Flexible GraphRAG Docker Deployment

This directory contains Docker Compose configuration for Flexible GraphRAG with modular database selection.

📖 Documentation Overview

This is a quick reference for Docker deployment. For detailed information, see:

../README.md - Main documentation with deployment scenarios (A, B, C, D, E) and complete setup instructions
DOCKER-ENV-SETUP.md - Environment configuration guide explaining the two-layer .env system
docker.env - Docker networking overrides with SCENARIO A and SCENARIO B configurations
docker-env-sample.txt - Template for docker.env with all database options
../docs/PORT-MAPPINGS.md - Complete port reference for all services
../docs/DEFAULT-USERNAMES-PASSWORDS.md - Database credentials and dashboard access

🏗️ Directory Structure

docker/
├── docker-compose.yaml         # Main compose file with modular includes
├── docker.env                  # Docker networking overrides (SCENARIO A/B)
├── docker-env-sample.txt       # Template for docker.env
├── neptune.env                 # Neptune Graph Explorer credentials (optional)
├── neptune-env-sample.txt      # Template for Neptune credentials
├── DOCKER-ENV-SETUP.md         # Environment configuration guide
├── README.md                   # This file
├── config/                     # Configuration files
│   └── kibana.yml              # Kibana configuration
├── lancedb-info/               # LanceDB info dashboard
│   └── index.html              # Static info page
├── nginx/                      # NGINX proxy configuration
│   └── nginx.conf              # Reverse proxy rules
├── pinecone-info/              # Pinecone info dashboard
│   └── index.html              # Static info page
├── postgres-init/              # PostgreSQL initialization scripts
│   └── 01-init-pgvector.sql    # pgvector extension setup
└── includes/                   # Modular service configurations
    ├── commons/                # Common/shared configurations
    │   └── base.yaml           # Base service definitions
    ├── neo4j.yaml              # Neo4j graph database
    ├── ladybug-explorer.yaml   # Ladybug Explorer web UI (optional; build image locally)
    ├── falkordb.yaml           # FalkorDB graph database
    ├── arcadedb.yaml           # ArcadeDB multi-model database
    ├── memgraph.yaml           # MemGraph graph database
    ├── nebula.yaml             # NebulaGraph distributed graph database
    ├── neptune.yaml            # Amazon Neptune Graph Explorer
    ├── qdrant.yaml             # Qdrant vector database
    ├── chroma.yaml             # Chroma vector database
    ├── milvus.yaml             # Milvus vector database
    ├── weaviate.yaml           # Weaviate vector database
    ├── pinecone.yaml           # Pinecone info dashboard
    ├── postgres-pgvector.yaml  # PostgreSQL with pgvector extension
    ├── lancedb.yaml            # LanceDB embedded vector database
    ├── elasticsearch-dev.yaml  # Elasticsearch (security disabled)
    ├── kibana-simple.yaml      # Kibana dashboard
    ├── opensearch.yaml         # OpenSearch search engine + dashboards
    ├── alfresco.yaml           # Alfresco Community (full stack)
    ├── app-stack.yaml          # Flexible GraphRAG backend + UIs
    └── proxy.yaml              # NGINX reverse proxy

🚀 Quick Start

1. Configure Environment Files

Main configuration (required):

# Navigate to backend directory
cd ../flexible-graphrag

# Windows
copy env-sample.txt .env

# macOS/Linux  
cp env-sample.txt .env

# Edit .env with your LLM provider, API keys, and database passwords
# Return to project root
cd ..

Docker overrides (required for Docker deployment):

# Navigate to docker directory
cd docker

# Windows
copy docker-env-sample.txt docker.env

# macOS/Linux
cp docker-env-sample.txt docker.env

# See DOCKER-ENV-SETUP.md for details on the two-layer configuration system

2. Start Services

From the docker directory:

# Start with project name
docker-compose -f docker-compose.yaml -p flexible-graphrag up -d

Common operations:

# Stop services
docker-compose -f docker-compose.yaml -p flexible-graphrag down

# Stop and remove volumes (⚠️ deletes all data)
docker-compose -f docker-compose.yaml -p flexible-graphrag down -v

# View logs
docker-compose -f docker-compose.yaml -p flexible-graphrag logs -f

# Check status
docker-compose -f docker-compose.yaml -p flexible-graphrag ps

See ../README.md for complete deployment scenario details (A, B, C, D, E).

🔧 Selective Database Deployment

The main deployment scenarios (covered in detail in ../README.md):

SCENARIO A (Default): Databases in Docker, app standalone - for development

Uncomment: neo4j, qdrant, elasticsearch-dev, kibana-simple
Comment out: app-stack, proxy
Use host.docker.internal in docker.env

SCENARIO B: Full stack in Docker - for production

Uncomment: neo4j, qdrant, elasticsearch-dev, kibana-simple, app-stack, proxy
Use service names in docker.env

General principle: Comment/uncomment services in docker-compose.yaml based on your needs. Only include:

The graph, vector, and search databases you're actually using
app-stack and proxy only if running backend/UIs in Docker (SCENARIO B)

For alternative databases, uncomment the appropriate include file:

Graph databases: ladybug-explorer, falkordb, arcadedb, memgraph, nebula, neptune
Vector databases: chroma, milvus, weaviate, pinecone, postgres-pgvector, lancedb
Search engines: opensearch (alternative to Elasticsearch)
Content sources: alfresco

🔄 Data Persistence

All database data is stored in named Docker volumes that persist across container restarts:

Graph databases:

neo4j_data - Neo4j graph data
falkordb_data - FalkorDB data
arcadedb_data - ArcadeDB data
memgraph_data - MemGraph data
nebula_data - NebulaGraph data

Vector databases:

qdrant_data - Qdrant vector data
chroma_data - Chroma vector data
milvus_data - Milvus vector data
weaviate_data - Weaviate vector data
postgres_data - PostgreSQL pgvector data
lancedb_data - LanceDB data

Search engines:

elasticsearch_data - Elasticsearch indices
opensearch_data - OpenSearch indices

Content management:

alfresco_data - Alfresco content store
alfresco_db_data - Alfresco PostgreSQL data

🔌 External LLM Access

For external Ollama access from Docker containers:

# Make sure Ollama is running on host
ollama serve

# Configure in docker.env (already configured by default)
OLLAMA_BASE_URL=http://host.docker.internal:11434

Note: Ollama typically runs on the host machine, not in a Docker container.

🛠️ Development

Building Custom Images

# Backend only
docker-compose -f docker-compose.yaml build flexible-graphrag-backend

# Specific UI
docker-compose -f docker-compose.yaml build flexible-graphrag-ui-react

# All services
docker-compose -f docker-compose.yaml build

Logs and Debugging

# All services
docker-compose -f docker-compose.yaml -p flexible-graphrag logs -f

# Specific service
docker-compose -f docker-compose.yaml -p flexible-graphrag logs -f flexible-graphrag-backend

# Database logs
docker-compose -f docker-compose.yaml -p flexible-graphrag logs -f neo4j

Health Checks

All services include health checks. Check status:

docker-compose -f docker-compose.yaml -p flexible-graphrag ps

🧹 Cleanup

# Stop services
docker-compose -f docker-compose.yaml -p flexible-graphrag down

# Remove volumes (⚠️ deletes all data)
docker-compose -f docker-compose.yaml -p flexible-graphrag down -v

# Remove images
docker-compose -f docker-compose.yaml -p flexible-graphrag down --rmi all

🚨 Troubleshooting

Common Issues

Port conflicts: Check ../docs/PORT-MAPPINGS.md and modify port mappings in individual YAML files if needed
Memory issues: Increase Docker memory limits (Settings → Resources) or reduce the number of services
Permission errors: Check volume mount permissions, especially on Linux
Network connectivity: Services use the default network flexible-graphrag_default
Connection refused: Ensure docker/docker.env exists and uses correct addresses for your scenario

Resource Requirements

Minimum requirements (basic setup - Neo4j + Qdrant + Elasticsearch):

RAM: 8GB
Disk: 20GB free space
Docker: 20.10+, Compose V2

Recommended for full stack (multiple databases + backend + UIs):

RAM: 16GB+
Disk: 50GB+ free space
CPU: 4+ cores

Note: Running multiple vector or graph databases simultaneously increases resource requirements proportionally.

📚 MCP Server Integration

The Docker setup works seamlessly with the MCP server for Claude Desktop integration, other MCP clients, etc.

Install MCP server: uvx flexible-graphrag-mcp
Configure to point to:
- SCENARIO A (standalone): http://localhost:8000
- SCENARIO B (Docker): http://localhost:8070/api/
All database services are accessible through the unified API

See ../flexible-graphrag-mcp/README.md for complete MCP server documentation.

📖 Additional Resources

Main README: ../README.md - Complete deployment scenarios and setup
Environment Setup: DOCKER-ENV-SETUP.md - Two-layer configuration system
Port Mappings: ../docs/PORT-MAPPINGS.md - All service ports
Credentials: ../docs/DEFAULT-USERNAMES-PASSWORDS.md - Default passwords
Deployment Options: ../docs/DEPLOYMENT-CONFIGURATIONS.md - Detailed scenarios
Architecture: ../docs/ARCHITECTURE.md - System architecture overview

This modular approach allows you to run exactly what you need while maintaining easy access to all Flexible GraphRAG features.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Flexible GraphRAG Docker Deployment

📖 Documentation Overview

🏗️ Directory Structure

🚀 Quick Start

1. Configure Environment Files

2. Start Services

🔧 Selective Database Deployment

🔄 Data Persistence

🔌 External LLM Access

🛠️ Development

Building Custom Images

Logs and Debugging

Health Checks

🧹 Cleanup

🚨 Troubleshooting

Common Issues

Resource Requirements

📚 MCP Server Integration

📖 Additional Resources

FilesExpand file tree

README.md

Latest commit

History

README.md

File metadata and controls

Flexible GraphRAG Docker Deployment

📖 Documentation Overview

🏗️ Directory Structure

🚀 Quick Start

1. Configure Environment Files

2. Start Services

🔧 Selective Database Deployment

🔄 Data Persistence

🔌 External LLM Access

🛠️ Development

Building Custom Images

Logs and Debugging

Health Checks

🧹 Cleanup

🚨 Troubleshooting

Common Issues

Resource Requirements

📚 MCP Server Integration

📖 Additional Resources