Comprehensive Guide to Linux File System Management

Most Linux server emergencies start with the disk: a full filesystem stops databases and mail, a filesystem error after a crash stops the server booting, and wrong permissions break websites or open them to attack. This guide covers the commands you need to watch disk space, find what is using it, check and repair filesystems, add a new disk, and set ownership and permissions correctly.

1. See how much space is left

df -hT        # size, used, free and type of every mounted filesystem
df -i         # inodes: the number of files a filesystem can still hold
findmnt       # what is mounted where, as a tree

A filesystem can report "No space left on device" while df -h still shows free space. That means it has run out of inodes, usually because of millions of tiny files such as PHP sessions, cache fragments or mail in a queue. df -i shows it straight away.

2. Find what is using the space

sudo du -xh --max-depth=1 / | sort -h            # biggest top-level folders, this filesystem only
sudo du -xh --max-depth=1 /var | sort -h         # then drill down
sudo find / -xdev -type f -size +500M -exec ls -lh {} \;   # single large files

ncdu (install it from your distribution's packages) does the same interactively and lets you delete as you go. The usual suspects on a web server:

If df still shows the disk full after you deleted a big file, a running process is holding it open. sudo lsof +L1 lists deleted-but-open files; restart that service to free the space.

3. List disks and partitions

lsblk -f      # disks, partitions, filesystem type, label, UUID and mount point
sudo blkid    # UUIDs and types, for /etc/fstab

On a KVM VPS the disk is usually vda (partitions vda1, vda2); on other systems it may be sda or nvme0n1. Read the output carefully before you run anything that writes to a device.

4. Check and repair a filesystem

ext4. Do a read-only check first, then repair:

sudo umount /data
sudo e2fsck -fn /dev/vdb1     # -n: report only, change nothing
sudo e2fsck -fy /dev/vdb1     # -y: fix, answering yes to every question

XFS (the default on AlmaLinux and Rocky Linux) uses its own tool; fsck does nothing on XFS:

sudo umount /data
sudo xfs_repair -n /dev/vdb1  # dry run
sudo xfs_repair /dev/vdb1     # repair

If xfs_repair refuses because of a dirty log, mount and unmount the filesystem once to replay the log. xfs_repair -L throws the log away and can lose recent changes: use it only as a last resort.

The root filesystem. On systemd distributions, add fsck.mode=force to the kernel command line for one boot, or check from a rescue system. Most distributions also check the root filesystem automatically at boot after an unclean shutdown.

5. Add a new disk or partition

On a VPS, extra space usually comes from moving to a bigger plan, which grows the existing disk. You then grow the partition and filesystem instead of creating a new one; Cloning and resizing your VPS shows how with growpart, resize2fs and xfs_growfs.

6. Ownership and permissions

Every file has an owner, a group, and permissions for the owner (u), the group (g) and everyone else (o).

sudo chown -R www-data:www-data /var/www/example   # owner:group (nginx/apache user varies by distribution)
chmod 640 wp-config.php                             # numeric
chmod u+x deploy.sh                                 # symbolic: add execute for the owner
find /var/www/example -type d -exec chmod 755 {} +  # directories
find /var/www/example -type f -exec chmod 644 {} +  # files

Avoid chmod -R 755 on a whole site: it makes every file executable. Use the two find commands instead. And never use 777; it lets any process on the server change your files, which is how many hacked sites get infected.

7. Keep the disk healthy

• Set up log rotation for every application log you create.
• Alert yourself when a filesystem passes about 80% full, before it becomes an outage.
• Keep backups off the server, not in the web root or on the same disk.
• After a crash or forced restart, read journalctl -b -p err for filesystem errors.

8. On shared hosting

On shared hosting (cPanel, DirectAdmin, Webuzo) the server's disks are managed for you. You can still:

• see your disk and inode usage in the control panel; see how to check disk and bandwidth usage;
• change permissions in the File Manager; see how to use the File Manager, and for DirectAdmin, where home folders are 710, domains is 711 and public_html is 755, managing file permissions in DirectAdmin;
• use du, find, chmod and ls over SSH. Jailed SSH access is available on every shared hosting plan; it is off by default, so ask support to enable it for your account. Logins use an SSH key. See enabling jailed SSH.

9. Where Domain India fits

A Domain India VPS gives you the root access these commands need, on KVM with NVMe storage. It is self-managed, so disk health and backups are yours to look after, and support helps with anything outside the server, such as a forced restart. Live price, excluding 18% GST:

Ready to manage your own server? Read the Linux shell commands guide, compare VPS plans, or open a ticket if a disk problem on your server needs our help.