docker-compose-ollama.yml

################################################################
# Ollama LLM Server and Web UI - Production Configuration
################################################################
#
# PRODUCTION SETUP NOTES:
#
# 1. SECURITY - Update these settings before deploying:
#    - Change WEBUI_ADMIN_USER and WEBUI_ADMIN_PASS to strong credentials
#    - Consider using Docker secrets or environment files for sensitive data
#    - Use a reverse proxy (nginx/traefik) with SSL/TLS in front
#    - Restrict port access using firewall rules (only expose what's needed)
#
# 2. MODELS - Customize for your use case:
#    - Update the preloaded model in command section (default: qwen3:0.6b)
#    - Add multiple models: ollama pull model1 && ollama pull model2
#    - For GPU support, add deploy.resources.reservations.devices config
#
# 3. RESOURCE LIMITS - Add resource constraints:
#    - Set memory limits (e.g., mem_limit: 8g)
#    - Set CPU limits (e.g., cpus: '4.0')
#    - Adjust OLLAMA_NUM_PARALLEL and OLLAMA_MAX_LOADED_MODELS based on hardware
#
# 4. DATA PERSISTENCE:
#    - All data stored in ./.ollama directory
#    - Ensure ./.ollama is backed up regularly
#    - Add .ollama/ to .gitignore to avoid committing model data
#
# 5. NETWORKING:
#    - Consider using a custom network for service isolation
#    - For production, bind to specific IPs instead of 0.0.0.0
#    - Update port mappings if needed (default: 11434 for Ollama, 3000 for WebUI)
#
# 6. MONITORING:
#    - Health checks are configured but consider adding logging drivers
#    - Integrate with monitoring solutions (Prometheus, Grafana, etc.)
#    - Set up alerting for service failures
#
# 7. UPDATES:
#    - Pin specific image versions instead of :latest for stability
#    - Example: ollama/ollama:0.1.26 instead of ollama/ollama:latest
#    - Test updates in staging before deploying to production
#
################################################################

services:
  ################################################################
  # Ollama LLM Server
  ###############################################################
  ollama:
    image: ollama/ollama:0.18.2
    container_name: ollama
    ports:
      - "11434:11434"
    volumes:
      - ./.ollama/server:/root/.ollama
    environment:
      - OLLAMA_NUM_PARALLEL=4
      - OLLAMA_MAX_LOADED_MODELS=3
      - OLLAMA_HOST=0.0.0.0:11434
    entrypoint: ["/bin/bash", "-c"]
    command: |
      "ollama serve &
      sleep 10
      # Preload models
      ollama pull qwen3:0.6b
      ollama pull gemma3
      ollama pull nomic-embed-text
      wait"
    restart: always
    healthcheck:
      test: ["CMD", "ollama", "list"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 40s

  webui:
    image: ghcr.io/open-webui/open-webui:main
    container_name: ollama-webui
    ports:
      - "3000:8080"
    volumes:
      - ./.ollama/webui:/app/backend/data
    depends_on:
      ollama:
        condition: service_healthy
    environment:
      - OLLAMA_BASE_URL=http://ollama:11434
      - WEBUI_AUTH=True
      - WEBUI_EMAIL=admin@boxlang.io
      - WEBUI_PASSWORD=rocks
    restart: always
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:8080/health"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 30s