From NixOS Wiki
Jump to: navigation, search

fileSystems is a NixOS option that allows the user to mount filesystems at specific mount points. The mounted filesystems may also be encrypted. Also see the fileSystem option documentation.

For boot mount options, check here.

Common example filesystem mount. You can put this in configuration.nix:

 fileSystems."/mnt/exampleDrive" = {
   device = "/dev/disk/by-uuid/4f999afe-6114-4531-ba37-4bf4a00efd9e";
   fsType = "exfat";
   options = [ # If you don't have this options attribute, it'll default to "defaults" 
     # boot options for fstab. Search up fstab mount options you can use
     "users" # Allows any user to mount and unmount
     "nofail" # Prevent system from failing if this drive doesn't mount

Making disk visible in your file explorer

You might not see the disk in your file explorer (ie GNOME Nautilus). Add to the options: x-gvfs-show and it'll show up.

Porting /etc/fstab

The options specified in /etc/fstab may not be fully compatible with NixOS fileSystems options. For example, here are some options NixOS doesn't recognize that are available on some Linux distributions:

  • iocharset
  • rw (but it seems to not be needed)
  • uid with username rather than actual uid

Mount order

Without any specification, the mount order is up to the implementation (probably alphabetic).

Should the order in which filesystems are mounted is important, users should make use of the fileSystems.<mount>.depends option. This is useful for example in #Bind mounts

Bind mounts

Bind mounting allows a filesystem hierarchy or a file to be mounted at a different mount point. Unlike a symbolic link, a bind mount does not exist on the filesystem itself.[3] In the following example, the path /olddir will be mounted in /newdir [1]

These are used to make files or folders available in other parts of the filesystem hierarchy. In order to do so both source and target filesystems have to be mounted first.

fileSystems."/mnt/datastore".label = "datastore";
fileSystems."/mnt/aggregator".label = "aggregator";

# Bind mounts

# Mount /mnt/datastore/applications/app1 on /mnt/aggregator/app1
# Accessing /mnt/aggregator/app1 will actually access /mnt/datastore/...
fileSystems."/mnt/aggregator/app1" = {
  depends = [
      # The mounts above have to be mounted in this given order
  device = "/mnt/datastore/applications/app1";
  fsType = "none";
  options = [
    "ro" # The filesystem hierarchy will be read-only when accessed from /mnt/aggregator/app1