langraphKG

Overview

This project utilizes a Neo4j database within a Docker container and includes a setup script for installing Miniconda and managing Python dependencies.

Prerequisites

• Docker installed on your system
• NVIDIA GPU and appropriate drivers (the container uses GPU support)
• Sufficient disk space for Neo4j data and Miniconda installation

Docker Container Setup

Container Command

docker run \
    -p 7474:7474 -p 7687:7687 \
    -v $PWD/data:/data \
    -v $PWD/plugins:/plugins \
    -v $PWD:/workspace \
    --name neo4jLangchain \
    -e NEO4J_apoc_export_file_enabled=true \
    -e NEO4J_apoc_import_file_enabled=true \
    -e NEO4J_apoc_import_file_useneo4jconfig=true \
    -itd --gpus all \
    -w /workspace \
    neo4j:5.24.1

Command Breakdown

• Ports:
  • 7474: Neo4j browser interface
  • 7687: Bolt protocol for database access
• Volumes:
  • $PWD/data:/data: Persists Neo4j data
  • $PWD/plugins:/plugins: Custom Neo4j plugins
  • $PWD:/workspace: Mounts current directory as workspace
• Environment Variables:
  • Enables APOC file import/export functionality
• GPU Support:
  • --gpus all: Enables all available GPUs
• Working Directory:
  • -w /workspace: Sets the container's working directory

Python Environment Setup

Setup Script (setup.sh)

#!/bin/bash

# Update system packages
apt-get update && apt-get install -y wget curl && rm -rf /var/lib/apt/lists/*
apt update && apt upgrade -y

# Download and install Miniconda
wget https://repo.anaconda.com/miniconda/Miniconda3-py311_24.7.1-0-Linux-x86_64.sh
bash Miniconda3-py311_24.7.1-0-Linux-x86_64.sh -b

# Initialize conda
. ~/miniconda3/etc/profile.d/conda.sh
conda init

# Cleanup
rm Miniconda3-py311_24.7.1-0-Linux-x86_64.sh
rm ~/miniconda.sh

Python Dependencies

Install required packages using:

pip install -r requirements.txt

Usage

1. Start the Docker container:

   docker start neo4jLangchain

2. Execute the setup script:

   docker exec neo4jLangchain bash setup.sh

3. Install Python dependencies:

   docker exec neo4jLangchain pip install -r requirements.txt

Accessing Neo4j

• Browser interface: http://localhost:7474
• Default credentials (if not changed):
  • Username: neo4j
  • Password: neo4j (you'll be prompted to change on first login)

Best Practices

1. Data Persistence:

   • All Neo4j data is persisted in the ./data directory
   • Back up this directory regularly

2. Plugin Management:

   • Place any Neo4j plugins in the ./plugins directory
   • Restart the container after adding new plugins

3. Python Environment:

   • Use virtual environments for isolation
   • Keep requirements.txt updated

Notes

• This setup uses Neo4j version 5.24.1
• Python version: 3.11 (via Miniconda)
• APOC core functionality is enabled for file operations