Adding new nodes

Provisioning the new node

While adding a new node to the cluster, users can modify the following:

  • The operating system

  • CUDA

  • OFED

A new node can be added using the following ways:

  1. When the discovery mechanism is switch-based:

    • Edit or append JSON list stored in switch-based-details in input/provision_config.yml.

    Note

    • All ports residing on the same switch should be listed in the same JSON list element.

    • Ports configured via Omnia should be not be removed from switch-based-details in input/provision_config.yml.

    • Run provision.yml.

  2. When the discovery mechanism is mapping:

    • Update the existing mapping file by appending the new entry (without the disrupting the older entries) or provide a new mapping file by pointing pxe_mapping_file_path in provision_config.yml to the new location.

    • Run provision.yml.

  3. When the discovery mechanism is snmpwalk:

    • Run provision.yml once the switch has discovered the potential new node.

  4. When the discovery mechanism is bmc:

    • Run provision.yml once the node has joined the cluster using an IP that exists within the provided range.

Alternatively, if a new node is to be added with no change in configuration, run the following commands:

cd provision
ansible-playbook discovery_provision.yml

Adding the new node to the cluster

  1. Update the existing inventory file with the new node details following the example provided here.

Note

Do not change the manager node in the existing inventory.

  1. To install security, job scheduler and storage tools (NFS, BeeGFS) on the node, run omnia.yml:

    ansible-playbook omnia.yml -i inventory

If you have any feedback about Omnia documentation, please reach out at omnia.readme@dell.com.