From cac269f160e249859adf42d8337d0e8de378dcdd Mon Sep 17 00:00:00 2001 From: Alexandre Pires Date: Mon, 9 Jun 2025 13:34:30 +0200 Subject: [PATCH] Minor fixes --- ansible/host_vars/gpu-01.lab.alexpires.me.yml | 3 +++ ansible/playbook_nvidia_cdi_generator.yml | 7 +++++++ ansible/playbook_podman_comfyui.yml | 7 ------- ansible/playbook_podman_kokoro.yml | 4 ---- ansible/playbook_podman_ollama.yml | 4 ---- ansible/tasks/podman.yml | 2 +- inventory/hosts | 5 ++++- 7 files changed, 15 insertions(+), 17 deletions(-) create mode 100644 ansible/playbook_nvidia_cdi_generator.yml diff --git a/ansible/host_vars/gpu-01.lab.alexpires.me.yml b/ansible/host_vars/gpu-01.lab.alexpires.me.yml index 5fc467a..4b6f978 100644 --- a/ansible/host_vars/gpu-01.lab.alexpires.me.yml +++ b/ansible/host_vars/gpu-01.lab.alexpires.me.yml @@ -63,6 +63,9 @@ custom_hosts_acls: true # This fixes polkit issues with sudoers file (CIS hardening) hide_pid: 0 +# CIS by default changes compilers permissions, Nvidia KMod needs to be build +compilers: + # aide install_aide: false diff --git a/ansible/playbook_nvidia_cdi_generator.yml b/ansible/playbook_nvidia_cdi_generator.yml new file mode 100644 index 0000000..4280022 --- /dev/null +++ b/ansible/playbook_nvidia_cdi_generator.yml @@ -0,0 +1,7 @@ +- name: Setup NVidia CDI Generator + hosts: nvidia + + tasks: + - name: Setup nvidia-cdi-generator + ansible.builtin.include_tasks: + file: tasks/nvidia-cdi-generator.yml diff --git a/ansible/playbook_podman_comfyui.yml b/ansible/playbook_podman_comfyui.yml index 7001728..c667b1f 100644 --- a/ansible/playbook_podman_comfyui.yml +++ b/ansible/playbook_podman_comfyui.yml @@ -44,15 +44,8 @@ - "{{ podman_user_home }}/output:/home/worker/output:Z" device: - "nvidia.com/gpu=all" - systemd: - enabled: false - state: "stopped" tasks: - - name: Setup nvidia-cdi-generator - ansible.builtin.include_tasks: - file: tasks/nvidia-cdi-generator.yml - - name: Setup Pods ansible.builtin.include_tasks: file: tasks/podman.yml diff --git a/ansible/playbook_podman_kokoro.yml b/ansible/playbook_podman_kokoro.yml index d630c7a..75418bc 100644 --- a/ansible/playbook_podman_kokoro.yml +++ b/ansible/playbook_podman_kokoro.yml @@ -18,10 +18,6 @@ - "nvidia.com/gpu=all" tasks: - - name: Setup nvidia-cdi-generator - ansible.builtin.include_tasks: - file: tasks/nvidia-cdi-generator.yml - - name: Setup Pods ansible.builtin.include_tasks: file: tasks/podman.yml diff --git a/ansible/playbook_podman_ollama.yml b/ansible/playbook_podman_ollama.yml index d2cf7a5..6f9591f 100644 --- a/ansible/playbook_podman_ollama.yml +++ b/ansible/playbook_podman_ollama.yml @@ -19,10 +19,6 @@ - "nvidia.com/gpu=all" tasks: - - name: Setup nvidia-cdi-generator - ansible.builtin.include_tasks: - file: tasks/nvidia-cdi-generator.yml - - name: Setup Pods ansible.builtin.include_tasks: file: tasks/podman.yml diff --git a/ansible/tasks/podman.yml b/ansible/tasks/podman.yml index 063a221..6145f9b 100644 --- a/ansible/tasks/podman.yml +++ b/ansible/tasks/podman.yml @@ -106,7 +106,7 @@ containers.podman.podman_container: name: "{{ item.name }}" image: "{{ item.repo.image | default(podman_user + '/' + item.name) }}" - state: "{{ item.state | default('present') }}" + state: "{{ item.state | default('started') }}" device: "{{ item.device | default(omit) }}" env: "{{ item.env | default(omit) }}" network: "{{ podman_network_name | default(omit) }}" diff --git a/inventory/hosts b/inventory/hosts index 13e00a7..5e117de 100644 --- a/inventory/hosts +++ b/inventory/hosts @@ -85,4 +85,7 @@ dev-01.lab.alexpires.me dev-01.lab.alexpires.me [certbot] -vh-01.lab.alexpires.me \ No newline at end of file +vh-01.lab.alexpires.me + +[nvidia] +gpu-01.lab.alexpires.me