Principle:Infiniflow Ragflow LLM Provider Initialization
| Knowledge Sources | |
|---|---|
| Domains | RAG, NLP, Configuration_Management |
| Last Updated | 2026-02-12 06:00 GMT |
Overview
A system initialization pattern that bootstraps all backend services including document store connections, storage backends, LLM model factories, and retrieval engines.
Description
LLM Provider Initialization via settings.init_settings() bootstraps the entire backend subsystem. It loads service_conf.yaml, initializes the document store connection (Elasticsearch/Infinity/OpenSearch/OceanBase), storage backend (MinIO/S3/Azure/GCS/OSS), the Dealer retrieval engine, and the LLM factory catalog (66+ providers from llm_factories.json).
Usage
Called automatically during server startup. Configuration is via service_conf.yaml and the UI settings page.
Theoretical Basis
Centralized initialization ensures consistent global state:
- Single initialization point: All connections established once at startup
- Factory pattern: LLM models resolved by provider name and model ID
- Backend abstraction: Document store and storage backends selected by configuration
Related Pages
Implemented By
Page Connections
Double-click a node to navigate. Hold to expand connections.
Principle
Implementation
Heuristic
Environment