Merge pull request #159 from stackhpc/feature/openhpc_config

openhpc role bump & dependency pinning

Author: sjpb

.github/workflows/smslabs.yml

@@ -1,5 +1,5 @@
 
-name: Test on OpenStack via smslabs
+name: Test on SMS-Labs OpenStack in stackhpc-ci
 on:
   push:
     branches:
@@ -22,7 +22,7 @@ jobs:
           SSH_KEY: ${{ secrets.SSH_KEY }}
 
       - name: Add bastion's ssh key to known_hosts
-        run: cat environments/smslabs-example/bastion_fingerprint >> ~/.ssh/known_hosts
+        run: cat environments/smslabs/bastion_fingerprint >> ~/.ssh/known_hosts
         shell: bash
 
       - name: Install ansible etc
@@ -33,7 +33,7 @@ jobs:
 
       - name: Initialise terraform
         run: terraform init
-        working-directory: ${{ github.workspace }}/environments/smslabs-example/terraform
+        working-directory: ${{ github.workspace }}/environments/smslabs/terraform
 
       - name: Write clouds.yaml
         run: |
@@ -47,7 +47,7 @@ jobs:
         id: provision
         run: |
           . venv/bin/activate
-          . environments/smslabs-example/activate
+          . environments/smslabs/activate
           cd $APPLIANCES_ENVIRONMENT_ROOT/terraform
           terraform apply -auto-approve
         env:
@@ -58,7 +58,7 @@ jobs:
         id: provision_failure
         run: |
           . venv/bin/activate
-          . environments/smslabs-example/activate
+          . environments/smslabs/activate
           cd $APPLIANCES_ENVIRONMENT_ROOT/terraform
           echo "::set-output name=messages::$(./getfaults.py)"
         env:
@@ -69,7 +69,7 @@ jobs:
       - name: Delete infrastructure if failed due to lack of hosts
         run: |
           . venv/bin/activate
-          . environments/smslabs-example/activate
+          . environments/smslabs/activate
           cd $APPLIANCES_ENVIRONMENT_ROOT/terraform
           terraform destroy -auto-approve
         env:
@@ -80,7 +80,7 @@ jobs:
       - name: Configure infrastructure
         run: |
           . venv/bin/activate
-          . environments/smslabs-example/activate
+          . environments/smslabs/activate
           ansible all -m wait_for_connection
           ansible-playbook ansible/adhoc/generate-passwords.yml
           ansible-playbook -vv ansible/site.yml
@@ -91,7 +91,7 @@ jobs:
       - name: Run MPI-based tests
         run: |
           . venv/bin/activate
-          . environments/smslabs-example/activate
+          . environments/smslabs/activate
           ansible-playbook -vv ansible/adhoc/hpctests.yml
         env:
           ANSIBLE_FORCE_COLOR: True
@@ -100,7 +100,7 @@ jobs:
       - name: Build control and compute images
         run: |
           . venv/bin/activate
-          . environments/smslabs-example/activate
+          . environments/smslabs/activate
           cd packer
           PACKER_LOG=1 PACKER_LOG_PATH=build.log packer build -var-file=$PKR_VAR_environment_root/builder.pkrvars.hcl openstack.pkr.hcl
         env:
@@ -110,7 +110,7 @@ jobs:
       - name: Reimage compute nodes via slurm and check cluster still up
         run: |
           . venv/bin/activate
-          . environments/smslabs-example/activate
+          . environments/smslabs/activate
           ansible-playbook -vv $APPLIANCES_ENVIRONMENT_ROOT/ci/reimage-compute.yml
           ansible-playbook -vv $APPLIANCES_ENVIRONMENT_ROOT/hooks/post.yml
         env:
@@ -120,7 +120,7 @@ jobs:
       - name: Reimage login nodes via openstack and check cluster still up
         run: |
           . venv/bin/activate
-          . environments/smslabs-example/activate
+          . environments/smslabs/activate
           ansible-playbook -vv $APPLIANCES_ENVIRONMENT_ROOT/ci/reimage-login.yml
           ansible-playbook -vv $APPLIANCES_ENVIRONMENT_ROOT/hooks/post.yml
         env:
@@ -130,7 +130,7 @@ jobs:
       - name: Delete infrastructure
         run: |
           . venv/bin/activate
-          . environments/smslabs-example/activate
+          . environments/smslabs/activate
           cd $APPLIANCES_ENVIRONMENT_ROOT/terraform
           terraform destroy -auto-approve
         env:

README.md

@@ -1,5 +1,4 @@
 ---
 
-# You must define this variable.
-#filebeat_config_path: undefined
-filebeat_podman_user: "{{ ansible_user }}"
+#filebeat_config_path: undefined # REQUIRED. Path to filebeat.yml configuration file template
+filebeat_podman_user: "{{ ansible_user }}" # User that runs the filebeat container

ansible/roles/filebeat/defaults/main.yml

@@ -34,7 +34,7 @@
 # Note this still doesn't fix any non-unique names but we should get a length mis-match at least with that.
 # although this looks a bit crazy:
 - name: Expand node list
-  shell: "scontrol show hostnames {{ openhpc_tests_nodes if openhpc_tests_nodes is defined else (hpctests_computes.stdout_lines | join(',')) }}"
+  shell: "scontrol show hostnames {{ hpctests_nodes if hpctests_nodes is defined else (hpctests_computes.stdout_lines | join(',')) }}"
   register: scontrol_hostnames
 
 - name: Create sorted node expression

ansible/roles/hpctests/tasks/pingmatrix.yml

@@ -9,6 +9,17 @@
         that: groups['control'] | length
         fail_msg: "no hosts found in group 'control' - has control node been deployed?"
 
+- name: Validate openhpc configuration
+  hosts: openhpc
+  tags: openhpc
+  tasks:
+    - assert:
+        that: "'enable_configless' in openhpc_config.SlurmctldParameters | default([])"
+        fail_msg: |
+          'enable_configless' not found in openhpc_config.SlurmctldParameters - is variable openhpc_config overridden?
+          Additional slurm.conf parameters should be provided using variable openhpc_config_extra.
+        success_msg: Checked Slurm will be configured for configless operation
+    
 - name: Validate podman configuration
   hosts: podman
   tags: podman

ansible/validate.yml

@@ -1,2 +1,7 @@
+---
+
+# See: ansible/roles/basic_users/README.md 
+# for variable definitions.
+
 basic_users_homedir: /home
 basic_users_users: []

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

@@ -1,4 +1,5 @@
 ---
+
 # Path to filebeat.yml configuration file template
 filebeat_config_path: "{{ appliances_repository_root }}/environments/common/files/filebeat/filebeat.yml"
 

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

@@ -1 +1,5 @@
+---
+# See: ansible/roles/hpctests/README.md
+# for variable definitions.
+
 hpctests_rootdir: "/home/hpctests" # Can't use centos's $HOME as that's not on /home and may not have another user

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

@@ -1,6 +1,7 @@
 ---
-# Path to template that specifies opendistro users.
 # See: https://opendistro.github.io/for-elasticsearch-docs/docs/security/configuration/yaml/
+
+# Path to template that specifies opendistro users
 opendistro_internal_users_path: "{{ appliances_repository_root }}/environments/common/files/opendistro/internal_users.yml"
 
 # The user what runs the opendistro container

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

@@ -1,4 +1,5 @@
 ---
+
 # See: https://github.com/stackhpc/ansible-role-openhpc
 # for variable definitions
 
@@ -22,5 +23,9 @@ openhpc_packages_default:
 openhpc_packages_extra: []
 openhpc_packages: "{{ openhpc_packages_default + openhpc_packages_extra }}"
 openhpc_munge_key: "{{ vault_openhpc_mungekey | b64decode }}"
-openhpc_slurm_configless: true
-openhpc_login_only_nodes: login
+openhpc_login_only_nodes: login
+openhpc_config_default:
+  SlurmctldParameters:
+    - enable_configless
+openhpc_config_extra: {}
+openhpc_config: "{{ openhpc_config_default | combine(openhpc_config_extra, list_merge='append') }}"

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

@@ -1,6 +1,9 @@
 ---
-## TODO: maybe we should arrange it following the osc.github.io docs
-## NB: Variables prefixed ood_ are all from https://github.com/OSC/ood-ansible
+
+# See: ansible/roles/openondemand/README.md 
+# for variable definitions.
+
+# NB: Variables prefixed ood_ are all from https://github.com/OSC/ood-ansible
 
 openondemand_servername: '' # Must be changed when using openondemand, but allows templating when openondemand group is empty
 

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

@@ -1,2 +1,2 @@
-podman_users: "{{ appliances_local_users_podman }}"
-podman_cidr: 10.0.2.0/24 # see slirp4netns:cidr= at https://docs.podman.io/en/latest/markdown/podman-run.1.html
+podman_users: "{{ appliances_local_users_podman }}" # user to use for podman
+podman_cidr: 10.0.2.0/24 # IP range to use for podman - see slirp4netns:cidr= at https://docs.podman.io/en/latest/markdown/podman-run.1.html

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

@@ -1,4 +1,5 @@
 ---
+
 # See: https://github.com/cloudalchemy/ansible-prometheus
 # for variable definitions
 

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

@@ -1,3 +1,4 @@
 ---
+
 selinux_state: permissive
 selinux_policy: targeted

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

@@ -1,3 +1,5 @@
+---
+
 update_enable: false
 # These variables define the packages updates and are passed to ansible's yum module parameters with the same names: https://docs.ansible.com/ansible/latest/collections/ansible/builtin/yum_module.html
 update_name: '*'

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

@@ -56,11 +56,6 @@ filebeat
 [node_exporter]
 # All hosts to monitor for hardware and OS metrics.
 
-[openhpc_tests:children]
-# For post-deploy MPI-based tests - see ansible/adhoc/test.yml
-login
-compute
-
 [selinux:children]
 # All hosts requiring control of SELinux status.
 cluster

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

@@ -1,4 +1,4 @@
-openhpc_config:
+openhpc_config_extra:
   SlurmctldDebug: debug
   SlurmdDebug: debug
 openhpc_slurm_partitions:

environments/common/inventory/groups

@@ -13,6 +13,10 @@ resource "openstack_compute_instance_v2" "control" {
     access_network = true
   }
 
+  metadata = {
+    environment_root = var.environment_root
+  }
+
 }
 
 resource "openstack_compute_instance_v2" "login" {
@@ -31,6 +35,10 @@ resource "openstack_compute_instance_v2" "login" {
     access_network = true
   }
 
+  metadata = {
+    environment_root = var.environment_root
+  }
+
 }
 
 resource "openstack_compute_instance_v2" "compute" {
@@ -49,4 +57,8 @@ resource "openstack_compute_instance_v2" "compute" {
     access_network = true
   }
 
+  metadata = {
+    environment_root = var.environment_root
+  }
+
 }

environments/smslabs-example/.gitignore renamed to environments/smslabs/.gitignore

@@ -66,3 +66,8 @@ variable "compute_images" {
     default = {}
     description = "Mapping to override compute images from compute_types: key ->(str) node name, value -> (str) image name"
 }
+
+variable "environment_root" {
+    type = string
+    description = "Path to environment root, automatically set by activate script"
+}

environments/smslabs-example/activate renamed to environments/smslabs/activate

@@ -1,20 +1,26 @@
 ---
 roles:
   - src: stackhpc.nfs
+    version: v21.2.1
   - src: https://github.com/stackhpc/ansible-role-openhpc.git
-    version: v0.9.0 # supports Rocky Linux
+    version: v0.12.0
     name: stackhpc.openhpc
   - src: https://github.com/stackhpc/ansible-node-exporter.git
     version: support-rhel-clones
     name: cloudalchemy.node_exporter
   - src: cloudalchemy.blackbox-exporter
+    version: 1.0.0
   - src: https://github.com/cloudalchemy/ansible-prometheus.git
     version: 4d2c8d742de39e50387e0aa6d5510b21c7451343 # need fix in preceeding commit for rocky
     name: cloudalchemy.prometheus
   - src: cloudalchemy.alertmanager
+    version: 0.19.1
   - src: cloudalchemy.grafana
+    version: 0.18.0
   - src: geerlingguy.mysql
+    version: 3.3.2
   - src: jriguera.configdrive
+    # No versions available
   - src: https://github.com/OSC/ood-ansible.git
     name: osc.ood
     version: v2.0.5