hdparm

Overview

hdparm gets and sets ATA/SATA disk parameters: identify data, power management, read-ahead, write-cache, and controversial performance timings. Useful for bare-metal disk troubleshooting. Wrong settings can risk data loss — especially write-cache and dangerous benchmark flags.

Prefer vendor tools / smartmontools for health; use hdparm carefully.

Syntax

hdparm [options] device

Device is typically /dev/sda (whole disk), not a partition.

Common Options

Option	Description
-I	Detailed identity information
-i	Kernel’s identify summary
-C	Power mode status
-M	AAM acoustic (legacy)
-B	APM level
-S	Standby timeout
-W	Write-cache on/off query/set
-a / -A	Read-ahead
-t / -T	Timing buffered disk / cache reads (benchmark load)
--security-*	ATA security (dangerous)
-y / -Y	Standby / sleep now
-z	Reread partition table

Examples with Explanations

Identify

sudo hdparm -I /dev/sda
sudo hdparm -i /dev/sda
sudo smartctl -a /dev/sda

Power status

sudo hdparm -C /dev/sda

Read-only info queries (safer)

sudo hdparm -W /dev/sda
sudo hdparm -a /dev/sda

Benchmarks (I/O heavy; maintenance window)

sudo hdparm -tT /dev/sda

Interprets sequential throughput roughly; not a substitute for fio.

Write cache (know the risk)

sudo hdparm -W /dev/sda
# enabling write cache without barriers/UPS can lose data on power loss

Prefer filesystem/mount and device defaults unless you know the storage stack.

Reread partitions

sudo hdparm -z /dev/sda
# often partprobe / blockdev --rereadpt also used

Notes / Pitfalls

• Never experiment with --security-erase / passwords on disks you care about without full understanding.
• USB bridges and NVMe devices may ignore or poorly implement ATA commands — NVMe has nvme CLI.
• Cloud block volumes often virtualize away hdparm usefulness.
• Timing tests stress disks — avoid on production peaks.
• Requires root.

2026-relevant notes

• NVMe dominates SSDs: learn nvme list, nvme smart-log alongside hdparm.
• For spinning rust power policies on laptops, distro power tools may be safer than manual -B/-S.
• Always have backups before storage experiments.

Related Commands

• smartctl — SMART health
• nvme — NVMe admin
• lsblk / fdisk / blkid — topology
• fio — proper benchmarks
• blockdev — low-level block ioctls

Additional Resources

• man hdparm (read warnings)