Production
Guidance for running the engine as a production workload. This page grows with the self-hosting docs; deployment topology is the load-bearing decision today.
Deployment topology
- Local: engine and nodes on the same machine. Zero network latency between nodes; API calls to LLM/vector store providers cross the internet normally.
- Self-hosted: engine in Docker or Kubernetes, co-located with vector store. Reduces vector store latency significantly.
- Cloud: managed engine. Best for production — automatic scaling, managed auth, no infrastructure to operate.
For high-throughput workloads, run the vector store and embedding service on the same network as the engine to minimise round-trip latency on the embedding and upsert steps.
Parameter token key
When the engine validates or saves a pipeline, it encrypts each node's secure
parameters into a token (secureParameters.token in the node's config). Set
RR_PARAM_TOKEN_SECRET in the engine's environment to give your deployment its
own key for these tokens:
- New tokens are encrypted with AES-256-GCM under a key the engine derives from the value (HKDF-SHA256), and are marked with a format version. A token that was changed after it was written fails to load with an error; it never decrypts into other values.
- Tokens saved before you set it still load. They move to the new key the next time the engine encrypts that pipeline's parameters, for example when the pipeline is validated and saved again.
- Use a random value of at least 32 bytes, for example
python -c "import secrets; print(secrets.token_urlsafe(32))". Store it the way you store provider credentials. - Give every engine replica the same value and keep it. A token written with one value cannot be read without it, so pipelines saved under a lost or changed value need their secure parameters entered again.
- Pipeline tasks never receive it. The engine decodes a pipeline's tokens before it starts the task.
Without the variable, tokens use the original format, which every engine reads.
Public pages
An engine that serves a public site builds its page map from the
publicPages of published platform apps (see
Public pages).
| Variable | Effect |
|---|---|
RR_APP_URL | The public origin, such as https://cloud.rocketride.ai: canonical links, absolute sitemap and feed URLs, and a robots.txt that allows crawling. Set it to make published pages indexable. Unset, robots.txt disallows all crawling and the sitemap and llms.txt answer 404, which suits local and staging engines; an engine that serves published pages without it logs a warning once. |
RR_PUBLIC_ROOT_APPS | The app ids allowed to declare pages, separated by commas. Default rocketride.home. An id outside the rocketride developer namespace never may, even when it is listed. |
RR_INDEXNOW_KEY | The IndexNow key: 8 to 128 letters, digits and -. With an https:// RR_APP_URL, the engine serves it at /<key>.txt and, whenever the page map changes, submits the listed pages that changed to IndexNow. Unset or malformed, it serves and submits nothing. |
Related
- Self-hosting overview: install and run the engine.
- Security: credentials, ports, and TLS before you expose anything.
- Performance: tuning levers inside the pipeline.