Configuration

Environment variables and configuration options for the TFG-Chatbot project.

---

Environment Variables

All configuration is done through environment variables, typically stored in a .env file.

LLM Configuration

Variable	Default	Description
LLM_PROVIDER	gemini	LLM provider: gemini, mistral, or vllm
GOOGLE_API_KEY	-	Gemini API key (required if provider=gemini)
MISTRAL_API_KEY	-	Mistral API key (required if provider=mistral)
VLLM_API_URL	http://localhost:8085/v1	vLLM endpoint URL
LLM_MODEL	gemini-2.0-flash	Model name for inference
LLM_TEMPERATURE	0.7	Generation temperature
LLM_MAX_TOKENS	2048	Maximum output tokens

Database Configuration

Variable	Default	Description
MONGO_HOSTNAME	localhost	MongoDB hostname
MONGO_PORT	27017	MongoDB port
MONGO_ROOT_USERNAME	root	MongoDB admin username
MONGO_ROOT_PASSWORD	example	MongoDB admin password
MONGO_DATABASE	tfg_chatbot	Database name

Vector Store Configuration

Variable	Default	Description
QDRANT_URL	http://localhost:6333	Qdrant server URL
QDRANT_COLLECTION	documents	Default collection name
OLLAMA_URL	http://localhost:11434	Ollama server URL
EMBEDDING_MODEL	nomic-embed-text	Embedding model name
EMBEDDING_DIMENSION	768	Vector dimension

Service URLs

Variable	Default	Description
CHATBOT_SERVICE_URL	http://localhost:8080	Chatbot service URL
RAG_SERVICE_URL	http://localhost:8081	RAG service URL
BACKEND_URL	http://localhost:8000	Backend gateway URL

Inside containers, use service names instead of localhost (e.g., http://chatbot:8080).

Security

Variable	Default	Description
SECRET_KEY	-	JWT signing key (min 32 chars)
ACCESS_TOKEN_EXPIRE_MINUTES	1440	JWT token expiration (24h)
ALGORITHM	HS256	JWT algorithm

---

Example .env File

# ============================================================================
# TFG-Chatbot Environment Configuration
# ============================================================================

# LLM Provider Configuration
LLM_PROVIDER=gemini
GOOGLE_API_KEY=your-google-api-key-here
LLM_MODEL=gemini-2.0-flash

# MongoDB
MONGO_ROOT_USERNAME=root
MONGO_ROOT_PASSWORD=example
MONGO_HOSTNAME=localhost
MONGO_PORT=27017
MONGO_DATABASE=tfg_chatbot

# Vector Store
QDRANT_URL=http://localhost:6333
OLLAMA_URL=http://localhost:11434
EMBEDDING_MODEL=nomic-embed-text

# Service URLs (local development)
CHATBOT_SERVICE_URL=http://localhost:8080
RAG_SERVICE_URL=http://localhost:8081

# Security
SECRET_KEY=your-super-secret-key-at-least-32-characters-long

# Logging
LOG_LEVEL=INFO
LOG_FORMAT=json

# Development
DEBUG=false

---

Service-Specific Configuration

Each service has its own config.py using Pydantic Settings:

Backend Configuration

# backend/config.py
class Settings:
    mongo_hostname: str = "localhost"
    mongo_port: int = 27017
    mongo_root_username: str = "root"
    mongo_root_password: str = "example"
    mongo_database: str = "tfg_chatbot"
    
    secret_key: str
    access_token_expire_minutes: int = 1440
    
    chatbot_service_url: str = "http://localhost:8080"

Chatbot Configuration

# chatbot/config.py
from pydantic_settings import BaseSettings

class Settings(BaseSettings):
    llm_provider: str = "gemini"
    google_api_key: str | None = None
    mistral_api_key: str | None = None
    vllm_api_url: str = "http://localhost:8085/v1"
    
    rag_service_url: str = "http://localhost:8081"
    
    class Config:
        env_file = ".env"

RAG Service Configuration

# rag_service/config.py
from pydantic_settings import BaseSettings

class Settings(BaseSettings):
    qdrant_url: str = "http://localhost:6333"
    ollama_url: str = "http://localhost:11434"
    embedding_model: str = "nomic-embed-text"
    embedding_dimension: int = 768
    
    class Config:
        env_file = ".env"

---

Docker Compose Configuration

The docker-compose.yml sets environment variables for containers:

services:
  backend:
    environment:
      - MONGO_HOSTNAME=mongo
      - CHATBOT_SERVICE_URL=http://chatbot:8080
      
  chatbot:
    environment:
      - RAG_SERVICE_URL=http://rag_service:8081
      - LLM_PROVIDER=${LLM_PROVIDER:-gemini}
      - GOOGLE_API_KEY=${GOOGLE_API_KEY}

Container hostnames use service names (mongo, chatbot, rag_service) instead of localhost.

---

Configuration Precedence

Settings are loaded in this order (later overrides earlier):

1. Default values in config.py
2. .env file in project root
3. Environment variables set in shell
4. Docker Compose environment section

---

Logging Configuration

Log Levels

Level	Description
DEBUG	Detailed diagnostic information
INFO	General operational messages
WARNING	Warning messages
ERROR	Error messages
CRITICAL	Critical errors

JSON Logging

Set LOG_FORMAT=json for structured logging:

{
  "timestamp": "2024-01-31T12:00:00Z",
  "level": "INFO",
  "message": "Request processed",
  "service": "backend",
  "request_id": "abc123"
}

---

Production Configuration

For production deployments:

# Production .env example
LLM_PROVIDER=vllm
VLLM_API_URL=http://vllm-server:8085/v1

MONGO_ROOT_PASSWORD=strong-production-password
SECRET_KEY=production-secret-key-64-characters-minimum

LOG_LEVEL=WARNING
LOG_FORMAT=json
DEBUG=false

Never commit .env files with real credentials to version control.