Changes to make the appliance work with the CaaS operator (#38)

* Update dependencies

* Bump Terraform collection

* Bump version of terraform collection

* Make partitions available to openstack group

* Allow the use of flavor IDs in partition definition

* Add the community.mysql dependency back in

* Fix errors with warn arg

* Update submodule to use branch with fixes

* Update node-exporter dependency after merge

* Terraform changes merged

* Slurm appliance changes merged

Author: Matt Pryor

group_vars/openhpc.yml

@@ -2,8 +2,4 @@ openhpc_cluster_name: "{{ cluster_name }}"
 
 # Provision a single "standard" compute partition using the supplied
 # node count and flavor
-openhpc_slurm_partitions:
-  - name: "standard"
-    count: "{{ compute_count }}"
-    flavor_name: "{{ compute_flavor }}"
-    default: "YES"
+openhpc_slurm_partitions: "{{ hostvars[groups['openstack'][0]]['openhpc_slurm_partitions'] }}"

group_vars/openstack.yml

@@ -19,3 +19,11 @@ cluster_ssh_user: rocky
 
 # Set the size of the state volume to metrics_db_maximum_size + 10
 state_volume_size: "{{ metrics_db_maximum_size + 10 }}"
+
+# Provision a single "standard" compute partition using the supplied
+# node count and flavor
+openhpc_slurm_partitions:
+  - name: "standard"
+    count: "{{ compute_count }}"
+    flavor: "{{ compute_flavor }}"
+    default: "YES"

requirements.yml

@@ -6,7 +6,7 @@ roles:
     version: v0.20.0 # Allow multiple empty partitions by @sjpb in #156
     name: stackhpc.openhpc
   - src: https://github.com/stackhpc/ansible-node-exporter.git
-    version: feature/no-install
+    version: c9ec08364b91d4097168528699829e6c9c0d38b6 # stackhpc-fixes
     name: cloudalchemy.node_exporter
   - src: https://github.com/cloudalchemy/ansible-prometheus.git
     version: 4d2c8d742de39e50387e0aa6d5510b21c7451343 # need fix in preceeding commit for rocky
@@ -15,27 +15,31 @@ roles:
     version: 0.19.1
   - src: https://github.com/stackhpc/ansible-grafana.git
     name: cloudalchemy.grafana
-    version: service-state
-    # No versions available
+    version: 27b57570400deda29dbe32ddcb07558f4e957e4e # service-state
   - src: https://github.com/OSC/ood-ansible.git
     name: osc.ood
     version: v2.0.8
 
 collections:
   - name: containers.podman
+    version: 1.10.2
   - name: community.grafana
+    version: 1.5.4
   - name: https://github.com/stackhpc/ansible_collection_slurm_openstack_tools
     type: git
     version: v0.2.0
   - name: ansible.posix
+    version: 1.5.4
   - name: ansible.netcommon
+    version: 5.1.1
   - name: community.general
-    version: 4.5.0  # https://github.com/ansible-collections/community.general/pull/4281
+    version: 7.1.0
   - name: community.crypto
     version: 2.10.0
+  - name: community.mysql
+    version: 3.7.2
   - name: openstack.cloud
-    version: 1.10.0
+    version: 2.1.0
   - name: https://github.com/stackhpc/ansible-collection-terraform
     type: git
-    version: dadf2e81b78cfd267821c568f92d3a8da6541a41
-
+    version: 1a8f5af0239de2bfedb37f51e20d973e05699b8a # main @ 20230627

roles/cluster_infra/templates/resources.tf.j2

@@ -379,7 +379,11 @@ resource "openstack_compute_instance_v2" "{{ partition.name }}" {
 
   name      = "{{ cluster_name }}-compute-{{ partition.name }}-${count.index}"
   image_id  = "{{ cluster_previous_image | default(cluster_image) }}"
+  {% if 'flavor_name' in partition %}
   flavor_name = "{{ partition.flavor_name }}"
+  {% else %}
+  flavor_id = "{{ partition.flavor }}"
+  {% endif %}
 
   network {
     port = openstack_networking_port_v2.{{ partition.name }}[count.index].id