Skip to content

Environment Configuration Guide

This document explains how to configure Flexible GraphRAG using environment variables and configuration files.

📁 Configuration Files

Primary Configuration

  • .env - Your main configuration file (copy from flexible-graphrag/env-sample.txt)
  • flexible-graphrag/env-sample.txt - Template with all options and examples

Frontend Configuration

  • Angular: Uses PROCESS_FOLDER_PATH
  • React/Vue: Uses VITE_PROCESS_FOLDER_PATH
  • See docs/DATA-SOURCES/SOURCE-PATH-EXAMPLES.md for details

🏗️ 5-Section Configuration Structure

The env-sample.txt follows a logical 5-section structure:

Section 1: Graph Database Configuration

  • Graph database selection (GRAPH_DB)
  • Knowledge graph extraction settings (ENABLE_KNOWLEDGE_GRAPH)
  • Schema configuration (SCHEMA_NAME, SCHEMAS)
  • Graph database connection configs (GRAPH_DB_CONFIG)

Section 2: Vector Database Configuration

  • Vector database selection (VECTOR_DB)
  • Vector database connection configs (VECTOR_DB_CONFIG)
  • Index/collection names for vector storage

Section 3: Search Database Configuration

  • Search database selection (SEARCH_DB)
  • Search database connection configs (SEARCH_DB_CONFIG)
  • Index names for fulltext search

Section 4: LLM Configuration

  • LLM provider selection (LLM_PROVIDER)
  • Provider-specific settings (OpenAI, Ollama, Azure, etc.)
  • API keys and model configurations
  • Timeout settings

Section 5: Content Sources Configuration

  • CMIS and Alfresco settings for document sources
  • Authentication credentials

🔧 Database Configuration Patterns

Selection Variables

These control which backend to use: 

VECTOR_DB=neo4j        # neo4j, qdrant, elasticsearch, opensearch, none
SEARCH_DB=elasticsearch # elasticsearch, opensearch, bm25, none  
GRAPH_DB=neo4j         # neo4j, ladybug, falkordb, arcadedb, memgraph, nebula, neptune, neptune_analytics, none

Connection Configuration

Each database type has a *_DB_CONFIG JSON configuration: 

# Vector database connections
VECTOR_DB_CONFIG={"host": "localhost", "port": 6333, "collection_name": "hybrid_search_vector"}

# Search database connections  
SEARCH_DB_CONFIG={"url": "http://localhost:9200", "index_name": "hybrid_search_fulltext"}

# Graph database connections
GRAPH_DB_CONFIG={"url": "bolt://localhost:7687", "username": "neo4j", "password": "password"}

Individual Database Settings

Traditional environment variables for specific databases: 

# Neo4j
NEO4J_URI=bolt://localhost:7687
NEO4J_USER=neo4j
NEO4J_PASSWORD=password

# Elasticsearch
ELASTICSEARCH_URL=http://localhost:9200
ELASTICSEARCH_USERNAME=
ELASTICSEARCH_PASSWORD=

# Qdrant
QDRANT_HOST=localhost
QDRANT_PORT=6333
QDRANT_API_KEY=

🎯 Easy Database Switching

The configuration is designed for easy switching:

# Current setup: OpenAI + Qdrant + Elasticsearch + Neo4j
LLM_PROVIDER=openai
VECTOR_DB=qdrant
SEARCH_DB=elasticsearch  
GRAPH_DB=neo4j

# Switch to: Ollama + Neo4j (all-in-one)
LLM_PROVIDER=ollama
VECTOR_DB=neo4j
SEARCH_DB=elasticsearch
GRAPH_DB=neo4j

# Switch to: OpenAI + Ladybug + OpenSearch
LLM_PROVIDER=openai  
VECTOR_DB=opensearch
SEARCH_DB=opensearch
GRAPH_DB=ladybug

📝 Configuration Best Practices

Development Setup

  1. Start simple: Use Neo4j for both vector and graph storage
  2. Use defaults: Copy env-sample.txt to .env and adjust API keys
  3. Test locally: Use localhost connections before cloud deployment

Production Setup

  1. Separate concerns: Use specialized databases (Qdrant for vectors, Elasticsearch for search)
  2. Secure connections: Use proper authentication and HTTPS where supported
  3. Performance tuning: Adjust timeout values and batch sizes

Schema Configuration

  1. Start with default: Use SCHEMA_NAME=default for comprehensive extraction
  2. Customize gradually: Create domain-specific schemas as needed
  3. Test thoroughly: Compare different schema approaches on your content
  • Source paths: docs/DATA-SOURCES/SOURCE-PATH-EXAMPLES.md
  • Schema examples: docs/CONFIGURATION/SCHEMA-EXAMPLES.md
  • Timeout configuration: docs/ADVANCED/TIMEOUT-CONFIGURATIONS.md
  • Neo4j URLs: docs/DATABASES/GRAPH-DATABASES/Neo4j-URLs.md
  • Vector dimensions: docs/DATABASES/VECTOR-DATABASES/VECTOR-DIMENSIONS.md

🚀 Quick Start

  1. Copy template: cp flexible-graphrag/env-sample.txt .env
  2. Set API key: Add your OpenAI API key to OPENAI_API_KEY
  3. Choose databases: Uncomment your preferred database options
  4. Update connections: Modify database URLs/credentials as needed
  5. Test configuration: Run a small test to verify everything works

The modular 5-section structure makes it easy to understand and modify any part of the configuration without affecting others.