$ guides / smartctl-disk-monitoring / smartctl-g-sense-error-rate ▌

Operations Guides

G-Sense_Error_Rate rising: shock and vibration reaching the drive

G-Sense_Error_Rate (SMART attribute ID 221 on most drives, ID 191 on some) is a cumulative counter of shock and vibration events that exceeded the drive’s internal threshold, as detected by its built-in accelerometer. The attribute is HDD-only. SSDs and NVMe drives do not report it. Some enterprise HDD models do not expose the attribute at all, which does not mean vibration is absent, only that the drive does not instrument it.

When this number climbs, the first question is not “is the drive failing?” but “is the vibration causing damage?” A high raw value with no correlated media errors means the accelerometer is registering environmental noise. A growing raw value alongside rising Seek_Error_Rate (ID 7) or Reallocated_Sector_Ct (ID 5) means the vibration is causing the read/write head to contact the platter surface (head slap), producing permanent media damage.

The second question is “what is the source?” The accelerometer cannot tell you whether the vibration comes from the drive next door, a failing HVAC system, a rigid mounting tray, or construction in the building. It only tells you the drive felt something.

What this means

Smartmontools reports G-Sense_Error_Rate as a raw48 value under either attribute ID 221 or ID 191, depending on the drive. Two columns in smartctl -A output matter:

RAW_VALUE: the cumulative count of detected events since manufacture.
VALUE: the normalized score the firmware computes from the raw data. On most drives this starts at 100 or 200 and trends toward THRESH (typically 000) as accumulated events reach levels the firmware considers harmful.

The raw value alone is not actionable without context. Some drives, particularly Seagate IronWolf and Exos enterprise HDDs, have accelerometers sensitive enough to trigger on normal chassis vibration from fans and adjacent drives. Raw values in the hundreds or low thousands are common on these drives in standard server chassis, with the normalized value steady at 100 and no media errors. That is sensor noise, not damage.

What is actionable:

Any growth from the last reading: investigate the physical environment. TICKET.
Rapid growth alongside rising Seek_Error_Rate or Reallocated_Sector_Ct: the vibration is causing head slap. PAGE.

The normalized value declining from its starting point (typically 100) toward the threshold is the firmware’s own assessment that accumulated shock events are exceeding the drive’s design tolerance. A declining normalized value is always significant, regardless of the raw value.

flowchart TD
    A["G-Sense_Error_Rate
growing"] --> B{"Normalized VALUE
declining?"}
    B -- "No, stable at 100/200" --> C{"Seek_Error_Rate (ID 7)
or Reallocated_Sector_Ct (ID 5)
also rising?"}
    C -- "No" --> D["Likely benign sensor noise
TICKET: investigate environment"]
    C -- "Yes" --> E["Vibration causing
positioning errors or head slap"]
    B -- "Yes, declining" --> E
    E --> F["PAGE: fix vibration source
and plan drive replacement"]
    D --> G["Fix mounting, resonance,
or environmental vibration"]

If smartctl -A does not list G-Sense_Error_Rate under either ID, the drive either lacks an accelerometer or does not expose it through SMART. Absence of the attribute means you have no telemetry for it.

Common causes

Cause	What it looks like	First thing to check
Dense-shelf resonance	Multiple drives in the same JBOD shelf show G-Sense growth simultaneously; seek errors may also rise	Check whether drives in the same physical shelf share the pattern
Missing anti-vibration mounts	Single drive shows growth after installation or relocation; no other drives affected	Inspect drive mounting hardware (grommets, rubber washers, sled rails)
Environmental vibration (HVAC, construction, nearby machinery)	Growth correlates with specific times of day or facility events	Ask facilities about HVAC cycling, construction, or equipment changes
Shipping damage	Brand-new drive with large raw value on first read; normalized value may already be degraded	Compare against baseline; RMA if normalized is degraded on a new drive
Seagate sensor sensitivity	High raw value (hundreds to thousands) but normalized stays at 100; no seek errors or reallocations	Verify normalized value and check for correlated damage signals

Quick checks

# Read the G-Sense_Error_Rate attribute (both possible IDs)
smartctl -A /dev/sdX | grep -i "g.sense"

# Check for correlated seek errors (vibration degrading head positioning)
smartctl -A /dev/sdX | grep -i "seek_error"

# Check for reallocated sectors (head slap damage)
smartctl -A /dev/sdX | grep -i "reallocat"

# Check for pending sectors (vibration-induced read failures)
smartctl -A /dev/sdX | grep -i "current_pending"

# Check whether other drives in the same chassis show similar patterns
for disk in /dev/sd?; do echo "=== $disk ==="; smartctl -A "$disk" 2>/dev/null | grep -i "g.sense"; done

# Review ATA error log for mechanical error signatures
smartctl -l error /dev/sdX | grep -iE "AMNF|CCTO|ABRT"

# Run a conveyance test to check for shipping/handling damage (~5 minutes, HDD only)
smartctl -t conveyance /dev/sdX

All commands except the conveyance test are read-only. The conveyance test runs a short surface scan designed to detect transport-related damage. It is safe to run on a production drive but will briefly compete with host I/O.

smartctl does print CCTO in ATA error-log records, but only for commands that use that error-bit mapping; it means Command Completion Timed Out, not a mechanical shock signature. The vibration-correlated signatures in this guide are AMNF, ABRT, and UNC.

How to diagnose it

Record the raw and normalized values. The raw value is cumulative since manufacture. What matters is whether it changed since the last collection, and whether the normalized value has moved from its starting point (typically 100 or 200).
Check for correlated damage signals. Pull Seek_Error_Rate (ID 7), Reallocated_Sector_Ct (ID 5), and Current_Pending_Sector (ID 197). If any are growing, the vibration is past noise and into active damage. Escalate from TICKET to PAGE.
Scope the problem across the chassis. Run G-Sense checks on every drive in the same physical enclosure. If multiple drives show growth simultaneously, the root cause is environmental (shelf resonance, HVAC, power supply vibration). If only one drive is affected, the problem is likely its mounting.
Compare against the deployment baseline. A new drive with a high raw value but normalized at 100 and no media errors likely accumulated sensor events during shipping. A drive that was zero and is now growing has an active vibration source.
Identify the vibration source. This is physical investigation. Check drive mounting hardware, chassis fan health, nearby equipment, and facility conditions. The accelerometer tells you the drive felt something, not what caused it.
Check the ATA error log for mechanical signatures. AMNF (Address Mark Not Found) errors indicate low-level servo or format problems that vibration can worsen. ABRT (Aborted Command) errors can indicate the drive aborted due to mechanical difficulty. If either appears alongside G-Sense growth, the vibration is actively degrading the drive.

Metrics and signals to monitor

Signal	Why it matters	Warning sign
G-Sense_Error_Rate raw value (ID 191/221)	Cumulative count of vibration events exceeding the drive’s threshold	Any growth from last reading
G-Sense_Error_Rate normalized value	Firmware’s assessment of accumulated shock severity	Decline from starting value (typically 100 or 200)
Seek_Error_Rate normalized value (ID 7)	Vibration directly degrades head positioning accuracy	Normalized value declining alongside G-Sense growth
Reallocated_Sector_Ct raw value (ID 5)	Head slap from vibration damages platter surface	Any increase
Current_Pending_Sector raw value (ID 197)	Vibration-induced read failures mark sectors as suspect	Any non-zero value
G-Sense_Error_Rate across chassis	Multiple drives growing together indicates environmental source	Two or more drives in same shelf showing simultaneous growth
Drive temperature (ID 194)	Failed fans increase chassis vibration and temperature	Temperature rising alongside G-Sense growth

Fixes

Fix the environment, not the drive. If the media is healthy (no seek errors, no reallocations, normalized G-Sense stable), replacing the drive will not solve the problem. The new drive will show the same growth because the vibration source is still present.

Dense-shelf resonance

In JBOD enclosures with many bays, adjacent spinning drives create sympathetic vibration that feeds back through the chassis:

Install anti-vibration grommets or rubber washers between the drive sled and the chassis.
Verify that drive rails or sleds are designed for vibration damping, not rigid metal-to-metal contact.
If the enclosure supports staggered spin-up, enable it to reduce transient vibration during power cycles.
Reduce drive density if the enclosure is overpopulated for its vibration isolation design.

Missing or degraded mounting hardware

A single drive showing G-Sense growth while neighbors are stable suggests a mounting problem. Inspect:

Anti-vibration grommets: present, not perished, correctly seated.
Drive sled rails: properly engaged, not loose.
Screws: torqued to spec. Over-tightening can compress rubber isolation solid, defeating its purpose.

Environmental vibration

If growth correlates with facility events (HVAC cycling, construction, nearby equipment):

Relocate the server or storage shelf away from the vibration source.
Install vibration isolation pads under the rack.
Coordinate with facilities to identify and mitigate the source.

Shipping damage on a new drive

A brand-new drive with a large G-Sense raw value and a degraded normalized value likely experienced shock during shipping. If the drive also shows media errors, file an RMA. Do not deploy a shipping-damaged drive.

When to replace the drive

Replace the drive only when vibration has caused confirmed media damage: Reallocated_Sector_Ct is growing, or Seek_Error_Rate normalized is declining alongside G-Sense growth. In that case:

Fix the vibration source first (mounting, resonance, environment).
Replace the damaged drive.
Monitor the replacement for G-Sense growth. If the new drive shows the same pattern, the environment was not fixed.

Prevention

Baseline at deployment. Capture a full SMART snapshot when each drive is installed. The G-Sense raw value at deployment tells you what accumulated during shipping and factory testing. Alert on growth from this baseline, not on absolute value.
Use vibration-damped mounting hardware in every HDD in every chassis. Rigid metal-to-metal contact transmits chassis vibration directly to the drive.
Monitor G-Sense across the chassis. A single drive showing growth is a mounting problem. Multiple drives growing is an environmental problem. The distinction determines whether you fix a sled or call facilities.
Alert on growth and normalized value, not raw thresholds. A drive that sits at raw value 500 with normalized 100 and zero seek errors for its entire service life is healthy.
Run conveyance tests on newly deployed drives before committing them to production.

How Netdata helps

Rate of change detection. Netdata tracks G-Sense_Error_Rate changes over time, distinguishing a counter that jumped 200 in an hour from one that grew 200 over six months.
Correlation across signals. G-Sense_Error_Rate growth appears alongside Seek_Error_Rate changes and Reallocated_Sector_Ct in a single view, the exact correlation that separates benign sensor noise from head slap.
Cross-drive comparison. When multiple drives in the same host show G-Sense growth simultaneously, Netdata surfaces the chassis-wide pattern, pointing to an environmental root cause rather than a drive-specific one.
Anomaly detection flags G-Sense growth that deviates from each drive’s established baseline, reducing false alerts from drives with naturally high but stable raw values.