deploy/docker/docker-compose.distributed.yml - main - AlexandrNarbaev/rag-system

deploy/docker/docker-compose.distributed.yml
386 строк13 KB
Alexandr Narbaev
RAG System Init
19 июл 2026, 13:55
19 июл 2026, 13:5559d6dd0
# docker-compose.distributed.yml
# Multi-machine distributed deployment for RAG System.
#
# ── Machine 1 (Proxy + Qdrant + Neo4j + Redis + MinIO) ──
#   docker compose -f docker-compose.distributed.yml up -d proxy qdrant neo4j redis minio
#
# ── Machine 2 (OpenWebUI) ──
#   docker compose -f docker-compose.distributed.yml up -d openwebui postgres tika
#
# ── Machine 3 (ETL Worker) ──
#   docker compose -f docker-compose.distributed.yml up -d etl
#
# Prerequisites:
#   - Shared network `rag-network` (Docker Swarm overlay or manually bridged)
#   - Shared volumes for Qdrant, Neo4j, Redis data (NFS or distributed FS)
#   - All machines must be in the same Docker Swarm or have routable IPs
#
# Environment: copy .env.example to .env and configure
#   QDRANT_HOST, NEO4J_URI, REDIS_URL must point to Machine 1

services:
  # ═══════════════════════════════════════════════════════════════════════════════
  # Machine 1 — Core Infrastructure
  # ═══════════════════════════════════════════════════════════════════════════════

  # ── Qdrant (vector database) ──────────────────────────────────────────────────
  qdrant:
    image: qdrant/qdrant:v1.12.1
    container_name: rag-qdrant
    volumes:
      - qdrant_data:/qdrant/storage
    ports:
      - "6333:6333"
      - "6334:6334"
    environment:
      - QDRANT__SERVICE__GRPC_PORT=6334
    deploy:
      resources:
        limits:
          cpus: "4"
          memory: 8G
        reservations:
          cpus: "1"
          memory: 2G
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:6333/healthz"]
      interval: 10s
      timeout: 5s
      retries: 3
      start_period: 15s
    restart: unless-stopped
    networks:
      - rag-network

  # ── Neo4j (graph database) ────────────────────────────────────────────────────
  neo4j:
    image: neo4j:5-community
    container_name: rag-neo4j
    volumes:
      - neo4j_data:/data
      - neo4j_logs:/logs
    ports:
      - "7474:7474"
      - "7687:7687"
    environment:
      - NEO4J_AUTH=${NEO4J_USER:-neo4j}/${NEO4J_PASSWORD:-password}
      - NEO4J_server_memory_pagecache_size=1G
      - NEO4J_server_memory_heap_initial__size=1G
      - NEO4J_server_memory_heap_max__size=2G
    deploy:
      resources:
        limits:
          cpus: "4"
          memory: 4G
        reservations:
          cpus: "0.5"
          memory: 2G
    healthcheck:
      test: ["CMD", "neo4j", "status"]
      interval: 15s
      timeout: 5s
      retries: 5
      start_period: 60s
    restart: unless-stopped
    networks:
      - rag-network

  # ── Redis (shared cache) ──────────────────────────────────────────────────────
  redis:
    image: redis:7-alpine
    container_name: rag-redis
    ports:
      - "6379:6379"
    volumes:
      - redis_data:/data
    command: redis-server --appendonly yes --maxmemory 2gb --maxmemory-policy allkeys-lru --save 900 1 300 10
    deploy:
      resources:
        limits:
          cpus: "1"
          memory: 2G
        reservations:
          cpus: "0.1"
          memory: 256M
    healthcheck:
      test: ["CMD", "redis-cli", "ping"]
      interval: 10s
      timeout: 3s
      retries: 3
      start_period: 5s
    restart: unless-stopped
    networks:
      - rag-network

  # ── MinIO (object storage) ────────────────────────────────────────────────────
  minio:
    image: minio/minio:RELEASE.2024-11-07T00-52-20Z
    container_name: rag-minio
    volumes:
      - minio_data:/data
    ports:
      - "9000:9000"
      - "9001:9001"
    environment:
      - MINIO_ROOT_USER=${MINIO_ROOT_USER:-CHANGE_ME}
      - MINIO_ROOT_PASSWORD=${MINIO_ROOT_PASSWORD:-CHANGE_ME}
    command: server /data --console-address ":9001"
    deploy:
      resources:
        limits:
          cpus: "2"
          memory: 4G
        reservations:
          cpus: "0.5"
          memory: 1G
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:9000/minio/health/live"]
      interval: 15s
      timeout: 5s
      retries: 3
      start_period: 30s
    restart: unless-stopped
    networks:
      - rag-network

  # ── RAG Proxy (FastAPI) ───────────────────────────────────────────────────────
  proxy:
    build:
      context: ../..
      dockerfile: proxy/Dockerfile
    container_name: rag-proxy
    ports:
      - "8080:8080"
    volumes:
      - ${MODEL_CACHE_PATH:-/opt/models/cache}:/app/cache:ro
      - ../../proxy/.env:/app/.env:ro
      - proxy_logs:/app/logs
    environment:
      - QDRANT_HOST=${QDRANT_HOST:-qdrant}
      - QDRANT_PORT=${QDRANT_PORT:-6333}
      - QDRANT_GRPC_ENABLED=${QDRANT_GRPC_ENABLED:-true}
      - QDRANT_GRPC_PORT=${QDRANT_GRPC_PORT:-6334}
      - QDRANT_QUANTIZATION_ENABLED=${QDRANT_QUANTIZATION_ENABLED:-true}
      - NEO4J_URI=${NEO4J_URI:-bolt://neo4j:7687}
      - REDIS_URL=${REDIS_URL:-redis://redis:6379/0}
      - LLM_ENDPOINT=${LLM_ENDPOINT:-http://vllm:8000/v1}
      - MINIO_ENDPOINT=${MINIO_ENDPOINT:-minio:9000}
      - USE_REDIS=true
      - USE_LANGGRAPH=true
      - GRAPH_ENABLED=true
      - METRICS_ENABLED=true
      - LOG_FORMAT=json
      - PREFIX_CACHING_ENABLED=true
    depends_on:
      qdrant:
        condition: service_healthy
      neo4j:
        condition: service_healthy
      redis:
        condition: service_healthy
    deploy:
      resources:
        limits:
          cpus: "4"
          memory: 8G
        reservations:
          cpus: "1"
          memory: 2G
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:8080/v1/health/live"]
      interval: 10s
      timeout: 5s
      retries: 3
      start_period: 60s
    restart: unless-stopped
    stop_grace_period: 45s
    networks:
      - rag-network

  # ═══════════════════════════════════════════════════════════════════════════════
  # Machine 2 — OpenWebUI Interface
  # ═══════════════════════════════════════════════════════════════════════════════

  openwebui:
    image: ghcr.io/open-webui/open-webui:main
    container_name: rag-openwebui
    ports:
      - "3000:8080"
    environment:
      - WEBUI_SECRET_KEY=${WEBUI_SECRET_KEY}
      - DATABASE_URL=postgresql://${OWUI_DB_USER:-openwebui}:${OWUI_DB_PASSWORD}@postgres:5432/${OWUI_DB_NAME:-openwebui}
      - REDIS_URL=redis://rag-redis:6379/1
      - OPENAI_API_BASE_URLS=${OPENAI_API_BASE_URLS:-http://rag-proxy:8080/v1}
      - OPENAI_API_KEYS=${OPENAI_API_KEYS:-sk-rag-proxy}
      - STORAGE_PROVIDER=${STORAGE_PROVIDER:-s3}
      - S3_ENDPOINT_URL=${S3_ENDPOINT_URL:-http://rag-minio:9000}
      - S3_ACCESS_KEY_ID=${S3_ACCESS_KEY_ID}
      - S3_SECRET_ACCESS_KEY=${S3_SECRET_ACCESS_KEY}
      - S3_BUCKET_NAME=${S3_BUCKET_NAME:-openwebui-files}
      - S3_REGION_NAME=${S3_REGION_NAME:-us-east-1}
      - S3_ADDRESSING_STYLE=${S3_ADDRESSING_STYLE:-path}
      - CONTENT_EXTRACTION_ENGINE=${CONTENT_EXTRACTION_ENGINE:-tika}
      - TIKA_SERVER_URL=${TIKA_SERVER_URL:-http://tika:9998}
    volumes:
      - openwebui_data:/app/backend/data
      - openwebui_tmp:/tmp
    depends_on:
      postgres:
        condition: service_healthy
      tika:
        condition: service_healthy
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:8080/health"]
      interval: 15s
      timeout: 10s
      retries: 5
      start_period: 45s
    restart: unless-stopped
    stop_grace_period: 30s
    deploy:
      resources:
        limits:
          cpus: "2"
          memory: 2G
        reservations:
          cpus: "0.5"
          memory: 512M
    networks:
      - rag-network

  # ── PostgreSQL (OpenWebUI metadata) ───────────────────────────────────────────
  postgres:
    image: postgres:16-alpine
    container_name: rag-openwebui-postgres
    ports:
      - "127.0.0.1:5432:5432"
    environment:
      - POSTGRES_USER=${OWUI_DB_USER:-openwebui}
      - POSTGRES_PASSWORD=${OWUI_DB_PASSWORD}
      - POSTGRES_DB=${OWUI_DB_NAME:-openwebui}
      - PGDATA=/var/lib/postgresql/data/pgdata
    volumes:
      - postgres_data:/var/lib/postgresql/data
    command: >
      postgres
      -c shared_buffers=256MB
      -c effective_cache_size=768MB
      -c maintenance_work_mem=64MB
      -c max_connections=50
      -c random_page_cost=1.1
    healthcheck:
      test: ["CMD-SHELL", "pg_isready -U ${OWUI_DB_USER:-openwebui} -d ${OWUI_DB_NAME:-openwebui}"]
      interval: 10s
      timeout: 5s
      retries: 5
      start_period: 20s
    restart: unless-stopped
    stop_grace_period: 60s
    deploy:
      resources:
        limits:
          cpus: "2"
          memory: 1G
        reservations:
          cpus: "0.25"
          memory: 256M
    networks:
      - rag-network

  # ── Apache Tika (document extraction) ─────────────────────────────────────────
  tika:
    image: apache/tika:latest
    container_name: rag-openwebui-tika
    environment:
      - TIKA_CHILD_JAVA_OPTS=-Xmx512m
      - JAVA_TOOL_OPTIONS=-Xmx512m
    tmpfs:
      - /tmp:size=512M,mode=1777
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:9998/tika"]
      interval: 30s
      timeout: 15s
      retries: 3
      start_period: 45s
    restart: unless-stopped
    deploy:
      resources:
        limits:
          cpus: "2"
          memory: 1G
        reservations:
          cpus: "0.25"
          memory: 256M
    networks:
      - rag-network

  # ═══════════════════════════════════════════════════════════════════════════════
  # Machine 3 — ETL Worker
  # ═══════════════════════════════════════════════════════════════════════════════

  etl:
    build:
      context: ../..
      dockerfile: etl/Dockerfile.etl
    container_name: rag-etl
    volumes:
      - ../../etl/config:/app/etl/config:ro
      - etl_data:/app/etl/data
    environment:
      - QDRANT_HOST=${QDRANT_HOST:-rag-qdrant}
      - QDRANT_PORT=${QDRANT_PORT:-6333}
      - NEO4J_URI=${NEO4J_URI:-bolt://rag-neo4j:7687}
      - NEO4J_USER=${NEO4J_USER:-neo4j}
      - NEO4J_PASSWORD=${NEO4J_PASSWORD}
      - ETL_SECRET=${ETL_SECRET}
      - CONFLUENCE_API_URL=${CONFLUENCE_API_URL}
      - CONFLUENCE_API_TOKEN=${CONFLUENCE_API_TOKEN}
      - JIRA_API_URL=${JIRA_API_URL}
      - JIRA_API_TOKEN=${JIRA_API_TOKEN}
      - GITLAB_API_URL=${GITLAB_API_URL}
      - GITLAB_API_TOKEN=${GITLAB_API_TOKEN}
    command: python scheduler/run_etl.py --config /app/etl/config/etl_config.yaml
    deploy:
      resources:
        limits:
          cpus: "4"
          memory: 8G
        reservations:
          cpus: "1"
          memory: 2G
    restart: unless-stopped
    networks:
      - rag-network

# ═════════════════════════════════════════════════════════════════════════════════
# Volumes
# ═════════════════════════════════════════════════════════════════════════════════
volumes:
  qdrant_data:
    driver: local
  neo4j_data:
    driver: local
  neo4j_logs:
    driver: local
  redis_data:
    driver: local
  minio_data:
    driver: local
  proxy_logs:
    driver: local
  postgres_data:
    driver: local
  openwebui_data:
    driver: local
  openwebui_tmp:
    driver: local
  etl_data:
    driver: local

# ═════════════════════════════════════════════════════════════════════════════════
# Network
# ═════════════════════════════════════════════════════════════════════════════════
networks:
  rag-network:
    name: rag-network
    external: true