RED HAT ENTERPRISE LINUX

NFS: Network
File Systems

Mount and unmount network file systems using NFS

CIS126RH | RHEL System Administration 1
Mesa Community College

NFS (Network File System) allows Linux systems to share directories over the network and mount them as if they were local filesystems. Shared home directories, centralised application data, and read-only software repositories are all common NFS use cases in enterprise environments. This module covers discovering available NFS exports, mounting them temporarily and persistently, and unmounting them cleanly. The client-side NFS skills are tested on the RHCSA exam.

Learning Objectives

1. Explain how NFS works — Describe the NFS client-server model, NFS versions, and what the client needs to mount a remote share
2. Discover and mount NFS shares — Use showmount to list available exports and mount with NFS-specific options to attach a remote share
3. Configure persistent NFS mounts in /etc/fstab — Write correct fstab entries for NFS filesystems including the _netdev option and verify them
4. Use autofs for on-demand mounting — Install and configure the autofs service to mount NFS shares automatically when they are accessed

How NFS Works

NFS is a client-server protocol. The server exports directories; the client mounts them and accesses the files as if they were local.

• The NFS server runs nfs-server.service and lists exported directories in /etc/exports
• The NFS client mounts the server's exports using the mount command or /etc/fstab
• File access goes over the network — reads and writes are sent to the server
• RHEL 9 supports NFSv4 (default) and NFSv3
• NFSv4 uses TCP port 2049 only; NFSv3 requires multiple RPC ports
• The client needs the nfs-utils package for NFS tools

Installing NFS Client Tools

The NFS client tools are in the nfs-utils package. This package provides mount.nfs, showmount, and the kernel NFS client support files.

# Install nfs-utils if not already present
$ sudo dnf install -y nfs-utils

# Verify key NFS client tools are available
$ which showmount
/usr/sbin/showmount

$ rpm -q nfs-utils
nfs-utils-2.5.4-26.el9.x86_64

# Check which NFS versions are supported by the kernel
$ cat /proc/fs/nfsfs/volumes

# The rpcbind service is needed for NFSv3 only
$ sudo systemctl enable --now rpcbind   # only if using NFSv3

Discovering NFS Exports: showmount

showmount queries an NFS server and lists the directories it is currently exporting — the prerequisite to mounting.

# List all exports from an NFS server
$ showmount -e servera
Export list for servera:
/exports/data    *
/exports/shared  192.168.1.0/24
/home            *(ro,sync)

# List exports using the server's IP address
$ showmount -e 192.168.1.10

# List which clients have each export mounted
$ showmount -a servera
All mount points on servera:
192.168.1.20:/exports/data
192.168.1.21:/exports/shared

# List only the exported directories (no client info)
$ showmount -d servera

Mounting an NFS Share

The NFS mount syntax is SERVER:/export/path — server hostname or IP, colon, and the exported directory path.

# Basic NFS mount — filesystem type and options auto-detected
$ sudo mkdir -p /mnt/nfsdata
$ sudo mount -t nfs servera:/exports/data /mnt/nfsdata

# Mount with explicit NFSv4
$ sudo mount -t nfs4 servera:/exports/data /mnt/nfsdata

# Mount with NFSv3
$ sudo mount -t nfs -o nfsvers=3 servera:/exports/data /mnt/nfsdata

# Mount read-only
$ sudo mount -t nfs -o ro servera:/exports/data /mnt/nfsdata

# Verify the mount succeeded
$ mount | grep nfs
servera:/exports/data on /mnt/nfsdata type nfs4 (rw,relatime,...)

$ df -hT /mnt/nfsdata
Filesystem             Type  Size  Used Avail Use% Mounted on
servera:/exports/data  nfs4   10G  3.2G  6.8G  32% /mnt/nfsdata

NFS Mount Options

NFS-specific mount options control performance, reliability, and behaviour when the server is unreachable.

Persistent NFS Mounts in /etc/fstab

For an NFS mount to survive reboots, add it to /etc/fstab with the _netdev option — which tells systemd to wait for the network before mounting.

# /etc/fstab entry for an NFS share
# SERVER:/path  MOUNTPOINT  nfs  OPTIONS  0 0

servera:/exports/data    /mnt/nfsdata  nfs  defaults,_netdev  0 0

# With explicit NFS options
servera:/exports/data    /mnt/nfsdata  nfs  rw,sync,hard,_netdev  0 0

# Read-only NFS share (e.g. software repository)
servera:/exports/software  /software   nfs  ro,_netdev             0 0

# Test the fstab entry
$ sudo mount -a
$ df -hT /mnt/nfsdata
$ findmnt --verify

Unmounting NFS Shares

NFS shares are unmounted with umount just like local filesystems, but there are additional considerations when the server is unavailable.

# Unmount by mount point
$ sudo umount /mnt/nfsdata

# Unmount when the NFS server is unreachable (lazy unmount)
$ sudo umount -l /mnt/nfsdata

# Force unmount even with open files (use with caution)
$ sudo umount -f /mnt/nfsdata

# Lazy + force for an unreachable server with processes using the mount
$ sudo umount -l -f /mnt/nfsdata

# If normal umount hangs — check for processes using the mount
$ sudo fuser -vm /mnt/nfsdata
$ sudo lsof /mnt/nfsdata

# Confirm unmount succeeded
$ mount | grep nfsdata
# No output = successfully unmounted

Introduction to autofs

autofs is a service that mounts NFS shares (and other filesystems) automatically when a user accesses them, and unmounts them after a period of inactivity.

• NFS shares are mounted on demand — only when accessed
• Shares are unmounted automatically after 5 minutes of inactivity (default)
• If the NFS server is down, autofs simply returns "no such file or directory" — it does not hang the boot
• autofs is the enterprise standard for home directory mounting in large organisations
• No /etc/fstab entries are needed — autofs has its own configuration files

Configuring autofs

autofs uses two levels of configuration: the master map defines mount points and points to map files; map files define the individual NFS shares.

# Step 1: Install autofs
$ sudo dnf install -y autofs

# Step 2: Edit the master map (/etc/auto.master)
$ cat /etc/auto.master
# Master map for autofs
/misc  /etc/auto.misc
/mnt   /etc/auto.mnt   --timeout=300
/net   -hosts

# Step 3: Create a map file (/etc/auto.mnt)
$ sudo vim /etc/auto.mnt
# Format: KEY  OPTIONS  SERVER:/EXPORT
data    -rw,sync,hard  servera:/exports/data
shared  -ro            servera:/exports/shared

# Step 4: Enable and start autofs
$ sudo systemctl enable --now autofs

# Step 5: Test — accessing the path triggers an automatic mount
$ ls /mnt/data           # this triggers the mount
$ mount | grep autofs   # confirm autofs mount

autofs Map Types

autofs supports indirect maps (a directory containing automounted subdirectories) and direct maps (specific absolute paths).

Indirect map (most common) — shares appear as subdirectories

# /etc/auto.master — indirect map mounted under /shares
/shares  /etc/auto.shares

# /etc/auto.shares — each key becomes /shares/KEY
data    -rw,sync  servera:/exports/data
logs    -ro       servera:/exports/logs
# Accessing /shares/data triggers mount of servera:/exports/data

Direct map — each share at its own absolute path

# /etc/auto.master — direct map
/-  /etc/auto.direct

# /etc/auto.direct — absolute paths as keys
/data      -rw,sync  servera:/exports/data
/var/share -ro       servera:/exports/shared
# Accessing /data triggers mount of servera:/exports/data

autofs for Home Directories

A common exam scenario: mount user home directories from an NFS server automatically when users log in.

# Scenario: servera exports /home/ldapuser* via NFS
# Goal: /home/ldapuser1 on the client should automount from the server

# Step 1: Add to /etc/auto.master
/home  /etc/auto.home

# Step 2: Create /etc/auto.home
$ sudo vim /etc/auto.home
# The wildcard * matches any username; & substitutes the matched value
*  -rw,sync,hard  servera:/home/&
# When a user accesses /home/ldapuser1:
#   autofs mounts servera:/home/ldapuser1 at /home/ldapuser1

# Step 3: Reload autofs
$ sudo systemctl reload autofs

# Step 4: Test — access the home directory
$ ls /home/ldapuser1   # triggers the mount
$ mount | grep ldapuser

Troubleshooting NFS Mounts

NFS mount failures have several common causes — most are diagnosed with a systematic approach.

# Test 1: Can the client reach the server?
$ ping -c 3 servera
$ nc -z servera 2049    # test port 2049 is reachable (NFSv4)

# Test 2: Is the export available?
$ showmount -e servera

# Test 3: View NFS client errors in the journal
$ sudo journalctl -u autofs --since "10 minutes ago"
$ sudo journalctl -k | grep nfs   # kernel NFS messages

# Test 4: Check RPC services (NFSv3 only)
$ rpcinfo -p servera

# Test 5: Try verbose mount
$ sudo mount -v -t nfs servera:/exports/data /mnt/nfsdata

NFS Quick Reference

Common Mistakes

Knowledge Check

Answer these before moving to the next slide.

1. Write the command to discover which NFS exports are available from a server named fileserver.
2. Write the command to mount fileserver:/exports/apps at /apps using NFSv4 with read-write access.
3. Write the /etc/fstab entry for the mount in question 2 to make it persistent. What two field values are always 0 for NFS?
4. What does the _netdev fstab option do, and why is it required for NFS mounts?
5. Write the autofs configuration (master map entry and map file entry) to make fileserver:/exports/apps automatically mount at /apps when accessed.
6. Write the autofs wildcard map file entry that would automatically mount any user's home directory from fileserver:/home/USERNAME when they access /home/USERNAME.

Knowledge Check — Answers

1. showmount -e fileserver — the -e flag lists all exported directories from the specified server.
2. sudo mkdir -p /apps && sudo mount -t nfs fileserver:/exports/apps /apps
   NFSv4 is the default — no additional flags needed. For explicit NFSv4: sudo mount -t nfs4 fileserver:/exports/apps /apps
3. fileserver:/exports/apps /apps nfs defaults,_netdev 0 0
   Both the dump field (5th) and the fsck order field (6th) are always 0 for NFS — the dump command cannot back up NFS, and fsck cannot check network filesystems.
4. _netdev tells systemd to defer the mount until after the network is fully initialised. NFS requires the network to contact the server. Without _netdev, the system tries to mount the NFS share during early boot before the network interface is configured, causing the mount to fail or the boot to hang.
5. In /etc/auto.master: /apps /etc/auto.apps
   In /etc/auto.apps: apps -rw,sync fileserver:/exports/apps
   Then: sudo systemctl enable --now autofs
6. In /etc/auto.master: /home /etc/auto.home
   In /etc/auto.home: * -rw,sync,hard fileserver:/home/&
   The * matches any username; the & substitutes the matched name into the server path.

Key Takeaways

1. NFS mount syntax: SERVER:/export as the device argument. Install nfs-utils. Use showmount -e SERVER to discover exports. Mount with sudo mount -t nfs SERVER:/export /mountpoint. Unmount with umount; use -l -f if the server is unreachable.
2. fstab NFS entries always use _netdev and 0 0. _netdev defers the mount until the network is ready — without it, boot fails. The dump (field 5) and fsck order (field 6) are both 0 for NFS — neither dump nor fsck works on network filesystems.
3. autofs mounts on demand and unmounts after inactivity. Configure the master map (/etc/auto.master) and a map file. Indirect map format: KEY -OPTIONS SERVER:/export. Reload after changes: sudo systemctl reload autofs.
4. The wildcard pattern handles many users from one map entry. * -rw,sync SERVER:/home/& mounts any user's home directory on demand. The * matches any key; & substitutes the matched value into the server path.

Graded Lab

• Install nfs-utils on the client. Use showmount -e to list the exports from your lab NFS server. Record the export path and confirm the client IP is allowed.
• Create /mnt/nfsshare and mount the NFS export temporarily with sudo mount -t nfs. Verify with df -hT that the filesystem type shows as nfs4. Create a test file in the mount.
• Unmount the share with sudo umount /mnt/nfsshare. Add a persistent fstab entry using _netdev and the correct 0 0 fields. Test with sudo mount -a and verify the test file is still accessible.
• Install autofs. Add an entry to /etc/auto.master pointing /shares to a new map file /etc/auto.shares. Create the map file with an entry to mount the NFS export at /shares/nfsdata.
• Enable and start autofs. Access /shares/nfsdata and confirm it is automatically mounted with mount | grep autofs. Wait 5 minutes and confirm the share is automatically unmounted.
• Reboot the system. After reboot, confirm the fstab NFS mount is present at /mnt/nfsshare with df -hT, and that autofs is running and serves the /shares/nfsdata mount on demand.

Next: Getting Help from Local Documentation