Filesystems, Formatting, Mounting, and /etc/fstab

1. A filesystem turns blocks into named, protected objects

Raw block devices provide addressable storage. A filesystem adds directories, filenames, inode or object metadata, allocation maps, timestamps, ownership, permissions, free-space accounting, crash-recovery structures, and consistency rules. The kernel exposes a common Virtual Filesystem interface so applications can use familiar operations even when the underlying filesystem implementations differ.

flowchart LR
  A["Application: open, read, write"] --> B["Linux VFS"]
  B --> C["ext4"]
  B --> D["XFS"]
  B --> E["Btrfs"]
  B --> F["tmpfs or network filesystem"]
  C --> G["Block device"]
  D --> G
  E --> G

Choosing a filesystem is an operational decision involving workload behavior, support lifecycle, recovery tools, growth and shrink capabilities, snapshots, checksumming, quotas, and the distribution's defaults. Avoid selecting solely from benchmark headlines.

2. Match filesystem capabilities to the operating model

Use findmnt -t ext4,xfs,btrfs to see what a system actually runs. The tools used to create and repair a filesystem are filesystem-specific; mkfs is commonly a dispatcher to programs such as mkfs.ext4 or mkfs.xfs.

3. Formatting creates metadata and destroys previous interpretation

# Inspect first; these commands do not create a filesystem.
lsblk -f
sudo blkid
sudo wipefs --no-act /dev/CONFIRMED_DEVICE

# Examples only—never substitute a real device casually.
# sudo mkfs.ext4 -L appdata /dev/CONFIRMED_EMPTY_DEVICE
# sudo mkfs.xfs  -L appdata /dev/CONFIRMED_EMPTY_DEVICE

A formatting command writes superblocks and allocation metadata. Existing data may become difficult or impossible to recover even when the command completes instantly. A change plan should capture the device identity, expected old signatures, intended filesystem type and label, ownership model after mounting, and restore or rollback route.

4. Mounts attach filesystem instances to directories

A mount point is an ordinary directory used as the attachment location for a filesystem. When mounted, the filesystem's root becomes visible at that directory and temporarily hides any pre-existing directory contents until unmounted.

# Inspect mount relationships
findmnt
findmnt --target /var/lib
findmnt --source UUID=YOUR-UUID

# Show kernel mount data and propagation where needed
findmnt -o TARGET,SOURCE,FSTYPE,OPTIONS,PROPAGATION

# A bind mount exposes an existing directory at another path
mkdir -p "$HOME/source-dir" "$HOME/view-dir"
# sudo mount --bind "$HOME/source-dir" "$HOME/view-dir"
# sudo umount "$HOME/view-dir"

Mount options affect correctness and security. Examples include ro, rw, noexec, nosuid, nodev, discard, and filesystem-specific performance or recovery options. Apply them from a threat and workload model rather than a universal hardening list.

5. /etc/fstab is declarative boot-time storage policy

Each non-comment entry has six fields:

# source                target       type  options                       dump pass
UUID=1111-2222          /srv/app     ext4  defaults,nodev,nosuid          0    2

# Review active fstab and resolve identifiers
cat /etc/fstab
findmnt --fstab --evaluate

# Validate syntax and semantics before rebooting
sudo findmnt --verify --verbose

# After an intentional fstab edit, test mounts without rebooting.
# sudo mount -av

On systemd systems, fstab entries are translated into mount units. Options such as x-systemd.automount, x-systemd.device-timeout=, and _netdev can express boot ordering or on-demand behavior, but they should be documented because they change failure semantics.

6. Hands-on lab: create and mount a disposable filesystem image

This lab never touches a real disk. It requires mkfs.ext4 and sudo for the temporary mount. Read every command before running it.

lab="$HOME/devops-academy/linux/chapter10/lesson02"
image="$lab/lab-filesystem.img"
mountpoint="$lab/mnt"
mkdir -p "$mountpoint"
cd "$lab"

# Create a sparse 128 MiB regular file.
truncate -s 128M "$image"

# Confirm it is a regular file, then create ext4 inside that file.
file "$image"
mkfs.ext4 -F -L devops-lab "$image"

# Mount through the kernel's loop mechanism.
sudo mount -o loop,nodev,nosuid "$image" "$mountpoint"
findmnt --target "$mountpoint"

sudo sh -c 'printf "storage lab\n" > "$1/README.txt"' sh "$mountpoint"
sudo chown "$USER":"$(id -gn)" "$mountpoint/README.txt"
stat "$mountpoint/README.txt"
df -hT "$mountpoint"

# Capture the loop device before cleanup.
findmnt -no SOURCE "$mountpoint" > loop-source.txt
sudo umount "$mountpoint"

# Verify that no mount remains.
if findmnt --target "$mountpoint" >/dev/null 2>&1; then
  echo 'ERROR: mount still active' >&2
  exit 1
fi
file "$image"
ls -lh "$image"

Validate a proposed fstab entry without modifying /etc/fstab

uuid=$(blkid -s UUID -o value "$image")
printf 'UUID=%s %s ext4 loop,nodev,nosuid,nofail 0 2\n' \
  "$uuid" "$mountpoint" > proposed-fstab

findmnt --verify --verbose --tab-file proposed-fstab
cat proposed-fstab

7. Common filesystem and mount mistakes

8. Knowledge check

9. Summary

Filesystems transform block storage into organized files and metadata. Formatting creates that structure; mounting attaches it to the directory tree; fstab declares persistent policy. Safe storage work separates identity, creation, attachment, validation, and cleanup, with every destructive action confined to verified disposable targets.

10. Further reading

• mount(8), umount(8), findmnt(8), fstab(5), and mkfs(8).
• ext4, XFS, and Btrfs administration documentation for the deployed distribution.
• systemd mount-unit and automount-unit documentation.