NFS, or Network File System, is a distributed file system protocol that allows you to mount remote directories on your server. This lets you manage storage space in a different location and write to that space from multiple clients. NFS provides a relatively quick and easy way to access remote systems over a network and works well in situations where the shared resources will be accessed regularly.
In this guide, we’ll cover how to configure NFS mounts.
We will be using two servers in this tutorial: one will share part of its filesystem with the other. To follow along, you will need:
Two Ubuntu 16.04 servers, each with a non-root user with
sudo privileges and private networking enabled.
For assistance setting up a user with these privileges, follow our Initial Server Setup with Ubuntu 16.04 guide.
For help setting up private networking, see How To Set Up And Use DigitalOcean Private Networking .
Throughout the tutorial, we refer to the server that shares its directories as the host and the server that mounts these directories as the client. In order to keep them straight, we’ll use the following IP addresses as stand-ins for the host and client values:
You should replace these values with your own host and client ip addresses.
We’ll begin by installing the necessary components on each server.
On the host server, we will install the
nfs-kernel-server package, which will allow us to share our directories. Since this is the first operation that we’re performing with
apt in this session, we’ll refresh our local package index before the installation:
- sudo apt-get update
- sudo apt-get install nfs-kernel-server
Once these packages are installed, switch to the client server.
On the client server, we need to install a package called
nfs-common, which provides NFS functionality without including unneeded server components. Again, we will refresh the local package index prior to installation to ensure that we have up-to-date information:
- sudo apt-get update
- sudo apt-get install nfs-common
Now that both servers have the necessary packages, we can start configuring them.
We’re going to share two separate directories, with different configuration settings, in order to illustrate two key ways that NFS mounts can be configured with respect to superuser access.
Superusers can do anything anywhere on their system. However, NFS-mounted directories are not part of the system on which they are mounted, so by default, the NFS server refuses to perform operations that require superuser privileges. This default restriction means that superusers on the client cannot write files as root, re-assign ownership, or perform any other superuser tasks on the NFS mount.
Sometimes, however, there are trusted users on the client system who need to be able to do these things on the mounted file system but who have no need for superuser access on the host. The NFS server can be configured to allow this, although it introduces an element of risk, as such a user could gain root access to the entire host system.
In the first example, we’ll create a general-purpose NFS mount that uses default NFS behavior to makes it difficult for a user with root privileges on the client machine to interact with the host using those client superuser privileges. You might use something like this to store the files uploaded using a content management system or to create space for users to easily share project files.
First, make a share directory called
- sudo mkdir /var/nfs/general -p
Since we’re creating it with
sudo, the directory is owned by root here on the host.
- ls -la /var/nfs/general
Output4 drwxr-xr-x 2 root root 4096 Jul 25 15:26 .
NFS will translate any
root operations on the client to the
nobody:nogroup credentials as a security measure. Therefore, we need to change the directory ownership to match those credentials.
- sudo chown nobody:nogroup /var/nfs/general
This directory is now ready for export.
In our second example, the goal is to make user home directories stored on the host available on client servers, while allowing trusted administrators of those client servers the access they need to conveniently manage users.
To do this, we’ll export the
/home directory. Since it already exists, we don’t need to create it. We won’t change the permissions, either. If we did, it would cause all kinds of issues for anyone with a home directory on the host machine.
Next, we’ll dive into the NFS configuration file to set up the sharing of these resources.
/etc/exports file in your text editor with root privileges:
- sudo nano /etc/exports
The file has comments showing the general structure of each configuration line. The syntax is basically:
We’ll need to create a line for each of the directories that we plan to share. Since our example client has an IP of
203.0.113.256, our lines will look like the following. Be sure to change the IPs to match your client:
/var/nfs/general 203.0.113.256(rw,sync,no_subtree_check) /home 203.0.113.256(rw,sync,no_root_squash,no_subtree_check)
We’re using the same configuration options for both directories with the exception of
no_root_squash. Let’s take a look at what each one means.
no_root_squashdisables this behavior for certain shares.
When you are finished making your changes, save and close the file. Then, to make the shares available to the clients that you configured, restart the NFS server with the following command:
- sudo systemctl restart nfs-kernel-server
Before you can actually use the new shares, however, you’ll need to be sure that traffic to the shares is permitted by firewall rules
First, let’s check the firewall status to see if it’s enabled and if so, to see what’s currently permitted:
- sudo ufw status
OutputStatus: active To Action From -- ------ ---- OpenSSH ALLOW Anywhere OpenSSH (v6) ALLOW Anywhere (v6)
On our system, only SSH traffic is being allowed, so we’ll need to add a rule for NFS traffic.
With many applications, you can use
sudo ufw app list and enable them by name, but
nfs is not one of those. Because
ufw also checks
/etc/services for the port and protocol of a service, we can still add NFS by name. Best practice recommends that you enable the most restrictive rule that will still allow the traffic you want to permit, so rather than enabling traffic from just anywhere, we’ll be specific.
Use the following command to open port 2049 on the host, being sure to substitute your client’s ip address:
- sudo ufw allow from 203.0.113.256 to any port nfs
You can verify the change by typing:
- sudo ufw status
You should see traffic allowed from port 2049 in the output:
OutputStatus: active To Action From -- ------ ---- OpenSSH ALLOW Anywhere 2049 ALLOW 203.0.113.256 OpenSSH (v6) ALLOW Anywhere (v6)
This confirms that UFW will only allow NFS traffic on port 2049 from our client machine.
Now that the host server is configured and serving its shares, we’ll prepare our client.
In order to make the remote shares available on the client, we need to mount the host directory on an empty client directory.
Note: If there are files and directories in your mount point, as soon as you mount the NFS share, they’ll be hidden. Be sure if you mount in a directory that already exists that the directory is empty.
We’ll create two directories for our mounts:
- sudo mkdir -p /nfs/general
- sudo mkdir -p /nfs/home
Now that we have some place to put the remote shares and we’ve opened the firewall, we can mount the shares by addressing our host server, which in this guide is
203.0.113.0, like this:
- sudo mount 203.0.113.0:/var/nfs/general /nfs/general
- sudo mount 203.0.113.0:/home /nfs/home
These commands should mount the shares from the host computer onto the client machine. You can double-check that they mounted successfully in several ways. You can check this with a plain
findmnt command, but
df -h will give you more human readable output illustrates how disk usage is displayed differently for the nfs shares:
- df -h
OutputFilesystem Size Used Avail Use% Mounted on udev 238M 0 238M 0% /dev tmpfs 49M 628K 49M 2% /run /dev/vda1 20G 1.2G 18G 7% / tmpfs 245M 0 245M 0% /dev/shm tmpfs 5.0M 0 5.0M 0% /run/lock tmpfs 245M 0 245M 0% /sys/fs/cgroup tmpfs 49M 0 49M 0% /run/user/0 203.0.113.0:/home 20G 1.2G 18G 7% /nfs/home 203.0.113.0:/var/nfs/general 20G 1.2G 18G 7% /nfs/general
Both of the shares we mounted appear at the bottom. Because they were mounted from the same file system, they show the same disk usage. To see how much space is actually being used under each mount point, use the disk usage command
du and the path of the mount. The
-s flag will provide a summary of usage rather than displaying the usage for every file. The
-h will print human readable output.
- du -sh /nfs/home
This shows us that the contents of the entire home directory is using only 20K of the available space.
Next, let’s test access to the shares by writing something to each of them.
First, write a test file to the
- sudo touch /nfs/general/general.test
Then, check its ownership:
- ls -l /nfs/general/general.test
Output-rw-r--r-- 1 nobody nogroup 0 Aug 1 13:31 /nfs/general/general.test
Because we mounted this volume without changing NFS’s default behavior and created the file as the client machine’s root user via the
sudo command, ownership of the file defaults to nobody:nogroup. Client superusers won’t be able to perform typical administrative actions, like changing the owner of a file or creating a new directory for a group of users, on this NFS-mounted share.
To compare the permissions of the General Purpose share with the Home Directory share, create a file Home Directory the same way:
- sudo touch /nfs/home/home.test
Then look at the ownership of the file:
- ls -l /nfs/home/home.test
Output-rw-r--r-- 1 root root 0 Aug 1 13:32 /nfs/home/home.test
home.test as root via the
sudo command, exactly the same way we created the
general.test file. However, in this case it is owned by root because we overrode the default behavior when we specified the
no_root_squash option on this mount. This allows our root users on the client machine to act as root and makes the administration of user accounts much more convenient. At the same time, it means we don’t have to give these users root access on the host.
We can mount the remote NFS shares automatically at boot by adding them to
/etc/fstab file on the client.
Open this file with root privileges in your text editor:
- sudo nano /etc/fstab
At the bottom of the file, we’re going to add a line for each of our shares. They will look like this:
. . . 203.0.113.0:/var/nfs/general /nfs/general nfs auto,nofail,noatime,nolock,intr,tcp,actimeo=1800 0 0 203.0.113.0:/home /nfs/home nfs auto,nofail,noatime,nolock,intr,tcp,actimeo=1800 0 0
Note: More information about the options we are specifying here can be found in the man page that describes NFS mounting in the
fstab with the
man nfs command.
The client server will automatically mount the remote partitions at boot, although it may take a few moments for the connection to be made and the shares to be available.
If you no longer want the remote directory to be mounted on your system, you can unmount it by moving out of the share’s directory structure and unmounting, like this:
- cd ~
- sudo umount /nfs/home
- sudo umount /nfs/general
This will remove the remote shares, leaving only your local storage accessible:
- df -h
OutputFilesystem Size Used Avail Use% Mounted on /dev/vda 59G 1.3G 55G 3% / none 4.0K 0 4.0K 0% /sys/fs/cgroup udev 2.0G 12K 2.0G 1% /dev tmpfs 396M 320K 396M 1% /run none 5.0M 0 5.0M 0% /run/lock none 2.0G 0 2.0G 0% /run/shm none 100M 0 100M 0% /run/user
If you also want to prevent them from being remounted on the next reboot, edit
/etc/fstab and either delete the line or comment it out by placing a # symbol at the beginning of the line. You can also prevent auto-mounting by removing the
auto option, which will allow you to mount it manually.
In this tutorial, we created an NFS host and illustrated some key NFS behaviours by creating two different NFS mounts, which we shared with our NFS client. If you’re looking to implement NFS in production, it’s important to note that the protocol itself is not encrypted. In cases where you’re sharing files that are intended to be publicly accessible, that doesn’t cause any serious problems.
If you’re using NFS for private data, however, you’ll need to decide how you want to protect that data. You might be able to route NFS over SSH or a VPN connection to create a more secure experience, but this often comes with a serious loss of performance. If performance is an issue, consider SSHFS. It’s slightly slower than unencrypted NFS traffic, but usually much faster than tunnelled NFS. Kerberos authenticated encryption for NFS is another option to explore.
Thanks for learning with the DigitalOcean Community. Check out our offerings for compute, storage, networking, and managed databases.