Check Server Disk Health with SMART and smartctl

SMART (Self-Monitoring, Analysis and Reporting Technology) is built into hard drives and SSDs and records signs of wear and failure. Checking it regularly on a dedicated server lets you replace a dying disk before it takes data with it. This guide uses smartctl to check disk health on SATA and NVMe drives.

Before you start

Root access to a dedicated or bare-metal server. On a VPS, virtual disks usually do not expose SMART data – the host monitors the physical drives.
Drives behind a hardware RAID controller may need extra options (for example -d megaraid,0).

Install smartmontools

Ubuntu/Debian: sudo apt install -y smartmontools
AlmaLinux/Rocky: sudo dnf install -y smartmontools

List drives and check overall health

Find the drives: sudo smartctl --scan or lsblk -d.
Quick health verdict:
```
sudo smartctl -H /dev/sda
```
You want PASSED (SATA) or OK (NVMe). A pass does not guarantee health – read the details too.
Full report:
```
sudo smartctl -a /dev/nvme0
```

SATA attributes that matter

ID	Attribute	Warning sign
5	Reallocated_Sector_Ct	Any non-zero raw value that keeps rising
187	Reported_Uncorrect	Non-zero
197	Current_Pending_Sector	Non-zero – sectors waiting to be remapped
198	Offline_Uncorrectable	Non-zero
199	UDMA_CRC_Error_Count	Rising – often a cable or backplane issue, not the disk
177/231/233	Wear leveling / SSD life (vendor-specific)	Approaching end of rated life

NVMe health fields

Critical Warning – should be 0x00.
Percentage Used – estimated wear; 100% means rated endurance reached.
Available Spare – should stay above the threshold.
Media and Data Integrity Errors – should be 0.
Temperature – sustained high temperatures shorten life and cause throttling.

Run self-tests

Short test (a few minutes): sudo smartctl -t short /dev/sda
Long test (hours on large HDDs): sudo smartctl -t long /dev/sda
Read results when done: sudo smartctl -l selftest /dev/sda

Self-tests run in the background while the server keeps working, though long tests add some I/O load.

Enable automatic monitoring with smartd

Edit /etc/smartd.conf (or /etc/smartmontools/smartd.conf) and replace the DEVICESCAN line with:
```
DEVICESCAN -a -o on -S on -s (S/../.././02|L/../../6/03) -m admin@example.com
```
This monitors all disks, runs a short test daily at 02:00, a long test on Saturdays at 03:00, and emails on problems.
Make sure the server can send mail, then enable the service: sudo systemctl enable --now smartd (named smartmontools on some Debian versions).

What to do when a disk is failing

Confirm your latest backup is good.
Check RAID state with cat /proc/mdstat or the controller tool.
Note the drive's serial number from smartctl -i so the right disk is replaced.
Open a ticket with your provider to replace it, then rebuild the array.

FAQ: SMART disk health

Can SMART predict every failure?

No. Many failures show warning signs first, but some drives fail suddenly. RAID and backups are still essential.

Why does smartctl show nothing on my VPS?

Virtual disks don't pass through SMART. Your provider monitors the physical hardware.

Is a rising CRC error count a bad disk?

Usually it points to a cable, connector or backplane issue rather than the disk itself.

Next steps

Was this guide helpful?

Still stuck?

Ask MdmHost AI for a quick answer, or open a ticket and our team will take a look. We are here 24/7.

Ask MdmHost AI Open a ticket