Add support for RockyLinux9 (#353)

sjpb · web-flow · commit a415036937e3 · 2024-03-14T13:12:23.000Z
* add changes from branch rl9

* fix unqualified names for container pulls

* fix openondemand install

* bugfix filebeat unit reload

* comment on required image

* bump fat image base to RL9.3

* fix dbus-launch command for OOD desktop

* fix OOD desktop launch

* fix useradd warning: {grafana,prometheus}'s uid * outside of the UID_MIN 1000 and UID_MAX 60000 range.

* fix manila support in RL9

* bump openhpc role after merge

* downsize arcus control node to try to avoid CI failures

* prevent ssh hanging after NFS server reimaged in CI

* fix 'prevent ssh hanging after NFS server reimaged in CI'

* fix import of osc GPG key for RL9

* make fatimage more like openondemand role

* enable fatimage build on either RL8 or RL9, inc on PRs

* get concurrent OS builds

* enable RL8 and RL9 for CI test workflow

* fix stackhpc workflow

* fix CI image selection

* fix os debug output

* fix podman systemd warnings

* bump CI image

* bump openhpc role for release

* fix OOD app partitions

* delete unused packer vars for compute image builds

* update packer README to remove compute image builds

* bump os-manila role after release
diff --git a/.github/workflows/fatimage.yml b/.github/workflows/fatimage.yml
@@ -1,12 +1,26 @@
 
 name: Build fat image
-on:
+'on':
   workflow_dispatch:
+    inputs:
+      use_RL9:
+        required: true
+        description: Include RL9 image build
+        type: boolean
+        default: false
 jobs:
   openstack:
     name: openstack-imagebuild
-    concurrency: ${{ github.ref }} # to branch/PR
     runs-on: ubuntu-20.04
+    concurrency: ${{ github.ref }}-{{ matrix.os_version }} # to branch/PR + OS
+    strategy:
+      matrix:
+        os_version: [RL8, RL9]
+        rl9_selected:
+          - ${{ inputs.use_RL9 == true }} # only potentially true for workflow_dispatch
+        exclude:
+          - os_version: RL9
+            rl9_selected: false
     env:
       ANSIBLE_FORCE_COLOR: True
       OS_CLOUD: openstack
@@ -48,6 +62,8 @@ jobs:
           cd packer/
           packer init .
           PACKER_LOG=1 packer build -only openstack.openhpc -on-error=${{ vars.PACKER_ON_ERROR }} -var-file=$PKR_VAR_environment_root/${{ vars.CI_CLOUD }}.pkrvars.hcl openstack.pkr.hcl
+        env:
+          PKR_VAR_os_version: ${{ matrix.os_version }}
 
       - name: Get created image name from manifest
         id: manifest
diff --git a/.github/workflows/stackhpc.yml b/.github/workflows/stackhpc.yml
@@ -2,19 +2,36 @@
 name: Test deployment and reimage on OpenStack
 on:
   workflow_dispatch:
+    inputs:
+      use_RL9:
+        required: true
+        description: Include RL9 tests
+        type: boolean
+        default: false
   push:
     branches:
       - main
   pull_request:
 jobs:
   openstack:
     name: openstack-ci
-    concurrency: ${{ github.ref }} # to branch/PR
+    concurrency: ${{ github.ref }}-{{ matrix.os_version }} # to branch/PR + OS
     runs-on: ubuntu-20.04
+    strategy:
+      matrix:
+        os_version: [RL8, RL9]
+        rl9_selected:
+          - ${{ inputs.use_RL9 == true }} # only potentially true for workflow_dispatch
+        rl9_branch:
+          - ${{ startsWith(github.head_ref, 'rl9') == true }} # only potentially for pull_request, always false on merge
+        exclude:
+          - os_version: RL9
+            rl9_selected: false
+            rl9_branch: false
     env:
       ANSIBLE_FORCE_COLOR: True
       OS_CLOUD: openstack
-      TF_VAR_cluster_name: ci${{ github.run_id }}
+      TF_VAR_cluster_name: slurmci-${{ matrix.os_version }}-${{ github.run_id }}
       CI_CLOUD: ${{ vars.CI_CLOUD }}
     steps:
       - uses: actions/checkout@v2
@@ -69,6 +86,8 @@ jobs:
           . environments/.stackhpc/activate
           cd $APPLIANCES_ENVIRONMENT_ROOT/terraform
           terraform apply -auto-approve -var-file="${{ vars.CI_CLOUD }}.tfvars"
+        env:
+          TF_VAR_os_version: ${{ matrix.os_version }}
 
       - name: Delete infrastructure if provisioning failed
         run: |
@@ -77,6 +96,8 @@ jobs:
           cd $APPLIANCES_ENVIRONMENT_ROOT/terraform
           terraform destroy -auto-approve -var-file="${{ vars.CI_CLOUD }}.tfvars"
         if: failure() && steps.provision_servers.outcome == 'failure'
+        env:
+          TF_VAR_os_version: ${{ matrix.os_version }}
 
       - name: Configure cluster
         run: |
@@ -175,6 +196,8 @@ jobs:
           cd $APPLIANCES_ENVIRONMENT_ROOT/terraform
           terraform destroy -auto-approve -var-file="${{ vars.CI_CLOUD }}.tfvars"
         if: ${{ success() || cancelled() }}
+        env:
+          TF_VAR_os_version: ${{ matrix.os_version }}
 
       # - name: Delete images
       #   run: |
diff --git a/ansible/bootstrap.yml b/ansible/bootstrap.yml
@@ -40,7 +40,13 @@
 - hosts: cluster
   gather_facts: false
   tasks:
-    - name: Add groups
+    - name: Prevent ssh hanging if shared home is unavailable
+      lineinfile:
+        path: /etc/profile
+        search_string: HOSTNAME=$(/usr/bin/hostnamectl --transient 2>/dev/null) || \
+        state: absent
+      become: yes
+    - name: Add system user groups
       ansible.builtin.group: "{{ item.group }}"
       loop: "{{ appliances_local_users }}"
       when:
@@ -50,7 +56,7 @@
       # Need to change working directory otherwise we try to switch back to non-existent directory.
       become_flags: '-i'
       become: true
-    - name: Add users
+    - name: Add system users
       ansible.builtin.user: "{{ item.user }}"
       loop: "{{ appliances_local_users }}"
       when: item.enable | default(true) | bool
diff --git a/ansible/fatimage.yml b/ansible/fatimage.yml
@@ -49,7 +49,7 @@
 
 - hosts: builder
   become: yes
-  gather_facts: no
+  gather_facts: yes
   tasks:
     # - import_playbook: slurm.yml:
     - name: Setup DB
@@ -61,15 +61,13 @@
         name: stackhpc.openhpc
         tasks_from: install.yml
 
-    - name: Include distribution variables for osc.ood
-      include_vars: "{{ appliances_repository_root }}/ansible/roles/osc.ood/vars/Rocky/8.yml"
-    # FUTURE: install-apps.yml - this is git clones
-
     # - import_playbook: portal.yml
     - name: Open Ondemand server
-      import_role:
+      include_role:
         name: osc.ood
         tasks_from: install-package.yml
+        vars_from: "Rocky/{{ ansible_distribution_major_version }}.yml"
+    # # FUTURE: install-apps.yml - this is git clones
     - name: Open Ondemand remote desktop
       import_role:
         name: openondemand
diff --git a/ansible/roles/filebeat/templates/filebeat.service.j2 b/ansible/roles/filebeat/templates/filebeat.service.j2
@@ -12,7 +12,7 @@ After=network-online.target
 [Service]
 Environment=PODMAN_SYSTEMD_UNIT=%n
 Restart=always
-ExecStart=/usr/bin/podman run \
+ExecStart=/usr/bin/podman --cgroup-manager=cgroupfs run \
     --network=host \
     --sdnotify=conmon \
     --cgroups=no-conmon \
diff --git a/ansible/roles/mysql/tasks/install.yml b/ansible/roles/mysql/tasks/install.yml
@@ -1,6 +1,12 @@
+- name: Install pip
+  dnf:
+    name: python3-pip
+
 - name: Install python mysql client
   pip:
-    name: pymysql
+    name:
+      - pymysql
+      - cryptography
     state: present
 
 - name: Create systemd mysql container unit file
@@ -11,6 +17,6 @@
 
 - name: Pull container image
   containers.podman.podman_image:
-    name: "mysql"
+    name: docker.io/library/mysql
     tag: "{{ mysql_tag }}"
   become_user: "{{ mysql_podman_user }}"
diff --git a/ansible/roles/mysql/templates/mysql.service.j2 b/ansible/roles/mysql/templates/mysql.service.j2
@@ -14,7 +14,7 @@ EnvironmentFile=/etc/sysconfig/mysqld
 # The above EnvironmentFile must define MYSQL_INITIAL_ROOT_PASSWORD
 ExecStartPre=+install -d -o {{ mysql_podman_user }} -g {{ mysql_podman_user }} -Z container_file_t {{ mysql_datadir }}
 ExecStartPre=+chown -R {{ mysql_podman_user }}:{{ mysql_podman_user }} {{ mysql_datadir }}
-ExecStart=/usr/bin/podman run \
+ExecStart=/usr/bin/podman --cgroup-manager=cgroupfs run \
     --network=host \
     --sdnotify=conmon \
     --cgroups=no-conmon \
@@ -26,7 +26,7 @@ ExecStart=/usr/bin/podman run \
     --volume {{ mysql_datadir }}:/var/lib/mysql:U \
     --publish 3306:3306 \
     --env MYSQL_ROOT_PASSWORD=${MYSQL_INITIAL_ROOT_PASSWORD} \
-    mysql:{{ mysql_tag }}{%- for opt in mysql_mysqld_options %} \
+    docker.io/library/mysql:{{ mysql_tag }}{%- for opt in mysql_mysqld_options %} \
     --{{ opt }}{% endfor %}
 
 ExecStop=/usr/bin/podman stop --ignore mysql -t 10
diff --git a/ansible/roles/openondemand/tasks/main.yml b/ansible/roles/openondemand/tasks/main.yml
@@ -10,7 +10,7 @@
 - include_role:
     name: osc.ood
     tasks_from: install-package.yml
-    vars_from: Rocky/8.yml
+    vars_from: "Rocky/{{ ansible_distribution_major_version }}.yml"
     public: yes # Expose the vars from this role to the rest of the play
   # can't set vars: from a dict hence the workaround above
 
diff --git a/ansible/roles/openondemand/tasks/vnc_compute.yml b/ansible/roles/openondemand/tasks/vnc_compute.yml
@@ -17,6 +17,7 @@
       - turbovnc-3.0.1
       - nmap-ncat
       - python3.9
+      - dbus-x11
 
 - name: Install Xfce desktop
   tags: install
diff --git a/ansible/roles/opensearch/tasks/install.yml b/ansible/roles/opensearch/tasks/install.yml
@@ -16,7 +16,7 @@
 
 - name: Pull container image
   containers.podman.podman_image:
-    name: "opensearchproject/opensearch"
+    name: docker.io/opensearchproject/opensearch
     tag: "{{ opensearch_version }}"
   become_user: "{{ opensearch_podman_user }}"
 
diff --git a/ansible/roles/opensearch/templates/opensearch.service.j2 b/ansible/roles/opensearch/templates/opensearch.service.j2
@@ -11,7 +11,7 @@ Environment=PODMAN_SYSTEMD_UNIT=%n
 Restart=always
 # paths below based on https://opensearch.org/docs/latest/opensearch/configuration/ and https://opensearch.org/docs/latest/security-plugin/configuration/yaml
 # see also https://opensearch.org/docs/2.0/opensearch/install/important-settings/
-ExecStart=/usr/bin/podman run \
+ExecStart=/usr/bin/podman --cgroup-manager=cgroupfs run \
     --network=host \
     --sdnotify=conmon \
     --cgroups=no-conmon \
@@ -29,7 +29,7 @@ ExecStart=/usr/bin/podman run \
     --env bootstrap.memory_lock=true \
     --env "OPENSEARCH_JAVA_OPTS=-Xms512m -Xmx512m" \
     --env DISABLE_INSTALL_DEMO_CONFIG=true \
-    opensearchproject/opensearch:{{ opensearch_version }}
+    docker.io/opensearchproject/opensearch:{{ opensearch_version }}
 ExecStop=/usr/bin/podman stop --ignore opensearch -t 10
 # note for some reason this returns status=143 which makes systemd show the unit as failed, not stopped
 ExecStopPost=/usr/bin/podman rm --ignore -f opensearch
diff --git a/environments/.stackhpc/ARCUS.pkrvars.hcl b/environments/.stackhpc/ARCUS.pkrvars.hcl
@@ -1,10 +1,8 @@
 flavor = "vm.ska.cpu.general.small"
 use_blockstorage_volume = true
-volume_size = 12 # GB. Compatible with SMS-lab's general.v1.tiny 
+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"]
diff --git a/environments/.stackhpc/LEAFCLOUD.pkrvars.hcl b/environments/.stackhpc/LEAFCLOUD.pkrvars.hcl
@@ -4,8 +4,6 @@ volume_size = 12 # GB. Compatible with SMS-lab's general.v1.tiny
 volume_type = "unencrypted"
 image_disk_format = "qcow2"
 networks = ["909e49e8-6911-473a-bf88-0495ca63853c"] # slurmapp-ci
-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"]
diff --git a/environments/.stackhpc/hooks/post-bootstrap.yml b/environments/.stackhpc/hooks/post-bootstrap.yml
@@ -3,14 +3,15 @@
   gather_facts: false
   tags: podman
   tasks:
-    - name: Configure container image registry for unqualified searches to avoid docker.io ratelimits
+    - name: Configure container image registry to avoid docker.io ratelimits
       copy:
-        dest: /etc/containers/registries.conf.d/003-arcus-unqualfied-overrides.conf
+        dest: /etc/containers/registries.conf.d/003-arcus-mirror.conf
         content: |
-          unqualified-search-registries = ['{{ podman_registry_address  | split('/') | first }}', 'registry.access.redhat.com', 'registry.redhat.io', 'docker.io']
-
           [[registry]]
-          prefix = "{{ podman_registry_address }}"
+          location="docker.io/library/"
+          prefix="docker.io/library/"
+          
+          [[registry.mirror]]
           location = "{{ podman_registry_address }}"
           insecure = true
       when: "ci_cloud == 'ARCUS'"
diff --git a/environments/.stackhpc/hooks/pre.yml b/environments/.stackhpc/hooks/pre.yml
@@ -2,6 +2,8 @@
   become: yes
   gather_facts: false
   tasks:
+    - name: Output OS version
+      command: cat /etc/redhat-release
     - name: Write CI-generated inventory and secrets for debugging
       ansible.builtin.copy:
         dest: /etc/ci-config/
diff --git a/environments/.stackhpc/inventory/group_vars/openondemand/overrides.yml b/environments/.stackhpc/inventory/group_vars/openondemand/overrides.yml
@@ -1,6 +1,6 @@
 openondemand_auth: basic_pam
-openondemand_jupyter_partition: small
-openondemand_desktop_partition: small
+openondemand_jupyter_partition: standard
+openondemand_desktop_partition: standard
 #openondemand_dashboard_support_url: 
 #openondemand_dashboard_docs_url:
 #openondemand_filesapp_paths:
diff --git a/environments/.stackhpc/terraform/main.tf b/environments/.stackhpc/terraform/main.tf
@@ -19,11 +19,19 @@ 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"
+}
+
 variable "cluster_image" {
-    description = "single image for all cluster nodes - a convenience for CI"
-    type = string
-    default = "openhpc-240308-1011-0f0291c0" # https://github.com/stackhpc/ansible-slurm-appliance/pull/364
-    # default = "Rocky-8-GenericCloud-Base-8.9-20231119.0.x86_64.qcow2"
+    description = "single image for all cluster nodes, keyed by os_version - a convenience for CI"
+    type = map(string)
+    default = {
+        # https://github.com/stackhpc/ansible-slurm-appliance/pull/353
+        RL8: "openhpc-RL8-240313-1028-15f9ab38"
+        RL9: "openhpc-RL9-240313-1057-15f9ab38"
+    }
 }
 
 variable "cluster_net" {}
@@ -60,23 +68,23 @@ module "cluster" {
     key_pair = "slurm-app-ci"
     control_node = {
         flavor: var.control_node_flavor
-        image: var.cluster_image
+        image: var.cluster_image[var.os_version]
     }
     login_nodes = {
         login-0: {
             flavor: var.other_node_flavor
-            image: var.cluster_image
+            image: var.cluster_image[var.os_version]
         }
     }
     compute_types = {
         standard: { # NB: can't call this default!
             flavor: var.other_node_flavor
-            image: var.cluster_image
+            image: var.cluster_image[var.os_version]
         }
         # Example of how to add another partition:
         # extra: {
         #     flavor: var.other_node_flavor
-        #     image: var.cluster_image
+        #     image: var.cluster_image[var.os_version]
         # }
     }
     compute_nodes = {
diff --git a/environments/common/inventory/group_vars/all/defaults.yml b/environments/common/inventory/group_vars/all/defaults.yml
@@ -58,6 +58,7 @@ appliances_local_users_default:
         uid: 981
         home: "{{ prometheus_db_dir }}"
         shell: /usr/sbin/nologin
+        system: true
       enable: "{{ 'prometheus' in group_names }}"
     
     - group:
@@ -69,6 +70,7 @@ appliances_local_users_default:
         uid: 984
         home: /usr/share/grafana
         shell: /sbin/nologin
+        system: true
       enable: "{{ 'grafana' in group_names }}"
 
 # Overide this to add extra users whilst keeping the defaults.
diff --git a/environments/common/inventory/group_vars/all/openondemand.yml b/environments/common/inventory/group_vars/all/openondemand.yml
@@ -49,6 +49,8 @@ openondemand_clusters:
             module purge
 
             export PATH=/opt/TurboVNC/bin:$PATH
+            # avoid "Failed to create secure directory (/run/user/*/pulse)"
+            export XDG_RUNTIME_DIR="$TMPDIR/xdg_runtime"
 
             # Workaround to avoid "Unable to contact settings server" when
             # lauching xfce4-session
diff --git a/packer/README.md b/packer/README.md
diff --git a/packer/compute_extravars.yml b/packer/compute_extravars.yml
diff --git a/packer/openstack.pkr.hcl b/packer/openstack.pkr.hcl
diff --git a/requirements.yml b/requirements.yml
