How to Fix Linux EIO
Quick Fix Summary
TL;DRCheck disk health with smartctl, verify cable connections, and run filesystem checks.
EIO (Input/Output Error) indicates the kernel failed to read from or write to a device, typically storage. This is a hardware-level failure signal that requires immediate investigation.
Diagnosis & Causes
Recovery Steps
Step 1: Immediate Diagnostics & Log Inspection
First, gather all available error data from the kernel and system logs to pinpoint the failing device.
dmesg -T | grep -i EIO
journalctl --since "5 minutes ago" | grep -i EIO
lsblk
cat /proc/mounts | grep <suspected_device>Step 2: Hardware Health Check & Cable Verification
Physically inspect connections and use SMART tools to assess the disk's health. For NVMe, use nvme-cli.
smartctl -a /dev/sdX
(For NVMe: nvme smart-log /dev/nvme0n1)
Check /sys/block/sdX/device/stateStep 3: Filesystem Check & Remount (If Read-Only)
If the system has remounted a filesystem as read-only (ro), attempt a safe check and remount. Unmount first if possible.
umount /failing/mountpoint
fsck -y /dev/sdXY
mount -o remount,rw /failing/mountpointStep 4: Failover & Data Recovery Strategy
If the device is failing, initiate failover to a standby system or disk. Use dd_rescue to attempt data salvage from the failing block device.
ddrescue -d -r3 /dev/failing_sdX /dev/healthy_sdY rescue.log
(For LVM: pvmove /dev/failing_sdX)Architect's Pro Tip
"EIO on cloud/VM storage often indicates underlying hypervisor or network issues. Check your provider's status and consider detaching/reattaching the volume."
Frequently Asked Questions
Can a Linux EIO error cause data loss?
Yes. EIO signals a failed physical write or read. Any data not yet committed to stable storage is at risk. Immediate read-only mode and backup are critical.
How do I distinguish a hardware EIO from a software one?
Check kernel logs (dmesg). Consistent errors on a specific block device (e.g., /dev/sdb) point to hardware. Isolated errors on a single file may indicate localized corruption.
The system is frozen with EIO errors. What's the safest next step?
If possible, trigger a kernel panic to get a crash dump for analysis: `echo c > /proc/sysrq-trigger`. Otherwise, a hard reboot may be necessary to regain control.