Make the resource consumtion more obvious

Author: Matt Pryor

group_vars/openstack.yml

@@ -17,9 +17,6 @@ terraform_project_path: "{{ playbook_dir }}/terraform"
 terraform_state: "{{ cluster_state | default('present') }}"
 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:

group_vars/prometheus.yml

@@ -8,4 +8,5 @@ openondemand_address: "{{ hostvars[groups['openondemand'].0].api_address if 'ope
 prometheus_scrape_configs: "{{ prometheus_scrape_configs_default + (openondemand_scrape_configs if ( 'openondemand' in groups ) else [] ) }}"
 
 # Set Prometheus storage retention size
-prometheus_storage_retention_size: "{{ metrics_db_maximum_size }}GB"
+# We reserve 10GB of the state volume for cluster state, the rest is for metrics
+prometheus_storage_retention_size: "{{ state_volume_size - 10 }}GB"

ui-meta/slurm-infra-fast-volume-type.yml

@@ -12,6 +12,24 @@ parameters:
     kind: cloud.ip
     immutable: true
 
+  - name: login_flavor
+    label: Login node size
+    description: The size to use for the login node.
+    kind: cloud.size
+    immutable: true
+    options:
+      min_ram: 2048
+      min_disk: 20
+
+  - name: control_flavor
+    label: Control node size
+    description: The size to use for the control node.
+    kind: cloud.size
+    immutable: true
+    options:
+      min_ram: 2048
+      min_disk: 20
+
   - name: compute_count
     label: Compute node count
     description: The number of compute nodes in the cluster.
@@ -23,16 +41,16 @@ parameters:
   - name: compute_flavor
     label: Compute node size
     description: The size to use for the compute node.
-    kind: "cloud.size"
+    kind: cloud.size
     immutable: true
     options:
       min_ram: 2048
       min_disk: 20
 
   - name: home_volume_size
     label: Home volume size (GB)
-    description: The size of the cloud volume to use for home directories
-    kind: integer
+    description: The size of the cloud volume to use for home directories.
+    kind: cloud.volume_size
     immutable: true
     options:
       min: 10
@@ -51,19 +69,20 @@ parameters:
     options:
       checkboxLabel: Put home directories on high-performance storage?
 
-  - name: metrics_db_maximum_size
-    label: Metrics database size (GB)
+  - name: state_volume_size
+    label: State volume size (GB)
     description: |
+      The size of the state volume, used to hold and persist important files and data. Of
+      this volume, 10GB is set aside for cluster state and the remaining space is used
+      to store cluster metrics.
+
       The oldest metrics records in the [Prometheus](https://prometheus.io/) database will be 
-      discarded to ensure that the database does not grow larger than this size.
-      
-      **A cloud volume of this size +10GB will be created to hold and persist the metrics 
-      database and important Slurm files.**
-    kind: integer
+      discarded to ensure that the database does not grow larger than this volume.
+    kind: cloud.volume_size
     immutable: true
     options:
-      min: 10
-    default: 10
+      min: 20
+    default: 20
 
   - name: cluster_run_validation
     label: Post-configuration validation

ui-meta/slurm-infra.yml

@@ -12,6 +12,24 @@ parameters:
     kind: cloud.ip
     immutable: true
 
+  - name: login_flavor
+    label: Login node size
+    description: The size to use for the login node.
+    kind: cloud.size
+    immutable: true
+    options:
+      min_ram: 2048
+      min_disk: 20
+
+  - name: control_flavor
+    label: Control node size
+    description: The size to use for the control node.
+    kind: cloud.size
+    immutable: true
+    options:
+      min_ram: 2048
+      min_disk: 20
+
   - name: compute_count
     label: Compute node count
     description: The number of compute nodes in the cluster.
@@ -23,34 +41,35 @@ parameters:
   - name: compute_flavor
     label: Compute node size
     description: The size to use for the compute node.
-    kind: "cloud.size"
+    kind: cloud.size
     immutable: true
     options:
       min_ram: 2048
       min_disk: 20
 
   - name: home_volume_size
     label: Home volume size (GB)
-    description: The size of the cloud volume to use for home directories
-    kind: integer
+    description: The size of the cloud volume to use for home directories.
+    kind: cloud.volume_size
     immutable: true
     options:
       min: 10
     default: 100
 
-  - name: metrics_db_maximum_size
-    label: Metrics database size (GB)
+  - name: state_volume_size
+    label: State volume size (GB)
     description: |
+      The size of the state volume, used to hold and persist important files and data. Of
+      this volume, 10GB is set aside for cluster state and the remaining space is used
+      to store cluster metrics.
+
       The oldest metrics records in the [Prometheus](https://prometheus.io/) database will be 
-      discarded to ensure that the database does not grow larger than this size.
-      
-      **A cloud volume of this size +10GB will be created to hold and persist the metrics 
-      database and important Slurm files.**
-    kind: integer
+      discarded to ensure that the database does not grow larger than this volume.
+    kind: cloud.volume_size
     immutable: true
     options:
-      min: 10
-    default: 10
+      min: 20
+    default: 20
 
   - name: cluster_run_validation
     label: Post-configuration validation