Experimental installation¶

It is recommended to visit the hardware and software prerequisites to learn more about deployment options and hardware requirements.

For the installation you must have two type of nodes:

• Control Node: The control node is where you install Ansible to manage other machines. It runs the playbooks and coordinates operations.

• Target Nodes: These are the machines Ansible manages. They receive and execute tasks sent by the control node.

You must have ssh capability from the Control node to the Target nodes.

Prerequisites on the target node(s)¶

The system can be installed on any OS that supports docker, however the recommended OS is Ubuntu 22.04. The steps described below assume an Ubuntu 22.04 installation.

1. System update¶

Make sure your system is up-to-date.

sudo apt update && \
sudo apt upgrade -y && \
sudo reboot

2. Add Docker GPG key and repository¶

curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo gpg --dearmor -o /usr/share/keyrings/docker-archive-keyring.gpg && \
echo "deb [arch=amd64 signed-by=/usr/share/keyrings/docker-archive-keyring.gpg] https://download.docker.com/linux/ubuntu $(lsb_release -cs) stable" | sudo tee /etc/apt/sources.list.d/docker.list > /dev/null

3. Add NVIDIA Docker GPG key and repository¶

curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit-keyring.gpg

4. Add the NVIDIA container toolkit repository¶

echo 'deb [signed-by=/usr/share/keyrings/nvidia-container-toolkit-keyring.gpg] https://nvidia.github.io/libnvidia-container/stable/deb/$(ARCH) /' \
| sudo tee /etc/apt/sources.list.d/nvidia-container-toolkit.list

5. Install required dependencies¶

sudo apt update && \
sudo apt -y install ca-certificates curl software-properties-common docker-ce docker-ce-cli containerd.io python3-pip python3-venv nvidia-driver-550 nvidia-docker2 && \
sudo reboot

Installation steps on the control node¶

1. Install Ansible¶

The installer requires Ansible on the Control node

sudo apt -y install sshpass && \
pip3 install ansible && \
source ~/.profile

2. Download and extract Ansible installer¶

 wget https://u-query.ultinous.com/docs/r9/ansible_playbook.tar -P /tmp && \
 mkdir -p ~/ansible-installer && \
 tar -xvf /tmp/ansible_playbook.tar -C ~/ansible-installer && \
 rm /tmp/ansible_playbook.tar && \
 cd ~/ansible-installer

3. Set up the Ansible inventory¶

You have to set up the Ansible inventory file under inventories/all.yaml:

• You have to replace YOUR_TARGET_NODE_NAME_HERE as you would name your target node. It'll be a reference during the installation.

• You have to add value for:

  • ansible_host: It has to be the target node's IP address or domain name.

  • ansible_port: It has to be the target node's SSH port number.

all:
  hosts:
    YOUR_TARGET_NODE_NAME_HERE:
      ansible_host:  # TARGET_NODE_IP_HERE
      ansible_port:  # TARGET_NODE_PORT_HERE

4. Set up the Target node specific variables¶

Create a directory for node/host specific variables in the document root of the installer and copy the example variables:

    mkdir -p host_vars/YOUR_TARGET_NODE_NAME_HERE && \
    cp examples/vars.yaml host_vars/YOUR_TARGET_NODE_NAME_HERE/

5. Start the installation¶

Start the Ansible playbook:

ansible-playbook start.yaml

It'll prompt you for ssh password and sudo password for the target node's authentication:

   SSH password:
   BECOME password[defaults to SSH password]:

The installation can take up to 30 minutes.