diff --git a/roles/mellanox_drivers/tasks/main.yml b/roles/mellanox_drivers/tasks/main.yml index c084b9dad15290738693e02c00253f3f739a1e48..4a023a18123ccaf558d99fc23e373e3d9ddece2e 100644 --- a/roles/mellanox_drivers/tasks/main.yml +++ b/roles/mellanox_drivers/tasks/main.yml @@ -114,7 +114,6 @@ - mlnx-ofa_kernel-modules when: install_now - - name: install drivers shell: ./mlnxofedinstall -q --add-kernel-support --force --skip-repo args: @@ -129,6 +128,18 @@ become_user: root register: service_file +- name: Set failover mode on DGX baremetal nodes + lineinfile: + path: /etc/modprobe.d/ko2iblnd.conf + state: present + line: "options ko2iblnd dev_failover=1" + insertafter: EOF + when: + - '"DGX" in ansible_product_name' + - '"RedHat" in ansible_distribution' + become: true + become_user: root + - name: Reload systemd shell: systemctl daemon-reload become: true