Learn FreeBSD - Filesystems, Mounts & Disks

#Introduction

In the previous episode 6, you managed services with the rc system and sysrc. Now we go one layer down to the foundation of storage: filesystems, mounts, and disks. Understanding this layer matters because every decision — from layout to disaster recovery — rests on it.

FreeBSD provides two main filesystems with very different personalities: UFS, classic and simple, and ZFS, modern with checksums and snapshots. On top of those are pseudo-filesystems for devices and memory. This episode maps it all out, then covers mounts, fstab, and swap.

#Main Filesystems: UFS and ZFS

#UFS: The Classic Berkeley Filesystem

UFS (Unix File System) is a Berkeley legacy with a long history in FreeBSD. Its modern version (UFS2) supports soft updates and snapshots. For simple systems, UFS remains a legitimate choice:

Memeriksa filesystem dengan mount dan df

mount
df -h

UFS is fast for small workloads and easy to understand. Its drawbacks: no end-to-end checksums, no multi-disk pooling, and limited snapshots compared to ZFS.

#ZFS: The Modern Filesystem

ZFS combines a volume manager and a filesystem into a single layer. Datasets, properties, checksums, compression, and snapshots make it far superior for production. ZFS details are covered fully in episodes 8-9, but you've already seen its traces since the bsdinstall stage.

Melihat dataset ZFS

zfs list

Info

Generally speaking, for a new installation in this day and age, use ZFS. UFS remains useful for special cases — such as small boot volumes or very constrained environments — but ZFS is the future of FreeBSD.

#Pseudo-Filesystems

FreeBSD mounts several virtual filesystems that don't store permanent data. Understanding them prevents confusion when mount displays many entries.

#devfs: Devices as Files

devfs provides device nodes in /dev. When you add a disk, new nodes appear automatically. This is why FreeBSD uses names like /dev/da0 for SCSI/SATA disks and /dev/ada0 for ATA disks.

#tmpfs: A Filesystem in Memory

tmpfs stores data in RAM and swap. It's well suited for /tmp, whose contents are easily discarded:

Memasang tmpfs di /tmp

mount -t tmpfs tmpfs /tmp

#procfs and fdescfs

procfs exposes processes as files (usually mounted at /proc for compatibility), and fdescfs provides file descriptors in /dev/fd. Both are rarely touched, but useful for compatibility with certain applications.

#Mount and Umount

#Mounting Filesystems

mount connects a filesystem to a directory (mount point):

Memasang filesystem UFS

mount -t ufs /dev/da1p1 /mnt
mount -t zfs tank/data /tank-data

mount -t specifies the filesystem type. Without a flag, FreeBSD tries to detect it automatically. View all mount points with mount or df.

#Unmounting Filesystems

umount detaches a filesystem. Make sure no process is using the mount point:

Melepas filesystem

umount /mnt
umount -f /mnt

The -f flag forces an unmount when a process is still using it. Use it only when truly necessary — you can lose data that hasn't been synced yet.

Warning

Before detaching a disk, make sure no process is writing. Use fstat to find processes that have files open on the mount point: fstat /mnt will show the PIDs that must be stopped first.

#/etc/fstab: Automatic Mounts

#Understanding the fstab Format

/etc/fstab determines which filesystems are mounted automatically at boot. Each line has six columns: device, mount point, type, options, dump flag, and passno:

Contoh isi /etc/fstab

cat /etc/fstab

On UFS systems, entries like /dev/ada0p2 / ufs rw 1 1 appear in this file. On ZFS systems, fstab is usually empty or minimal — ZFS mounts its own datasets at boot through the pool.

#Adding an Entry

To add a permanent filesystem, edit /etc/fstab with the right options. The noauto option prevents automatic mounting:

Menambahkan entri fstab

mount -a

mount -a mounts all fstab entries. After editing, run it to verify the syntax before rebooting.

#Swap Configuration

Swap is configured via an fstab entry with type swap, or managed manually with swapon and swapoff:

Mengelola swap

swapon /dev/da1p3
swapoff /dev/da1p3
swapinfo

swapinfo displays swap status. On a default ZFS installation, swap is created as a ZVOL in the pool.

Success

Check fstab with mount -a before rebooting every time you change this file. An error in fstab isn't always fatal — FreeBSD drops to single-user mode — but it's better to catch problems early.

#Checking Disk Health

A few commands for monitoring disks and filesystems:

Memantau disk

df -h
pstat -s
fsck -f /dev/ada0p2

df -h shows used capacity, pstat -s shows swap, and fsck checks and repairs UFS filesystems. ZFS doesn't need fsck — its integrity is maintained by checksums and zpool scrub (episode 9).

#Closing

In this episode 7, you understood FreeBSD filesystems: UFS and ZFS as the two main choices, the pseudo-filesystems devfs, tmpfs, procfs, and fdescfs, the mount and umount commands, the format and use of /etc/fstab, and swap management with swapon.

Key takeaways:

UFS is classic and simple; ZFS is modern with checksums and snapshots.
devfs exposes devices in /dev; tmpfs stores data in RAM.
mount -t <type> and umount control filesystem mounting.
/etc/fstab determines automatic mounts; verify with mount -a.
Swap is configured in fstab or via swapon/swapoff.

In the next episode, episode 8, we'll cover ZFS fundamentals — pools and vdevs, datasets, properties like compression and checksum, plus snapshots, clones, and boot environments with bectl. This is the storage foundation most widely used in FreeBSD production.