Make eessi configure gpu node automatically (#841)

* Adding tasks to essi/configure.yml to make eessi configure
gpu node automatically

* Adding tasks to essi/configure.yml to make eessi configure
gpu node automatically

* applying edits to task names

* replacing EESSI block with running the configure.yml task directly

* Bump CI imageX

* removed v from trivyscan.yml

* fix eessi compute-init tasks to include role defaults

* removed eessi tasks from compute-init/tasks/export.yml

* bump CI image

Author: claudia-lola

.github/workflows/trivyscan.yml

@@ -102,7 +102,7 @@ jobs:
         run: sudo guestmount -a /mnt/images/${{ steps.manifest.outputs.image-name }}.qcow2 -i --ro -o allow_other './${{ steps.manifest.outputs.image-name }}'
 
       - name: Run Trivy vulnerability scanner
-        uses: aquasecurity/trivy-action@v0.33.1
+        uses: aquasecurity/trivy-action@0.33.1
         with:
           scan-type: fs
           scan-ref: "${{ steps.manifest.outputs.image-name }}"
@@ -122,7 +122,7 @@ jobs:
           category: "${{ matrix.build }}"
 
       - name: Fail if scan has CRITICAL vulnerabilities
-        uses: aquasecurity/trivy-action@v0.33.1
+        uses: aquasecurity/trivy-action@0.33.1
         with:
           scan-type: fs
           scan-ref: "${{ steps.manifest.outputs.image-name }}"

ansible/roles/compute_init/files/compute-init.yml

@@ -277,22 +277,13 @@
         name: basic_users
       when: enable_basic_users
 
-    - name: EESSI
-      when: enable_eessi
       # NB: don't need conditional block on enable_compute as have already exited
       # if not the case
-      block:
-        - name: Copy cvmfs config
-          ansible.builtin.copy:
-            src: /var/tmp/cluster/cvmfs/default.local
-            dest: /etc/cvmfs/default.local
-            owner: root
-            group: root
-            mode: "0644"
-
-        - name: Ensure CVMFS config is setup # noqa: no-changed-when
-          ansible.builtin.command:
-            cmd: "cvmfs_config setup"
+    - name: Configure EESSI
+      ansible.builtin.include_role:
+        name: eessi
+        tasks_from: configure.yml
+      when: enable_eessi
 
     - name: Configure VGPUs
       ansible.builtin.include_role:

ansible/roles/compute_init/tasks/export.yml

@@ -62,17 +62,6 @@
   run_once: true
   delegate_to: "{{ groups['control'] | first }}"
 
-- name: Copy EESSI CVMFS config to /exports/cluster
-  ansible.builtin.copy:
-    src: /etc/cvmfs/default.local
-    dest: /exports/cluster/cvmfs/default.local
-    owner: slurm
-    group: root
-    mode: "0644"
-    remote_src: true
-  run_once: true
-  delegate_to: "{{ groups['control'] | first }}"
-
 - name: Export cacerts
   ansible.builtin.include_role:
     name: cacerts

ansible/roles/compute_init/tasks/install.yml

@@ -54,6 +54,8 @@
       dest: roles/
     - src: ../../nhc
       dest: roles/
+    - src: ../../eessi
+      dest: roles/
 
 - name: Add filter_plugins to ansible.cfg
   ansible.builtin.lineinfile:

ansible/roles/eessi/tasks/configure.yml

@@ -15,3 +15,20 @@
 - name: Ensure CVMFS config is setup # noqa: no-changed-when
   ansible.builtin.command:
     cmd: "cvmfs_config setup"
+
+# configure gpus
+- name: Check for NVIDIA GPU
+  ansible.builtin.stat:
+    path: /dev/nvidia0
+  register: nvidia_driver
+
+- name: Set fact if NVIDIA GPU is present
+  ansible.builtin.set_fact:
+    has_nvidia_driver: "{{ nvidia_driver.stat.exists | default(false) }}"
+
+- name: Expose GPU drivers
+  ansible.builtin.shell: |
+    source /cvmfs/software.eessi.io/versions/2023.06/init/bash
+    /cvmfs/software.eessi.io/versions/2023.06/scripts/gpu_support/nvidia/link_nvidia_host_libraries.sh
+  when: has_nvidia_driver
+  changed_when: true

environments/.stackhpc/tofu/cluster_image.auto.tfvars.json

@@ -1,6 +1,6 @@
 {
   "cluster_image": {
-    "RL8": "openhpc-RL8-251027-1123-d389c00b",
-    "RL9": "openhpc-RL9-251027-1123-d389c00b"
+    "RL8": "openhpc-RL8-251119-1202-332ac921",
+    "RL9": "openhpc-RL9-251119-1202-332ac921"
   }
 }