Add a new disk: partition, format and mount it by UUID

You attached a new disk to a server (a physical drive or a virtual disk on a VM) and want it mounted at /srv/data. The steps are the same on Debian, Ubuntu, RHEL, Rocky and AlmaLinux: identify the disk, create a partition, put a filesystem on it, and add one line to /etc/fstab.

Warning: partitioning and formatting destroy everything on the target device. Picking the wrong disk is the classic mistake. Do every check in step 1 before running any command that writes.

1. Identify the new disk

lsblk -o NAME,SIZE,TYPE,FSTYPE,MOUNTPOINTS
lsblk -f

Example output:

NAME   SIZE TYPE FSTYPE MOUNTPOINTS
sda     40G disk
├─sda1   1G part xfs    /boot
└─sda2  39G part LVM2_member
  ├─rl-root 35G lvm xfs /
  └─rl-swap  4G lvm swap [SWAP]
sdb    100G disk

The new disk is the one with the expected size, no partitions, no filesystem and no mount point; here /dev/sdb. On VMs it may be /dev/vdb, on NVMe /dev/nvme1n1 (partitions are then nvme1n1p1). Confirm the disk carries no data signatures; with no options, wipefs only lists them and changes nothing:

sudo wipefs /dev/sdb        # no output = no known filesystem or partition table

If it lists anything, stop and find out what is on that disk before continuing.

2. Create a partition table and one partition

parted can do this in one non-interactive command. GPT is the right choice for new disks, and starting at 1MiB keeps the partition aligned:

sudo parted -s -a optimal /dev/sdb mklabel gpt mkpart data ext4 1MiB 100%
sudo parted /dev/sdb print

data is the GPT partition name and ext4 is only a type hint; parted does not format the partition. That happens in step 3.

If you prefer an interactive tool, sudo fdisk /dev/sdb works too: g creates a GPT table, n a new partition (accept the defaults to use the whole disk), p prints, and w writes and quits. Nothing is written before w.

3. Create a filesystem

RHEL and its rebuilds default to XFS; Debian and Ubuntu default to ext4. Both are fine. Remember that XFS can grow but cannot be shrunk.

# ext4
sudo mkfs.ext4 -L data /dev/sdb1

# or XFS
sudo mkfs.xfs -L data /dev/sdb1

4. Find the UUID

Device names such as /dev/sdb can change when disks are added or removed, so fstab should refer to the filesystem’s UUID, which stays the same.

sudo blkid /dev/sdb1
# /dev/sdb1: LABEL="data" UUID="0f1e2d3c-4b5a-6978-8a9b-0c1d2e3f4a5b" TYPE="ext4" ...

5. Add the fstab entry

sudo mkdir -p /srv/data
sudo cp /etc/fstab /etc/fstab.bak
sudo nano /etc/fstab

Add one line (use your UUID):

UUID=0f1e2d3c-4b5a-6978-8a9b-0c1d2e3f4a5b  /srv/data  ext4  defaults,nofail  0  2

The six fields are: what to mount, where, filesystem type, mount options, dump flag (0) and fsck order (1 for the root filesystem, 2 for others, 0 to skip). nofail lets the server finish booting even if this disk is missing, which is a good idea for data disks; leave it off for disks the system cannot run without.

6. Test before you reboot

A broken fstab line can stop the server at boot and drop it into emergency mode. Test now, while you still have a shell:

sudo findmnt --verify          # checks /etc/fstab for errors
sudo systemctl daemon-reload   # systemd builds mount units from fstab
sudo mount -a                  # mounts everything in fstab that is not mounted yet
findmnt /srv/data
df -h /srv/data

If mount -a or findmnt --verify reports an error, fix the line (or restore /etc/fstab.bak) before rebooting.

7. Set ownership

sudo chown appuser:appuser /srv/data

Should I use LVM instead?

If the data may grow, consider putting LVM on the new disk instead of a plain partition. It lets you extend the volume later, also across several disks, without unmounting. See the LVM concept entry.

Official documentation

• fstab(5)
• lsblk(8)
• GNU Parted manual