Skip to main content

Deploy with an agent

Copy agent prompt

Before you start

Connect your AWS account and choose a project. Review the upstream deployment instructions for the release you plan to pin.

Data dependencies

MySQL for the documented MySQL layout, a document index, object storage and any cache or queue services required by the pinned release. See SQL databases for private writer access and connection details.

Deployment order

  1. Pin a release and use its Compose profiles to list required services. For the MySQL layout, create a compatible private MySQL database first. Wait for the writer and create the application database and role if required.
  2. Set the mysql host, port, name, user and password inputs for service_conf.yaml.template as defined by that release. Replace Compose hostnames with actual private endpoints when services run separately.
  3. Start the selected document engine and durable object storage. Start the cache, queue and any other enabled services before the RAGFlow processes that use them. Keep data services private.
  4. For the full Compose layout, use a persistent container host. If you split the HTTP service onto Spring Winter, verify private reachability, startup commands and all required companion processes first. Wait for schema initialization before serving requests.
Use Runtime variables for runtime credentials. For a Spring Winter web server, set an app-supported health path that returns HTTP 200 without credentials. Check History before running the application check.

Verify the deployment

Upload a non-sensitive document, wait for indexing and confirm a query retrieves a relevant passage.