28 lines
1002 B
YAML

---
- name: Check nvidia-smi availability
ansible.builtin.command: nvidia-smi --query-gpu=name,memory.total,driver_version --format=csv,noheader
register: ai_node_nvidia_smi
changed_when: false
failed_when: false
- name: Optionally fail when NVIDIA tooling is required but unavailable
ansible.builtin.fail:
msg: >-
NVIDIA GPU tooling is unavailable. Install a compatible NVIDIA driver and
nvidia-utils package, then re-run onboarding.
when:
- ai_node_require_nvidia_tooling | bool
- ai_node_nvidia_smi.rc != 0
- name: Warn when nvidia-smi is unavailable
ansible.builtin.debug:
msg: >-
nvidia-smi is not available yet. This is common on fresh hosts before driver install.
Continue onboarding now, then install validated drivers separately.
when: ai_node_nvidia_smi.rc != 0
- name: Capture GPU info lines
ansible.builtin.set_fact:
ai_node_gpu_lines: "{{ ai_node_nvidia_smi.stdout_lines | default([]) }}"
when: ai_node_nvidia_smi.rc == 0