SMART is predictive, not prophetic — a PASSED status does not guarantee the disk survives tomorrow. Replace drives with rising reallocated or pending sectors before they take your data with them.
What you will achieve
Install smartmontools on Ubuntu or Debian, run SMART health checks on HDDs and SSDs, and interpret the attributes that signal imminent failure.
1) Install smartmontools
sudo apt update
sudo apt install smartmontools2) Check if SMART is supported and enabled
sudo smartctl -i /dev/sda
sudo smartctl -i /dev/nvme0n1Look for SMART support is: Enabled. USB adapters often hide SMART — results may be unavailable for external drives.
3) Overall health summary
sudo smartctl -H /dev/sda
sudo smartctl -H /dev/nvme0n1PASSED or OK is good. FAILED means replace the drive now, not after lunch.
4) Full attribute report (HDD/SATA SSD)
sudo smartctl -a /dev/sdaWatch these attributes:
- Reallocated_Sector_Ct — rising count means bad sectors swapped out
- Current_Pending_Sector — sectors waiting to remap; investigate immediately
- UDMA_CRC_Error_Count — often cable/port issue, not disk media
- Power_On_Hours — age context for wear decisions
5) NVMe specific checks
sudo smartctl -a /dev/nvme0n1
sudo nvme smart-log /dev/nvme0n1 # if nvme-cli installedMonitor Percentage Used (wear), Media Errors, and Critical Warning flags.
6) Run self-tests
sudo smartctl -t short /dev/sda # ~2 minutes
sudo smartctl -t long /dev/sda # hours — schedule off-peak
sudo smartctl -l selftest /dev/sdaSelf-test log shows pass/fail and which LBA failed if problematic.
7) Enable SMART monitoring daemon
sudo systemctl enable --now smartd
sudo smartctl -s on /dev/sdaConfigure /etc/smartd.conf to email on attribute changes — especially for servers without hands-on disk access.
8) When to replace
Replace on FAILED health, steadily climbing reallocated sectors, or NVMe critical warnings. Plan replacement on elevated wear above 80% on production SSDs. Keep backups regardless — SMART catches many failures, not all.
9) Schedule regular checks
echo 'DEVICESCAN -a -o on -s (S/../.././02|L/../../7/04)' | sudo tee /etc/smartd.confShort self-test every day at 2 AM, long test weekly — adjust mail address in smartd config for alerts on attribute threshold breaches.
Verify
sudo smartctl -H /dev/sda
sudo smartctl -A /dev/sda | grep -E 'Reallocated|Pending|Power_On'Baseline attributes after install; compare monthly to spot trends before catastrophic failure.