Deploy RocketRide
RocketRide AI pipeline engine with pgvector, Milvus, etcd and MinIO
Just deployed
/etcd
RocketRide
Just deployed
/opt/data
Just deployed
/var/lib/postgresql/data
Just deployed
/var/lib/milvus
Just deployed
/minio_data
Deploy and Host RocketRide on Railway
RocketRide is an open-source (MIT) AI pipeline engine with a C++ core and 50+ Python nodes for LLMs, OCR, embeddings, vector stores and agents. You build pipelines in its web UI or VS Code extension and run them from the TypeScript or Python SDKs, the CLI or an MCP client.
About Hosting RocketRide
This template runs RocketRide engine 3.6.0 with the data stores from upstream's Docker stack, each as its own Railway service on the private network: PostgreSQL with pgvector, and Milvus with its etcd metadata store and MinIO object storage. Every service keeps its data on a volume. The engine has a generated API key, and its start script refuses to run without one, because an engine with no key accepts anyone. The deploy form asks for nothing. After deploy, open the RocketRide URL and sign in with ROCKETRIDE_APIKEY from the service's Variables tab. Model-provider keys go on the RocketRide service as ROCKETRIDE_… variables.
Common Use Cases
- Retrieval pipelines that parse, OCR, chunk and embed documents into the bundled pgvector or Milvus store
- Agent workflows that call LLMs, vector search and other tools, run from your app through the SDKs
- Pipelines exposed to AI assistants over MCP or to other services as webhook and chat endpoints
Dependencies for RocketRide Hosting
- RocketRide engine (
ghcr.io/rocketride-org/rocketride-engine:3.6.0, pinned by digest) with web UI and WebSocket API - PostgreSQL with pgvector (Railway Postgres), plus Milvus 3.0.6 with etcd 3.5.36 and MinIO
Deployment Dependencies
- RocketRide source and releases: https://github.com/rocketride-org/rocketride-server/releases/tag/server-v3.6.0
- RocketRide self-hosting docs: https://docs.rocketride.org/operate/self-hosting
- Milvus standalone with etcd and MinIO: https://milvus.io/docs/install_standalone-docker-compose.md
- Template repo (Dockerfiles, start scripts, changelog): https://github.com/aalfath/rocketride-railway-template
Implementation Details
| Service | What it does | Volume |
|---|---|---|
| RocketRide | Engine, web UI and WebSocket API on port 5565 (the only public service) | /opt/data: projects, deployed apps, task data |
| Postgres | Vector store for the PostgreSQL (pgvector) node; the engine enables vector at start | Railway Postgres volume |
| Milvus | Vector database for the Milvus node, gRPC on 19530 | /var/lib/milvus |
| etcd | Milvus metadata store | /etcd |
| MinIO | Milvus object storage, with generated credentials | /minio_data |
Pipelines can read only ROCKETRIDE_-prefixed variables, so the template sets them for you. In a node's config, use ${ROCKETRIDE_POSTGRES_HOST}, ${ROCKETRIDE_POSTGRES_USER}, ${ROCKETRIDE_POSTGRES_PASSWORD} and ${ROCKETRIDE_POSTGRES_DATABASE} for pgvector, or ${ROCKETRIDE_MILVUS_HOST} with port 19530 for Milvus (profile "Your own Milvus server"). SDK and VS Code clients connect to wss://YOUR-DOMAIN with the same API key.
Every service builds from the public template repo, so Railway shows you an update when we bump a version there. Read CHANGELOG.md before you apply it; data stays on the volumes.
Why Deploy RocketRide on Railway?
Railway is a singular platform to deploy your infrastructure stack. Railway will host your infrastructure so you don't have to deal with configuration, while allowing you to vertically and horizontally scale it.
By deploying RocketRide on Railway, you are one step closer to supporting a complete full-stack application with minimal burden. Host your servers, databases, AI agents, and more on Railway.
Template Content

