Mounting and Automounting Drives on Linux with fstab: Internal Disks and NAS Shares

A field guide I put together while wiring up storage on my own desktop. It covers two jobs people usually look up separately: mounting a secondary internal drive so it comes back after every reboot and mounting network shares from a NAS so they are there when you need them but do not hang your boot when the NAS is off.

If you have ever added a second drive and watched it mount at a different letter next time, or added a NAS line to /etc/fstab and then could not boot because the server was unplugged, this is the post for you. Everything here is done on a systemd-based Ubuntu or Kubuntu system, and the same approach works on most modern distributions.

All IP addresses and share names below are examples. Swap in your own.

What fstab actually is

/etc/fstab (the “file systems table”) is the list your system reads at boot to decide what gets mounted, where and with which options. Each line has six fields:

<what to mount>   <where>   <type>   <options>   <dump>   <pass>

what to mount: a device (better: its UUID) for local disks, or server:/export for a network share.
where: the mount point, an empty directory you create.
type: the filesystem, for example ext4 for a local disk or nfs for a network share.
options: a comma-separated list that controls how and when it mounts. This is where most of the useful behavior lives.
dump: almost always 0.
pass: the fsck order at boot. 1 for the root filesystem, 2 for other local disks, 0 for network shares.

The golden rule for local disks: mount by UUID, not by /dev/sdX. Device names like /dev/sdb are assigned in the order the kernel finds disks, so adding a drive or changing a cable can rename them and break your mounts. A UUID is baked into the filesystem and never changes.

Part 1: A secondary internal drive

Say you added a second SSD or hard drive and you want it mounted at /mnt/data on every boot.

Step 1: Find the drive

lsblk -o NAME,SIZE,FSTYPE,MOUNTPOINT

Look for your new drive (for example /dev/sdb) and note whether it already has a filesystem. If the FSTYPE column is blank, the drive is unformatted.

Step 2: Format it, only if it is new

Skip this step if the drive already has data you want to keep. Formatting erases everything on the target.

# DANGER: this wipes /dev/sdb1. Triple-check the device name first.
sudo mkfs.ext4 /dev/sdb1

If the drive has no partition table yet, create one first with sudo fdisk /dev/sdb (make a single primary partition), or use the Disks GUI (gnome-disks) if you prefer clicking to typing.

Step 3: Get the UUID

sudo blkid /dev/sdb1

You will get a line with UUID="xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx". Copy that value.

Step 4: Create the mount point

sudo mkdir -p /mnt/data

Step 5: Add the fstab line

sudoedit /etc/fstab

Add:

# Secondary internal data drive
UUID=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx   /mnt/data   ext4   defaults,nofail   0   2

defaults is a sensible baseline (read-write, mount at boot and so on).
nofail means the system still boots if the drive is missing or dies, instead of dropping you into an emergency shell.
0 2 means no dump and fsck this disk after the root filesystem.

Step 6: Apply and test without rebooting

sudo systemctl daemon-reload
sudo mount -a

mount -a mounts everything in fstab that is not already mounted. If it returns with no errors, you are good. Confirm:

df -h /mnt/data

If you want the files owned by your user rather than root, either sudo chown -R $USER:$USER /mnt/data after mounting, or for filesystems without Linux permissions (like NTFS or exFAT on an external drive) set ownership in the options, for example uid=1000,gid=1000.

Part 2: NAS shares over NFS

Now the network side. The goal here is different from a local disk. You want the share available on demand, but you never want a missing NAS to hang your boot. The trick is to let systemd mount it the moment something touches the folder, rather than at boot.

For this walkthrough the NAS lives at 192.168.1.50 and exports a few shares we will mount under /mnt/nas/.

Step 1: Install the NFS client

sudo apt install -y nfs-common

Confirm the mount helper is present and check what the NAS is offering:

which mount.nfs
showmount -e 192.168.1.50

showmount lists the exports the server is willing to share. If your share is not in that list, fix it on the NAS side before touching fstab.

Step 2: Map out the shares

NAS export	Local mount point
`192.168.1.50:/volume1/Media`	`/mnt/nas/Media`
`192.168.1.50:/volume1/Projects`	`/mnt/nas/Projects`
`192.168.1.50:/volume1/My Files`	`/mnt/nas/My_Files`
`192.168.1.50:/volume1/Backups`	`/mnt/nas/Backups`

Step 3: Create the mount points

sudo mkdir -p /mnt/nas/{Media,Projects,My_Files,Backups}

Step 4: Add the fstab block

# NAS NFS mounts from 192.168.1.50 (on-demand automount)
192.168.1.50:/volume1/Media       /mnt/nas/Media      nfs  noauto,vers=4.1,_netdev,nofail,x-systemd.automount,x-systemd.requires=network-online.target  0  0
192.168.1.50:/volume1/Projects    /mnt/nas/Projects   nfs  noauto,vers=4.1,_netdev,nofail,x-systemd.automount,x-systemd.requires=network-online.target  0  0
192.168.1.50:/volume1/My\040Files /mnt/nas/My_Files   nfs  noauto,vers=4.1,_netdev,nofail,x-systemd.automount,x-systemd.requires=network-online.target  0  0
192.168.1.50:/volume1/Backups     /mnt/nas/Backups    nfs  noauto,vers=4.1,_netdev,nofail,x-systemd.automount,x-systemd.requires=network-online.target  0  0

Note the third line. The export path My Files contains a space, and a space would break the fstab field parsing, so it is written as My\040Files. The \040 is the octal escape for a space. Any space in a path needs this.

Option reference

Option	Meaning
`noauto`	Do not mount at boot. Paired with `x-systemd.automount`, the share mounts on first access instead.
`vers=4.1`	Force NFS protocol version 4.1. Change to `vers=3` if your NAS does not support 4.x.
`_netdev`	Marks this as a network mount so systemd orders it after the network is up.
`nofail`	Do not block boot if the mount cannot be made.
`x-systemd.automount`	Creates an automount unit so the share mounts when you open the folder (for example in your file manager).
`x-systemd.requires=network-online.target`	Holds the mount until the network is actually online.
`0 0`	No dump, no `fsck` (never `fsck` a network share).

Step 5: Apply and verify

sudo systemctl daemon-reload
sudo systemctl restart 'mnt-nas-*.automount'

Systemd turns each mount point into a unit named after its path, so /mnt/nas/Media becomes mnt-nas-Media.automount. Check they are active:

systemctl list-units --type=automount | grep mnt-nas

Then trigger a mount just by listing the folder:

ls /mnt/nas/Media

The first access is slightly slow because that is when the real mount happens. After that it behaves like any other folder.

Manual test if something misbehaves

Before trusting fstab, you can always mount a share by hand to prove the server and path are right:

sudo mount -t nfs -o vers=4.1 192.168.1.50:/volume1/Media /mnt/nas/Media
ls /mnt/nas/Media
sudo umount /mnt/nas/Media

If the manual mount works but the automount does not, the problem is in your fstab options, not the network.

Troubleshooting

See why an automount or mount failed

systemctl status mnt-nas-Media.automount
systemctl status mnt-nas-Media.mount
journalctl -u mnt-nas-Media.mount -n 50

The journal output almost always names the real reason: wrong path, wrong NFS version, permission denied or no route to the host.

`mount-start-limit-hit`

This means the .mount unit failed several times quickly and systemd stopped retrying. Common causes:

nfs-common is not installed.
The export path is wrong (recheck with showmount -e).
The NFS version does not match what the server supports.

Fix the cause, then clear the failed state:

sudo systemctl reset-failed 'mnt-nas-*.automount' 'mnt-nas-*.mount'
sudo systemctl restart 'mnt-nas-*.automount'

The drive mounts but comes up at the wrong place, or vanishes after a reboot

This is the classic /dev/sdX trap on a local disk. Switch the fstab entry to UUID= (see Part 1, Step 3). Device letters are not stable, UUIDs are.

`mount -a` throws an error and I cannot boot

If you added a bad line and the machine drops to an emergency shell, remount root as writable, fix the file and continue:

mount -o remount,rw /
sudoedit /etc/fstab   # correct or comment out the bad line

This is exactly why nofail matters. With it on every non-root line, a missing drive or offline NAS becomes a warning instead of a failed boot. Add it before you reboot, not after.

NFS maps by user ID, not name. If your user ID on the desktop does not match the one that owns the files on the NAS, permissions will look off. Line the UIDs up on both ends, or set share-level permissions on the NAS.

Summary

Local disks: mount by UUID, use defaults,nofail and pass of 2.
NAS shares: install nfs-common, then use noauto with x-systemd.automount for on-demand mounting.
Use _netdev and x-systemd.requires=network-online.target so network mounts wait for the network.
Put nofail on everything except root so a missing drive never blocks boot.
Escape spaces in paths as \040.
Always run sudo systemctl daemon-reload after editing fstab and test with sudo mount -a before you trust it across a reboot.

References

Do not take my word or anyone else’s for how these options behave. The official documentation is the source of truth:

fstab(5) man page - the file format and fields
mount(8) man page - mount options, including filesystem-independent ones like nofail
nfs(5) man page - NFS-specific mount options like vers= and _netdev
systemd.mount(5) man page - the x-systemd.* options and how fstab becomes systemd units
systemd.automount(5) man page - on-demand mounting behavior
blkid(8) man page - finding UUIDs
Ubuntu: install and configure an NFS client - the official Ubuntu server guide
Ubuntu community wiki: Fstab and NFSv4Howto

Related reading:

Running a different NAS or a distribution that is not Ubuntu-based? Post your setup below and we can work through the options.