pulumi-self-hosted-installers

What the repository provides

This repository collects installation guides for deploying the Self-Hosted Pulumi Service into a range of target environments. The Pulumi Service is the control plane that stores state and coordinates runs for infrastructure defined with the Pulumi SDK, and self hosting lets an organization keep that service inside its own cloud accounts. Each guide in the repository explains how to deploy the supporting cloud infrastructure on which the service runs, and then how to deploy the container images that make up the service itself. The README points readers to the separate Self-Hosted Pulumi Service documentation for the component architecture and for general operating guidance, so this repository is the practical entry point rather than the full reference manual. The split keeps the deployment recipes close to the infrastructure they target while the conceptual material lives in the main docs site. Organizations evaluating self hosting can start here to see which environment matches their existing footprint before reading deeper operational detail. By keeping the recipes in one repository, Pulumi gives operators a consistent starting point across clouds, and the guides reduce the risk of omitting a required component during a first install. Teams that already use one of the supported clouds can therefore start from a known good configuration instead of assembling the pieces from scattered docs.

Supported deployment targets

The guides currently cover several common platforms. A Quickstart uses Docker Compose for a fast local or single host deployment. AWS is covered through both EKS and ECS hosted options. Azure is covered through AKS. A plain Docker Engine path is provided for local Docker installs. Google Cloud is covered through GKE. The README also lists VMware as coming soon, which tells readers that an on premises virtualization path is planned but not yet shipped. Each platform directory in the repository carries its own guidance, and the root covers repository wide concerns. This breadth means a team can pick the environment it already operates and follow a path written for that provider instead of adapting a generic tutorial. The README closes by pointing back to the main self hosted documentation for organizations that want to understand capacity, backups, and upgrades before committing to a specific target. The VMware entry is listed as coming soon rather than available, which sets an honest expectation for teams on premises who should not assume the guide exists yet. Each guide follows the target provider's own conventions, so an AWS user follows EKS or ECS, an Azure user follows AKS, and a Google Cloud user follows GKE, keeping instructions aligned with ordinary platform operation.

Contribution and review workflow

The repository uses AGENTS.md files to capture review guidance. The root AGENTS.md covers concerns that apply across the whole repository, and each platform directory such as eks-hosted, ecs-hosted, aks-hosted, gke-hosted, and components-microstacks carries its own file with a platform specific Code Review Guidance section. The README states that AI reviewers and coding agents pick these files up automatically, while human reviewers can use the same checklists. This design makes the review expectations explicit and machine readable, which is useful for a repository whose content is deployment configuration and scripts where a small mistake can be costly. Contributors working on a platform guide therefore have a documented standard to meet before a change is merged. The approach also lowers the barrier for outside contributors, because the expectations are stated in the repo rather than only in a reviewer's head. The repository complements the upstream Pulumi product and is maintained alongside the company's official self hosted documentation. The AGENTS.md convention also lets automated coding agents produce changes that match the project's expectations, which matters because the guides are infrastructure definitions where a subtle error can break a deployment. Contributors therefore get a checklist instead of an unwritten standard. This makes the repository easier to contribute to for first time authors who might otherwise guess at the expected structure.

Editorial conclusion

The repository is published under the Apache-2.0 license and its most recent commit was on 2026-08-26.