Filesystems: Formatting the Disk

In the previous lesson, we created the "Rooms" (Partitions) on our disk. But a room with just a dirt floor isn't useful. You need to put down a floor and a filing system so you can save and find your documents.

In Linux, this is called Formatting or "Making a Filesystem."

A filesystem is a digital ledger that tracks which bits on the physical platter belong to which file. Linux supports dozens of filesystems, but in the modern server world, three kings rule.

---

1. The Three Kings of Linux Filesystems

I. ext4 (Fourth Extended Filesystem)

• The Standard: This is the default for Ubuntu and Debian.
• Why use it? It is arguably the most tested and reliable filesystem on the planet. It is simple, fast, and handles power failures gracefully using a "Journal."
• Best for: Standard desktops and low-to-medium load servers.

II. XFS

• The Workhorse: This is the default for Red Hat (RHEL).
• Why use it? It was designed for "High-Performance" and massive files. It handles parallel operations better than ext4. It cannot be "shrunk" (only grown), but it is rock solid for big databases.
• Best for: Enterprise servers and extremely large storage arrays.

III. Btrfs (B-Tree Filesystem)

• The Next-Gen: Used by Fedora and SUSE.
• Why use it? It includes features usually reserved for high-end hardware: Snapshots, Compression, and Self-Healing. If a file gets corrupted, Btrfs can detect it and fix it automatically from a mirror.
• Best for: Servers where data integrity is the #1 priority.

---

2. Practical: Formatting a Partition

Once you have a partition (like /dev/sdb1), you format it using the mkfs (Make FileSystem) command.

# Format as ext4 (Reliable)
sudo mkfs.ext4 /dev/sdb1

# Format as XFS (High Performance)
sudo mkfs.xfs /dev/sdb1

# Format as Btrfs (Snapshots and Checksums)
sudo mkfs.btrfs /dev/sdb1

---

3. The Identity Problem: UUID vs. Device Name

In the old days, we mounted disks by their name: mount /dev/sdb1 /data.

The Danger: If you unplug one disk and plug in another, Linux might rename the disks. /dev/sdb1 might become /dev/sdc1, and your server will fail to boot or, worse, mount the wrong disk to the wrong folder.

The Solution: The UUID. The UUID is a unique 128-bit string assigned to the filesystem when it is formatted. It never changes, even if you move the disk to a different server.

# Find the UUID of your disks
lsblk -f
# OR
sudo blkid

---

4. Manual Mounting: Bringing the Disk Online

Formatting isn't enough. You must "Attach" the filesystem to a folder in your hierarchy.

# Create the attachment point (Mount Point)
sudo mkdir /mnt/data

# Mount the disk manually
sudo mount /dev/sdb1 /mnt/data

# Verify it is working
df -h

---

5. Metadata: Labels and Optimization

You can give your disk a human-readable name (a Label). This is helpful when you are staring at a server with 24 identical drives.

# Give an ext4 partition a label
sudo e2label /dev/sdb1 BACKUP_DISK

# Check information about an ext4 filesystem
sudo tune2fs -l /dev/sdb1

---

6. Example: Filesystem Health Monitor (Python)

A filesystem isn't static. It can "Frag" or run out of "Inodes" (the limit on how many files you can have, even if you have GBs of space). Here is a Python script that audits the technical health of your filesystems.

import shutil

def check_disk_health(path="/"):
    """
    Checks total, used, and free space on a path.
    """
    total, used, free = shutil.disk_usage(path)
    
    usage_percent = (used / total) * 100
    
    print(f"--- Health Report for {path} ---")
    print(f"Total Capacity: {total // (2**30)} GB")
    print(f"Used Space:     {used // (2**30)} GB ({usage_percent:.1f}%)")
    print(f"Free Space:     {free // (2**30)} GB")
    
    if usage_percent > 90:
        print("[!!!] ALERT: Critical Space Shortage!")
    elif usage_percent > 75:
        print("[WA] Warning: Usage is exceeding 75%.")
    else:
        print("[OK] Space is healthy.")

if __name__ == "__main__":
    check_disk_health("/mnt/data")

---

7. Professional Tip: Check 'Reserved Blocks'

By default, ext4 reserves 5% of your disk space for the 'root' user. On a 10TB disk, that's 500GB of wasted space that you can't use for your data! If the disk is just for media or backups, you can reduce this reserve to 1%.

# Reduce reserved blocks to 1% to save space
sudo tune2fs -m 1 /dev/sdb1

---

8. Summary

Formatting is the act of preparing the disk's ledger.

• ext4 is the reliable veteran.
• XFS is the modern enterprise workhorse.
• Btrfs is the visionary architect with snapshots and self-healing.
• UUIDs are the only safe way to identify disks in a production system.
• mkfs creates the ledger, and mount opens it.

In the next lesson, we will learn how to make these mounts Permanent so they survive a reboot: Mastering /etc/fstab.

Quiz Questions

1. Why is XFS preferred over ext4 for very large files like database images?
2. What is the benefit of using a UUID over a device path like /dev/sdb1 in your configuration?
3. What happens to the data on a partition when you run a mkfs command on it?

---

Continue to Lesson 3: Persistent Mounts—Mastering /etc/fstab.