While piping echo commands to fdisk (like echo -e "n\\np\\n1\\n\\n\\nt\\nc\\a\\1\\nw" | fdisk /dev/sdb) technically works for creating a single partition using all available disk space, it raises several concerns:

• Fragile string escaping that's prone to errors
• No proper error handling or validation
• Poor readability and maintainability
• Potential version compatibility issues across different fdisk implementations

Using sfdisk (Recommended Approach)

The sfdisk utility is specifically designed for scripted partitioning:

# Wipe existing partition table
wipefs -a /dev/sdb

# Create single partition using all space
echo 'label: gpt
,,' | sfdisk /dev/sdb

This creates a GPT partition table with a single partition spanning the entire disk. For MBR:

echo 'label: dos
,,' | sfdisk /dev/sdb

Using parted

GNU parted offers another clean solution:

parted /dev/sdb --script \
    mklabel gpt \
    mkpart primary 0% 100%

Using sgdisk (for GPT specifically)

sgdisk -n 0:0:0 -t 0:8300 -c 0:"Linux Filesystem" /dev/sdb

For production scripts, add verification steps:

# Verify partition was created
if ! lsblk /dev/sdb1 > /dev/null 2>&1; then
    echo "Partition creation failed" >&2
    exit 1
fi

Complete the process by adding filesystem creation (example with ext4):

mkfs.ext4 /dev/sdb1

For XFS:

mkfs.xfs -f /dev/sdb1

Here's a complete script that handles all steps safely:

#!/bin/bash

DISK="/dev/sdb"

# Verify disk exists
if [ ! -b "$DISK" ]; then
    echo "Error: Disk $DISK not found" >&2
    exit 1
fi

# Wipe existing partitions
wipefs -a "$DISK"

# Create partition table and single partition
if ! echo 'label: gpt
,,' | sfdisk "$DISK"; then
    echo "Partitioning failed" >&2
    exit 1
fi

# Create filesystem
if ! mkfs.ext4 "${DISK}1"; then
    echo "Filesystem creation failed" >&2
    exit 1
fi

echo "Successfully prepared ${DISK}1"

---

When automating server provisioning or storage configuration, the traditional interactive fdisk approach becomes problematic. The common workaround of piping commands via echo feels hacky and lacks proper error handling. Here's why we need better solutions:

# The "echo pipe" method everyone uses but hates
echo -e "n\\np\\n1\\n\\n\\nt\\nc\\a\\1\\nw" | fdisk /dev/sdb

For Linux systems, we have several production-grade alternatives:

1. Using sfdisk (The Preferred Method)

sfdisk is specifically designed for scripting and handles edge cases better:

# Clear existing partition table
wipefs -a /dev/sdb

# Create new partition using all space
echo 'type=c' | sfdisk /dev/sdb

# Alternative with explicit sector calculation
echo 'start=2048, size=, type=c' | sfdisk /dev/sdb

2. Using parted (For Advanced Filesystems)

When working with GPT or advanced filesystems, parted offers better control:

parted /dev/sdb --script \
  mklabel msdos \
  mkpart primary 0% 100%

3. Using sgdisk (For GPT Partitions)

For UEFI systems or disks larger than 2TB:

sgdisk -n 0:0:0 -t 0:8300 -c 0:"Linux FS" /dev/sdb

Always include proper validation in your scripts:

#!/bin/bash

DEVICE="/dev/sdb"

# Verify block device exists
if [ ! -b "$DEVICE" ]; then
  echo "Error: Device $DEVICE not found" >&2
  exit 1
fi

# Partition with error checking
if ! echo 'type=83' | sfdisk "$DEVICE"; then
  echo "Partitioning failed" >&2
  exit 1
fi

# Verify partition creation
if [ ! -b "${DEVICE}1" ]; then
  echo "Partition not created properly" >&2
  exit 1
fi

After partitioning, you'll typically want to create a filesystem:

mkfs.ext4 -L "data_disk" /dev/sdb1

# For XFS
mkfs.xfs -f /dev/sdb1

# For btrfs
mkfs.btrfs -f /dev/sdb1