docker-compose.yaml

version: '3.8'

services:

  # ─────────────────────────────────────────────
  # Frontend — React app served by nginx on :3000
  # nginx also proxies /api/* → backend:8000
  # ─────────────────────────────────────────────
  frontend:
    build:
      context: ./frontend
      dockerfile: Dockerfile
    container_name: political-chatbot-frontend
    ports:
      - "3000:80"
    depends_on:
      - backend
    networks:
      - chatbot-network
    restart: unless-stopped

  # ─────────────────────────────────────────────
  # Backend — FastAPI on :8000
  # ─────────────────────────────────────────────
  backend:
    build:
      context: ./backend
      dockerfile: Dockerfile
    container_name: political-chatbot-backend
    ports:
      - "8000:8000"
    env_file:
      - .env.docker
    environment:
      # Override just the Ollama host so it uses the Docker service name
      OLLAMA_HOST: "http://ollama:11434"
      CORS_ORIGINS: "http://localhost:3000,http://frontend:80"
    depends_on:
      - ollama
    networks:
      - chatbot-network
    restart: unless-stopped
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:8000/health"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 15s

  # ─────────────────────────────────────────────
  # Ollama — local LLM inference on :11434
  # ─────────────────────────────────────────────
  ollama:
    image: ollama/ollama:latest
    container_name: political-chatbot-ollama
    ports:
      - "11434:11434"
    volumes:
      - ollama-models:/root/.ollama
    networks:
      - chatbot-network
    restart: unless-stopped
    # Limit CPU so Ollama doesn't saturate the host during inference.
    # 4.0 = up to 4 logical cores. Raise if responses feel too slow,
    # lower further (e.g. 2.0) if the machine is still struggling.
    deploy:
      resources:
        limits:
          cpus: "4.0"
          memory: 6G
    environment:
      # Keep only 1 model loaded in memory at a time
      OLLAMA_MAX_LOADED_MODELS: "1"
      # Limit parallel inference requests
      OLLAMA_NUM_PARALLEL: "1"
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:11434/api/tags"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 60s

volumes:
  ollama-models:
    driver: local

networks:
  chatbot-network:
    driver: bridge