Skip to main content
Version: v0.1.x

Self-Hosted Quick Start

Use this guide for an end-to-end self-hosted setup that starts with the appliance installation and ends with importing and deploying your first profile bundle.

This quickstart follows the appliance-to-first-deployment workflow and maps it to the current public docs structure. Use it as the main sequence, then proceed to the linked task pages when you need detailed field-by-field guidance.

Overview​

The self-hosted quickstart has four phases:

  1. Install self-hosted Palette and PaletteAI on the appliance.

  2. Complete post-install Palette setup.

  3. Complete post-install PaletteAI setup.

  4. Import and deploy your first profile bundle.

Install the Appliance​

Complete the appliance installation first:

  1. Follow Appliance Installation.

  2. Complete Prepare Infrastructure.

  3. Complete Deploy PaletteAI.

During appliance setup, make sure you record the following values because you use them later in this guide:

  • The root domain or VIP used to access the appliance
  • The Palette system console URL
  • The PaletteAI URL
  • The internal Zot registry address and port

If you use a root domain, access PaletteAI over HTTPS. The login flow does not work correctly over HTTP.

Complete Palette Setup​

Activate a Tenant​

After the appliance is running:

  1. Log in to the Palette system console.

  2. Create or activate the tenant you plan to use with PaletteAI.

  3. Confirm you can access the tenant in Palette before continuing.

For more background on Palette integration requirements, refer to Integrate with Palette.

Configure Pack Registries​

Connected Workflow​

For connected environments:

  1. Add the required system-level OCI pack registries in Palette.

  2. Wait for the registry sync to complete before you build cluster profiles or import bundles that depend on those packs.

Air-gapped Workflow​

For air-gapped environments:

  1. Identify the packs your cluster profiles require. If every required pack is available in Artifact Studio, download them and skip the next step.

  2. For each pack that is not available in Artifact Studio, obtain a Pack tarball from your Spectro Cloud representative. On a host that can reach upstream chart and image registries, issue paletteai mirror export-pack to build a mirror bundle from that tarball, then, from a host that can reach your air-gapped OCI registry, issue paletteai mirror push to upload the pack archive and its declared chart and image dependencies into your internal OCI registry. Repeat this sequence once per pack. Refer to PaletteAI CLI - mirror and Deploy Profile Bundles in Air-Gapped Environments.

  3. Upload the required pack content in Palette Local UI and sync the OCI pack registry in the Palette system console.

  4. If your workload depends on non-FIPS content, enable the tenant setting that allows non-FIPS packs before continuing.

Use Deploy Profile Bundles in Air-Gapped Environments for the detailed pack-identification, upload, and sync workflow.

Create Credentials and Tokens​

In Palette, create the credentials you need for the remaining setup:

  1. Create a Palette API key.

  2. Create a registration token for edge host registration if you plan to provision edge hosts for Compute resources.

Register Compute Hosts​

Provision or register the hosts that PaletteAI will discover as Compute resources.

You can use your existing edge host workflow, or follow the guidance in Integrate with Palette for deployment prerequisites such as:

  • registering edge nodes
  • labeling nodes for discovery
  • creating cluster profiles

Complete PaletteAI Setup​

Sign In to PaletteAI​

  1. Open the PaletteAI URL on the appliance.

  2. Sign in with the user or admin credentials assigned to your environment.

  3. Open the tenant and project that you plan to use for imports and deployments.

Configure the Palette Integration​

Configure a Settings resource that points PaletteAI to the correct Palette tenant, Palette project, and API key.

  1. In the PaletteAI UI, go to the project settings area.

  2. Create or update the project's Settings reference.

  3. Provide the Palette host URL, tenant, project identifier, and API key.

  4. Verify that discovered Compute resources appear in PaletteAI after the integration is configured and your hosts are registered.

For the underlying integration requirements, refer to Integrate with Palette and Settings.

Import Your First Profile Bundle​

Connected Workflow​

For connected environments:

  1. Download the profile bundle archive from Artifact Studio.

  2. Import the bundle into your project namespace by following Import Profile Bundles.

  3. Verify that the imported bundle reaches Ready in PaletteAI.

Air-Gapped Workflow​

For air-gapped environments, repeat the following steps for each profile bundle you wish to import:

  1. Download the profile bundle archive from Artifact Studio.

  2. Prepare any required Palette pack content before you import the bundle.

  3. Import the bundle into the hub cluster by following Import Profile Bundles.

  4. Mirror the required Helm chart and images for the profile bundle into the internal Zot registry.

  5. Complete any required project admission, variable, and CA-secret preparation.

  6. Start the deployment from the PaletteAI UI or apply the corresponding manifests.

Use Deploy Profile Bundles in Air-Gapped Environments for the full step-by-step air-gapped procedure.

Next Steps​

After the environment is ready:

  1. Create or import additional profile bundles.

  2. Create an App Deployment or Model Deployment.

  3. Create a dedicated or shared Compute Pool as needed.

  4. Review Create and Manage Projects for project-level settings, access control, and deployment defaults.