Skip to main content

OS with Open WebUI application

Last update:

On a dedicated server, you can install an OS with a ready-made Open WebUI application.

The OS image with Open WebUI is based on SelectOS - see SelectOS documentation for more information about the OS. The parameters of available models (LLM, Large Language Model) depend on the hardware characteristics of the dedicated server. The model will not run if its size exceeds the size of the video memory on the GPU. See the Machine Learning Models subsection of the Select Components for a Dedicated Server of Any Configuration tutorial for more information on selecting a configuration for LLM.

The OS image contains:

  • Open WebUI - A web interface that allows you to manage and interact with the LLM through Ollama;

  • Ollama - a platform for running LLM locally on a server. Runs in a Docker container ollama which does not have direct access to the internet, but communicates only with the Docker container openweb-ui through the Docker network llm-net;

  • Docker - a containerization platform for developing and running containerized applications;

  • settings for network components:

    • nginx - web server and reverse proxy. Open WebUI does not support HTTPS directly. Nginx is used as a proxy server and provides access to Open WebUI over HTTPS, for this purpose a self-signed TLS(SSL)-certificate is generated on the system by IP address.
      For long-term work with the server, we recommend replacing the self-signed certificate with a trusted certificate for your domain and configuring nginx to work with it;

    • UFW (Uncomplicated Firewall) is a simplified command line interface for managing the Netfilter firewall on Linux-based operating systems. UFW allows incoming connections only on ports 22, 80 and 443, other traffic is blocked.

Install OS with Open WebUI

For Open WebUI to work, the dedicated server must be accessible from the Internet.

You can install the OS with Open WebUI immediately with the LLM or install the desired LLM after the OS is installed.

  1. In the Control Panel, on the top menu, click Products and select Dedicated Servers.

  2. In the Servers section, open the Server page → Operating System tab.

  3. Click Reinstall OS.

  4. Optional: To securely connect to the server using the encrypted SSH protocol, select a previously downloaded and hosted SSH key or add a new one. For more information, see Create and upload an SSH key on a dedicated server.

  5. Enter user data. With user data you can automate the configuration of servers. You can only specify user data for servers without hardware HBAs and RAID controllers.

    To install the LLM, you need to insert a script in the user data field:

    #cloud-config
    runcmd:
      - ollama_pull hf.co/<llm_name>

    Specify <llm_name> - manufacturer and model name, for example Qwen/Qwen3-30B-A3B-GGUF. Learn more about available models on the Hugging Face website.

  6. In the Distribution field, select Open WebUI.

  7. Optional: partition the disks. By default, the optimal RAID type is determined and only partitioned disks and disks in RAID are partitioned.

  8. Click Install. The installation may take from 7 to 60 minutes, depending on server performance and the distribution selected. If it takes longer, create a ticket.

  9. After the OS is installed, a password will be generated to connect to the server. The password can be copied in the control panel: in the top menu, click Products → Dedicated Servers → Server page → Operating System tab → in the Password field, click . The password is available for 24 hours from the start of the OS installation or configuration change. If you forget the server password, you can reset and restore it.

Connect to the Open WebUI panel

  1. Open the page in your browser:

    https://<ip_address>

    Specify <ip_address> - public IP address of the server. It can be copied in the control panel: in the top menu click Products → Dedicated Servers → Server page → Operating System tab → in the IP field click .

  2. Click Get started.

  3. Create an administrator account:

    3.1 In the Name field, enter the name of the administrator, for example Admin.

    3.2 Enter the Email address in the Email field.

    3.3 In the Password field, enter the account password.

    3.4 Click Create Admin Account.

Install LLM

  1. Connect to the server via SSH or via KVM console.

  2. Install the model:

ollama_pull hf.co/<llm_name>

Specify <llm_name> - manufacturer and model name, for example Qwen/Qwen3-30B-A3B-GGUF. Learn more about available models on the Hugging Face website.