Chandra OCR 2的Docker Compose配置：多容器部署方案

Chandra OCR 2的Docker Compose配置：多容器部署方案

【免费下载链接】chandraOCR model that handles complex tables, forms, handwriting with full layout.项目地址: https://gitcode.com/GitHub_Trending/ch/chandra

Chandra OCR 2是一款强大的OCR模型，能够处理复杂表格、表单和手写文档，同时保持完整的布局信息。这款先进的OCR工具支持90多种语言，在数学公式、表格识别和多语言文档处理方面表现出色。对于需要大规模文档处理的企业用户来说，使用Docker Compose进行多容器部署是最佳选择，可以轻松实现高可用、可扩展的生产环境配置。🚀

为什么选择Docker Compose部署Chandra OCR 2？

传统的单机部署方式存在诸多限制，而Docker Compose多容器方案提供了以下优势：

• 资源隔离：每个服务在独立的容器中运行，避免资源冲突
• 易于扩展：可以根据负载动态调整vLLM推理容器的数量
• 快速部署：一键启动完整的OCR处理流水线
• 环境一致性：确保开发、测试、生产环境完全一致
• 监控管理：集成Prometheus、Grafana等监控工具

完整的Docker Compose配置文件

创建docker-compose.yml文件，配置完整的Chandra OCR 2服务栈：

version: '3.8' services: # vLLM推理服务 - 核心OCR引擎 chandra-vllm: image: datalab-to/chandra-ocr-2:latest container_name: chandra-vllm restart: unless-stopped deploy: resources: reservations: devices: - driver: nvidia count: all capabilities: [gpu] ports: - "8000:8000" environment: - MODEL_NAME=datalab-to/chandra-ocr-2 - MAX_MODEL_LEN=12384 - TENSOR_PARALLEL_SIZE=1 - GPU_MEMORY_UTILIZATION=0.9 volumes: - ./models:/models command: > python -m vllm.entrypoints.openai.api_server --model datalab-to/chandra-ocr-2 --max-model-len 12384 --tensor-parallel-size 1 --gpu-memory-utilization 0.9 --port 8000 healthcheck: test: ["CMD", "curl", "-f", "http://localhost:8000/health"] interval: 30s timeout: 10s retries: 3 # API网关服务 - 提供RESTful接口 chandra-api: build: context: . dockerfile: Dockerfile.api container_name: chandra-api restart: unless-stopped ports: - "8080:8080" depends_on: - chandra-vllm environment: - VLLM_API_BASE=http://chandra-vllm:8000/v1 - VLLM_MODEL_NAME=chandra - MAX_OUTPUT_TOKENS=12384 - REDIS_HOST=redis volumes: - ./uploads:/app/uploads - ./outputs:/app/outputs # 批处理服务 - 处理大批量文档 chandra-worker: build: context: . dockerfile: Dockerfile.worker container_name: chandra-worker restart: unless-stopped depends_on: - chandra-vllm - redis - postgres environment: - VLLM_API_BASE=http://chandra-vllm:8000/v1 - REDIS_HOST=redis - POSTGRES_HOST=postgres volumes: - ./batch_input:/app/input - ./batch_output:/app/output deploy: replicas: 2 # Redis缓存 - 提高性能 redis: image: redis:7-alpine container_name: chandra-redis restart: unless-stopped ports: - "6379:6379" command: redis-server --appendonly yes volumes: - redis-data:/data # PostgreSQL数据库 - 存储处理结果 postgres: image: postgres:15-alpine container_name: chandra-postgres restart: unless-stopped environment: - POSTGRES_DB=chandra - POSTGRES_USER=chandra - POSTGRES_PASSWORD=chandra123 ports: - "5432:5432" volumes: - postgres-data:/var/lib/postgresql/data - ./init.sql:/docker-entrypoint-initdb.d/init.sql # 监控服务 - Prometheus + Grafana prometheus: image: prom/prometheus:latest container_name: chandra-prometheus restart: unless-stopped ports: - "9090:9090" volumes: - ./prometheus.yml:/etc/prometheus/prometheus.yml - prometheus-data:/prometheus grafana: image: grafana/grafana:latest container_name: chandra-grafana restart: unless-stopped ports: - "3000:3000" environment: - GF_SECURITY_ADMIN_PASSWORD=admin volumes: - grafana-data:/var/lib/grafana - ./grafana/provisioning:/etc/grafana/provisioning volumes: redis-data: postgres-data: prometheus-data: grafana-data:

环境变量配置文件

创建.env文件管理配置：

# Docker Compose环境变量 COMPOSE_PROJECT_NAME=chandra-ocr COMPOSE_PROFILE=production # vLLM服务配置 VLLM_MODEL_NAME=datalab-to/chandra-ocr-2 VLLM_API_BASE=http://chandra-vllm:8000/v1 VLLM_GPUS=all MAX_OUTPUT_TOKENS=12384 # 数据库配置 POSTGRES_DB=chandra POSTGRES_USER=chandra POSTGRES_PASSWORD=chandra123 POSTGRES_HOST=postgres POSTGRES_PORT=5432 # Redis配置 REDIS_HOST=redis REDIS_PORT=6379 REDIS_PASSWORD= # API服务配置 API_PORT=8080 API_WORKERS=4 UPLOAD_FOLDER=/app/uploads OUTPUT_FOLDER=/app/outputs # 监控配置 PROMETHEUS_PORT=9090 GRAFANA_PORT=3000

一键启动和管理的Shell脚本

创建deploy.sh脚本简化部署流程：

#!/bin/bash # 颜色定义 RED='\033[0;31m' GREEN='\033[0;32m' YELLOW='\033[1;33m' NC='\033[0m' # No Color echo -e "${GREEN}Chandra OCR 2 Docker Compose 部署脚本${NC}" echo "========================================" # 检查Docker和Docker Compose check_dependencies() { echo -e "${YELLOW}检查依赖...${NC}" if ! command -v docker &> /dev/null; then echo -e "${RED}错误: Docker未安装${NC}" exit 1 fi if ! command -v docker-compose &> /dev/null; then echo -e "${RED}错误: Docker Compose未安装${NC}" exit 1 fi echo -e "${GREEN}✓ 依赖检查通过${NC}" } # 检查NVIDIA GPU check_gpu() { echo -e "${YELLOW}检查GPU支持...${NC}" if docker run --rm --gpus all nvidia/cuda:11.8.0-base-ubuntu22.04 nvidia-smi &> /dev/null; then echo -e "${GREEN}✓ NVIDIA GPU可用${NC}" export GPU_ENABLED=true else echo -e "${YELLOW}⚠ 未检测到NVIDIA GPU，将使用CPU模式${NC}" export GPU_ENABLED=false fi } # 创建必要目录 create_directories() { echo -e "${YELLOW}创建目录结构...${NC}" mkdir -p {uploads,outputs,models,batch_input,batch_output,logs,grafana/provisioning} echo -e "${GREEN}✓ 目录创建完成${NC}" } # 启动服务 start_services() { echo -e "${YELLOW}启动Chandra OCR 2服务...${NC}" if [ "$GPU_ENABLED" = true ]; then echo -e "${GREEN}使用GPU加速模式${NC}" docker-compose up -d else echo -e "${YELLOW}使用CPU模式${NC}" docker-compose -f docker-compose.yml -f docker-compose.cpu.yml up -d fi echo -e "${GREEN}✓ 服务启动完成${NC}" } # 显示服务状态 show_status() { echo -e "${YELLOW}服务状态:${NC}" echo "----------------------------------------" docker-compose ps echo -e "\n${YELLOW}访问地址:${NC}" echo "API服务: http://localhost:8080" echo "vLLM服务: http://localhost:8000" echo "Grafana监控: http://localhost:3000 (admin/admin)" echo "Prometheus: http://localhost:9090" } # 主流程 main() { check_dependencies check_gpu create_directories start_services show_status echo -e "\n${GREEN}部署完成！Chandra OCR 2已启动并运行。${NC}" echo "使用 'docker-compose logs -f' 查看日志" echo "使用 'docker-compose down' 停止服务" } main

性能优化配置

针对不同场景创建专门的Docker Compose配置文件：

docker-compose.prod.yml- 生产环境配置：

services: chandra-vllm: deploy: resources: limits: cpus: '8' memory: 32G reservations: devices: - driver: nvidia count: 2 capabilities: [gpu] environment: - TENSOR_PARALLEL_SIZE=2 - PIPELINE_PARALLEL_SIZE=1 - MAX_NUM_BATCHED_TOKENS=16384

docker-compose.dev.yml- 开发环境配置：

services: chandra-vllm: environment: - DEV_MODE=true - LOG_LEVEL=DEBUG volumes: - ./chandra:/app/chandra - ./tests:/app/tests

监控和日志配置

Prometheus配置 (prometheus.yml)：

global: scrape_interval: 15s scrape_configs: - job_name: 'chandra-vllm' static_configs: - targets: ['chandra-vllm:8000'] labels: service: 'ocr-inference' - job_name: 'chandra-api' static_configs: - targets: ['chandra-api:8080'] labels: service: 'api-gateway' - job_name: 'chandra-worker' static_configs: - targets: ['chandra-worker:8081'] labels: service: 'batch-worker'

日志收集配置：

version: '3.8' services: loki: image: grafana/loki:latest container_name: chandra-loki ports: - "3100:3100" promtail: image: grafana/promtail:latest container_name: chandra-promtail volumes: - /var/log:/var/log - ./promtail-config.yml:/etc/promtail/config.yml command: -config.file=/etc/promtail/config.yml

扩展和负载均衡配置

对于高并发场景，可以配置多个vLLM实例和负载均衡：

services: chandra-vllm-1: image: datalab-to/chandra-ocr-2:latest environment: - INSTANCE_ID=1 deploy: replicas: 2 chandra-vllm-2: image: datalab-to/chandra-ocr-2:latest environment: - INSTANCE_ID=2 deploy: replicas: 2 nginx: image: nginx:alpine ports: - "8000:80" volumes: - ./nginx.conf:/etc/nginx/nginx.conf depends_on: - chandra-vllm-1 - chandra-vllm-2

故障排除和常见问题

1. GPU内存不足

# 降低批次大小 docker-compose exec chandra-vllm vllm config --max-batch-size 4 # 调整GPU内存使用率 environment: - GPU_MEMORY_UTILIZATION=0.8

2. 服务启动失败

# 查看详细日志 docker-compose logs chandra-vllm # 检查GPU驱动 nvidia-smi # 验证模型下载 docker-compose exec chandra-vllm ls -la /models

3. 性能调优

# 监控资源使用 docker stats # 调整vLLM参数 environment: - MAX_NUM_SEQS=256 - MAX_NUM_BATCHED_TOKENS=32768 - BLOCK_SIZE=16

最佳实践建议

1. 资源分配策略

• 开发环境：单GPU，16GB内存
• 测试环境：双GPU，32GB内存
• 生产环境：四GPU，64GB内存以上

2. 数据持久化

volumes: - type: bind source: /data/chandra/uploads target: /app/uploads - type: volume source: chandra-data target: /app/data

3. 安全配置

environment: - API_KEY=${API_KEY} - ENABLE_AUTH=true - CORS_ORIGINS=https://yourdomain.com - RATE_LIMIT=100/分钟

4. 备份策略

# 数据库备份 docker-compose exec postgres pg_dump -U chandra chandra > backup.sql # 模型备份 docker cp chandra-vllm:/models ./backup/models

总结

通过Docker Compose部署Chandra OCR 2，您可以获得一个完整、可扩展的OCR处理平台。这种多容器架构不仅提供了高可用性和弹性扩展能力，还简化了部署和维护流程。无论是处理少量文档的个人用户，还是需要大规模批处理的企业用户，这种部署方案都能满足您的需求。

核心优势：

• 🚀一键部署：简化复杂的OCR系统配置
• 📊监控完善：集成Prometheus和Grafana
• 🔄弹性扩展：根据负载动态调整资源
• 🔒安全可靠：完善的权限和访问控制
• 💾数据持久：确保处理结果不丢失

现在就开始使用Docker Compose部署Chandra OCR 2，体验高效、稳定的文档OCR处理服务吧！

【免费下载链接】chandraOCR model that handles complex tables, forms, handwriting with full layout.项目地址: https://gitcode.com/GitHub_Trending/ch/chandra