Merge branch 'main' into fix/multi-images

Author: sjpb

ansible/fatimage.yml

@@ -52,7 +52,11 @@
   become: yes
   gather_facts: no
   tasks:
-    # - import_playbook: slurm.yml
+    # - import_playbook: slurm.yml:
+    - name: Setup DB
+      include_role:
+        name: mysql
+        tasks_from: install.yml
     - name: OpenHPC
       import_role:
         name: stackhpc.openhpc
@@ -77,10 +81,10 @@
         name: opensearch
         tasks_from: install.yml
       become: true
-
-    #   opensearch - containerised, nothing to do
     # slurm_stats - nothing to do
-    # filebeat - containerised - nothing to do
+    - import_role:
+        name: filebeat
+        tasks_from: install.yml
 
     - import_role:
       # can't only run cloudalchemy.node_exporter/tasks/install.yml as needs vars from preflight.yml and triggers service start

ansible/monitoring.yml

@@ -26,19 +26,12 @@
           # Collection currently requires root for all tasks.
           become: true
 
-- name: Setup filebeat
+- name: Deploy filebeat
   hosts: filebeat
   tags: filebeat
   tasks:
     - import_role:
         name: filebeat
-        tasks_from: config.yml
-      tags: config
-
-    - import_role:
-        name: filebeat
-        tasks_from: deploy.yml
-      tags: deploy
 
 - name: Deploy node_exporter
   hosts: node_exporter

ansible/roles/filebeat/tasks/deploy.yml

@@ -0,0 +1,17 @@
+---
+- name: Create systemd unit file
+  template:
+    dest: /etc/systemd/system/filebeat.service
+    src: filebeat.service.j2
+  become: true
+  register: _filebeat_unit
+
+- name: Pull container image
+  containers.podman.podman_image:
+    name: "docker.elastic.co/beats/filebeat-oss"
+    tag: "{{ filebeat_version }}"
+  become_user: "{{ filebeat_podman_user }}"
+
+- name: Reload filebeat unit file
+  command: systemctl daemon-reload
+  when: _filebeat_unit.changed

ansible/roles/filebeat/tasks/install.yml

@@ -0,0 +1,2 @@
+- import_tasks: install.yml
+- import_tasks: runtime.yml

ansible/roles/filebeat/tasks/main.yml

@@ -27,3 +27,13 @@
       mode: 0600
   notify: Restart filebeat container
   become: true
+
+- name: Flush handlers
+  meta: flush_handlers
+
+- name: Ensure filebeat service state
+  systemd:
+    name: filebeat.service
+    state: started
+    enabled: true
+  become: true

ansible/roles/filebeat/tasks/post.yml

@@ -8,3 +8,9 @@
     dest: /etc/systemd/system/mysql.service
     src: mysql.service.j2
   register: _mysql_unitfile
+
+- name: Pull container image
+  containers.podman.podman_image:
+    name: "mysql"
+    tag: "{{ mysql_tag }}"
+  become_user: "{{ mysql_podman_user }}"

ansible/roles/filebeat/tasks/config.yml renamed to ansible/roles/filebeat/tasks/runtime.yml

@@ -14,6 +14,12 @@
     src: opensearch.service.j2
   register: _opensearch_unit
 
+- name: Pull container image
+  containers.podman.podman_image:
+    name: "opensearchproject/opensearch"
+    tag: "{{ opensearch_version }}"
+  become_user: "{{ opensearch_podman_user }}"
+
 - name: Reload opensearch unit file
   command: systemctl daemon-reload
   when: _opensearch_unit.changed

ansible/roles/mysql/tasks/install.yml

@@ -74,11 +74,6 @@
   notify: Restart opensearch service
   become: true
 
-- name: Pull container
-  containers.podman.podman_image:
-    name: "opensearchproject/opensearch:{{ opensearch_version }}"
-  become_user: "{{ opensearch_podman_user }}"
-
 - name: Flush handlers
   meta: flush_handlers
 

ansible/roles/opensearch/tasks/install.yml

@@ -34,52 +34,3 @@
   with_items: "{{ podman_users }}"
   register: podman_user_info
   become: yes
-
-- name: Define tmp directories on tmpfs
-  blockinfile:
-    path: /etc/tmpfiles.d/podman.conf
-    create: yes
-    block: |
-      d {{ podman_tmp_dir_root }}/{{ item.name }}/libpod/tmp 0755 {{ item.name }} {{ item.name }}
-      Z {{ podman_tmp_dir_root }}/{{ item.name }}            0755 {{ item.name }} {{ item.name }}
-  become: yes
-  loop: "{{ podman_users }}"
-  register: podman_tmp_dirs
-
-- name: Create tmp directories
-  command: systemd-tmpfiles --create
-  become: true
-  when: podman_tmp_dirs.results | selectattr('changed') | list | length > 0 # when: any changed
-
-- name: Create podman configuration directories
-  file:
-    path: "{{ item.home }}/.config/containers/"
-    state: directory
-    owner: "{{ item.name }}"
-    group: "{{ item.name }}"
-  become: yes
-  loop: "{{ podman_user_info.results }}"
-
-- name: Set podman to use temp directories
-  community.general.ini_file:
-    path: "{{ item.home }}/.config/containers/containers.conf"
-    section: engine
-    option: tmp_dir
-    value: '"{{ podman_tmp_dir_root }}/{{ item.name }}/libpod/tmp"'
-    owner: "{{ item.name }}"
-    group: "{{ item.name }}"
-    create: yes
-  loop: "{{ podman_user_info.results }}"
-  become: yes
-  register: podman_tmp
-
-- name: Reset podman database
-  # otherwise old config overrides!
-  command:
-    cmd: podman system reset --force
-  become: yes
-  become_user: "{{ item.item.name }}"
-  when: item.changed
-  loop: "{{ podman_tmp.results }}"
-  loop_control:
-    label: "{{ item.item.name }}"

ansible/roles/opensearch/tasks/runtime.yml

@@ -1,12 +1,11 @@
 flavor = "vm.ska.cpu.general.small"
 use_blockstorage_volume = true
-volume_size = 10 # GB
+volume_size = 12 # GB. Compatible with SMS-lab's general.v1.tiny 
 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-8.9-20231119.0.x86_64.qcow2"
 ssh_keypair_name = "slurm-app-ci"
 ssh_private_key_file = "~/.ssh/id_rsa"
 security_groups = ["default", "SSH"]
-ssh_bastion_host = "128.232.222.183"
-ssh_bastion_username = "slurm-app-ci"
+floating_ip_network = "CUDN-Internet" # Use FIP to avoid docker ratelimits on portal-internal outbound IP

ansible/roles/podman/tasks/config.yml

@@ -0,0 +1 @@
+#update_enable: false # Can uncomment for speed debugging non-update related build issues

environments/.stackhpc/ARCUS.pkrvars.hcl

@@ -13,7 +13,7 @@ variable "cluster_name" {
 variable "cluster_image" {
     description = "single image for all cluster nodes - a convenience for CI"
     type = string
-    default = "openhpc-240102-1025-e533fd70" # https://github.com/stackhpc/ansible-slurm-appliance/pull/346
+    default = "openhpc-240116-1156-aa8dba7d" # https://github.com/stackhpc/ansible-slurm-appliance/pull/351
     # default = "Rocky-8-GenericCloud-Base-8.9-20231119.0.x86_64.qcow2"
 }
 

environments/.stackhpc/inventory/group_vars/builder.yml

@@ -2,6 +2,7 @@
 # NOTE: Might be better of as extra vars or in a builder specific inventory as
 # as dependent on alphabetical ordering of groups, so if these variables are
 # defined elsewhere the group that is ordered lower will determine the values.
+update_enable: true
 openhpc_slurm_service_started: false
 nfs_client_mnt_state: present
 block_devices_partition_state: skip

environments/.stackhpc/terraform/main.tf

@@ -1 +1 @@
-update_enable: true
+workaround_ansible_issue_61497: yes # extravars files can't be empty

environments/common/inventory/group_vars/builder/defaults.yml

@@ -92,12 +92,12 @@ variable "image_visibility" {
 
 variable "ssh_bastion_host" {
   type = string
-  default = ""
+  default = null
 }
 
 variable "ssh_bastion_username" {
   type = string
-  default = ""
+  default = null
 }
 
 variable "ssh_bastion_private_key_file" {