Ansible Playbook: manage_hpc_nodes¶
Provision high-performance computing environments.
Supported features in the current version:
- Deploy OpenCL environment
- Deploy AMD GPU drivers
Deploy high-performance computing environment¶
- Verify that target nodes are registered in the inventory file: hpc_nodes.ini
- Verify that target endstate is set: manage_hpc_nodes.yml
- Run the playbook. Use the
-s <SITE>parameter to select the target site.
/opt/aplatform64/bin/ap64.sh -n -p manage_hpc_nodes -s <SITE>
Register the hosts that will consume the service in the Ansible Inventory file:
- Host Group:
A dedicated group_vars directory is used to store end-state configuration settings for both the playbook and related Ansible Roles.
Set playbook specific settings in the file:
manage_hpc_nodes_apps: opencl: gpu_amd:
|Parameter||Required?||Type||Default||Purpose / Value|
|manage_hpc_nodes_apps||yes||dictionary||Define what applications will be deployed|
|manage_hpc_nodes_apps.opencl||no||boolean|| ||Deploy the application?|
|manage_hpc_nodes_apps.gpu_amd||no||boolean|| ||Deploy the application?|
Additional role specific settings are available to further customize the playbook:
|A:Platform64 role||group_vars file|
OS compatibility is role dependant. Please refer to the respective role documentation.
Dependencies in this section are automatically solved during the installation of A:Platform64.
- Ansible Collections:
- Control Node: A:Platform64 installed and configured.
- Managed Nodes: target hosts prepared for A:Platform64 control.
The playbook is automatically deployed during the A:Platform64 installation process.
Help on implementing new features and maintaining the code base is welcomed.
Please see the guidelines for further details.