Skip to content

Environment Preparation ​

Select the appropriate environment installation script based on your deployment architecture.

System Requirements ​

Before deployment, verify your server's operating system version using the following command:

bash
lsb_release -a

Example output:

bash
No LSB modules are available.
Distributor ID: Ubuntu
Description:    Ubuntu 22.04.5 LTS
Release:    22.04
Codename:   jammy

Note: The cluster supports Ubuntu 22.04 and Ubuntu 24.04. If your version differs from the recommended specifications, forced installation may cause program crashes or service interruptions. For optimal experience, we strongly recommend using the officially supported versions.

Disable Kernel Auto-Update ​

If kernel auto-update is not disabled, the system may automatically upgrade the kernel version after reboot, causing a mismatch between the current kernel and the kernel version required by the graphics driver during installation, rendering the graphics driver unusable. We recommend disabling kernel auto-update before installing graphics drivers.

10periodic Configuration ​

Open the configuration file /etc/apt/apt.conf.d/10periodic:

bash
vim /etc/apt/apt.conf.d/10periodic

Add or modify the following content:

bash
APT::Periodic::Update-Package-Lists "0";
APT::Periodic::Download-Upgradeable-Packages "0";
APT::Periodic::AutocleanInterval "0";

20auto-upgrades Configuration ​

Open the configuration file /etc/apt/apt.conf.d/20auto-upgrades:

bash
vim /etc/apt/apt.conf.d/20auto-upgrades

Add or modify the following content:

bash
APT::Periodic::Update-Package-Lists "0";
APT::Periodic::Unattended-Upgrade "0";

Disable Nouveau Open-Source Driver ​

To prevent nouveau (Linux open-source NVIDIA driver) from interfering with the installation and use of official drivers, it needs to be disabled.

Edit Configuration ​

Open the configuration file /etc/modprobe.d/blacklist-nouveau.conf:

bash
vim /etc/modprobe.d/blacklist-nouveau.conf

Add the following content:

bash
blacklist nouveau
options nouveau modeset=0

Update and Reboot ​

Update initramfs to apply the configuration changes:

bash
update-initramfs -u

After completion, reboot the server for changes to take effect:

bash
reboot

Check Drivers ​

Before proceeding with installation, first check whether the server has the corresponding GPU driver installed. If already installed, skip to Environment Installation

bash
nvidia-smi

Example output:

If GPU information is displayed, the driver is already installed, and you can skip the following installation steps.

bash
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 570.153.02             Driver Version: 570.153.02     CUDA Version: 12.8     |
|-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA GeForce RTX 3090        Off |   00000000:01:00.0 Off |                  N/A |
| 36%   34C    P0            111W /  350W |       0MiB /  24576MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+
                                                                                          
+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI              PID   Type   Process name                        GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
|  No running processes found                                                             |
+-----------------------------------------------------------------------------------------+

Tip: If the query result matches the examples above (driver information displayed successfully), skip the Download Driver and Install Driver steps below and proceed directly to Download Installation Script.

Download Drivers ​

Visit the NVIDIA Driver Download Page and select and download the Runfile installer based on your GPU model and operating system version.

Enter the GPU model (e.g., GeForce RTX 3090, Linux 64-bit).

GPU Driver - Search Click Search.

GPU Driver - Search

Select Recommended Driver/Certified Driver type and click View.

GPU Driver - Download

Click Download and upload the corresponding driver to your server.

Install Drivers ​

If already installed, skip to Environment Installation

Replace the example filename NVIDIA-Linux-x86_64-570.144.run with your actual downloaded filename.

Grant execute permission to the installation file:

bash
chmod +x NVIDIA-Linux-x86_64-570.144.run

Run the installer, skipping X Server check and not installing OpenGL files:

bash
./NVIDIA-Linux-x86_64-570.144.run -no-x-check -no-opengl-files

GPU Driver - Installation GPU Driver - Installation If the installation prompts missing tools:

Execute the following commands to install compilation tools:

bash
apt update
apt install make gcc -y

After tools installation is complete, run the installer again:

bash
./NVIDIA-Linux-x86_64-570.144.run -no-x-check -no-opengl-files

Check the driver after installation:

bash
nvidia-smi

Environment Installation ​

Replace <CustomerID> in the URL with your actual customer ID. For ID retrieval details, refer to Console - Profile.

Professional Architecture ​

bash
wget -O manager.sh https://product-deploy.titanai.cloud/admin/deploy/env/professional/manager/install/<CustomerID> && bash manager.sh

Converged Architecture ​

NVIDIA

bash
wget -O computation.sh https://product-deploy.titanai.cloud/admin/deploy/env/install/<客户ID> && bash computation.sh

CPU

bash
wget -O computation.sh https://product-deploy.titanai.cloud/admin/deploy/env/cpu/install/<客户ID> && bash computation.sh

:::

When you see the following logs, installation has completed successfully:

[2025-04-07 01:46:41] Installation completed successfully
[2025-04-07 01:46:41]   WARNING: System will reboot in 1 minute. Use 'shutdown -c' to cancel.
[2025-04-07 01:46:41] Executing: shutdown -r +1 'Reboot By InfraX'
Installation log has been saved to: /var/log/infra_install_20250407_014609.log

Please wait for the server to reboot before proceeding with subsequent operations.