Translation(s): English – Português (Brasil)


NFS Server Setup

The following do not specify NFS version 2 versus 3 versus 4; the steps below worked for me using NFS version 3 support built into the kernels of the server and the client (server is a Debian 4.0 Etch machine, the client was another Linux distribution, PLD "rescue".)

If you're looking to set up NFSv4 with Kerberos, please see: NFS/Kerberos.

Make sure you have NFS server support in your server's kernel (kernel module named "knfsd.ko" or "nfsd.ko" under your /lib/modules/uname -r/ directory structure). The config file will confirm if the kernel was compiled with NFS enabled.

  $ grep NFSD /boot/config-`uname -r`

Note that the config file might be found somewhere else in the directory structure (perhaps in /usr/src/linux/.config).

Then, note that there are at least two mainstream NFS server implementations at present (excluding those implemented in Python and similar): one implemented in user space (slower, easier to debug), and the other implemented in kernel space (faster.) The below shows the setup of the kernel-space one. If you wish to use the user-space server, then install the similarly-named package.

First, the packages to begin with:

 $ apt install nfs-kernel-server rpcbind

NOTE: The rpcbind package is only required if you want to run an NFSv2 or NFSv3 server. If all your clients support NFSv4, you can disable NFSv2 and NFSv3 (as described in the "NFSv4 only" section below) and skip the "Rpcbind" section below.

Rpcbind

Note that rpcbind defaults to only listening for NFS connection attempts on 127.0.0.1 (localhost), so if you wish to allow connections on your local network, then you need to edit /etc/default/rpcbind, to comment out the "OPTIONS" line. Also, we need to ensure that the /etc/hosts.allow file allows connections to the rpcbind port. For example:

 $ perl -pi -e 's/^OPTIONS/#OPTIONS/' /etc/default/rpcbind
 $ echo "rpcbind: 192.168.1." >> /etc/hosts.allow
 $ systemctl restart rpcbind.service

See 'man hosts.allow' for examples on the syntax. But in general, specifying only part of the IP address like this (leaving the trailing period) treats the specified IP address fragment as a wildcard, allowing all IP addresses in the range 192.168.1.0 to 192.168.1.255 (in this example.) You can do more "wildcarding" using DNS names, and so on too.

Exports

Edit the /etc/exports file, which lists the server's filesystems to export over NFS to client machines. And create the NFS table with "exportfs -a". The following example shows the addition of a line which adds the path "/example", for access by any machine on the local network (here 192.168.1.*).

 $ echo "/example 192.168.1.0/255.255.255.0(rw,no_root_squash,subtree_check)" >> /etc/exports
 $ exportfs -a
 $ /etc/init.d/nfs-kernel-server reload

This tells the server to serve up that path, readable/writable, with root-user-id connecting clients to use root access instead of being mapped to 'nobody', and to use the 'subtree_check' to silence a warning message. Then, reloads the server.

Note: If you want /etc/exports to share an NFS share to multiple, discrete entries, use space-separate entries. The syntax looks like this:

 $ echo "/multi-example    192.168.1.0/24(rw) 172.16.1.0/24(ro)  10.11.12.0/24(rw)" >> /etc/exports

Check which versions of NFS the server is running this way:

    root@deb9:/# cat /proc/fs/nfsd/versions
    -2 -3 +4 +4.1 +4.2
In this case, nfs2, nfs3 are disabled. NFS 4, 4.1, and 4.2 are enabled.  Any client trying to connect with NFSv2 or NFSv3 should get an error.

From the client side (for example, I used a PLD "rescue" disk to boot up a Windows machine for some diagnostics, and used its built-in NFS client support to mount a path from my other computer), you need to ensure that portmap is running, and that the NFS client support is built into the kernel or the proper module ("nfs.ko") is loaded, and then mount the server's path like so:

 $ mount 192.168.1.100:/example /mnt/example

To help debug:

NFSv4 only

To enable only NFSv4 or higher (and thus disabling v2 and v3), set the following variables in /etc/default/nfs-common:

NEED_STATD="no"
NEED_IDMAPD="yes"

And the following in /etc/default/nfs-kernel-server. Please note that RPCNFSDOPTS is not present by default, and needs to be added.

RPCNFSDOPTS="-N 2 -N 3"
RPCMOUNTDOPTS="--manage-gids -N 2 -N 3"

Additionally, rpcbind is not strictly needed by NFSv4 but will be started as a prerequisite by nfs-server.service. This can be prevented by masking rpcbind.service and rpcbind.socket.

sudo systemctl mask rpcbind.service
sudo systemctl mask rpcbind.socket

If you later choose to unmask rpcbind.service and rpcbind.socket execute the commands below. For example choose to unmask if 'sudo showmount' returns the error 'clnt_create: RPC: Unable to receive'.

sudo systemctl unmask rpcbind.service
sudo systemctl unmask rpcbind.socket

NFSv4 only requires a single port (TCP/UDP 2049) and does not require the portmap service to be installed.

Listening only on particular IPs

By default, NFS will listen for connections on all ports. To only listen for NFS(v4) connections on a particular IP address, add the -H option to RPCNFSDOPTS in /etc/default/nfs-kernel-server:

RPCNFSDOPTS="-N 2 -N 3 -H 10.0.1.1"

Alternatively, a hostname may be provided instead of an IP address. The -H option can also be provided multiple times to listen on multiple different IPs.

Refer to the rpc.nfsd manpage to see all the options available for RPCNFSDOPTS.


See Also:


CategoryNetwork CategorySoftware CategorySystemAdministration