Fix an NTFS external drive that Files refuses to mount: "volume is dirty and \"force\" flag is not set"
Checked the dirty-flag claim against the kernel source, the Arch wiki, the Arch package index, the udisks API documentation and this workstation (omarchy 4.0.2-1, kernel 7.1.9, udisks2 2.11.2-1, udiskie 2.7.0-2). Confirmed from source: fs/ntfs3/super.c emits It is recommended to use chkdsk. when VOLUME_FLAG_DIRTY is set and then refuses the mount with volume is dirty and "force" flag is not set! only when the mount is not read-only and force is absent, which is exactly the record's two dmesg lines and which proves the read-only mount advice. The Arch wiki NTFS page independently states that udisks prefers the ntfs3 driver, quotes the same dmesg line, and names ntfsfix --clear-dirty as the remedy. The Arch package index confirms ntfsprogs is still 2026.7.7-1 in extra today, pkgbase ntfs-3g, and that it ships /usr/bin/ntfsfix. Confirmed on this machine: neither ntfsprogs nor ntfs-3g is installed, so the default driver really is kernel ntfs3, udisksctl mount does accept -o, --options, /usr/share/omarchy/bin/omarchy-pkg-add exists and runs pacman -S --noconfirm --needed with no -Sy, so it trips no partial upgrade and no ALPM guard (the guard in /usr/bin/omarchy-update-pacman-guard fires only when both a sync and a sysupgrade flag are present). Issue 8725, read in full with comments, supports the symptom, the cause and the fix, and the reporter confirmed ntfsfix worked. It is still open today, which the record does not misstate. One thing needed correcting. The record stops at the dirty flag, but the same thread shows the reporter then hit a root-owned mount caused by a leftover /etc/fstab entry written by the Disks app, and the udisks Filesystem API documentation confirms the mechanism: if a device is referenced in /etc/fstab, udisks calls mount directly as root and ignores the options given. That makes two of the record's own steps unreliable as written, the -o ro mount and the verify line promising a /run/media/<user> target, so I rewrote the fix to check fstab first and rewrote verify to name the three failure outputs. I also moved the package date from 2026-09-07 to 2026-09-11, the day I checked it. NOT exercised: I have no NTFS device and no sudo here, so I ran no mount, no dmesg and no ntfsfix. The dirty-flag behaviour and the fstab behaviour are taken from the kernel source, the udisks documentation and the thread.
ntfsfix is not chkdsk. It clears the dirty flag and repairs a few known inconsistencies, and if it reports errors it cannot fix, stop and let Windows chkdsk run on the drive. Copy irreplaceable data off through the read-only mount before clearing the flag.
Symptom
A USB hard drive formatted NTFS (a WD Elements in the report) shows up in the sidebar of Files (Nautilus) but clicking it gives a generic mount error. The kernel log has the real reason:
$ sudo dmesg | tail -n 20
[ 118.870318] ntfs3(sda1): It is recommended to use chkdsk.
[ 118.934869] ntfs3(sda1): volume is dirty and "force" flag is not set!
Reported on Omarchy 4 on an Intel NUC13, one machine.
Cause
Omarchy mounts NTFS through udisks, which uses the kernel ntfs3 driver. The volume carries the NTFS dirty flag because Windows did not detach it cleanly (Fast Startup, hibernation, or unplugging without Safely Remove), and ntfs3 refuses a read/write mount of a dirty volume unless force is passed. Files shows only its generic message. The flag is bookkeeping and the files are normally intact. This diagnosis was posted by a helper who reproduced the error text on a test image, and the reporter confirmed the dmesg line and the fix on their drive.
Fix
Confirm the cause first. Retry the mount in Files, then read the kernel log and look for the volume is dirty line:
sudo dmesg | tail -n 20
lsblk -f # find the NTFS partition if it is not /dev/sda1
Check for an /etc/fstab entry before anything else. udisks hands a device that is named in fstab straight to mount as root and ignores the options you pass it, so both the read-only mount below and the later mount from Files come back owned by root:
grep -n sda1 /etc/fstab
If that prints a line, delete it, reload, and unmount what is already there, so udisks manages the drive again:
sudo nvim /etc/fstab # delete the sda1 line
sudo systemctl daemon-reload
sudo umount /mnt/sda1 # repeat until findmnt /mnt/sda1 prints nothing
If anything on the drive is irreplaceable, copy it off first. A read-only mount works while the flag is set, because ntfs3 refuses only a read/write mount:
udisksctl mount -b /dev/sda1 -o ro
Clear the flag. ntfsfix is in the ntfsprogs package (a split package of the ntfs-3g pkgbase, extra repo, version 2026.7.7-1 on 2026-09-11), which Omarchy does not install by default.
Omarchy 4:
omarchy pkg add ntfsprogs
sudo ntfsfix --clear-dirty /dev/sda1
Plain Arch:
sudo pacman -S --needed ntfsprogs
sudo ntfsfix --clear-dirty /dev/sda1
Then mount from Files again, or unplug and replug so udiskie automounts it. Never put sudo in front of udisksctl for a removable drive. ntfs3 has no SID to uid mapping and takes ownership from the uid= and gid= mount options udisks sets from the caller, so a mount made as root lands in /run/media/root/<label>, which your user cannot enter.
If you also use Windows, eject with Safely Remove and turn off Fast Startup (Control Panel, Hardware and Sound, Power Options, "Choose what the power buttons do", untick "Turn on fast startup"), which was offered as the alternative fix but not tested in the thread.
Verify
sudo dmesg | tail -n 5 # no new 'volume is dirty' line after the mount
findmnt -no SOURCE,TARGET,FSTYPE,OPTIONS /dev/sda1
id -u; id -g
A mount made by Files lands under /run/media/<user>/<label> with uid= and gid= matching your own ids. Three other outputs mean the job is not finished. A /run/media/root/<label> target means something ran udisksctl or mount under sudo. A /mnt/... target with uid=0,gid=0 means an /etc/fstab entry is still in play, which is what the reporter hit after clearing the flag. Two lines for the same target mean the device is mounted twice and needs unmounting until findmnt prints nothing. ro in the options is the read-only mount from the earlier step, so unmount and mount again without -o ro. The reporter confirmed the drive mounted and all files were readable after ntfsfix --clear-dirty.