Disk Errors

A collection of disk-related errors and their fixes.

Block Size Mismatch Warning

After writing an ISO to a flash drive, partitioning tools may complain:

1

Warning: The driver descriptor says the physical block size is 2048 bytes, but Linux says it is 512 bytes.

This happens when a tool like dd writes at the ISO's native 2048-byte block size onto a device that uses 512-byte sectors. Zeroing the first few blocks sometimes helps:

1

$ dd if=/dev/zero of=/dev/sdX bs=2048 count=32 oflag=sync

If that doesn't clear it, recreating the partition table with fdisk or gdisk will.

SMART Monitoring

Check the health of a drive with smartctl:

1

$ smartctl -a /dev/sdX

Key attributes to watch for trouble:

• Reallocated_Sector_Ct (ID 5): sectors the drive has remapped due to read errors. Any non-zero value means the drive is actively degrading.
• Current_Pending_Sector (ID 197): sectors waiting to be remapped on next write. Non-zero means the drive is struggling.
• Offline_Uncorrectable (ID 198): sectors that couldn't be read during offline tests.
• UDMA_CRC_Error_Count (ID 199): usually indicates a bad cable or connection rather than a dying drive.

Run a short self-test:

1

$ smartctl -t short /dev/sdX

And check results after a few minutes:

1

$ smartctl -l selftest /dev/sdX

Filesystem Repair

For ext4 filesystems, unmount first then check:

1
2

$ umount /dev/sdX1
$ fsck.ext4 -f /dev/sdX1

The -f flag forces a check even if the filesystem appears clean. For XFS:

1

$ xfs_repair /dev/sdX1

If the filesystem is too damaged to mount normally, try mounting read-only first to recover what you can:

1

$ mount -o ro,norecovery /dev/sdX1 /mnt/recovery

Bad Blocks

Test a drive for bad blocks (this is a destructive write test, data will be lost):

1

$ badblocks -wsv /dev/sdX

For a non-destructive read-only test:

1

$ badblocks -sv /dev/sdX

These tests take a long time on large drives but can confirm whether a drive has physical media issues beyond what SMART reports.