docs / Adopter Guide
Audiences: adopter (deploy)
This guide covers configuring, deploying, and managing ML Deployment Toolkit environments. You configure two files per environment (config.yaml and .env), run Terraform via Make, and get a Kubernetes cluster with FluxCD reconciling platform services.
You never fork or edit the platform bundle -- all personalization flows through your local configuration.
For architecture details and design rationale, see the Architecture section. This guide focuses on how to deploy and manage.
| Page | Description |
|---|---|
| Prerequisites | Required tools, provider accounts, credentials |
| Provider setup | How to provision provider accounts, tokens, and DNS zones |
| Configuration | Environment config, secrets, named environments, OCI settings |
| Deployment | Shared deployment workflow (init, plan/apply, commands, destroy) |
| Deploy a Tooling Cluster (CC) | CC-specific config, verification, accessing services |
| Deploy a Switch (SW) | SW-specific config, verification, Harbor proxy cache |
| Upgrading | Upgrading platform services and infrastructure |
| Known issues | Deployment-level known issues and workarounds |
-
Clone the repository
git clone <repo-url> && cd ml-deployment-toolkit
-
Create your environment configuration
Copy the sample environment and fill in your values. See Configuration for details on each section.
cp config/environments/ml-cc/.env.sample config/environments/myenv/.env # Edit config/environments/myenv/config.yaml and .env -
Deploy
make init ENV=myenv make plan-apply ENV=myenv
-
Publish the GitOps artifact (if using OCI-based Flux)
make push-gitops ENV=myenv
Each step is covered in depth in the pages linked above. Start with Prerequisites if this is your first deployment.