Using an additional disk on your server: partition, format and mount permanently

1. What you should know first

A disk added from the panel arrives on your server as a raw, empty disk. The server sees it, but it has no file system, so it is not mounted anywhere. You will do three things:

1. Find the disk (it appears as sdb, sdc and so on).
2. Create a file system on it (format it).
3. Mount it on a folder and make that permanent.

The commands below are the same on Ubuntu, Debian, AlmaLinux and Rocky Linux; differences are noted where they matter. An HDD (large storage) additional disk is used with exactly the same steps.

2. Step by step: preparing the disk

1. Find the disk. Connect to the server and list disks:

   bash

   lsblk -o NAME,SIZE,TYPE,FSTYPE,MOUNTPOINT
   

   Example output (25 GB root disk, 20 GB new disk):

   code

   NAME    SIZE TYPE FSTYPE MOUNTPOINT
   sda      25G disk
   ├─sda1   24G part ext4   /
   ├─sda14   4M part
   ├─sda15 106M part vfat   /boot/efi
   └─sda16 913M part ext4   /boot
   sdb      20G disk
   

   The new disk has the size you entered in the panel, no partitions below it, and empty FSTYPE and MOUNTPOINT columns (here sdb). In the following commands use your own disk name instead of sdb.

2. Partition it (recommended). Write a GPT partition table and create a single partition:

   bash

   parted -s /dev/sdb mklabel gpt mkpart data 1MiB 100%
   lsblk /dev/sdb          # you should now see sdb1 underneath
   

   If parted is missing, install it with apt install parted (Ubuntu/Debian) or dnf install parted (AlmaLinux/Rocky).

   Note

   You can also format the disk directly without partitioning (use /dev/sdb instead of /dev/sdb1 in Step 3). It is simpler and makes growing easier later, but some tools may mistake an unpartitioned disk for an empty one and write over it. If unsure, partition.

3. Format it. Choose a file system:

   bash

   mkfs.ext4 -L data /dev/sdb1     # ext4: available everywhere, reliable, the common choice
   # or
   mkfs.xfs -L data /dev/sdb1      # xfs: for large files and heavy writes; the AlmaLinux/Rocky default
   

   On Ubuntu/Debian xfs may need apt install xfsprogs. If unsure, choose ext4. Changing the file system later means formatting again, which erases the data.

4. Mount it. Create a folder and mount the disk on it:

   bash

   mkdir -p /mnt/data
   mount /dev/sdb1 /mnt/data
   df -hT /mnt/data
   

   If df shows the disk size and file system, the disk is ready. To try it, write echo hello > /mnt/data/file.txt.

5. Make it permanent. mount only lasts for that session; after a reboot the mount is gone. To make it permanent, add a line to /etc/fstab. Use the disk's UUID in that line, not /dev/sdb1 (device names can change across reboots, the UUID does not):

   bash

   echo "UUID=$(blkid -s UUID -o value /dev/sdb1) /mnt/data ext4 defaults,nofail 0 2" >> /etc/fstab
   systemctl daemon-reload
   mount -a && echo "fstab ok"
   

   If you used xfs, write xfs instead of ext4 in that line. The nofail option at the end matters: if the disk cannot be mounted one day, the server still boots instead of hanging (see Section 5).

6. Test it. To verify it survives a reboot, restart the server and check the disk is still mounted:

   bash

   reboot
   # reconnect
   df -hT /mnt/data
   cat /mnt/data/file.txt
   

3. File permissions

When you format and mount as root, the folder is owned by root. To let another user (for example your web application's user) write to it, give ownership:

chown -R appuser:appuser /mnt/data     # replace appuser with your own user name
chmod 750 /mnt/data

4. Common uses

• Database data: stop the service, copy the existing data folder to the new disk (rsync -a), then change the data folder in the service's configuration. Do not rush; take a snapshot first.
• Backups: point your backup script's target to /mnt/data/backup. A backup on the server's own disk is lost when the server is deleted; copy important backups off the server too.
• Several disks: repeat Steps 1-6 for each new disk, with a different folder (/mnt/media, /mnt/logs) and a separate fstab line.

5. Common problems

The disk does not show in lsblk. Check in the panel that the disk was added and the task finished. If the server was not told about the new disk, ask it to rescan:

for h in /sys/class/scsi_host/host*/scan; do echo "- - -" > $h; done
lsblk

mount: wrong fs type, bad option, bad superblock error. The disk is probably not formatted yet, or you are mounting the wrong partition. Check the FSTYPE column with lsblk -f; if it is empty, do Step 3.

mkfs says the disk "contains a ... file system" or asks for confirmation. The disk may have been used before. If you are sure it is the right disk and you do not need its content, continue; if not, stop and check the disk again with lsblk -f.

The disk is not mounted after a reboot. Check the fstab line: if mount -a prints an error, there is a typo. Make sure the UUID matches the output of blkid /dev/sdb1.

The server does not boot because of an fstab error. The nofail option largely prevents this: if the disk cannot be mounted the system still boots and only that folder stays empty. If the server still stops in "emergency mode", open the Console in the panel, log in as root and fix the faulty line, or comment it out with # (nano /etc/fstab), then run reboot. Copying fstab before editing is a good habit: cp /etc/fstab /etc/fstab.bak.

I deleted the disk in the panel; how does the server react? Before deleting, unmount the disk on the server (umount /mnt/data) and remove the fstab line or disable it with #. The data of a deleted disk cannot be recovered. Thanks to nofail, a forgotten line does not stop the server from booting.

Disk names changed (sdc instead of sdb). With several disks, names can shift when a disk is added or removed. That is why we always use the UUID in fstab.

6. Frequently asked questions