Documentation
Enroute is a fast, programmable Git server that is easy to operate. It is built like a modern database: compute is separate from storage, and repositories are in an object store. Enroute handles Git over HTTP, Git data, and a gRPC API. Your application owns repository names, users, permissions, and policy.
Choose a starting point
| If you need to… | Read… |
|---|---|
| Run Enroute locally | Quickstart |
| Build a code-hosting application | Guide |
| Understand responsibilities and boundaries | Concepts |
| Assess the current limitations | Limitations |
Guides
Build a code-hosting application is a seven-part TypeScript and Next.js tutorial. It creates an application that can create repositories, serve Git traffic, and display files, history, and diffs. The same API and hook contract work with any stack that supports gRPC and HTTP.
After the tutorial, use Patterns for common product features such as branch protection, merge requests, CI, and mirroring.
Concepts
See Concepts for the data model, interfaces, hooks, and storage design.
Operating Enroute
See Operate Enroute for deployment, configuration, security, storage, cluster care, and observability. To choose between one node, one ledger with compute nodes, and a replicated ledger, see Deployment profiles.
Reference
See Reference for API, hook, protocol, configuration, and limitation details.
Architecture
See Architecture for how Enroute works inside. The design pages are the overview, consensus, ledger and compute, storage, write path, read path, serving, lifecycle, and maintenance. Jepsen tests describes how the suite tests consistency under faults.