Merge branch 'main' into feat/ansible-ssh-template

Author: sjpb

ansible/roles/cacerts/tasks/export.yml

@@ -3,9 +3,9 @@
   ansible.builtin.copy:
     src: "{{ item }}"
     dest: /exports/cluster/cacerts/
-    owner: slurm
-    group: root
-    mode: "0644"
+    owner: ansible-init
+    group: ansible-init
+    mode: u=rw,go=
   with_fileglob:
     - "{{ cacerts_cert_dir }}/*"
   delegate_to: "{{ groups['control'] | first }}"

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

@@ -83,7 +83,7 @@
         - ansible.builtin.meta: end_play
     - name: Check if hostvars exist
       become: true
-      become_user: slurm
+      become_user: ansible-init # share is root-squashed
       ansible.builtin.stat:
         path: "/mnt/cluster/hostvars/{{ ansible_hostname }}/hostvars.yml"
       register: hostvars_stat
@@ -98,7 +98,7 @@
         - ansible.builtin.meta: end_play
     - name: Sync /mnt/cluster to /var/tmp
       become: true
-      become_user: slurm
+      become_user: ansible-init # share is root-squashed
       ansible.posix.synchronize:
         src: "/mnt/cluster/"
         dest: "/var/tmp/cluster/"

ansible/roles/compute_init/tasks/export.yml

@@ -1,41 +1,47 @@
 ---
-- name: Ensure the /exports/cluster directory exists
+- name: Ensure /exports/cluster directory structure exists
   ansible.builtin.file:
-    path: /exports/cluster
+    path: "{{ item }}"
     state: directory
-    owner: slurm
-    group: root
+    owner: ansible-init
+    group: ansible-init
     mode: u=rX,g=rwX,o=
   run_once: true
+  loop:
+    - /exports/cluster
+    - /exports/cluster/hostvars
+    - /exports/cluster/cacerts
+    - /exports/cluster/cvmfs
+    - /exports/cluster/hostconfig
   delegate_to: "{{ groups['control'] | first }}"
 
 - name: Copy /etc/hosts to /exports/cluster
   ansible.builtin.copy:
     src: /etc/hosts
     dest: /exports/cluster/hosts
-    owner: slurm
-    group: root
-    mode: u=r,g=rw,o=
+    owner: ansible-init
+    group: ansible-init
+    mode: u=rw,go=r
     remote_src: true
   run_once: true
   delegate_to: "{{ groups['control'] | first }}"
 
-- name: Create hostvars directory
+- name: Create per-host hostvars directory
   ansible.builtin.file:
     path: /exports/cluster/hostvars/{{ inventory_hostname }}/
     state: directory
-    owner: slurm
-    group: root
-    mode: u=rX,g=rwX,o=
+    owner: ansible-init
+    group: ansible-init
+    mode: u=rwX,go=
   delegate_to: "{{ groups['control'] | first }}"
 
 - name: Template out hostvars
   ansible.builtin.template:
     src: hostvars.yml.j2
     dest: /exports/cluster/hostvars/{{ inventory_hostname }}/hostvars.yml
-    owner: slurm
-    group: root
-    mode: u=r,g=rw,o=
+    owner: ansible-init
+    group: ansible-init
+    mode: u=rw,go=
   delegate_to: "{{ groups['control'] | first }}"
 
 - name: Copy manila share info to /exports/cluster
@@ -52,29 +58,19 @@
     os_manila_mount_share_info_var:
       os_manila_mount_share_info: "{{ os_manila_mount_share_info }}"
 
-- name: Ensure /exports/cluster/cvmfs directory exists
-  ansible.builtin.file:
-    path: /exports/cluster/cvmfs
-    state: directory
-    owner: slurm
-    group: root
-    mode: "0755"
-  run_once: true
-  delegate_to: "{{ groups['control'] | first }}"
-
 - name: Export cacerts
   ansible.builtin.include_role:
     name: cacerts
     tasks_from: export.yml
   when: "'cacerts' in group_names"
 
-- name: Create hostconfig directory
+- name: Create per-host hostconfig directory
   ansible.builtin.file:
     path: "/exports/cluster/hostconfig/{{ inventory_hostname }}/"
     state: directory
-    owner: slurm
-    group: root
-    mode: u=rX,g=rwX,o=
+    owner: ansible-init
+    group: ansible-init
+    mode: u=rwX,go=
   delegate_to: "{{ groups['control'] | first }}"
 
 - name: Template sssd config

ansible/roles/cuda/defaults/main.yml

@@ -4,7 +4,7 @@ cuda_repo_url: "https://developer.download.nvidia.com/compute/cuda/repos/rhel{{
 cuda_nvidia_driver_stream: '580-open'
 cuda_nvidia_driver_version: '580.105.08-1'
 cuda_nvidia_driver_pkg: "nvidia-open-3:{{ cuda_nvidia_driver_version }}.el{{ ansible_distribution_major_version }}"
-cuda_package_version: '13.0.2-1'
+cuda_package_version: '13.1.0-1'
 cuda_version_short: "{{ (cuda_package_version | split('.'))[0:2] | join('.') }}" # major.minor
 cuda_packages_default:
   - "cuda-toolkit-{{ cuda_package_version }}"

ansible/roles/hpctests/tasks/pingpong.yml

@@ -64,9 +64,10 @@
   become: false
   when: hpctests_pingpong_plot | bool
 
-- ansible.builtin.debug:
-    # yamllint disable rule:line-length
-    msg: |
+- name: Set fact for summary
+  ansible.builtin.set_fact:
+  # yamllint disable rule:line-length
+    hpctests_pingpong_summary: |
       Summary for pingpong using 2x scheduler-selected nodes in '{{ hpctests_partition }}' partition, job ID {{ _pingpong_jobid }}, device '{{ hpctests_ucx_net_devices }}':
 
       Nodes: {{ hpctests_pingpong_run_nodes.stdout.split()[1] }}
@@ -78,3 +79,15 @@
       {{ _pingpong_plot.stdout }}
       {% endif %}
     # yamllint enable rule:line-length
+
+- ansible.builtin.debug:
+    msg: |
+      {{ hpctests_pingpong_summary }}
+
+- name: Write summary
+  ansible.builtin.copy:
+    dest: "{{ _pingpong_local_output | dirname }}/summary.txt"
+    content: "{{ hpctests_pingpong_summary }}"
+    mode: "0644"
+  delegate_to: localhost
+  become: false

ansible/roles/nhc/tasks/export.yml

@@ -3,5 +3,7 @@
   ansible.builtin.template:
     src: "{{ nhc_config_template }}"
     dest: "/exports/cluster/hostconfig/{{ inventory_hostname }}/nhc.conf"
-    mode: "0644"
+    owner: ansible-init
+    group: ansible-init
+    mode: u=rw,go=
   delegate_to: "{{ groups['control'] | first }}"

ansible/roles/sssd/tasks/export.yml

@@ -4,7 +4,7 @@
   ansible.builtin.template:
     src: "{{ sssd_conf_src }}"
     dest: "/exports/cluster/hostconfig/{{ inventory_hostname }}/sssd.conf"
-    owner: root
-    group: root
+    owner: ansible-init
+    group: ansible-init
     mode: u=rw,go=
   delegate_to: "{{ groups['control'] | first }}"

dev/image-set-properties.sh

@@ -17,7 +17,6 @@ openstack image set \
 --property hw_architecture=x86_64 \
 --property hw_vif_multiqueue_enabled=true \
 --property hw_firmware_type=uefi \
---property os_distro=rocky \
 --property os_type=linux \
 --property os_admin_user=rocky \
 "$image"

environments/.caas/README.md

@@ -12,7 +12,9 @@ Non-standard things for this environment:
         azimuth_caas_stackhpc_slurm_appliance_template:
         ...
         envVars:
-            ANSIBLE_INVENTORY: environments/common/inventory,environments/.caas/inventory
+            ANSIBLE_INVENTORY: environments/common/inventory,environments/site/inventory,environments/.caas/inventory
+
+  ([Source](https://github.com/azimuth-cloud/ansible-collection-azimuth-ops/blob/main/roles/azimuth_caas_operator/defaults/main.yml#L199))
 
   Ansible then defines `ansible_inventory_sources` which contains absolute paths, and
   that is used to derive the `appliances_environment_root` and

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

@@ -1,6 +1,6 @@
 {
   "cluster_image": {
-    "RL8": "openhpc-RL8-251211-0951-51b93e3f",
-    "RL9": "openhpc-RL9-251211-0951-51b93e3f"
+    "RL8": "openhpc-RL8-251213-1133-31273766",
+    "RL9": "openhpc-RL9-251213-1133-31273766"
   }
 }