Making mounts permanent with /etc/fstab

A mount you made by hand lasts until the next reboot. The file that tells the system what to mount at boot is /etc/fstab. It is also one of the files where a small typo can stop a server from booting, so the skill here is as much about the routine for changing it safely as about the syntax.

Read and write /etc/fstab lines, using UUIDs.
Test changes with findmnt --verify and mount -a before rebooting.
Add a swap area to fstab, and use mount options such as noexec and nofail.

The six fields

[root@servera ~]# tail -4 /etc/fstab
LABEL=UEFI    /boot/efi vfat  defaults  0 0
UUID=c64946eb-fc27-4485-8020-db4119b686eb  /srv/data     xfs   defaults          0 0
UUID=cd0b3883-1d6d-4271-a89e-2f48a2dac5b7  /srv/archive  ext4  defaults,noexec   0 0
UUID=89858a47-2255-4fa3-a2a9-64711ac2d782  none          swap  defaults          0 0

A line of /etc/fstab: Select a field.

Two words about the first field. Identify the device by UUID (or LABEL), not by /dev/sdb1: device letters follow the order in which disks are found, and adding a disk can reorder them, so /dev/sdb1 might one day be a different partition. blkid prints the UUIDs.

A safe routine

Make a mount permanent, safely: Step through the routine that avoids a failed boot.

[root@servera ~]# blkid -s UUID -o value /dev/sdb1
c64946eb-fc27-4485-8020-db4119b686eb
[root@servera ~]# mkdir -p /srv/data /srv/archive
[root@servera ~]# cp /etc/fstab /root/fstab.bak
[root@servera ~]# vim /etc/fstab
[root@servera ~]# systemctl daemon-reload
[root@servera ~]# findmnt --verify | tail -3
0 parse errors, 0 errors, 1 warning
/
   [W] recommended root FS passno is 1 (current is 0)
[root@servera ~]# mount -a; echo "rc=$?"
rc=0
[root@servera ~]# df -h /srv/data /srv/archive
Filesystem      Size  Used Avail Use% Mounted on
/dev/sdb1       960M   39M  922M   5% /srv/data
/dev/sdb2       974M   24K  907M   1% /srv/archive

systemctl daemon-reload is needed because systemd turns every fstab line into a mount unit. findmnt --verify checks the file for errors, and mount -a mounts all entries that are not mounted yet; if the line is wrong, you see the error now and not at boot. (The warning about the root file system is only a recommendation.) The step that proves everything is a reboot.

Options worth knowing

Option	Effect
`defaults`	`rw,suid,dev,exec,auto,nouser,async`
`ro` / `rw`	Read-only / read-write
`noexec`	Programs on this file system cannot be run: good for uploads, archives, `/tmp`
`nosuid`, `nodev`	Ignore setuid bits / device files
`nofail`	A missing device does not stop the boot (secondary disks)
`noauto`	Not mounted by `mount -a` or at boot
`x-systemd.device-timeout=10s`	Give up waiting for the device after 10 seconds

For the fifth and sixth fields use 0 0: the first is an obsolete backup flag, the second the boot-time check order, which XFS does not need.

Swap in fstab

A swap area is a line with type swap and mount point none; swapon -a turns on all of them (the boot process does it automatically):

[root@servera ~]# swapon -a
[root@servera ~]# swapon --show
NAME      TYPE      SIZE USED PRIO
/dev/sdb3 partition 512M   0B   -2
[root@servera ~]# free -h | tail -1
Swap:          511Mi          0B       511Mi

pri=N in the options sets the priority when there are several swap areas. How much swap does a server need? It depends on the workload; a small swap lets the kernel move rarely used memory away, but a system that actively swaps all the time is too short on memory.

Try it

Check your understanding

Why use UUID= instead of /dev/sdb1 in /etc/fstab?
Which command tests all fstab entries without rebooting?
A typo in fstab can cause ...
Which fstab option makes a missing disk not stop the boot?
What is the mount point field for a swap entry?
Which option prevents programs on the file system from being executed?