When automating disk preparation in Linux systems, traditional methods like dd if=/dev/zero leave the kernel's partition table cache out of sync. Here's what actually happens under the hood:

# Typical MBR wipe command
dd if=/dev/zero of=/dev/sdX bs=512 count=1 conv=notrunc

The kernel maintains an in-memory copy of partition tables. Any active handles (mounts, LVM, mdadm, swap) will block partition table updates:

# Common error patterns
sudo blockdev --rereadpt /dev/sda
# BLKRRPART: Device or resource busy

partprobe
# Error: Partition(s) on /dev/sda are in use

This bash function handles all common disk usage scenarios:

#!/bin/bash
function disk_reset() {
  local disk=$1
  [[ -b "$disk" ]] || { echo "Invalid block device"; return 1; }

  # Unmount all partitions
  grep "^$disk" /proc/mounts | awk '{print $2}' | sort -r | xargs -r umount

  # Deactivate LVM
  if command -v vgchange &>/dev/null; then
    vgchange -an 2>/dev/null
    for vg in $(vgs --noheadings -o vg_name 2>/dev/null); do
      vgremove -f "$vg" 2>/dev/null
    done
  fi

  # Stop MD arrays
  if command -v mdadm &>/dev/null; then
    mdadm --stop --scan
    mdadm --remove --scan
  fi

  # Clear swap
  swapoff -a

  # Wipe MBR and partition table
  dd if=/dev/zero of="$disk" bs=1M count=10 conv=fsync

  # Force kernel to reload
  partx -d "$disk" 2>/dev/null
  partprobe "$disk"
  blockdev --rereadpt "$disk" 2>/dev/null
}

# Usage: disk_reset /dev/sda

For systems with sg3_utils package:

# Physical disk reset (like power cycle)
sg_reset /dev/sda

# Combined with wipe:
sg_format --format /dev/sda

For advanced cases where device handlers persist:

# List all open handles
lsof /dev/sda*

# Force-close file descriptors
for fd in $(ls /proc/*/fd/* 2>/dev/null | grep -P '/dev/sda'); do
  kill -9 $(echo $fd | cut -d'/' -f3)
done

Remember that some operations (like filesystem journal replays) may require a full reboot for complete cleanup.

When working with storage devices in Linux, you often need to completely wipe a disk's partition table and MBR (Master Boot Record) to start fresh. While tools like dd can overwrite these structures, the kernel often keeps old partition information cached, leading to "device busy" errors when trying to repartition.

The fundamental issue is that Linux maintains device mappings in memory even after you've destroyed the partition table. This is particularly problematic when dealing with:

• Mounted filesystems
• LVM volumes
• MD RAID arrays
• Swap partitions

Here's a comprehensive bash script that handles all these cases:

#!/bin/bash

# Check root privileges
if [ "$(id -u)" != "0" ]; then
    echo "This script must be run as root" 1>&2
    exit 1
fi

# Validate input
if [ -z "$1" ]; then
    echo "Usage: $0 /dev/sdX"
    exit 1
fi

DEVICE="$1"

# Verify device exists
if [ ! -b "$DEVICE" ]; then
    echo "Error: $DEVICE is not a block device"
    exit 1
fi

# Unmount all partitions
echo "Unmounting any mounted partitions..."
umount -l ${DEVICE}* 2>/dev/null

# Stop any LVM activity
echo "Stopping LVM volumes..."
vgchange -an 2>/dev/null
for vg in $(vgs --noheadings -o vg_name 2>/dev/null); do
    vgremove -f "$vg" 2>/dev/null
done
pvremove -ff "$DEVICE"* 2>/dev/null

# Stop MD RAID arrays
echo "Stopping MD RAID devices..."
for md in $(ls /dev/md/* 2>/dev/null); do
    mdadm --stop "$md" 2>/dev/null
    mdadm --remove "$md" 2>/dev/null
done

# Clear swap
echo "Deactivating swap..."
swapoff -a

# Wipe partition table
echo "Wiping partition table..."
dd if=/dev/zero of="$DEVICE" bs=512 count=1 conv=notrunc 2>/dev/null

# Force kernel to reread partition table
echo "Reloading partition table..."
blockdev --rereadpt "$DEVICE" 2>/dev/null || partprobe "$DEVICE" 2>/dev/null

echo "Disk $DEVICE has been completely reset"

For different scenarios, consider these alternatives:

Using sgdisk (GPT disks)

sgdisk -Z /dev/sdX
sgdisk -o /dev/sdX
partprobe /dev/sdX

Using wipefs

wipefs -a /dev/sdX
partprobe /dev/sdX

If you still encounter "device busy" errors, try removing device mapper references:

dmsetup remove_all
partprobe

For data center environments, you might want to reset all non-system disks:

for disk in $(lsblk -dn -o NAME | grep -v $(lsblk -no PKNAME $(findmnt -n / -o SOURCE))) 
do
    /path/to/reset_script.sh "/dev/$disk"
done

• Always double-check the device path before running destructive commands
• This process doesn't securely erase data - use proper wiping tools for that
• Some hardware RAID controllers may require additional steps
• NVMe devices might need nvme format instead