Use community_images

.github/workflows/extra.yml

@@ -24,11 +24,11 @@ jobs:
       matrix: # build RL8, RL9
         build:
           - image_name: openhpc-extra-RL8
-            source_image_name_key: RL8 # key into environments/.stackhpc/tofu/cluster_image.auto.tfvars.json
+            appliances_distro_version: RL8
             inventory_groups: doca,cuda # lustre disabled due to https://github.com/stackhpc/ansible-slurm-appliance/pull/759 
             volume_size: 35 # needed for cuda
           - image_name: openhpc-extra-RL9
-            source_image_name_key: RL9
+            appliances_distro_version: RL9
             inventory_groups: doca,cuda,lustre
             volume_size: 35 # needed for cuda
     env:
@@ -37,23 +37,15 @@ jobs:
       CI_CLOUD: ${{ vars.CI_CLOUD }} # default from repo settings
       ARK_PASSWORD: ${{ secrets.ARK_PASSWORD }}
       PACKER_GITHUB_API_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+      APPLIANCES_DISTRO_VERSION: ${{ matrix.build.appliances_distro_version}}
 
     steps:
       - uses: actions/checkout@v4
 
-      - name: Load current fat images into GITHUB_ENV
-        # see https://docs.github.com/en/actions/writing-workflows/choosing-what-your-workflow-does/workflow-commands-for-github-actions#example-of-a-multiline-string
-        run: |
-          {
-            echo 'FAT_IMAGES<<EOF'
-            cat environments/.stackhpc/tofu/cluster_image.auto.tfvars.json
-            echo EOF
-          } >> "$GITHUB_ENV"
-
       - name: Record settings
         run: |
           echo CI_CLOUD: ${{ env.CI_CLOUD }}
-          echo "FAT_IMAGES: ${FAT_IMAGES}"
+          echo APPLIANCES_DISTRO_VERSION: ${{ env.APPLIANCES_DISTRO_VERSION }}
 
       - name: Setup ssh
         run: |
@@ -93,7 +85,6 @@ jobs:
           PACKER_LOG=1 packer build \
           -on-error=${{ vars.PACKER_ON_ERROR }} \
           -var-file="$PKR_VAR_environment_root/${{ env.CI_CLOUD }}.pkrvars.hcl" \
-          -var "source_image_name=${{ fromJSON(env.FAT_IMAGES)['cluster_image'][matrix.build.source_image_name_key] }}" \
           -var "image_name=${{ matrix.build.image_name }}" \
           -var "inventory_groups=${{ matrix.build.inventory_groups }}" \
           -var "volume_size=${{ matrix.build.volume_size }}" \

.github/workflows/main.yml

@@ -99,14 +99,14 @@ jobs:
               - 'dev/setup-env.sh'
               - '.github/workflows/stackhpc.yml'
             extra_on_push:
-              - 'environments/.stackhpc/tofu/cluster_image.auto.tfvars.json'
+              - 'environments/.stackhpc/inventory/group_vars/all/community_image_names.json'
               - 'ansible/roles/doca/**'
               - 'ansible/roles/cuda/**'
               - 'ansible/roles/slurm_recompile/**' # runs on cuda group
               - 'ansible/roles/lustre/**'
               - '.github/workflows/extra.yml'
             extra_on_pull_request:
-              - 'environments/.stackhpc/tofu/cluster_image.auto.tfvars.json'
+              - 'environments/.stackhpc/inventory/group_vars/all/community_image_names.json'
               - 'ansible/roles/doca/**'
               - 'ansible/roles/cuda/**'
               - 'ansible/roles/lustre/**'

ansible/adhoc/image-pull.yml

@@ -0,0 +1,17 @@
+- hosts: localhost
+  become: false
+  gather_facts: false
+  tasks:
+    - name: Ensure images exist on cloud
+      ansible.builtin.include_role:
+        name: azimuth_cloud.azimuth_ops.community_images
+        tasks_from: upload_image_and_store_id.yml
+      loop: "{{ community_images | dict2items | list }}"
+      loop_control:
+        label: "{{ item.value.name }}"
+    - name: Write image IDs
+      ansible.builtin.copy:
+        dest: "{{ appliances_repository_root }}/environments/site/images/community_image_ids.json"
+        content: "{{ community_images_image_ids | to_nice_json }}"
+        mode: ug=rw,o=r
+      delegate_to: localhost

docs/image-build.md

@@ -81,7 +81,7 @@ For either a site-specific fat-image build or an extra-build:
      - For a site-specific fatimage build: A RockyLinux GenericCloud or
        compatible image.
      - For an extra-build image: Usually the appropriate StackHPC fat image,
-       as defined in `environments/.stackhpc/tofu/cluster_image.auto.tfvars.json` at the
+       as defined in `environments/.stackhpc/inventory/group_vars/all/community_image_names.json` at the
        checkout's current commit. See the [GitHub release page](https://github.com/stackhpc/ansible-slurm-appliance/releases)
        for download links. In some cases extra builds may be chained, e.g.
        one extra build adds a Lustre client, and the resulting image is used

environments/.stackhpc/inventory/group_vars/all/community_images.yml

@@ -0,0 +1,5 @@
+# CI sets env from matrix for RL8 or RL8:
+appliances_distro_version: "{{ lookup('env', 'APPLIANCES_DISTRO_VERSION', default='RL9') }}"
+
+# Use prerelease bucket only for stackhpc env:
+community_images_stackhpc_source_url: https://leafcloud.store/swift/v1/AUTH_f39848421b2747148400ad8eeae8d536/openhpc-images-prerelease

environments/.stackhpc/tofu/main.tf

@@ -20,17 +20,6 @@ variable "cluster_name" {
   description = "Name for cluster, used as prefix for resources - set by environment var in CI"
 }
 
-variable "os_version" {
-  type        = string
-  description = "RL8 or RL9"
-  default     = "RL9"
-}
-
-variable "cluster_image" {
-  description = "single image for all cluster nodes, keyed by os_version - a convenience for CI"
-  type        = map(string)
-}
-
 # tflint-ignore: terraform_typed_variables
 variable "cluster_networks" {}
 
@@ -60,19 +49,13 @@ variable "volume_backed_instances" {
   default = false
 }
 
-data "openstack_images_image_v2" "cluster" {
-  name        = var.cluster_image[var.os_version]
-  most_recent = true
-}
-
 module "cluster" {
   source = "../../site/tofu/"
 
   cluster_name        = var.cluster_name
   cluster_networks    = var.cluster_networks
   vnic_types          = var.vnic_types
   key_pair            = "slurm-app-ci"
-  cluster_image_id    = data.openstack_images_image_v2.cluster.id
   control_node_flavor = var.control_node_flavor
 
   login = {

environments/.stackhpc/tofu/cluster_image.auto.tfvars.json → environments/common/inventory/group_vars/all/community_image_names.json

@@ -1,5 +1,5 @@
 {
-  "cluster_image": {
+  "stackhpc_image_names": {
     "RL8": "openhpc-RL8-251213-1133-31273766",
     "RL9": "openhpc-RL9-251213-1133-31273766"
   }

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

@@ -0,0 +1,38 @@
+appliances_distro_version: RL9
+
+# Default StackHPC images - usually won't need overriding:
+# (either used directly for cluster or as source images for Packer builds)
+# Image names are provided in adjacent json file as used to drive StackHPC CI:
+community_images_stackhpc_source_url: https://leafcloud.store/swift/v1/AUTH_f39848421b2747148400ad8eeae8d536/openhpc-images
+community_images_stackhpc:
+  stackhpc:
+    name: "{{ stackhpc_image_names[appliances_distro_version] }}"
+    source_url: "{{ community_images_stackhpc_source_url }}/{{ stackhpc_image_names[appliances_distro_version] }}"
+    source_disk_format: qcow2
+    container_format: bare
+community_images_default: "{{ community_images_stackhpc }}"
+
+# Per-cloud image settings - may be overriden:
+community_images_disk_format: raw # format required on the cloud
+community_images_custom_properties_default:
+  - hw_machine_type=q35
+  - hw_architecture=x86_64
+  - hw_vif_multiqueue_enabled=true
+  - hw_firmware_type=uefi
+  - os_type=linux
+  - "os_admin_user={{ ansible_user }}"
+community_images_custom_properties_disk_format:
+  raw:
+    - hw_scsi_model=virtio-scsi
+    - hw_disk_bus=scsi
+  qcow2:
+    - hw_disk_bus=virtio
+community_images_custom_properties: |
+  {{ community_images_custom_properties_default +
+      community_images_custom_properties_disk_format[community_images_disk_format]
+  }}
+
+# Additional images - usually won't need overriding
+community_images_extra: {}
+
+community_images: "{{ community_images_default | combine(community_images_extra) }}"

environments/site/images/community_image_ids.json

@@ -0,0 +1 @@
+{}

environments/site/tofu/additional.tf

@@ -15,7 +15,7 @@ module "additional" {
   config_drive          = var.config_drive
 
   # can be set for group, defaults to top-level value:
-  image_id                = lookup(each.value, "image_id", var.cluster_image_id)
+  image_id                = local.image_ids[lookup(each.value, "image_key", var.cluster_image_key)]
   vnic_types              = lookup(each.value, "vnic_types", var.vnic_types)
   volume_backed_instances = lookup(each.value, "volume_backed_instances", var.volume_backed_instances)
   root_volume_size        = lookup(each.value, "root_volume_size", var.root_volume_size)

environments/site/tofu/compute.tf

@@ -15,7 +15,7 @@ module "compute" {
   config_drive          = var.config_drive
 
   # can be set for group, defaults to top-level value:
-  image_id                     = lookup(each.value, "image_id", var.cluster_image_id)
+  image_id                     = local.image_ids[lookup(each.value, "image_key", var.cluster_image_key)]
   vnic_types                   = lookup(each.value, "vnic_types", var.vnic_types)
   volume_backed_instances      = lookup(each.value, "volume_backed_instances", var.volume_backed_instances)
   root_volume_size             = lookup(each.value, "root_volume_size", var.root_volume_size)

environments/site/tofu/control.tf

@@ -13,6 +13,7 @@ locals {
       environment_name      = basename(var.environment_root)
     }
   )
+  cluster_image_id = local.image_ids[var.cluster_image_key]
 }
 
 resource "openstack_networking_port_v2" "control" {
@@ -46,13 +47,13 @@ resource "openstack_networking_port_v2" "control" {
 resource "openstack_compute_instance_v2" "control" {
 
   name        = split(".", local.control_fqdn)[0]
-  image_id    = var.cluster_image_id
+  image_id    = local.cluster_image_id
   flavor_name = var.control_node_flavor
   key_pair    = var.key_pair
 
   # root device:
   block_device {
-    uuid                  = var.cluster_image_id
+    uuid                  = local.cluster_image_id
     source_type           = "image"
     destination_type      = var.volume_backed_instances ? "volume" : "local"
     volume_size           = var.volume_backed_instances ? var.root_volume_size : null

environments/site/tofu/images.tf

@@ -0,0 +1,3 @@
+locals {
+  image_ids = jsondecode(file("${path.module}/../images/community_image_ids.json"))
+}

environments/site/tofu/login.tf

@@ -15,7 +15,7 @@ module "login" {
   config_drive          = var.config_drive
 
   # can be set for group, defaults to top-level value:
-  image_id                     = lookup(each.value, "image_id", var.cluster_image_id)
+  image_id                     = local.image_ids[lookup(each.value, "image_key", var.cluster_image_key)]
   vnic_types                   = lookup(each.value, "vnic_types", var.vnic_types)
   volume_backed_instances      = lookup(each.value, "volume_backed_instances", var.volume_backed_instances)
   root_volume_size             = lookup(each.value, "root_volume_size", var.root_volume_size)

environments/site/tofu/variables.tf

@@ -60,7 +60,7 @@ variable "login" {
       nodes: List of node names
       flavor: String flavor name
     Optional:
-      image_id: Overrides variable cluster_image_id
+      image_key: Overrides variable cluster_image_key
       extra_networks: List of mappings in same format as cluster_networks
       vnic_types: Overrides variable vnic_types
       volume_backed_instances: Overrides variable volume_backed_instances
@@ -106,13 +106,13 @@ variable "login" {
     ))
   }
     EOF
-
 }
 }
 
-variable "cluster_image_id" {
+variable "cluster_image_key" {
   type        = string
-  description = "ID of default image for the cluster"
+  description = "Key in community_images of default image for the cluster"
+  default     = "stackhpc"
 }
 
 variable "compute" {
@@ -129,11 +129,11 @@ variable "compute" {
       nodes: List of node names
       flavor: String flavor name
     Optional:
-      image_id: Overrides variable cluster_image_id
+      image_key: Overrides variable cluster_image_key
       extra_networks: List of mappings in same format as cluster_networks
       vnic_types: Overrides variable vnic_types
       compute_init_enable: Toggles compute-init rebuild (see compute-init role docs)
-      ignore_image_changes: Ignore changes to the image_id parameter (see docs/experimental/compute-init.md)
+      ignore_image_changes: Ignore changes to the image ID (see docs/experimental/compute-init.md)
       volume_backed_instances: Overrides variable volume_backed_instances
       root_volume_size: Overrides variable root_volume_size
       extra_volumes: Mapping defining additional volumes to create and attach

packer/openstack.pkr.hcl

@@ -24,6 +24,7 @@ locals {
     git_commit = data.git-commit.cwd-head.hash
     timestamp = formatdate("YYMMDD-hhmm", timestamp())
     image_name_version = var.image_name_version == "auto" ? "-${local.timestamp}-${substr(local.git_commit, 0, 8)}" : var.image_name_version
+    community_image_ids = jsondecode(file("${path.root}/environments/site/images/community_image_ids.json"))
 }
 
 # Path pointing to root of repository - automatically set by environment variable PKR_VAR_repo_root
@@ -44,13 +45,13 @@ variable "networks" {
 variable "source_image_name" {
   type = string
   default = null
-  description = "name of source image"
+  description = "Name of source image"
 }
 
-variable "source_image" {
+variable "source_image_key" {
   type = string
-  default = null
-  description = "UUID of source image"
+  default = "stackhpc"
+  description = "Name of community_images key defining source image"
 }
 
 variable "flavor" {
@@ -177,9 +178,9 @@ source "openstack" "openhpc" {
   floating_ip = var.floating_ip
   security_groups = var.security_groups
 
-  # Input image:
-  source_image = "${var.source_image}"
-  source_image_name = "${var.source_image_name}" # NB: must already exist in OpenStack
+  # Input image - NB must already exist in OpenStack:
+  source_image = local.community_image_ids[var.source_image_key]
+  source_image_name = "${var.source_image_name}"
 
   # SSH:
   ssh_username = var.ssh_username

requirements.yml

@@ -30,11 +30,11 @@ collections:
   - name: community.grafana
     version: 1.5.4
   - name: ansible.posix
-    version: 1.5.4
+    version: 2.1.0
   - name: ansible.netcommon
     version: 5.1.1
   - name: community.general
-    version: 7.1.0
+    version: 11.2.1
   - name: community.crypto
     version: 2.10.0
   - name: community.mysql
@@ -55,3 +55,7 @@ collections:
     version: 0.5.5
   - name: stackhpc.linux
     version: 1.5.0
+  # for community_images role:
+  - name: https://github.com/azimuth-cloud/ansible-collection-azimuth-ops
+    type: git
+    version: 0.20.0