nmdctl: Unable to stop array when using Luks encryption

[sumitc96]

Hello,

I am not able to stop nonraid array using nmdctl stop command.

I am getting this message:

root@pve:~# nmdctl stop
Warning: Some filesystems are mounted. Unmount them before stopping the array.
Do you want to unmount all filesystems? (y/N): y
=== Unmounting NonRAID Array Disks ===

Slot 1 (nmd1p1): Unmounting XFS from /mnt/disk1... SUCCESS
Slot 1 (nmd1p1): Closing LUKS device... FAILED

Unmount summary:
1 disks unmounted successfully
0 disks skipped (already unmounted or no filesystem)
1 disks failed to unmount

Error: Failed to unmount some filesystems
Please unmount them manually before stopping the array
root@pve:~#

I have 1 disk array, the data disk is encrypted with LUKS and here's the encryption details:

root@pve:~# cryptsetup status /dev/mapper/nmd1p1
/dev/mapper/nmd1p1 is active and is in use.
type: LUKS2
cipher: aes-xts-plain64
keysize: 512 bits
key location: keyring
device: /dev/nmd1p1
sector size: 512
offset: 32768 sectors
size: 27344732056 sectors
mode: read/write

Any ideas why nmdctl stop is failing? Any precautions/actions i need to take to safely stop the array?

Thanks!

[qvr]

nmdctl just calls cryptsetup luksClose "nmd1p1" at that stage, so trying that manually might give more information why cryptsetup is not able to close the device.

[sumitc96]

Thanks for your reply, after trying the luksClose command and encountering errors, i did some further research and it seems the LXC takes a little about 15secs to shut down and luksClose being called during this time fails as disk is stll in use.

I need to find a way to add a timedelay or dependency on lxc shutdown state before calling nmdctl stop to ensure the disk is not being utiliized.

Maybe i can add it in the serviced definition? any ideas?

[qvr]

Yeah, you should be able to add a systemd override to either the LXC service to Require= nonraid.service, or to the nonraid service to have it start Before= your LXC service (or both, to make the ordering as explicit as possible).

So something like: systemctl edit your-lxc-service

[Unit]
Requires=nonraid.service
After=nonraid.service

[sumitc96]

Worked exactly as suggested! Edited lxc-service and added the unit definition, now when i reboot the machine nonraid waits for lxc to stop and does a clean stop. As a result no more parity checks getting triggered on a restart! I can use this as my 1 disk array until next year when hopefully disk prices come down and i can add a parity disk. Thanks so much!

…

On Mon, Dec 22, 2025 at 11:41 AM Matti Hiljanen ***@***.***> wrote: *qvr* left a comment (qvr/nonraid#97) <#97 (comment)> Yeah, you should be able to add a systemd override to either the LXC service to Require= nonraid.service, or to the nonraid service to have it start Before= your LXC service (or both, to make the ordering as explicit as possible). So something like: systemctl edit your-lxc-service [Unit] Requires=nonraid.service After=nonraid.service — Reply to this email directly, view it on GitHub <#97 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/APDCNTV25YXVTGPCPUJEAKD4DAULFAVCNFSM6AAAAACPW3RCUSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTMOBTGA4TSMBQHA> . You are receiving this because you authored the thread.Message ID: ***@***.***>