fdisk

What it does

fdisk views and edits the partition table on a block storage device — the structure that defines how a disk is divided into partitions, their sizes, and their types. Running it with -l is a safe, read-only way to inspect existing partitions. Running it against a device with no flags drops you into an interactive prompt where you can create, delete, resize (via delete-and-recreate), and change the type of partitions before writing those changes to disk.

Beginner examples

• sudo fdisk -l — list partition tables for every detected disk
• sudo fdisk -l /dev/sda — list the partition table for one specific disk
• sudo fdisk /dev/sdb — open /dev/sdb in the interactive partition editor
• Inside the interactive prompt: p — print the current (in-memory) partition table
• Inside the interactive prompt: m — show the help menu of available commands

sudo fdisk -l /dev/sda

Advanced examples

• Create a new partition interactively: run sudo fdisk /dev/sdb, then type n, follow the prompts for partition type/number/size, then w to commit.
• Delete a partition: inside the interactive session, type d, select the partition number, then w to commit — or q to back out without saving.
• Change a partition's type ID (e.g. to Linux LVM or swap): type t inside the session, select the partition, and enter the new type code.
• Check a specific partition's size in blocks without entering interactive mode: sudo fdisk -s /dev/sda1.
• Inspect sector-level layout (useful for alignment checks): sudo fdisk -u -l /dev/sda.

sudo fdisk /dev/sdb
# then: n, p, 1, <accept defaults>, w

Common mistakes

• Running fdisk against the wrong device (e.g. /dev/sda, the system disk, instead of /dev/sdb, an attached secondary disk) and writing changes — this can destroy the operating system's partition table.
• Forgetting that changes are only staged in memory until w is pressed, then being surprised that a previous "test" session actually had no effect (because q was used) — or the opposite, forgetting to press q and accidentally committing an experiment.
• Assuming fdisk formats the partition too — it only edits the partition table; you still need mkfs (e.g. mkfs.ext4) to create a filesystem on the new partition before it's usable.
• Editing a partition table on a disk that has partitions currently mounted and in use, which can lead to a kernel view of the disk that's out of sync until a reboot or partprobe.

Tips

• Always run fdisk -l first to confirm you have the right device name before opening it interactively.
• Use lsblk or df -h alongside fdisk -l to cross-check which device holds which mounted filesystem before making changes.
• After creating a partition, remember it still needs a filesystem (mkfs.ext4, mkfs.xfs, etc.) before it can be mounted and used.
• If the kernel doesn't immediately see a partition table change, run partprobe or reboot to force it to re-read the table.

Best practices

• Never run fdisk's interactive write (w) command on a production disk without a current backup — partition table mistakes can make all data on the disk unreadable.
• Double- and triple-check the device path (/dev/sdX) before committing changes; typos between similarly named devices are the single most common cause of accidental data loss with fdisk.
• Prefer fdisk -l (read-only) for routine inspection, and reserve interactive editing for deliberate, planned partitioning work.
• For GPT disks or advanced resizing needs, consider parted or gdisk, which handle GPT semantics more directly than traditional fdisk.

Real-world use cases

• Inspecting the partition layout of a newly attached disk before deciding how to partition and format it.
• Creating a new partition on a freshly added virtual disk in a cloud VM, then formatting and mounting it for extra storage.
• Diagnosing a boot issue by checking whether the expected boot/EFI partition still exists and has the correct type.
• Auditing partition types across disks before a migration to confirm nothing unexpected (e.g. a leftover swap partition) is present.

Common interview questions

• What's the difference between fdisk -l and running fdisk interactively? -l is a read-only listing of the partition table that exits immediately; running fdisk on a device with no other flags opens an interactive session where you can create, delete, or modify partitions before writing changes.
• Does fdisk format partitions? No — fdisk only manages the partition table (the boundaries and types of partitions). You still need a tool like mkfs to put a filesystem on a partition before it can store files.
• Why is fdisk considered risky? Because writing an incorrect partition table (via the w command) can overwrite the layout of an entire disk, potentially making existing data inaccessible, especially if you targeted the wrong device.
• What does macOS use instead of fdisk? macOS uses diskutil for partition and volume management rather than the Linux fdisk utility.