Administer an XFS file system

XFS is the default file system of RHEL, AlmaLinux and Rocky: built for large volumes and concurrent access. This guide shows how to format with mkfs.xfs, read the parameters with xfs_info, grow online with xfs_growfs and repair with xfs_repair. For beginner to intermediate administrators. Outputs captured on an AlmaLinux test disk. The choice between XFS, ext4 and Btrfs is covered in the dedicated decision guide.

What you will learn

• Format a volume as XFS with mkfs.xfs
• Read the parameters with xfs_info
• Grow online with xfs_growfs (and why no shrinking)
• Check and repair with xfs_repair

Prerequisites

• A volume to format (partition or LVM logical volume)
• The xfsprogs package (installed by default on RHEL; apt install xfsprogs on Debian/Ubuntu)

Why XFS

XFS was designed by Silicon Graphics to handle very large volumes with high throughput. Available on Linux since the early 2000s, it is the default of the Red Hat family. Its architecture in allocation groups allows several write operations in parallel, which makes it excellent for databases, multimedia and busy file servers.

Two limits to know before choosing XFS:

• No shrinking in practice: an XFS volume can only grow. Plan the margin at creation time.
• No native snapshots or compression: for that, look at Btrfs or the LVM layer.

Create an XFS file system

The mkfs.xfs command formats the volume. With a label (-L):

sudo mkfs.xfs -L donnees /dev/sda2

meta-data=/dev/sda2     isize=512    agcount=4, agsize=32768 blks
         =              crc=1        finobt=1, sparse=1, rmapbt=1
         =              reflink=1    bigtime=1 inobtcount=1
data     =              bsize=4096   blocks=131072, imaxpct=25
naming   =version 2     bsize=4096   ascii-ci=0, ftype=1

The output summarises the geometry: 4 allocation groups (agcount=4), checksums on metadata (crc=1, the modern default) and a block size of 4096 bytes. For a volume on RAID, align with -d su=<stripe>,sw=<disks> to optimise writes:

sudo mkfs.xfs -L donnees -d su=64k,sw=4 /dev/sda2

Formatting erases everything

mkfs.xfs destroys the contents of the volume. Check the target with lsblk and the absence of mounts beforehand. The -f option forces overwriting an existing signature: only use it knowingly.

Check the result

xfs_info shows the geometry of a mounted XFS volume. It is the check after formatting, and the command to read before growing:

sudo xfs_info /mnt/data

The output repeats agcount, bsize and blocks. The label is read or changed with xfs_admin, without unmounting:

sudo xfs_admin -l /dev/sda2      # label = "donnees"
sudo xfs_admin -L data-prod /dev/sda2

Grow online with xfs_growfs

It is XFS's great strength in production: growing without unmounting. The operation is always done in two steps, container first, file system next:

1. Extend the container (partition with parted, or logical volume with lvextend).
2. Extend the mounted file system:

sudo xfs_growfs /mnt/data

data blocks changed from 131072 to 262144

The line confirms the doubling of the number of blocks. No need to give a size: XFS takes all the available space in its container.

XFS does not shrink

In practice, you do not shrink an XFS volume. If you need to recover space, the only way is to back up, recreate smaller, restore. It is the major difference with ext4 and its resize2fs. With LVM, lvextend -r chains extending the LV and xfs_growfs in one command.

The absolute wording "no command shrinks an XFS" deserves to be nuanced, however, since xfsprogs 6.x. A path exists, explicitly marked experimental, and it fails as soon as the shrink reaches into an allocation group other than the last. On AlmaLinux 10 with xfsprogs 6.16, trying to bring a volume from 102,400 down to 51,200 blocks, that is half of its groups, gives:

[EXPERIMENTAL] try to shrink unused space 51200, old size is 102400
xfs_growfs: XFS_IOC_FSGROWFSDATA xfsctl failed: Invalid argument

Exit code 1, number of blocks unchanged. Conversely, a shrink that stays within the last group succeeds: measured with xfsprogs 6.13, a volume of 262,144 blocks went down to 222,144 (data blocks changed from 262144 to 222144). The gain is therefore capped by the free space at the end of the volume. The practical conclusion of the guide fully holds: do not count on it, plan the size at creation time. But if you come across this message, you will know it is not a mistake on your part.

Check and repair with xfs_repair

XFS is not checked at boot (passno column at 0 in /etc/fstab). In case of an incident, the dedicated tool is xfs_repair, and the volume must be unmounted:

sudo umount /mnt/data
sudo xfs_repair /dev/sda2

For a diagnosis without changes (check without repairing), add -n:

sudo xfs_repair -n /dev/sda2

If the volume refuses to mount because of a corrupted journal and xfs_repair asks for it, -L erases the journal. It is a last resort: it can lose the most recent writes.

sudo xfs_repair -L /dev/sda2   # last resort, possible loss

Mount options and quotas

Mount options are added in /etc/fstab (complete configuration in the mount and persistence guide):

Option	Effect	For
noatime	does not write the access date	very busy volumes
nodev,nosuid,noexec	hardens the mount	/tmp, external disks
uquota / gquota / pquota	enables XFS quotas	shared volumes

Swipe to see more

XFS has built-in quotas per user, group and project. After mounting with uquota, you drive them with xfs_quota:

sudo xfs_quota -x -c 'limit bsoft=50g bhard=60g alice' /mnt/data
sudo xfs_quota -x -c 'report -h' /mnt/data

Troubleshooting

XFS incidents are recognised by the tool that refuses to act: xfs_repair facing an unreplayed journal, mkfs.xfs facing an existing signature, or the kernel logging errors in dmesg. The table starts from that message.

Symptom	Probable cause	Solution
xfs_repair refuses (dirty log)	journal not replayed	mount/unmount cleanly, otherwise xfs_repair -L
Impossible to shrink the volume	XFS does not shrink	back up, recreate smaller, restore
XFS errors in dmesg	failing disk	journalctl -k (filter XFS), check smartctl
mkfs.xfs refuses to format	existing signature	check the target, then -f if certain

Swipe to see more

Recommended reading

XFS is the default file system on RHEL: mastering it (mkfs.xfs, xfs_growfs, xfs_repair) is part of the RHCSA exam objectives. The RHCSA Red Hat Enterprise Linux 9 Cert Guide covers these objectives with exam-style exercises.

Amazon affiliate link. Buying through this link supports the site at no extra cost to you.

Put it into practice

Formatting is the founding action of the whole storage chain: best to do it once for real. The lab has you create an XFS file system with mkfs.xfs on a prepared partition, give it a label, then mount it and prove the result through the type, the label and the active mount.

Guided lab Practice

Learn how to do it

Create and label an XFS filesystem, then mount it

$ dsoxlab run l2-filesystem-create-xfs

Environment: VM·Duration: ~25 min

How it is checked: the state of the system is checked after a reboot, not the command you typed.

First time? What DSOXLab is and how to install it

DSOXLab is the command line that prepares a disposable environment (an isolated shell or a VM), runs the lab scenario, then checks the real state of the system and computes a score. Nothing is installed or broken on your workstation.

It needs Python 3.11 or later, uv or pipx, and git. The installation guide details these prerequisites; on Windows, or on a Mac without Python, use the appliance instead.

This lab starts virtual machines: it also needs ansible-playbook, a hypervisor (KVM or Incus), Terraform and hardware virtualization (/dev/kvm), which inside a VirtualBox VM means nested virtualization. The first time, dsoxlab start l2-filesystem-create-xfs checks all of it before creating the machines. Prepare your machine for vm labs.

Install the CLI and add the catalogue, once:

uv tool install dsoxlab   # or: pipx install dsoxlab
dsoxlab catalog add https://github.com/stephrobert/linux-dsoxlab-training

Before you run the lab, check:

• DSOXLab installed
• catalogue available
• compatible environment (isolated shell or VM)

To check, nothing is probed from this page: run dsoxlab doctor. Read the diagnosis

Run this lab, then validate your work:

dsoxlab run l2-filesystem-create-xfs
dsoxlab check l2-filesystem-create-xfs

See every lab: dsoxlab list-labs. Track your progress: dsoxlab progress. The full dsoxlab guide covers scoring and the pitfalls.

Key points

• XFS is the RHEL default: high performance, large volumes, concurrent access.
• mkfs.xfs formats; -L sets a label, -d su,sw aligns on RAID.
• xfs_growfs grows online; XFS does not shrink in practice (experimental shrinking limited to the last allocation group), plan the margin.
• xfs_repair repairs unmounted; -n diagnoses, -L (erases the journal) is a last resort.
• xfs_info reads the geometry; xfs_admin changes the label without unmounting.
• Built-in quotas with xfs_quota; no snapshots or compression (see Btrfs).

Knowledge check

Check that the essentials of this guide are acquired. The questions only cover what has just been explained here.

Knowledge check

Check what you have learned with this interactive quiz

6 questions

6 min

70% to pass

How it works

• The timer starts when you click Start
• Multiple choice, true/false and short answer questions
• You can move between questions
• Detailed results are shown at the end

Start the quiz

Starts the quiz and the timer

×

See in the glossary →