Support software raid root disks in stackhpc images (#785)

* support raid root disks in stackhpc-built images

* clarify image requirements

* bump CI image

* fixup grub for RL8

* fix linter issues

* fix raid kernel commandline configuration for RL8

[no ci]

* bump CI image

* fix handler ansible-lint errors

* bump CI image

Author: sjpb

ansible/.gitignore

@@ -98,3 +98,5 @@ roles/*
 !roles/eessi/**
 !roles/topology/
 !roles/topology/**
+!roles/raid/
+!roles/raid/**

ansible/extras.yml

@@ -1,4 +1,11 @@
----
+- hosts: raid
+  become: true
+  tags: raid
+  gather_facts: true
+  tasks:
+    - ansible.builtin.include_role:
+        name: raid
+
 - hosts: k3s_server:!builder
   become: true
   tags: k3s

ansible/roles/raid/README.md

@@ -0,0 +1,17 @@
+# raid
+
+Configure an image to support software raid (via [mdadm](https://github.com/md-raid-utilities/mdadm)).
+
+RockyLinux genericcloud images already have the necessary `mdraid` dracut
+module installed, as well as kernel modules for `raid0`, `raikd1`, `raid10` and
+`raid456` [^1]. This covers all raid modes [supported by Ironic](https://docs.openstack.org/ironic/latest/admin/raid.html#software-raid)
+hence this role does not support extending this.
+
+This role changes the command line for the current kernel. It does not reboot
+the instance so generally is only useful during image builds.
+
+Note that the `rootfs_uuid` image property described in the [Ironic raid documentation](https://docs.openstack.org/ironic/latest/admin/raid.html#image-requirements)
+is not required; the root partition is the first (non-boot) partition and this
+is sufficent for Ironic to find the root file system.
+
+[^1]: As shown by `lsinitrd /boot/initramfs-$(uname -r).img | grep raid`

ansible/roles/raid/handlers/main.yml

@@ -0,0 +1,3 @@
+- name: Update GRUB configuration file
+  ansible.builtin.command: "grub2-mkconfig -o /boot/grub2/grub.cfg {{ '--update-bls-cmdline' if ansible_distribution_major_version == '9' else '' }}"
+  changed_when: true

ansible/roles/raid/tasks/main.yml

@@ -0,0 +1,17 @@
+- name: Enable autoassembly of mdraid devices
+  # adds rd.auto=1 - see `man dracut.cmdline`
+  ansible.builtin.lineinfile:
+    path: /etc/default/grub
+    regexp: >
+      ^{{ grub_cmdline_var[ansible_distribution_major_version] }}="((?:(?!rd.auto=1).)*?)"$
+    line: >
+      {{ grub_cmdline_var[ansible_distribution_major_version] }}="\1 rd.auto=1"
+    backup: true
+    backrefs: true
+  register: update_grub
+  notify:
+    - Update GRUB configuration file
+  vars:
+    grub_cmdline_var:
+      '8': GRUB_CMDLINE_LINUX
+      '9': GRUB_CMDLINE_LINUX_DEFAULT

environments/.stackhpc/inventory/extra_groups

@@ -50,3 +50,7 @@ cluster
 
 [compute_init:children]
 compute
+
+[raid:children]
+# Configure fatimage for raid
+builder

environments/.stackhpc/tofu/cluster_image.auto.tfvars.json

@@ -1,6 +1,6 @@
 {
   "cluster_image": {
-    "RL8": "openhpc-RL8-250924-1502-e9afbfe5",
-    "RL9": "openhpc-RL9-250924-1536-e9afbfe5"
+    "RL8": "openhpc-RL8-250925-1639-62d67ae3",
+    "RL9": "openhpc-RL9-250925-1639-62d67ae3"
   }
 }

environments/common/inventory/groups

@@ -217,3 +217,6 @@ extra_packages
 # separately from the appliance. e.g
 # pulp_host ansible_host=<VM-ip-address>
 # Note the host name can't conflict with group names i.e can't be called `pulp` or `pulp_server`
+
+[raid]
+# Add `builder` to configure image for software raid

environments/site/inventory/groups

@@ -169,3 +169,6 @@ compute
 # separately from the appliance. e.g
 # pulp_host ansible_host=<VM-ip-address>
 # Note inventory host name cannot conflict with group names i.e can't be called `pulp` or `pulp_server`.
+
+[raid]
+# Add `builder` to configure image for software raid