Learn NetBSD - Filesystems, Mounts & Disks

#Introduction

In episode 6 we brought services to life with the rc system. But everything we do — users, packages, services — is stored somewhere: the filesystem. In this episode we'll dissect NetBSD's storage layer: the FFS filesystem as the foundation, special filesystems like tmpfs and cd9660, the mount/umount commands, disk management with disklabel and newfs, and keeping data healthy with fsck and swapctl.

#The Main Filesystem: FFS

FFS (Fast File System) is NetBSD's default filesystem — the heir to the BSD tradition that has evolved over decades. FFS is known for its stability, maturity, and support for modern features like snapshots.

Let's see how your current system is organized:

Viewing the mounted filesystems

mount

Example mount output

/dev/wd0a on / type ffs (local)
/dev/wd0e on /usr type ffs (local)
/dev/wd0f on /var type ffs (local)
ptyfs on /dev/pts type ptyfs
procfs on /proc type procfs
tmpfs on /tmp type tmpfs (local)

Notice the different filesystems: ffs for persistent data, tmpfs for /tmp (in RAM), procfs and ptyfs for kernel interfaces. NetBSD uses special filesystems for special needs.

#Getting to Know Other Filesystems

Filesystem	Type	Purpose
`ffs`	Disk	Main filesystem for persistent data
`tmpfs`	RAM	Temporary data; lost on reboot
`mfs`	RAM	Traditional-style memory filesystem
`cd9660`	Optical	CD/DVD/ISO media
`procfs`	Pseudo	Kernel process interface
`ptyfs`	Pseudo	Pseudo-terminals

tmpfs is the modern choice for /tmp — fast because it lives in RAM, and automatically empty at reboot. mfs is its predecessor; tmpfs is now the recommended option.

#Mount and Umount

Attaching a filesystem into the directory tree is done with mount. To attach a disk or partition to a specific mount point:

Mounting a partition manually

mount /dev/wd0g /mnt

Checking the mount result

df -h /mnt

Mounting an ISO image is just one command — handy when you need to access installation media:

Mounting an ISO image (cd9660)

mount -t cd9660 -o loop NetBSD-11.0-amd64.iso /mnt/iso

Unmounting is done with umount:

Unmounting a filesystem

umount /mnt

Warning

Always umount before removing media. If the system complains of "device busy", a process is using files inside it — look for it with fstat -m /mnt or lsof /mnt before unmounting.

#/etc/fstab: The Permanent Mount List

Mounts that must exist at every boot are recorded in /etc/fstab. This is the system's permanent map:

Contents of /etc/fstab

/dev/wd0a   /            ffs     rw              1 1
/dev/wd0b   none         swap    sw              0 0
/dev/wd0e   /usr         ffs     rw              1 2
/dev/wd0f   /var         ffs     rw              1 2
tmpfs       /tmp         tmpfs   rw,-s=512m      0 0

Six columns per line: device, mount point, filesystem type, options, dump flag, and fsck pass. Notice two important lines: partition wd0b is swap (mounted with type sw), and /tmp uses tmpfs with a 512 MB limit.

#Managing Disks: disklabel & newfs

#Reading and Editing the disklabel

Every NetBSD disk has a disklabel — the BSD-level partition table we met in episode 3. Reading it:

Viewing the disklabel of the first disk

disklabel wd0

Example disklabel output

# /dev/rwd0d:
type: ESDI
disk: wd0
sectors/track: 63
...
 
a:   4096000  63    4.2BSD  2048 16384 12192 # (Cyl. 0 - 202)
b:   2048000  4096063 swap                   # (Cyl. 202 - 303)
c:  20971520  0     unused  0     0         # (Cyl. 0 - 1039)
e:   4194304  6144063 4.2BSD 2048 16384 12192 # (Cyl. 303 - 510)
f:  10633153  10338367 4.2BSD 2048 16384 12192 # (Cyl. 510 - 1039)

Partition c is the whole disk (BSD convention), and the other partitions stack on top of it. Type 4.2BSD marks FFS, and swap marks a swap partition.

#Creating a Filesystem with newfs

After the disklabel is set up, format a partition with newfs to create an FFS filesystem:

Creating an FFS filesystem on a partition

newfs /dev/rwd0g

Example newfs output

/dev/rwd0g: 2048.0MB (4194304 sectors) block size 32768, fragment size 4096
    using 4 cylinder groups of 512.00MB, 124 blks, 968 inodes.
super-block backups (for fsck -b #) at:
 32, 1048576, ...

Always use the raw device (with a leading r: rwd0g) for newfs — that's the BSD convention for buffer-less device access.

Danger

newfs erases the entire contents of a partition. Check the disklabel twice before running it, and make sure you're pointing at the right partition. There's no undo button.

#Staying Healthy: fsck

fsck checks and repairs filesystem consistency. After a system crash or power loss, NetBSD automatically runs fsck at boot for filesystems marked dirty. Manually:

Checking the root filesystem

fsck /dev/rwd0a

Example clean fsck output

/dev/rwd0a: 512 files, 4096 used, 299468 free

Modern FFS also supports snapshots — consistent copies of a filesystem for backup. Creating a snapshot uses fsck -t:

Creating a filesystem snapshot

fsck -t /dev/rwd0e

Example snapshot output

Snapshot of /dev/rwd0e created at /var/snapshot/rwd0e.0

#Managing Swap: swapctl

NetBSD swap is managed with swapctl — not swapon/swapoff as on Linux. Viewing the swap status:

Viewing active swap

swapctl -l

Example swapctl -l output

Device      512-blocks     Used    Avail Capacity  Priority
/dev/wd0b       2048000    51200  1996800     3%    0

Adding a new swap partition manually:

Activating a swap partition

swapctl -a /dev/wd0b

#Closing

In this episode 7, you've mastered NetBSD's storage layer: the FFS filesystem as the foundation, special filesystems (tmpfs, mfs, cd9660), the mount and umount commands, the permanent mount list in /etc/fstab, disk management with disklabel and newfs, and keeping data healthy with fsck and swapctl.

Key takeaways:

FFS is the default filesystem; use tmpfs for /tmp and temporary data.
Permanent mounts are recorded in /etc/fstab (six columns: device, mount, type, options, dump, pass).
newfs creates an FFS filesystem — always check the partition before running it.
fsck keeps the filesystem healthy; FFS supports snapshots with fsck -t.
Swap is managed with swapctl, not swapon.

In the next episode, episode 8, we'll connect the system to the outside world: networking and network configuration — configuring interfaces with ifconfig, routing, /etc/resolv.conf, and advanced features like VLAN, bonding (lagg), and bridges. See you in episode 8!