SMART Pushover Monitor

A lightweight Ubuntu-friendly Bash script that monitors disk health using smartctl and sends Pushover notifications when SMART indicates trouble.

Designed for real servers, not laptops:

• Works correctly with HBAs, backplanes, and SATA bridges
• Automatically applies the correct smartctl -d <type> per disk
• Prevents alert spam with per-device cooldowns
• Fails loudly if alerting is misconfigured (no silent failures)

---

Features

• Discovers disks via smartctl --scan-open
• Correctly handles:
  • SATA disks behind HBAs (-d sat)
  • NVMe devices (-d nvme)
• Alerts on:
  • SMART overall health failures
  • Critical SMART attributes:
    • Reallocated_Sector_Ct
    • Current_Pending_Sector
    • Offline_Uncorrectable
  • NVMe wear indicators:
    • Percentage Used ≥ 90%
    • Available Spare ≤ 10%
  • Non-empty SMART error logs (when supported)
• Notifications include:
  • Hostname
  • Device path
  • Device type
  • Model and serial number
• Cooldown logic prevents repeated alerts for the same issue

---

Requirements

• Ubuntu (tested on modern LTS releases)
• Root privileges
• Packages:

sudo apt-get install -y smartmontools curl

• A Pushover account with:
  • User Key
  • Application Token

---

Installation

1. Install the script

sudo install -m 0755 smart_pushover.sh /usr/local/sbin/smart_pushover.sh

2. Create the environment file

sudo nano /etc/smart_pushover.env

Example contents:

export PUSHOVER_USER_KEY="uXXXXXXXXXXXXXXXXXXXX"
export PUSHOVER_APP_TOKEN="aXXXXXXXXXXXXXXXXXXXX"

# Optional tuning
export COOLDOWN_SECONDS=21600   # 6 hours
export STATE_DIR=/var/tmp/smart_pushover_state

Secure the file (important):

sudo chown root:root /etc/smart_pushover.env
sudo chmod 600 /etc/smart_pushover.env

---

Manual Test

Run once manually to verify everything works:

sudo bash -c 'source /etc/smart_pushover.env; /usr/local/sbin/smart_pushover.sh'

Expected behavior

Healthy disks:

OK: /dev/sda
OK: /dev/sdb
...

Unhealthy disks:

• Immediate Pushover notification
• Script output noting that an alert was sent

No notification means no problems.

---

Cron Setup (Recommended)

Add an hourly cron job as root:

sudo crontab -e

0 * * * * /bin/bash -c 'source /etc/smart_pushover.env && /usr/local/sbin/smart_pushover.sh' >/dev/null 2>&1

---

Device Detection

The script uses:

smartctl --scan-open

Example output:

/dev/sda -d sat
/dev/sdb -d sat
/dev/nvme0 -d nvme

---

Cooldown Logic

To prevent alert storms:

• Each unique device + issue signature creates a cooldown stamp
• Default cooldown: 6 hours
• Controlled via COOLDOWN_SECONDS

This ensures:

• New problems alert immediately
• Repeated alerts for the same issue are suppressed
• Escalations still trigger alerts when the failure mode changes

---

Failure Behavior (Intentional)

The script exits with an error if:

• Pushover credentials are missing or placeholders
• Required tools (smartctl, curl) are unavailable

Silent disk-monitoring failures are worse than noisy ones.

---

Notes

• Must be run as root to access SMART data on many devices
• USB/SATA bridges and HBAs are fully supported
• NVMe wear thresholds follow common industry guidance

---

Possible Enhancements

This script is intentionally minimal, but can be extended to:

• Map /dev/sdX → /dev/disk/by-id → physical bay
• Track SMART attribute trends over time
• Escalate alert priority as values increase
• Export JSON for LibreNMS or Prometheus
• Schedule and monitor SMART self-tests