Deploy and Host Open Glean on Railway

Open Glean is a self-hosted AI workspace for knowledge work. It connects to Hydra DB, retrieves relevant memories and files, and produces cited answers through a modern browser interface. Deep Research, collections, context management, integrations, mindmaps, and OpenAI-compatible model support make it a practical second brain for teams and individuals.

About Hosting Open Glean on Railway

Railway runs the official Open Glean production container as a single HTTP service with automatic HTTPS, a generated public domain, environment variables, logs, and managed deployments. The template uses a pre-built Docker image and does not perform a source-code or Railpack build at deployment time.

Tech Stack

Next.js 16 standalone server
React 19
Tailwind CSS v4
Hydra DB SDK for retrieval and integrations
OpenAI-compatible LLM providers, including OpenRouter
Optional MongoDB or DocumentDB-compatible chat persistence
Docker on Railway

Why Deploy Open Glean on Railway

Railway provides a fast one-click path to a private, self-hosted AI workspace with HTTPS and simple configuration. The service is ready to run immediately, while optional Hydra, LLM, and MongoDB settings can be added as environment variables without changing the image. Logs, redeploys, and runtime configuration are available from one project dashboard.

Common Use Cases

Search and ask questions across Hydra DB memories and files
Run cited Deep Research workflows
Host a team knowledge workspace behind a Railway domain
Connect Hydra integrations and browse collections
Use OpenRouter or another OpenAI-compatible model for answer synthesis
Run a lightweight deployment with browser-local chat fallback when MongoDB is not configured

Deployment Notes

The service listens on container port 3000 and Railway routes the generated domain to that port through PORT=3000. The image's own node server.js command is preserved; no custom start command is required. OPEN_GLEAN_SESSION_SECRET is generated automatically by the template and is required for production session-cookie encryption. HYDRA_BASE_URL defaults to https://api.hydradb.com. Add HYDRA_API_KEY for a deployment-level Hydra key, or users can enter their key in Settings. Add OPENROUTER_API_KEY and OPEN_GLEAN_LLM_MODEL to enable server-level answer synthesis. MONGODB_URI and optional MONGODB_PROXY_KEY enable chat persistence; without them, the app uses browser localStorage. No volume is mounted by default because the default persistence mode is browser-local and the container image is non-root.

Template Variables

Variable	Default	Purpose
`PORT`	`3000`	Railway HTTP routing port
`OPEN_GLEAN_SESSION_SECRET`	`${{secret(32)}}`	Encrypts session cookies and signs browser subjects
`HYDRA_BASE_URL`	`https://api.hydradb.com`	Hydra DB API endpoint
`MONGODB_DB`	`open_glean`	MongoDB database name when persistence is configured
`OPEN_GLEAN_MAX_CONCURRENT_RESEARCH`	`3`	Concurrent Deep Research runs
`HYDRA_API_KEY`	unset	Optional shared Hydra DB API key
`OPENROUTER_API_KEY`	unset	Optional server-level LLM API key
`OPEN_GLEAN_LLM_MODEL`	unset	Optional default model identifier
`MONGODB_URI`	unset	Optional MongoDB or DocumentDB-compatible URI
`MONGODB_PROXY_KEY`	unset	Optional Lambda proxy shared secret

Dependencies for Open Glean on Railway

Open Glean is deployed as one web service. Hydra DB and an LLM provider are external integrations configured by environment variables; MongoDB is optional for server-side chat persistence. No additional Railway service is required for the default browser-local persistence mode.

Deployment Dependencies

Service	Image	Port	Volume
open-glean	xiaosong233/open-glean-railway:latest	3000	-
Hydra DB	external `https://api.hydradb.com`	443	-
MongoDB	external, optional	27017	-