8.2 Persistent Mounts: /etc/fstab, UUIDs & systemd mount units (104.3)

8.2 Persistent Mounts: /etc/fstab, UUIDs & systemd mount units

Quick Summary: Filesystem attachments made manually using the mount command are temporary and do not survive a system reboot. To make mounts persistent, Linux reads the /etc/fstab (File System Table) configuration file during system startup. Every line in /etc/fstab defines a storage device, its designated mount point, its filesystem type, specific mount options, dump backup scheduling, and filesystem integrity check (fsck) order. Modern Linux distributions convert /etc/fstab entries dynamically into native systemd .mount units. Mastering /etc/fstab column structure, persistent identifiers (UUID, LABEL), and verification techniques (mount -a) is a core LPIC-1 competency.

---

1. Anatomy of the /etc/fstab Configuration File

The /etc/fstab file is plain text. Blank lines and lines beginning with a hash mark (#) are treated as comments and ignored by the parser. Each active entry consists of exactly six whitespace-separated columns (spaces or tabs):

# Sample /etc/fstab entry:
# [1. Device]                                [2. Mount]   [3. Type]  [4. Options]             [5. Dump] [6. Pass]
UUID=4a92c3d1-81f2-4e02-990a-1123456789ab    /            ext4       defaults,noatime         0         1
UUID=9f8e7d6c-5b4a-3210-fedc-ba9876543210    /home        xfs        defaults,nodev,nosuid    0         0
LABEL=BACKUP_DATA                            /mnt/backup  ext4       defaults,noauto,user     0         2
/dev/sr0                                     /media/cdrom iso9660    ro,user,noauto           0         0
/dev/sda3                                    none         swap       sw                       0         0
192.168.1.50:/exports/data                   /mnt/nfs     nfs        _netdev,defaults         0         0

The 6 /etc/fstab Columns Reference Table

---

2. Field-by-Field Technical Breakdown

Field 1: Device Identifiers (UUID vs. LABEL vs. Device Nodes)

Using traditional device paths like /dev/sda1 in /etc/fstab is dangerous in modern systems. When storage hardware changes, controllers reinitialize, or USB drives are inserted, kernel device enumeration order can shift, causing /dev/sda1 to become /dev/sdb1 and leading to boot failure.

To prevent this, Linux supports persistent device naming:

1. UUID= (Universally Unique Identifier): A 128-bit hex string stored in the filesystem superblock (e.g., UUID=4a92c3d1-81f2-4e02-990a-1123456789ab). It remains constant even if the drive is moved to another SATA/NVMe port.
2. LABEL=: A human-readable volume label assigned during filesystem creation or via tools like e2label or xfs_admin -L (e.g., LABEL=ROOT_OS or LABEL=BACKUP).
3. PARTUUID= / PARTLABEL=: Identifiers stored in the GPT partition table rather than the filesystem superblock.

# Inspect UUIDs and Filesystem Types using blkid:
$ sudo blkid
/dev/sda1: UUID="4a92c3d1-81f2-4e02-990a-1123456789ab" BLOCK_SIZE="4096" TYPE="ext4" PARTUUID="000a12bc-01"
/dev/sda2: UUID="9f8e7d6c-5b4a-3210-fedc-ba9876543210" BLOCK_SIZE="512" TYPE="xfs" PARTUUID="000a12bc-02"

# Inspect block devices, UUIDs, and mount points using lsblk:
$ lsblk -f
NAME   FSTYPE FSVER LABEL  UUID                                 FSAVAIL FSUSE% MOUNTPOINTS
sda                                                                            
├─sda1 ext4   1.0   ROOT   4a92c3d1-81f2-4e02-990a-1123456789ab   18.2G    48% /
└─sda2 xfs          HOME   9f8e7d6c-5b4a-3210-fedc-ba9876543210   82.4G    12% /home

Field 4: Critical Mount Options in /etc/fstab

• defaults: Expands automatically to: rw,suid,dev,exec,auto,nouser,async.
• nofail: Prevents system boot failure and avoids dropping to emergency mode if the device is disconnected or fails to mount.
• _netdev: Instructs systemd and init scripts to delay mounting until network connectivity is active (mandatory for NFS, iSCSI, and CIFS shares).
• usrquota / grpquota: Enables user or group disk quota enforcement on ext4/XFS filesystems.

Field 6: Fsck Pass Order Rules (Exam Hotspot)

The sixth column determines how fsck behaves at boot:

• 1: Root filesystem (/) ONLY. The root filesystem must be checked first before any other services or mounts initialize.
• 2: All other local filesystems requiring boot-time integrity scans (e.g., /home, /var). All filesystems marked 2 are checked in parallel across physical spindles.
• 0: Disable fsck scanning. Mandatory for:
  • Swap partitions (swap)
  • Virtual filesystems (proc, sysfs, tmpfs)
  • Optical media and ISOs (iso9660)
  • Network shares (nfs, cifs)
  • Filesystems with internal journal replay engines that do not use fsck at boot, such as XFS and Btrfs.

⚠️ LPIC-1 Trap — Setting Fsck Pass Order for XFS: XFS has its own built-in metadata journaling and repair mechanisms executed during kernel mount. It does not run traditional fsck at boot. Setting column 6 to 1 or 2 for an XFS filesystem is an error; XFS entries in /etc/fstab should always have pass order set to 0.

---

3. Safe Workflow for Modifying /etc/fstab

A syntax error in /etc/fstab (such as a missing column, a typo in a UUID, or an invalid mount option) can prevent the system from completing boot, forcing it into an emergency root shell.

Best Practice Procedure:

1. Always create a backup before editing: sudo cp /etc/fstab /etc/fstab.bak.
2. Add or modify the entry using a text editor.
3. Test the configuration immediately without rebooting by executing:

   $ sudo mount -a
   

4. If mount -a returns zero errors and mounts the target successfully, the configuration is safe for reboot.

💡 LPIC-1 Exam Fill-in-the-Blank Alert: Which command reads /etc/fstab and mounts all listed filesystems that do not have the noauto option, allowing administrators to verify syntax without rebooting? Answer: mount -a

---

4. systemd Mount Units and systemd-fstab-generator

In modern systemd-based Linux systems, filesystem mounting is managed natively by systemd unit files ending in .mount and .automount.

How systemd Integrates with /etc/fstab

During early boot, systemd invokes an internal binary called systemd-fstab-generator. This tool parses /etc/fstab and automatically generates native .mount unit files inside the transient runtime directory /run/systemd/generator/.

systemd-fstab-generator Workflow:
 ┌──────────────┐      systemd-fstab-generator      ┌────────────────────────────────┐
 │  /etc/fstab  │ ─────────────────────────────────>│  /run/systemd/generator/       │
 └──────────────┘         (Executed at boot)        │  ├── var-log.mount             │
                                                    │  └── home.mount                │
                                                    └────────────────────────────────┘

Mount Unit Naming Convention (systemd-escape)

Systemd requires that .mount unit filenames strictly match the path of the mount point directory, with leading slashes stripped and all internal slashes replaced by dashes (-):

• Mount point / → -.mount
• Mount point /home → home.mount
• Mount point /var/log → var-log.mount
• Mount point /srv/web/data → srv-web-data.mount

Administrators can generate properly escaped unit names using the systemd-escape command:

$ systemd-escape --path /var/log
var-log

$ systemd-escape --path --suffix=mount /srv/web/data
srv-web-data.mount

Inspecting and Managing Mount Units via systemctl

# List all active mount units:
$ systemctl list-units --type=mount

# View the status of the /var/log mount unit generated from fstab:
$ systemctl status var-log.mount

# Reload systemd generators after editing /etc/fstab without rebooting:
$ sudo systemctl daemon-reload