Skip to content

d.run AI OS Installation Guide

This page is the main entry point for installing d.run AI OS (hereinafter "d.run"): it outlines the full installation chain, the document for each stage, and the initialization order after installation.

d.run is installed offline: the installer dce5-installer runs on a machine called the Bootstrap Node, uses Kubean to initialize the operating system and create a Kubernetes cluster, and then deploys the product modules declared in manifest-cloud.yaml to the global service cluster. Therefore, four types of offline packages must be prepared beforehand: the offline image package, ISO operating system images, osPackage offline packages, and optionally Addon offline packages.

Version note

This page uses installer v0.45.0 as an example. The versions of the installer, the offline image package, and the osPackage packages must match each other. For version changes, see the installer Release Notes. To install Token Factory instead, see the Token Factory installation guide; the manifest file and default modules differ.

There are eight stages from environment preparation to daily operations. If the installation fails, locate the cause first, then retry from the corresponding stage. Click a stage in the diagram to jump to its document.

flowchart TB
    A["1 Prepare environment<br/>Architecture · Hardware · Ports"] --> B["2 Install dependencies<br/>podman · helm · skopeo · kind"]
    B --> C["3 Download offline packages<br/>Image package · ISO · osPackage"]
    C --> D["4 Edit clusterConfig.yaml<br/>Deployment mode and nodes"]
    D --> E["5 Confirm manifest-cloud.yaml<br/>Choose the modules to install"]
    E --> F["6 Run the installer<br/>dce5-installer cluster-create"]
    F --> G{"Installation succeeded?"}
    G -- "No" --> H["Troubleshoot the installation<br/>Fix the cause and retry"]
    H --> F
    G -- "Yes" --> I["7 Access the console and request a license<br/>admin/changeme"]
    I --> J["8 Daily operations<br/>Upgrade version by version"]
    click A "commercial/deploy-requirements/"
    click B "dependency/"
    click C "start-install/"
    click D "commercial/cluster-config/"
    click E "commercial/manifest/"
    click F "start-install/"
    click H "faq/"
    click I "start-install/"
    click J "upgrade/"

Once installed and logged in with admin/changeme, initialize the platform in this order:

  1. Create users, user groups, and workspaces in Global Management, and plan roles and permissions.
  2. Create worker clusters in Container Management to host workloads; use GPU nodes for AI training and inference.
  3. Go to LLM Studio to deploy or connect models, and configure Tokens and API access.
  4. Verify metrics and log collection in Observability, and check metering and billing in Wallet.

Download Center Learn about d.run Token Factory Learn about d.run AI OS

Comments