feat(ansible): install amdgpu drivers and packages on sles

.vscode/settings.json

@@ -75,6 +75,7 @@
         "airgap",
         "airgapped",
         "alertmanager",
+        "amdgpu",
         "apiserver",
         "asciicast",
         "asciinema",
@@ -294,6 +295,7 @@
         "ripsecrets",
         "rmmod",
         "roboto",
+        "rocm",
         "roff",
         "rootlv",
         "rootvg",
@@ -312,6 +314,7 @@
         "setfattr",
         "sidekiq",
         "siem",
+        "sles",
         "slurm",
         "snmp",
         "spiffe",
@@ -373,6 +376,7 @@
         "yellowdog",
         "ykman",
         "yubikey",
-        "zstd"
+        "zstd",
+        "zypp"
     ]
 }

ansible/playbooks/amdgpu.install.yml

@@ -0,0 +1,49 @@
+---
+
+- name: Install AMDGPU drivers and applications
+  hosts: all
+  vars:
+    drivers_version: 6.3
+    distribution_version: "{{ ansible_distribution_version }}"
+    architecture: "{{ ansible_architecture }}"
+  tasks:
+    - name: Install AMDGPU drivers and applications on SLES
+      # refer https://rocm.docs.amd.com/projects/install-on-linux/en/latest/install/native-install/sles.html
+      tags: sles
+      when:
+        - ansible_pkg_mgr | lower in ['zypper']
+        - ansible_distribution_file_variety | lower in ['suse']
+      block:
+        - name: Add the AMDGPU kernel-mode driver repository on SLES
+          tags: repository
+          become: true
+          ansible.builtin.copy:
+            dest: /etc/zypp/repos.d/amdgpu.repo
+            content: |
+              [amdgpu]
+              name=amdgpu
+              baseurl=https://repo.radeon.com/amdgpu/{{ drivers_version }}/sle/{{ distribution_version }}/main/{{ architecture }}/
+              enabled=1
+              gpgcheck=1
+              gpgkey=https://repo.radeon.com/rocm/rocm.gpg.key
+        - name: Add the AMDGPU ROCm packages repository on SLES
+          tags: repository
+          become: true
+          ansible.builtin.copy:
+            dest: /etc/zypp/repos.d/rocm.repo
+            content: |
+              [ROCm-{{ drivers_version }}]
+              name=ROCm{{ drivers_version }}
+              baseurl=https://repo.radeon.com/rocm/zyp/{{ drivers_version }}/main
+              enabled=1
+              gpgcheck=1
+              gpgkey=https://repo.radeon.com/rocm/rocm.gpg.key
+        - name: Install AMDGPU packages on SLES
+          tags: package
+          become: true
+          community.general.zypper:
+            name:
+              - amdgpu-dkms
+              - rocm
+            disable_recommends: false
+            update_cache: true

knowledge base/ansible.md

@@ -28,6 +28,7 @@
     1. [Ansible Vault](#ansible-vault)
 1. [Best practices](#best-practices)
 1. [Troubleshooting](#troubleshooting)
+    1. [ERROR: Ansible could not initialize the preferred locale: unsupported locale setting](#error-ansible-could-not-initialize-the-preferred-locale-unsupported-locale-setting)
     1. [Print all known variables](#print-all-known-variables)
     1. [Force notified handlers to run at a specific point](#force-notified-handlers-to-run-at-a-specific-point)
     1. [Time tasks execution](#time-tasks-execution)
@@ -1139,6 +1140,17 @@ Or even edit their content with `ansible-vault edit 'path/to/file'`.
 
 ## Troubleshooting
 
+### ERROR: Ansible could not initialize the preferred locale: unsupported locale setting
+
+`ansible-core` requires the locale to have `UTF-8` encoding [since 2.14.0][ansible v2.14 CHANGELOG]:
+
+> ansible - At startup the filesystem encoding and locale are checked to verify they are UTF-8. If not, the process
+> exits with an error reporting the errant encoding.
+
+```sh
+LANG='C.UTF-8' ansible …
+```
+
 ### Print all known variables
 
 Print the special variable `vars` as a task:
@@ -1494,6 +1506,7 @@ Solution: use a version of `ansible-core` lower than 2.17.
 - [Patterns: targeting hosts and groups]
 - [How to use ansible with S3 - Ansible aws_s3 examples]
 - [How to run Ansible with_fileglob in alphabetical order?]
+- [Ansible v2.14 CHANGELOG]
 
 <!--
   Reference
@@ -1517,6 +1530,7 @@ Solution: use a version of `ansible-core` lower than 2.17.
 [ansible galaxy user guide]: https://docs.ansible.com/ansible/latest/galaxy/user_guide.html
 [ansible navigator]: https://ansible.readthedocs.io/projects/navigator/en/stable/
 [ansible runner]: https://ansible.readthedocs.io/projects/runner/en/stable/
+[ansible v2.14 changelog]: https://github.com/ansible/ansible/blob/7bb078bd740fba8ad43cc69e18fc8aeb4719180a/changelogs/CHANGELOG-v2.14.rst#id11
 [asynchronous actions and polling]: https://docs.ansible.com/ansible/latest/playbook_guide/playbooks_async.html
 [automating helm using ansible]: https://www.ansible.com/blog/automating-helm-using-ansible
 [blocks]: https://docs.ansible.com/ansible/latest/playbook_guide/playbooks_blocks.html