DNM: Feat/no ohpc mergeable rl9

ansible/fatimage.yml

@@ -58,9 +58,9 @@
         name: mysql
         tasks_from: install.yml
     - name: OpenHPC
-      import_role:
+      include_role:
         name: stackhpc.openhpc
-        tasks_from: install.yml
+        tasks_from: "install-{{ openhpc_install_type }}.yml"
 
     - name: Include distribution variables for osc.ood
       include_vars: "{{ appliances_repository_root }}/ansible/roles/osc.ood/vars/Rocky/8.yml"

ansible/roles/filebeat/tasks/install.yml

@@ -15,3 +15,4 @@
 - name: Reload filebeat unit file
   command: systemctl daemon-reload
   when: _filebeat_unit.changed
+  become: true

ansible/roles/hpctests/README.md

@@ -38,6 +38,8 @@ The following variables should not generally be changed:
 - `hpctests_pingpong_plot`: Whether to plot pingpong results. Default `yes`.
 - `hpctests_hpl_modules`: As above but for hpl tests.
 - `hpctests_hpl_version`: Version of HPL
+- `hpctests_extra_paths`: List of additional paths to add to $PATH in `pingpong` and `pingmatrix` sbatch scripts.
+- `hpctests_pingpong_command`: Command to use to run IMB-MPI1 pingpong.
 
 Dependencies
 ------------

ansible/roles/hpctests/defaults/main.yml

@@ -1,7 +1,9 @@
 ---
 hpctests_rootdir:
+hpctests_extra_paths: []
 hpctests_pingmatrix_modules: [gnu12 openmpi4]
 hpctests_pingpong_modules: [gnu12 openmpi4 imb]
+hpctests_pingpong_command: 'mpirun IMB-MPI1 pingpong' # NB 'srun --mpi=pmi2 IMB-MPI1 pingpong'  doesn't work in ohpc v2.1
 hpctests_pingpong_plot: yes
 hpctests_hpl_modules: [gnu12 openmpi4 openblas]
 hpctests_outdir: "{{ lookup('env', 'APPLIANCES_ENVIRONMENT_ROOT') }}/hpctests"

ansible/roles/hpctests/templates/pingmatrix.sh.j2

@@ -12,7 +12,8 @@ export UCX_NET_DEVICES={{ hpctests_ucx_net_devices }}
 echo SLURM_JOB_NODELIST: $SLURM_JOB_NODELIST
 echo SLURM_JOB_ID: $SLURM_JOB_ID
 echo UCX_NET_DEVICES: $UCX_NET_DEVICES
-module load {{ hpctests_pingmatrix_modules | join(' ' ) }}
+{% if hpctests_pingmatrix_modules %}module load {{ hpctests_pingmatrix_modules | join(' ' ) }}{% endif %}
+{% if hpctests_extra_paths %}export PATH={{ hpctests_extra_paths | join(':') }}:$PATH{% endif %}
 
 mpicc -o nxnlatbw mpi_nxnlatbw.c
 mpirun nxnlatbw

ansible/roles/hpctests/templates/pingpong.sh.j2

@@ -12,7 +12,7 @@ export UCX_NET_DEVICES={{ hpctests_ucx_net_devices }}
 echo SLURM_JOB_NODELIST: $SLURM_JOB_NODELIST
 echo SLURM_JOB_ID: $SLURM_JOB_ID
 echo UCX_NET_DEVICES: $UCX_NET_DEVICES
-module load {{ hpctests_pingpong_modules | join(' ' ) }}
+{% if hpctests_pingpong_modules %}module load {{ hpctests_pingpong_modules | join(' ' ) }}{% endif %}
+{% if hpctests_extra_paths %}export PATH={{ hpctests_extra_paths | join(':') }}:$PATH{% endif %}
 
-#srun --mpi=pmi2 IMB-MPI1 pingpong # doesn't work in ohpc v2.1
-mpirun IMB-MPI1 pingpong
+{{ hpctests_pingpong_command }}

ansible/roles/mysql/tasks/install.yml

@@ -1,6 +1,12 @@
+- name: Install pip
+  dnf:
+    name: python3-pip
+
 - name: Install python mysql client
   pip:
-    name: pymysql
+    name:
+      - pymysql
+      - cryptography
     state: present
 
 - name: Create systemd mysql container unit file
@@ -11,6 +17,6 @@
 
 - name: Pull container image
   containers.podman.podman_image:
-    name: "mysql"
+    name: docker.io/library/mysql
     tag: "{{ mysql_tag }}"
   become_user: "{{ mysql_podman_user }}"

ansible/roles/mysql/templates/mysql.service.j2

@@ -26,7 +26,7 @@ ExecStart=/usr/bin/podman run \
     --volume {{ mysql_datadir }}:/var/lib/mysql:U \
     --publish 3306:3306 \
     --env MYSQL_ROOT_PASSWORD=${MYSQL_INITIAL_ROOT_PASSWORD} \
-    mysql:{{ mysql_tag }}{%- for opt in mysql_mysqld_options %} \
+    docker.io/library/mysql:{{ mysql_tag }}{%- for opt in mysql_mysqld_options %} \
     --{{ opt }}{% endfor %}
 
 ExecStop=/usr/bin/podman stop --ignore mysql -t 10

ansible/roles/openondemand/tasks/main.yml

@@ -10,7 +10,7 @@
 - include_role:
     name: osc.ood
     tasks_from: install-package.yml
-    vars_from: Rocky/8.yml
+    vars_from: "Rocky/{{ ansible_distribution_major_version }}.yml"
     public: yes # Expose the vars from this role to the rest of the play
   # can't set vars: from a dict hence the workaround above
 

ansible/roles/openondemand/tasks/vnc_compute.yml

@@ -17,6 +17,7 @@
       - turbovnc-3.0.1
       - nmap-ncat
       - python3.9
+      - dbus-x11
 
 - name: Install Xfce desktop
   tags: install

ansible/roles/opensearch/tasks/install.yml

@@ -16,7 +16,7 @@
 
 - name: Pull container image
   containers.podman.podman_image:
-    name: "opensearchproject/opensearch"
+    name: docker.io/opensearchproject/opensearch
     tag: "{{ opensearch_version }}"
   become_user: "{{ opensearch_podman_user }}"
 

ansible/roles/opensearch/templates/opensearch.service.j2

@@ -29,7 +29,7 @@ ExecStart=/usr/bin/podman run \
     --env bootstrap.memory_lock=true \
     --env "OPENSEARCH_JAVA_OPTS=-Xms512m -Xmx512m" \
     --env DISABLE_INSTALL_DEMO_CONFIG=true \
-    opensearchproject/opensearch:{{ opensearch_version }}
+    docker.io/opensearchproject/opensearch:{{ opensearch_version }}
 ExecStop=/usr/bin/podman stop --ignore opensearch -t 10
 # note for some reason this returns status=143 which makes systemd show the unit as failed, not stopped
 ExecStopPost=/usr/bin/podman rm --ignore -f opensearch

ansible/slurm.yml

@@ -25,8 +25,14 @@
   tags:
     - openhpc
   tasks:
-    - import_role:
+    - include_role:
+        name: stackhpc.openhpc
+        tasks_from: "install-{{ openhpc_install_type }}.yml"
+      tags: install
+    - include_role:
         name: stackhpc.openhpc
+        tasks_from: runtime.yml
+      tags: runtime
 
 - name: Set locked memory limits on user-facing nodes
   hosts:

environments/.stackhpc/ARCUS.pkrvars.hcl

@@ -4,7 +4,7 @@ volume_size = 12 # GB. Compatible with SMS-lab's general.v1.tiny
 image_disk_format = "qcow2"
 networks = ["4b6b2722-ee5b-40ec-8e52-a6610e14cc51"] # portal-internal (DNS broken on ilab-60)
 source_image_name = "openhpc-230804-1754-80b8d714" # https://github.com/stackhpc/ansible-slurm-appliance/pull/298
-fatimage_source_image_name = "Rocky-8-GenericCloud-Base-8.9-20231119.0.x86_64.qcow2"
+fatimage_source_image_name = "Rocky-9-GenericCloud-Base-9.3-20231113.0.x86_64.qcow2"
 ssh_keypair_name = "slurm-app-ci"
 ssh_private_key_file = "~/.ssh/id_rsa"
 security_groups = ["default", "SSH"]

environments/.stackhpc/SMS.pkrvars.hcl

@@ -1,7 +1,7 @@
 flavor = "general.v1.tiny"
 networks = ["26023e3d-bc8e-459c-8def-dbd47ab01756"] # stackhpc-ipv4-geneve
 source_image_name = "openhpc-230503-0944-bf8c3f63" # https://github.com/stackhpc/ansible-slurm-appliance/pull/252
-fatimage_source_image_name = "Rocky-8-GenericCloud-Base-8.9-20231119.0.x86_64.qcow2"
+fatimage_source_image_name = "Rocky-9-GenericCloud-Base-9.3-20231113.0.x86_64.qcow2"
 ssh_keypair_name = "slurm-app-ci"
 ssh_private_key_file = "~/.ssh/id_rsa"
 security_groups = ["default", "SSH"]

environments/.stackhpc/hooks/post-bootstrap.yml

@@ -3,14 +3,15 @@
   gather_facts: false
   tags: podman
   tasks:
-    - name: Configure container image registry for unqualified searches to avoid docker.io ratelimits
+    - name: Configure container image registry to avoid docker.io ratelimits
       copy:
-        dest: /etc/containers/registries.conf.d/003-arcus-unqualfied-overrides.conf
+        dest: /etc/containers/registries.conf.d/003-arcus-mirror.conf
         content: |
-          unqualified-search-registries = ['{{ podman_registry_address  | split('/') | first }}', 'registry.access.redhat.com', 'registry.redhat.io', 'docker.io']
-
           [[registry]]
-          prefix = "{{ podman_registry_address }}"
+          location="docker.io/library/"
+          prefix="docker.io/library/"
+
+          [[registry.mirror]]
           location = "{{ podman_registry_address }}"
           insecure = true
       when: "ci_cloud == 'ARCUS'"

environments/.stackhpc/terraform/main.tf

@@ -13,8 +13,8 @@ variable "cluster_name" {
 variable "cluster_image" {
     description = "single image for all cluster nodes - a convenience for CI"
     type = string
-    default = "openhpc-240116-1156-aa8dba7d" # https://github.com/stackhpc/ansible-slurm-appliance/pull/351
-    # default = "Rocky-8-GenericCloud-Base-8.9-20231119.0.x86_64.qcow2"
+    # default = "openhpc-240116-1156-aa8dba7d" # https://github.com/stackhpc/ansible-slurm-appliance/pull/351
+    default = "Rocky-9-GenericCloud-Base-9.3-20231113.0.x86_64.qcow2" # TODO: create packer build
 }
 
 variable "cluster_net" {}
@@ -62,8 +62,6 @@ module "cluster" {
     compute_nodes = {
         compute-0: "small"
         compute-1: "small"
-        compute-2: "extra"
-        compute-3: "extra"
     }
     volume_backed_instances = var.volume_backed_instances
 

environments/common/inventory/group_vars/all/defaults.yml

@@ -58,6 +58,7 @@ appliances_local_users_default:
         uid: 981
         home: "{{ prometheus_db_dir }}"
         shell: /usr/sbin/nologin
+        system: true
       enable: "{{ 'prometheus' in group_names }}"
 
     - group:
@@ -69,6 +70,7 @@ appliances_local_users_default:
         uid: 984
         home: /usr/share/grafana
         shell: /sbin/nologin
+        system: true
       enable: "{{ 'grafana' in group_names }}"
 
 # Overide this to add extra users whilst keeping the defaults.

environments/common/inventory/group_vars/all/openhpc.yml

@@ -2,7 +2,7 @@
 
 # See: https://github.com/stackhpc/ansible-role-openhpc
 # for variable definitions
-
+openhpc_install_type: ohpc # use "ohcp" for an OpenHPC-based system or "generic" if providing binaries
 openhpc_enable:
   control: "{{ inventory_hostname in groups['control'] }}"
   batch: "{{ inventory_hostname in groups['compute'] }}"

environments/common/inventory/group_vars/all/openondemand.yml

@@ -49,6 +49,8 @@ openondemand_clusters:
             module purge
 
             export PATH=/opt/TurboVNC/bin:$PATH
+            # avoid "Failed to create secure directory (/run/user/*/pulse)"
+            export XDG_RUNTIME_DIR="$TMPDIR/xdg_runtime"
 
             # Workaround to avoid "Unable to contact settings server" when
             # lauching xfce4-session

requirements.yml

@@ -3,7 +3,7 @@ roles:
   - src: stackhpc.nfs
     version: v23.12.1 # Tolerate state nfs file handles
   - src: https://github.com/stackhpc/ansible-role-openhpc.git
-    version: v0.23.0 # https://github.com/stackhpc/ansible-role-openhpc/pull/165
+    version: feat/no-ohpc # https://github.com/stackhpc/ansible-role-openhpc/pull/162
     name: stackhpc.openhpc
   - src: https://github.com/stackhpc/ansible-node-exporter.git
     version: stackhpc