Merge pull request #341 from stackhpc/feat/base-RL89

Update fatimage base to RL8.9 with robust volume mounts

Author: sjpb

ansible/roles/cluster_infra/tasks/main.yml

@@ -57,29 +57,6 @@
     - terraform_state == "present"
     - cluster_upgrade_system_packages is not defined or not cluster_upgrade_system_packages
 
-- name: Detect volume device prefix from image metadata
-  block:
-    - name: Get image metadata from OpenStack API
-      openstack.cloud.image_info:
-        image: "{{ cluster_previous_image | default(cluster_image) }}"
-      register: cluster_image_info
-    - name: Check only single image found
-      assert:
-        that: cluster_image_info.images | length == 1
-        fail_msg: "Multiple images found for 'cluster_image' {{ cluster_image }}"
-    - name: Set volume_device_prefix fact
-      set_fact:
-        block_device_prefix: >-
-           {{
-              'sd' if (cluster_image_info.images | first).hw_scsi_model is defined and
-              (cluster_image_info.images | first).hw_scsi_model in scsi_models
-              else 'vd'
-           }}
-  # Only run when block_device_prefix isn't set as an extravar
-  when:
-    - block_device_prefix is not defined
-    - cluster_image is defined
-
 - name: Template Terraform files into project directory
   template:
     src: >-

ansible/roles/cluster_infra/templates/resources.tf.j2

@@ -358,18 +358,13 @@ resource "openstack_compute_instance_v2" "control" {
     {%- for ssh_key in cluster_deploy_ssh_keys_extra %}
       - {{ ssh_key }}
     {%- endfor %}
-    fs_setup:
-      - label: state
-        filesystem: ext4
-        device: /dev/{{ block_device_prefix }}b
-        partition: auto
-      - label: home
-        filesystem: ext4
-        device: /dev/{{ block_device_prefix }}c
-        partition: auto
+    bootcmd:
+    %{for volume in [openstack_blockstorage_volume_v3.state, openstack_blockstorage_volume_v3.home]}
+    - BLKDEV=$(readlink -f $(ls /dev/disk/by-id/*${substr(volume.id, 0, 20)}* | head -n1 )); blkid -o value -s TYPE $BLKDEV ||  mke2fs -t ext4 -L ${lower(split(" ", volume.description)[0])} $BLKDEV
+    %{endfor}
     mounts:
-        - [LABEL=state, /var/lib/state, auto, "x-systemd.required-by=nfs-server.service,x-systemd.before=nfs-server.service"]
-        - [LABEL=home, /exports/home, auto, "x-systemd.required-by=nfs-server.service,x-systemd.before=nfs-server.service"]
+        - [LABEL=state, {{ appliances_state_dir }}, auto]
+        - [LABEL=home, /exports/home, auto]
   EOF
 }
 

environments/.stackhpc/ARCUS.pkrvars.hcl

@@ -4,7 +4,7 @@ volume_size = 10 # GB
 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-8.8-20230518.0.x86_64.qcow2"
+fatimage_source_image_name = "Rocky-8-GenericCloud-Base-8.9-20231119.0.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-8.8-20230518.0.x86_64.qcow2"
+fatimage_source_image_name = "Rocky-8-GenericCloud-Base-8.9-20231119.0.x86_64.qcow2"
 ssh_keypair_name = "slurm-app-ci"
 ssh_private_key_file = "~/.ssh/id_rsa"
 security_groups = ["default", "SSH"]

environments/.stackhpc/terraform/ARCUS.tfvars

@@ -3,5 +3,3 @@ cluster_subnet = "portal-internal"
 vnic_type = "normal"
 control_node_flavor = "vm.ska.cpu.general.quarter"
 other_node_flavor = "vm.ska.cpu.general.small"
-state_volume_device_path = "/dev/sdb"
-home_volume_device_path = "/dev/sdc"

environments/.stackhpc/terraform/SMS.tfvars

@@ -3,5 +3,3 @@ cluster_subnet = "stackhpc-ipv4-geneve-subnet"
 vnic_type = "normal"
 control_node_flavor = "general.v1.medium"
 other_node_flavor = "general.v1.tiny"
-state_volume_device_path = "/dev/vdb"
-home_volume_device_path = "/dev/vdc"

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-231206-1648-9d6aa4e4" # https://github.com/stackhpc/ansible-slurm-appliance/pull/340
-    # default = "Rocky-8-GenericCloud-Base-8.8-20230518.0.x86_64.qcow2"
+    default = "openhpc-231208-1207-b69af6e2" # https://github.com/stackhpc/ansible-slurm-appliance/pull/341
+    # default = "Rocky-8-GenericCloud-Base-8.9-20231119.0.x86_64.qcow2"
 }
 
 variable "cluster_net" {}
@@ -31,10 +31,6 @@ variable "volume_backed_instances" {
     default = false
 }
 
-variable "state_volume_device_path" {}
-
-variable "home_volume_device_path" {}
-
 module "cluster" {
     source = "../../skeleton/{{cookiecutter.environment}}/terraform/"
 
@@ -76,6 +72,4 @@ module "cluster" {
     state_volume_size = 10
     home_volume_size = 20
 
-    state_volume_device_path = var.state_volume_device_path
-    home_volume_device_path = var.home_volume_device_path
 }

environments/skeleton/{{cookiecutter.environment}}/terraform/nodes.tf

@@ -126,19 +126,14 @@ resource "openstack_compute_instance_v2" "control" {
     #cloud-config
     fqdn: ${var.cluster_name}-${each.key}.${var.cluster_name}.${var.cluster_domain_suffix}
     
-    fs_setup:
-      - label: state
-        filesystem: ext4
-        device: ${var.state_volume_device_path}
-        partition: auto
-      - label: home
-        filesystem: ext4
-        device: ${var.home_volume_device_path}
-        partition: auto
+    bootcmd:
+      %{for volume in [openstack_blockstorage_volume_v3.state, openstack_blockstorage_volume_v3.home]}
+      - BLKDEV=$(readlink -f $(ls /dev/disk/by-id/*${substr(volume.id, 0, 20)}* | head -n1 )); blkid -o value -s TYPE $BLKDEV ||  mke2fs -t ext4 -L ${lower(split(" ", volume.description)[0])} $BLKDEV
+      %{endfor}
 
     mounts:
       - [LABEL=state, ${var.state_dir}]
-      - [LABEL=home, /exports/home, auto, "x-systemd.required-by=nfs-server.service,x-systemd.before=nfs-server.service"]
+      - [LABEL=home, /exports/home]
   EOF
 
 }

environments/skeleton/{{cookiecutter.environment}}/terraform/variables.tf

@@ -55,18 +55,6 @@ variable "environment_root" {
     description = "Path to environment root, automatically set by activate script"
 }
 
-variable "state_volume_device_path" {
-    type = string
-    description = "Path to block device for state"
-    default = "/dev/sdb"
-}
-
-variable "home_volume_device_path" {
-    type = string
-    description = "Path to block device name for home directories"
-    default = "/dev/sdc"
-}
-
 variable "state_dir" {
     type = string
     description = "Path to state directory on control node"

environments/skeleton/{{cookiecutter.environment}}/terraform/volumes.tf

@@ -1,11 +1,11 @@
 resource "openstack_blockstorage_volume_v3" "state" {
     name = "${var.cluster_name}-state"
-    description = "State for control node"
+    description = "State for control node" # first word used to label filesystem
     size = var.state_volume_size
 }
 
 resource "openstack_blockstorage_volume_v3" "home" {
     name = "${var.cluster_name}-home"
-    description = "Home for control node"
+    description = "Home for control node" # first word used to label filesystem
     size = var.home_volume_size
 }