nvidia-base vsnvidia frfom kubernetes

playbook.yml

@@ -6,7 +6,7 @@
     - vaultwarden
     - kubernetes
     - kube-master
-    - nvidia
+    - nvidia-base
 
 - name: Configure Kubernetes Nodes
   hosts: workers
@@ -16,12 +16,13 @@
     - vaultwarden
     - kubernetes
     - kube-node
-    - nvidia
+    - nvidia-base
     
 - name: Configure Kubernetes Data
   hosts: master
   become: yes
   roles:
+    - nvidia
     - stolon
     - cloudflared
     - qbittorrent

roles/nvidia-base/tasks/main.yml

@@ -0,0 +1,66 @@
+- name: Instalar driver NVIDIA
+  become: yes
+  ansible.builtin.apt:
+    name: nvidia-driver-550
+    state: present
+
+- name: Instalar utilitários NVIDIA
+  become: yes
+  ansible.builtin.apt:
+    name: nvidia-utils-550
+    state: present
+
+
+- name: Reboot host e esperar voltar
+  become: yes
+  ansible.builtin.reboot:
+    reboot_timeout: 600   # tempo máximo para o host voltar (segundos)
+    test_command: whoami  # comando usado para validar que o host voltou
+
+- name: Adicionar chave GPG da NVIDIA
+  become: yes
+  become_user: fenix 
+  ansible.builtin.shell: |
+    curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit.gpg
+  args:
+    creates: /usr/share/keyrings/nvidia-container-toolkit.gpg
+
+- name: Adicionar repositório NVIDIA
+  become: yes
+  ansible.builtin.shell: |
+    curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit-keyring.gpg \
+      && curl -s -L https://nvidia.github.io/libnvidia-container/stable/deb/nvidia-container-toolkit.list | \
+        sed 's#deb https://#deb [signed-by=/usr/share/keyrings/nvidia-container-toolkit-keyring.gpg] https://#g' | \
+        sudo tee /etc/apt/sources.list.d/nvidia-container-toolkit.list
+  args:
+    creates: /etc/apt/sources.list.d/nvidia-container-toolkit.list
+
+- name: Atualizar cache de pacotes
+  become: yes
+  ansible.builtin.apt:
+    update_cache: yes
+
+- name: Instalar NVIDIA Container Toolkit
+  become: yes
+  ansible.builtin.apt:
+    name: nvidia-container-toolkit
+    state: present
+
+
+- name: Configurar runtime containerd
+  become: yes
+  ansible.builtin.shell: |
+    nvidia-ctk runtime configure --runtime=containerd
+
+- name: Reiniciar containerd
+  become: yes
+  ansible.builtin.systemd:
+    name: containerd
+    state: restarted
+
+
+- name: Mostrar resultado nvidia-smi
+  become: yes
+  ansible.builtin.debug:
+    var: nvidia_smi_output.stdout
+

roles/nvidia/tasks/main.yml

@@ -1,94 +1,7 @@
-- name: Instalar as nvidia drivers
+- name: Aplicar o stolon
   become: yes
   become_user: fenix 
   shell: |
-    sudo apt install nvidia-driver-550
+    kubectl apply -f https://raw.githubusercontent.com/NVIDIA/k8s-device-plugin/v0.17.1/deployments/static/nvidia-device-plugin.yml
-    sudo apt install nvidia-utils-550
   environment:
     KUBECONFIG:  /home/fenix/.kube/config
-
-- name: Reboot remoto e continuar
-  become: yes
-  become_user: fenix 
-  tasks:
-    - name: Reboot host e esperar voltar
-      ansible.builtin.reboot:
-        reboot_timeout: 600   # tempo máximo para o host voltar (segundos)
-        test_command: whoami  # comando usado para validar que o host voltou
-
-
-- name: Adicionar chave GPG da NVIDIA
-  become: yes
-  become_user: fenix 
-  ansible.builtin.shell: |
-    curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit.gpg
-  args:
-    creates: /usr/share/keyrings/nvidia-container-toolkit.gpg
-
-
-- name: Adicionar repositório NVIDIA
-  become: yes
-  become_user: fenix 
-  ansible.builtin.shell: |
-    curl -s -L https://nvidia.github.io/libnvidia-container/stable/ubuntu22.04/amd64/nvidia-container-toolkit.list | sed 's#deb https://#deb [signed-by=/usr/share/keyrings/nvidia-container-toolkit.gpg] https://#' | tee /etc/apt/sources.list.d/nvidia-container-toolkit.list
-  args:
-    creates: /etc/apt/sources.list.d/nvidia-container-toolkit.list
-
-
-- name: Atualizar cache de pacotes
-  become: yes
-  become_user: fenix 
-  ansible.builtin.apt:
-    update_cache: yes
-
-- name: Instalar NVIDIA Container Toolkit
-  become: yes
-  become_user: fenix 
-  ansible.builtin.apt:
-    name: nvidia-container-toolkit
-    state: present
-
-
-- name: Configurar runtime Docker
-  become: yes
-  become_user: fenix 
-  ansible.builtin.shell: |
-    nvidia-ctk runtime configure --runtime=docker
-  when: "'docker.io' in ansible_facts.packages"
-
-- name: Reiniciar Docker
-  become: yes
-  become_user: fenix 
-  ansible.builtin.systemd:
-    name: docker
-    state: restarted
-  when: "'docker.io' in ansible_facts.packages"
-
-- name: Configurar runtime containerd
-  become: yes
-  become_user: fenix 
-  ansible.builtin.shell: |
-    nvidia-ctk runtime configure --runtime=containerd
-  when: "'containerd' in ansible_facts.packages"
-
-- name: Reiniciar containerd
-  become: yes
-  become_user: fenix 
-  ansible.builtin.systemd:
-    name: containerd
-    state: restarted
-  when: "'containerd' in ansible_facts.packages"
-
-- name: Validar GPU dentro de container
-  become: yes
-  become_user: fenix 
-  ansible.builtin.shell: |
-    docker run --rm --runtime=nvidia --gpus all nvidia/cuda:12.2-base nvidia-smi
-  register: nvidia_smi_output
-  ignore_errors: true
-
-- name: Mostrar resultado nvidia-smi
-  become: yes
-  become_user: fenix 
-  ansible.builtin.debug:
-    var: nvidia_smi_output.stdout

roles/nvidia/vars/main.yml

@@ -1,4 +0,0 @@
-bw_password: "{{ lookup('env', 'BW_PASSWORD') }}"
-VAULTWARDEN_LINK: "{{ lookup('env', 'VAULTWARDEN_LINK') }}"
-BW_CLIENTID: "{{ lookup('env', 'BW_CLIENTID') }}"
-BW_CLIENTSECRET : "{{ lookup('env', 'BW_CLIENTSECRET') }}"