Configuring custom repositories

Use the local repository feature to create a customized set of local repositories on the OIM for the cluster nodes to access.

Ensure the custom entry is included in the software_config.json file.

{
    "cluster_os_type": "ubuntu",
    "cluster_os_version": "22.04",
    "repo_config": "partial",
    "softwares": [
        {"name": "amdgpu", "version": "6.3.1"},
        {"name": "cuda", "version": "12.8.0"},
        {"name": "bcm_roce", "version": "232.1.133.2"},
        {"name": "ofed", "version": "24.01-0.3.3.1"},
        {"name": "openldap"},
        {"name": "secure_login_node"},
        {"name": "nfs"},
        {"name": "beegfs", "version": "7.4.5"},
        {"name": "k8s", "version":"1.31.4"},
        {"name": "roce_plugin"},
        {"name": "jupyter"},
        {"name": "kubeflow"},
        {"name": "kserve"},
        {"name": "pytorch"},
        {"name": "tensorflow"},
        {"name": "vllm"},
        {"name": "telemetry"},
        {"name": "ucx", "version": "1.15.0"},
        {"name": "openmpi", "version": "4.1.6"},
        {"name": "intelgaudi", "version": "1.19.2-32"},
        {"name": "csi_driver_powerscale", "version":"v2.13.0"}
    ],
    "bcm_roce": [
        {"name": "bcm_roce_libraries", "version": "232.1.133.2"}
    ],
    "amdgpu": [
        {"name": "rocm", "version": "6.3.1" }
    ],
    "intelgaudi": [
        {"name": "intel"}
    ],
    "vllm": [
        {"name": "vllm_amd"},
        {"name": "vllm_nvidia"}
    ],
    "pytorch": [
        {"name": "pytorch_cpu"},
        {"name": "pytorch_amd"},
        {"name": "pytorch_nvidia"},
        {"name": "pytorch_gaudi"}
    ],
    "tensorflow": [
        {"name": "tensorflow_cpu"},
        {"name": "tensorflow_amd"},
        {"name": "tensorflow_nvidia"}
    ]
}

Note

For Rocky Linux OS, the cluster_os_type in the above sample will be rocky.

Create a custom.json file in the following directory: input/config/<cluster_os_type>/<cluster_os_version> to define the repositories. For example, For a cluster running RHEL 8.8, go to input/config/rhel/8.8/ and create the file there. The file is a JSON list consisting of the package name, repository type, URL (optional), and version (optional). Below is a sample version of the file:

{
  "custom": {
    "cluster": [
      {
        "package": "ansible==5.3.2",
        "type": "pip_module"
      },
      {
        "package": "docker-ce-24.0.4",
        "type": "rpm",
        "repo_name": "docker-ce-repo"
      },

      {
        "package": "gcc",
        "type": "rpm",
        "repo_name": "appstream"
      },
      {
        "package": "community.general",
        "type": "ansible_galaxy_collection",
        "version": "4.4.0"
      },

      {
        "package": "perl-Switch",
        "type": "rpm",
        "repo_name": "codeready-builder"
      },
      {
        "package": "prometheus-slurm-exporter",
        "type": "git",
        "url": "https://github.com/vpenso/prometheus-slurm-exporter.git",
        "version": "master"
      },
      {
        "package": "ansible.utils",
        "type": "ansible_galaxy_collection",
        "version": "2.5.2"
      },
      {
        "package": "prometheus-2.23.0.linux-amd64",
        "type": "tarball",
        "url": "https://github.com/prometheus/prometheus/releases/download/v2.23.0/prometheus-2.23.0.linux-amd64.tar.gz"
      },
      {
        "package": "metallb-native",
        "type": "manifest",
        "url": "https://raw.githubusercontent.com/metallb/metallb/v0.13.4/config/manifests/metallb-native.yaml"
      },
      {
        "package": "registry.k8s.io/pause",
        "version": "3.9",
        "type": "image"
      }

    ]
  }
}

Enter the parameters required in input/local_repo_config.yml as explained here.

Run the following commands:

cd local_repo
ansible-playbook local_repo.yml

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