S.M.A.R.T metrics

This fork exists to keep Scrutiny alive and growing. The original AnalogJ/scrutiny project development slowed significantly in 2024, while community contributions and feature requests continued to grow. This fork picks up where the original left off, merging pending community PRs and adding new features.

Full credit for the original vision and architecture goes to AnalogJ. I started this fork as a learning project, so contributions from more experienced developers are greatly appreciated. Full disclosure: I use Claude to assist with development, but all code is manually reviewed by me before merging.

	Original	This Fork
Latest Version	v0.8.1 (Apr 2024)
Frontend	Angular 13	Modern Angular
Status	Minimal updates	Actively maintained
Community PRs	Many pending	Merged

• ZFS Pool Monitoring - Monitor ZFS pool health alongside individual drives
• Prometheus Metrics - Export metrics to Prometheus for advanced monitoring
• Device Archiving - Hide decommissioned drives without deleting history
• Per-Device Notification Control - Mute notifications for specific devices
• Device Labels - Add custom labels to drives for easier identification
• Day-Resolution Temperature Graphs - More granular temperature history
• SAS Temperature Support - Proper temperature readings for SAS drives
• SCT Temperature History Toggle - Control SCT ERC settings per drive
• S.M.A.R.T Attribute Overrides - Override manufacturer thresholds via UI or config
• Improved Dashboard Layout - Sidebar navigation moved to top for better attribute visibility
• Enhanced Mobile UI - Optimized layout for mobile devices
• Performance Benchmarking - Run fio benchmarks and track drive throughput, IOPS, and latency over time
• Scheduled Reports [WIP] - Automated daily/weekly/monthly health reports via email with HTML formatting
• Missed Ping Digest - Consolidated notification when multiple collectors miss pings (instead of one email per device)
• HTML Email Notifications - Rich HTML emails for reports and missed ping alerts via SMTP
• Enhanced Seagate Drive Support - Better timeout handling for Seagate drives
• Workload Insights - Visualize daily read/write rates, I/O intensity, SSD endurance, and activity spike detection
• SHA256 Checksums - Verify release binary integrity

If you run a server with more than a couple of hard drives, you're probably already familiar with S.M.A.R.T and the smartd daemon. If not, it's an incredible open source project described as the following:

smartd is a daemon that monitors the Self-Monitoring, Analysis and Reporting Technology (SMART) system built into many ATA, IDE and SCSI-3 hard drives. The purpose of SMART is to monitor the reliability of the hard drive and predict drive failures, and to carry out different types of drive self-tests.

These S.M.A.R.T hard drive self-tests can help you detect and replace failing hard drives before they cause permanent data loss. However, there's a couple issues with smartd:

• There are more than a hundred S.M.A.R.T attributes, however smartd does not differentiate between critical and informational metrics
• smartd does not record S.M.A.R.T attribute history, so it can be hard to determine if an attribute is degrading slowly over time.
• S.M.A.R.T attribute thresholds are set by the manufacturer. In some cases these thresholds are unset, or are so high that they can only be used to confirm a failed drive, rather than detecting a drive about to fail.
• smartd is a command line only tool. For head-less servers a web UI would be more valuable.

Scrutiny is a Hard Drive Health Dashboard & Monitoring solution, merging manufacturer provided S.M.A.R.T metrics with real-world failure rates.

• Web UI Dashboard - focused on Critical metrics
• smartd integration (no re-inventing the wheel)
• Auto-detection of all connected hard-drives
• S.M.A.R.T metric tracking for historical trends
• Customized thresholds using real world failure rates
• Temperature tracking with configurable resolution
• Provided as an all-in-one Docker image (but can be installed manually)
• Configurable Alerting/Notifications via Webhooks

• ZFS Pool Monitoring - Track pool health, capacity, and status
• Prometheus Metrics Endpoint - /api/metrics for Grafana integration
• Device Archiving - Archive old drives to declutter the dashboard
• Per-Device Notification Muting - Control which drives trigger alerts
• Custom Device Labels - Add meaningful names to your drives
• Day-Resolution Graphs - View temperature trends at daily granularity
• SAS Drive Support - Full temperature support for SAS devices
• S.M.A.R.T Attribute Overrides - Override thresholds per device via UI
• Improved UI Layout - Top navigation for better S.M.A.R.T attribute visibility
• Mobile-Optimized Interface - Better experience on mobile devices
• API Timeout Configuration - Adjust timeouts for slow storage systems
• Performance Benchmarking - fio-based benchmarks for throughput, IOPS, and latency with historical tracking
• Scheduled Reports [WIP] - Automated health reports on daily/weekly/monthly schedules with HTML emails and PDF export
• Missed Ping Digest - Batch notification when multiple collectors go unreachable
• HTML Email Notifications - Rich HTML formatting for SMTP notifications (reports and missed pings)
• Workload Insights - Daily read/write rates, R/W ratio, I/O intensity classification, SSD endurance tracking, and activity spike detection
• Heartbeat Notifications - Periodic "all clear" alerts for uptime monitoring integration

If you're currently using the original AnalogJ/scrutiny, migrating is straightforward:

1. Update your image reference from ghcr.io/analogj/scrutiny to ghcr.io/starosdev/scrutiny
2. Data is compatible - Your existing SQLite database and InfluxDB data will work without changes
3. Config files are compatible - No changes needed to scrutiny.yaml or collector.yaml

That's it! The fork maintains full backwards compatibility with the original project.

Scrutiny uses smartctl --scan to detect devices/drives.

• All RAID controllers supported by smartctl are automatically supported by Scrutiny.
  • While some RAID controllers support passing through the underlying SMART data to smartctl others do not.
  • In some cases --scan does not correctly detect the device type, returning incomplete SMART data. Scrutiny supports overriding detected device type via the config file: see example.collector.yaml
• If you use docker, you must pass though the RAID virtual disk to the container using --device (see below)
  • This device may be in /dev/* or /dev/bus/*.
  • If you're unsure, run smartctl --scan on your host, and pass all listed devices to the container.

See docs/TROUBLESHOOTING_DEVICE_COLLECTOR.md for help

If you're using Docker, getting started is as simple as running the following command:

See docker/example.omnibus.docker-compose.yml for a docker-compose file.

docker run -p 8080:8080 -p 8086:8086 --restart unless-stopped \
  -v `pwd`/scrutiny:/opt/scrutiny/config \
  -v `pwd`/influxdb2:/opt/scrutiny/influxdb \
  -v /run/udev:/run/udev:ro \
  --cap-add SYS_RAWIO \
  --device=/dev/sda \
  --device=/dev/sdb \
  --name scrutiny \
  ghcr.io/starosdev/scrutiny:latest-omnibus

• /run/udev is necessary to provide the Scrutiny collector with access to your device metadata
• --cap-add SYS_RAWIO is necessary to allow smartctl permission to query your device SMART data
  • NOTE: If you have NVMe drives, you must add --cap-add SYS_ADMIN as well.
• --device entries are required to ensure that your hard disk devices are accessible within the container.
• ghcr.io/starosdev/scrutiny:latest-omnibus is an omnibus image, containing both the webapp server (frontend & api) as well as the S.M.A.R.T metric collector. (see below)

In addition to the Omnibus image (available under the latest tag) you can deploy in Hub/Spoke mode using the following Docker images:

• ghcr.io/starosdev/scrutiny:latest-collector - Contains the Scrutiny data collector, smartctl binary and cron-like scheduler. You can run one collector on each server.
• ghcr.io/starosdev/scrutiny:latest-collector-zfs - ZFS pool collector for monitoring ZFS health. Run alongside or instead of the standard collector if you use ZFS. See docs/ZFS_POOL_MONITORING.md for setup instructions.
• ghcr.io/starosdev/scrutiny:latest-collector-performance - Performance benchmark collector using fio. Runs periodic benchmarks and tracks throughput, IOPS, and latency over time. See Performance Benchmarking for details.
• ghcr.io/starosdev/scrutiny:latest-web - Contains the Web UI and API. Only one container necessary
• influxdb:2.2 - InfluxDB image, used by the Web container to persist SMART data. Only one container necessary. See docs/TROUBLESHOOTING_INFLUXDB.md

See docker/example.hubspoke.docker-compose.yml for a docker-compose file.

docker run -p 8086:8086 --restart unless-stopped \
  -v `pwd`/influxdb2:/var/lib/influxdb2 \
  --name scrutiny-influxdb \
  influxdb:2.2

docker run -p 8080:8080 --restart unless-stopped \
  -v `pwd`/scrutiny:/opt/scrutiny/config \
  --name scrutiny-web \
  ghcr.io/starosdev/scrutiny:latest-web

docker run --restart unless-stopped \
  -v /run/udev:/run/udev:ro \
  --cap-add SYS_RAWIO \
  --device=/dev/sda \
  --device=/dev/sdb \
  -e COLLECTOR_API_ENDPOINT=http://SCRUTINY_WEB_IPADDRESS:8080 \
  --name scrutiny-collector \
  ghcr.io/starosdev/scrutiny:latest-collector

While the easiest way to get started with Scrutiny is using Docker (see above), it is possible to run it manually without much work. You can even mix and match, using Docker for one component and a manual installation for the other.

See docs/INSTALL_MANUAL.md for instructions.

Once scrutiny is running, you can open your browser to http://localhost:8080 and take a look at the dashboard.

If you're using the omnibus image, the collector should already have run, and your dashboard should be populate with every drive that Scrutiny detected. The collector is configured to run once a day, but you can trigger it manually by running the command below.

For users of the docker Hub/Spoke deployment or manual install: initially the dashboard will be empty. After the first collector run, you'll be greeted with a list of all your hard drives and their current smart status.

docker exec scrutiny /opt/scrutiny/bin/scrutiny-collector-metrics run

By default Scrutiny looks for its YAML configuration files in /opt/scrutiny/config

There are four configuration files available:

• Webapp/API config via scrutiny.yaml - example.scrutiny.yaml.
• Collector config via collector.yaml - example.collector.yaml.
• ZFS Collector config via collector-zfs.yaml - example.collector-zfs.yaml. See docs/ZFS_POOL_MONITORING.md for setup instructions.
• Performance Collector config via collector-performance.yaml - example.collector-performance.yaml. Falls back to collector.yaml if not found.

None of these files are required, however if provided, they allow you to configure how Scrutiny functions.

Unfortunately the Cron schedule cannot be configured via the collector.yaml (as the collector binary needs to be triggered by a scheduler/cron). However, if you are using the official ghcr.io/starosdev/scrutiny:latest-collector or ghcr.io/starosdev/scrutiny:latest-omnibus docker images, you can use the COLLECTOR_CRON_SCHEDULE environmental variable to override the default cron schedule (daily @ midnight - 0 0 * * *).

docker run -e COLLECTOR_CRON_SCHEDULE="0 0 * * *" ...

Scrutiny exposes a Prometheus metrics endpoint at /api/metrics. You can scrape this endpoint to integrate with Grafana or other monitoring tools.

Example Prometheus scrape config:

scrape_configs:
  - job_name: 'scrutiny'
    static_configs:
      - targets: ['scrutiny:8080']
    metrics_path: '/api/metrics'

Scrutiny can run periodic fio benchmarks on your drives and track performance over time. This helps detect drive degradation before S.M.A.R.T failures appear -- a drive that is suddenly 50% slower may be failing even if S.M.A.R.T attributes look normal.

1. The performance collector (scrutiny-collector-performance) runs fio benchmarks on configured devices
2. Results are uploaded to the Scrutiny API and stored as time-series data in InfluxDB
3. The web UI displays performance history charts and summary cards on the device detail page
4. A baseline is computed from the last 5 results, and current results are compared against it
5. Degradation detection flags warnings (>20% throughput drop or >30% latency increase) and failures (>40% / >60%)

The performance collector is available as a separate Docker image:

docker run --restart unless-stopped \
  --device=/dev/sda \
  --device=/dev/sdb \
  -e COLLECTOR_PERF_API_ENDPOINT=http://SCRUTINY_WEB_IPADDRESS:8080 \
  --name scrutiny-perf-collector \
  ghcr.io/starosdev/scrutiny:latest-collector-performance

The collector requires direct device access (not virtualized). Running benchmarks will temporarily increase I/O on the target drives, so schedule accordingly.

Performance data appears on the device detail page when benchmark results are available. The UI shows:

• Summary cards with latest values and baseline comparison badges
• Throughput chart -- sequential read/write bandwidth over time
• IOPS chart -- random read/write and mixed IOPS over time
• Latency chart -- read latency (average, P95, P99) over time

Use the duration selector to view day, week, month, or year ranges.

Scrutiny computes drive workload statistics from existing S.M.A.R.T attribute history. No additional collector configuration is required -- once at least two data points exist for a device, workload insights are available.

1. Scrutiny queries cumulative SMART counters (Total LBAs Written/Read for ATA, Data Units Written/Read for NVMe) from InfluxDB
2. The delta between the first and last data points in the selected time range is used to compute daily rates
3. Intensity is classified by total daily I/O: idle (<1 GB/day), light (1-20 GB), medium (20-100 GB), heavy (>100 GB)
4. SSD endurance is estimated from wear-leveling or percentage-used SMART attributes combined with power-on hours
5. Spike detection compares the most recent daily rate against the long-term baseline

• ATA: Uses SMART attributes 241/242 (Total LBAs Written/Read) or DeviceStats 1.24/1.40 (Logical Sectors Written/Read)
• NVMe: Uses Data Units Written/Read counters
• SCSI: Limited support (cumulative byte counters are not stored as SMART attributes)

Navigate to the Workload page from the top navigation bar. Use the duration selector (Day, Week, Month, Year, All) to adjust the analysis window. Click any row to navigate to the device detail page.

Scrutiny supports sending SMART device failure notifications via the following services:

• Custom Script (data provided via environmental variables)
• Email
• Webhooks
• Discord
• Gotify
• Hangouts
• IFTTT
• Join
• Mattermost
• ntfy
• Pushbullet
• Pushover
• Slack
• Teams
• Telegram
• Tulip

Check the notify.urls section of example.scrutiny.yml for examples.

For more information and troubleshooting, see the TROUBLESHOOTING_NOTIFICATIONS.md file

Scrutiny can send periodic "all clear" heartbeat notifications to confirm the monitoring system is running and all drives are healthy. This is useful for integration with uptime monitoring tools like Uptime Kuma.

• Disabled by default -- enable via Settings in the web UI or the /api/settings API
• Configurable interval -- defaults to every 24 hours
• Suppressed during failures -- heartbeat is not sent if any drive has active failures (failure notifications take priority)

You can mute notifications for specific devices through the web UI. This is useful for drives that are known to have issues but are being monitored before replacement.

Scrutiny can generate and email periodic health reports summarizing device status, temperature, alerts, and ZFS pool health. Reports are sent via your configured notification URLs (HTML formatting for SMTP, plain text for other services).

Note: This feature is a work in progress. It is functional and tested, but the UI and report content may change based on feedback. We'd appreciate hearing about your experience -- please open an issue with suggestions or bug reports.

Configuration is done via the Settings page in the web UI, or via the /api/settings API:

Example API call to enable daily reports:

curl -X POST http://localhost:8080/api/settings \
  -H "Content-Type: application/json" \
  -d '{"metrics": {"report_enabled": true, "report_daily_enabled": true, "report_daily_time": "07:00"}}'

On-demand report generation:

# Generate and send a report immediately
curl -X POST 'http://localhost:8080/api/reports/generate?period=daily&test=true'

# Generate a PDF report
curl -X POST 'http://localhost:8080/api/reports/generate?period=daily&format=pdf'

Report content includes:

• Overall health status (passed/warning/failed) with color-coded banner
• Summary counts (total, passed, warning, failed devices)
• Failure and warning details per device
• Device table with status, temperature, power-on hours, and alert counts
• Temperature summary (hottest/coldest devices)
• ZFS pool health (if applicable)

When multiple collectors miss their expected check-in within the configured timeout, Scrutiny sends a single consolidated notification listing all affected devices, instead of flooding your inbox with one email per device.

You can test that your notifications are configured correctly by posting an empty payload to the notifications health check API.

curl -X POST http://localhost:8080/api/health/notify

Scrutiny provides various methods to change the log level and generate log files. The web server and collector have independent log configurations and can be set separately.

The following log levels are supported (case-insensitive), listed from highest to lowest severity:

Setting a level includes all messages at that level and above (higher severity). For example, setting WARN will show WARN, ERROR, FATAL, and PANIC messages, but not INFO, DEBUG, or TRACE.

You can use environmental variables to enable debug logging and/or log files for the web server:

DEBUG=true
SCRUTINY_LOG_FILE=/tmp/web.log

You can configure the log level and log file in the config file:

log:
  file: '/tmp/web.log'
  level: DEBUG

Or if you're not using docker, you can pass CLI arguments to the web server during startup:

scrutiny start --debug --log-file /tmp/web.log

Any web server configuration key can be overridden via environment variables using the SCRUTINY_ prefix. Dots and dashes in key names become underscores.

Environment variables take precedence over config file values. This is useful for containerized deployments where you want to override specific settings without modifying the config file.

Example:

docker run -e SCRUTINY_WEB_LISTEN_PORT=9090 \
  -e SCRUTINY_WEB_INFLUXDB_HOST=influxdb.local \
  -e SCRUTINY_LOG_LEVEL=DEBUG \
  ghcr.io/starosdev/scrutiny:web

You can use environmental variables to enable debug logging and/or log files for the collector:

DEBUG=true
COLLECTOR_LOG_FILE=/tmp/collector.log

Or if you're not using docker, you can pass CLI arguments to the collector during startup:

scrutiny-collector-metrics run --debug --log-file /tmp/collector.log

Any collector configuration key can be overridden via environment variables using the COLLECTOR_ prefix. Dots and dashes in key names become underscores.

Environment variables take precedence over config file values. This is useful for containerized deployments where you want to override specific settings without modifying the config file.

Example:

docker run -e COLLECTOR_COMMANDS_METRICS_SMART_ARGS="--xall --json -T permissive" \
  -e COLLECTOR_API_ENDPOINT=http://scrutiny-web:8080 \
  ghcr.io/starosdev/scrutiny:collector

These environment variables are only available when running the collector in Docker containers (handled by the entrypoint script, not Viper configuration):

The performance collector is a separate binary (scrutiny-collector-performance) that runs fio benchmarks. It can use its own config file (collector-performance.yaml) or fall back to the main collector.yaml.

DEBUG=true
COLLECTOR_PERF_LOG_FILE=/tmp/performance.log

Or via CLI:

scrutiny-collector-performance run --debug --log-file /tmp/performance.log --profile quick

The performance collector checks COLLECTOR_PERF_ prefixed variables first, then falls back to COLLECTOR_ prefixed variables.

Example:

docker run --restart unless-stopped \
  --device=/dev/sda \
  --device=/dev/sdb \
  -e COLLECTOR_PERF_API_ENDPOINT=http://scrutiny-web:8080 \
  -e COLLECTOR_PERF_PROFILE=quick \
  -e COLLECTOR_PERF_CRON_SCHEDULE="0 2 * * 0" \
  ghcr.io/starosdev/scrutiny:latest-collector-performance

Please see the CONTRIBUTING.md for instructions for how to develop and contribute to the scrutiny codebase.

Work your magic and then submit a pull request. We love pull requests!

If you find the documentation lacking, help us out and update this README.md. If you don't have the time to work on Scrutiny, but found something we should know about, please submit an issue.

We use SemVer for versioning. For the versions available, see the tags on this repository.

Original Author: Jason Kulatunga (@AnalogJ) -- Created Scrutiny and built the foundation this fork builds upon.

Fork Maintainer: @Starosdev -- Maintaining this fork with continued development and community contributions.

• MIT
• Logo: Glasses by matias porta lezcano