I am sure you can, but for my part I prefer a dedicated Docker stacks application like dockge or portainer.
The UI is so much better than the Truenas one.
As always with Linux - there are 1000 ways that lead to Rome.
I migrated my whole docker host today to Truenas Scale using the way above and it works like a charm.

Currently my TrueNAS server is an underpowered i5 6500 on an old Lenovo board so I have been debating how I want to migrate as I prefer not to use a network to transfer data to containers on another machine, even if it is a dedicated 10Gb line separate from the main network. However, I don't think 4 ancient threads would be near sufficient for my workload. Hell even my test Jellyfin container peaked the 2 cores I gave it for hours while is scraped my collection.

I could also move everything over slowly and then do a quick transplant of hardware when I am done so I can keep my R5 2600. Hmmm choices. 🤔

I'm running everything on ASRock Rack B650D4U-2L2T with a AMD Ryzen 7 PRO 5750GE with Radeon Graphics (35W TDP), 128GB ECC RAM virtualized on Proxmox.
Proxmox has its own 2VDEV Mirrored ZFS Pool with 4x Micron 5300 Max SSDs on the MBs SATA Ports.
16x HBA is passed through to Truenas (for spinning rust).
Docker containers previously ran on a dedicated Arch VM using an NFS mount on Truenas Core.
After the upgrade to Truenas Scale now the Arch VM is gone and all Docker containers are running directly on Truenas Scale.
My Jellyfin is running in a dedicated non privileged Proxmox LXC with iGPU passed through with only with one CPU core due to the passed through iGPU doing the heavy load.

@Lepelot If you are using the TrueNAS Scale Docker Compose form, you only need to create the Nextcloud datasets you want to use for storage and make sure to set them for Apps at creation (or add the appropriate permissions manually) so the apps user can access them. Replace DIR_TO_NC_DATASET with the path to the dataset you just created. Make sure to also pick which interface you want to expose by using the IP_TO_EXPOSE_TO_PROXY.

"Make sure to also pick which interface you want to expose by using the IP_TO_EXPOSE_TO_PROXY."

In this case would this be the TrueNAS Scale IP adress? something like TRUENASIP:8080 ??

I appreciate your response, thank you so much.

Yes, for example my TrueNAS box has a Mellanox ConnectX-4 card and an integrated Intel gigabit NIC. The ConnectX-4 is physically connected to 1. my main 10Gb switch and 2. my Proxmox server. The line going out to the main switch is on my management LAN (xx.xx.10.1/24), the other line is on a direct connection with an IP address on the same subnet as my Proxmox box (e.g. xx.50.50.1/24). I also have 2 VLAN addresses set on my connection out to the switch. For example:

So I would have the options of:

• IP_TO_EXPOSE_TO_PROXY = 192.168.10.5
• IP_TO_EXPOSE_TO_PROXY = 192.168.20.5
• IP_TO_EXPOSE_TO_PROXY = 192.168.30.5
• IP_TO_EXPOSE_TO_PROXY = 192.50.50.5

This is the IP that your reverse proxy would point to and the IP address Nextcloud will listen to for client requests. Since my APPS subnet is where I point my proxy to, I would use 192.168.30.5. If you have no proxy (not recommended) that would be the address you type in to access Nextcloud.

My network tab in the web UI looks something like this. (yes I know its just local IP addresses but I redacted them anyway 🙃)

Ok, thank you, I'll give it a try.

@NikoKS Sorry for the late reply, you can use the AIO container interface of your existing installation to create a backup. On Nextcloud create the dataset for DIR_TO_NC_DATASET, point the docker compose to that and spin it up. Navigate to the AIO interface and restore from backup.

I'm going to be migrating my nextcloud data from a vanilla installation on truenas core to AIO on truenas scale. Can I simply point DIR_TO_NC_DATASET to my "files" dataset, and it work once I recreate the users? Or do I have to spin it all up and recreate the users first like it says in the migration.md file, then copy the data?

I am not 100% sure in your case. What I did was create a new Nextcloud dataset, spin up Nextcloud, create the users I wanted, and then manually copy the files from the old dataset into the new dataset for each user. Since I use Authentik OIDC/SSO the user folders were even named the same and it was incredibly simple. I then restarted the NC stack and NC slowly started to see the 'new' files. You may have to mess around with permissions (not familiar with the permission system on Core/BSD) when copying over to make sure the apps user can access the dataset.

It's a crude method but it worked and it made sure I had something to fall back on if the migration went poorly. Though I am fairly certain that you will lose metadata like link and user shares. This might be helpful for a more comprehensive migration.

Migrating the database is a nightmare if your previous instance did not use postgres, see discussion here. If by some miracle you resolve this, here's the official guidelines on migrating to AIO. TL;dr - if you are migrating only your files, you're likely fine-ish. If you want the database too, godspeed.

I have been struggling with the official TrueNAS NextCloud app for some time, and I must confess that the TrueNAS official installation tutorial is useless. I did, eventually, manage to install it using some clues provided here, though I am still in the dark regarding various configuration parameters. Posting a complete installation and configuration tutorial would be of immense help to all of us struggling with TrueNAS' NextCloud app.

+1

In the mean time I managed to the Nextcloud install to use my HDD array as it's data directory with:

      - NEXTCLOUD_DATADIR=/mnt/HDDs/Nextcloud/nc-data
      - NEXTCLOUD_MOUNT=/mnt/HDDs/Nextcloud

However, I think this means that the instance itself is also running on the HDD array. I am new to this, so probably missing, so any tips on how to get the container instance to run from SSD?

HI, I tried deploy Nextcloud AIO in TrueNAS Scale 25.10 with "Install via YAML", it's fonctional but GPU is not detected.

I went in circles on this for a while too, before eventually figuring out that Nextcloud wasn't gonna use the GPU for basically anything anyway. You can do it, but trust me, it's kind of a dead end.

Hi, with this config, it's work, the container Memory can see my Nvidia card:

`name: nextcloud-aio
services:
nextcloud-aio-mastercontainer:
container_name: nextcloud-aio-mastercontainer
deploy:
resources:
limits:
cpus: '4'
memory: 4096M
reservations:
devices:
- capabilities:
- gpu
device_ids:
- GPU-eb6b4f4e-144d-67d3-7ef5-6b6139cf2272
driver: nvidia
devices:
- /dev/dri:/dev/dri
environment:
APACHE_PORT: 11000
FULLTEXTSEARCH_JAVA_OPTIONS: '-Xms1024M -Xmx1024M'
NEXTCLOUD_DATADIR: /mnt/DATA/UserDATA
NEXTCLOUD_ENABLE_DRI_DEVICE: True
NEXTCLOUD_ENABLE_NVIDIA_GPU: True
NVIDIA_DRIVER_CAPABILITIES: all
NVIDIA_VISIBLE_DEVICES: GPU-5418fb24-4219-12b3-8d20-cbbfd6de0336
group_add:
- 44
- 107
- 568
image: ghcr.io/nextcloud-releases/all-in-one:latest
init: True
network_mode: bridge
networks: {}
ports:
- '8080:8080'
restart: always
runtime: nvidia
volumes:
- nextcloud_aio_mastercontainer:/mnt/docker-aio-config
- /var/run/docker.sock:/var/run/docker.sock:ro
volumes:
nextcloud_aio_mastercontainer:
name: nextcloud_aio_mastercontainer
x-portals:

• host: 0.0.0.0
  name: Web UI
  path: /
  port: 8080
  scheme: https`