main:新增健康检查支持和服务优化

- 在 Worker 中引入轻量级 HTTP 服务器，支持健康检查和就绪检查端点。
- 在 Kubernetes 和 Docker 配置中新增健康检查探针，提升服务稳定性。
- 更新依赖，引入 `aiohttp` 用于健康检查服务。
- 优化部署配置，调整 Redis 主机配置和镜像地址以适配新环境。

.env.example

@@ -1,32 +1,87 @@
 # Environment Configuration
 # Copy this file to .env and fill in your values
 
-# Application
+# =============================================================================
+# 应用信息
+# =============================================================================
 APP_NAME=FunctionalScaffold
 APP_VERSION=1.0.0
 APP_ENV=development
 
-# Server
+# =============================================================================
+# 服务器配置
+# =============================================================================
 HOST=0.0.0.0
 PORT=8000
 WORKERS=4
 
-# Logging
+# =============================================================================
+# 日志配置
+# =============================================================================
 LOG_LEVEL=INFO
 LOG_FORMAT=json
+# 日志文件配置（可选，默认禁用）
+LOG_FILE_ENABLED=false
+LOG_FILE_PATH=/var/log/app/app.log
 
-# Metrics
+# =============================================================================
+# 指标配置
+# =============================================================================
 METRICS_ENABLED=true
+METRICS_CONFIG_PATH=config/metrics.yaml
+# 指标实例 ID（可选，默认使用 hostname）
+# METRICS_INSTANCE_ID=my-instance
 
-# Tracing
+# =============================================================================
+# 追踪配置
+# =============================================================================
 TRACING_ENABLED=false
-JAEGER_ENDPOINT=http://localhost:14268/api/traces
+# JAEGER_ENDPOINT=http://localhost:14268/api/traces
 
-# External Services (examples)
+# =============================================================================
+# Redis 配置
+# =============================================================================
+REDIS_HOST=localhost
+REDIS_PORT=6379
+REDIS_DB=0
+REDIS_PASSWORD=your_redis_password
+
+# =============================================================================
+# 异步任务配置
+# =============================================================================
+# 任务结果缓存时间（秒），默认 30 分钟
+JOB_RESULT_TTL=1800
+# Webhook 最大重试次数
+WEBHOOK_MAX_RETRIES=3
+# Webhook 超时时间（秒）
+WEBHOOK_TIMEOUT=10
+# 最大并发任务数
+MAX_CONCURRENT_JOBS=10
+
+# =============================================================================
+# Worker 配置
+# =============================================================================
+# Worker 轮询间隔（秒）
+WORKER_POLL_INTERVAL=1.0
+# 任务队列 Redis Key
+JOB_QUEUE_KEY=job:queue
+# 全局并发计数器 Redis Key
+JOB_CONCURRENCY_KEY=job:concurrency
+# 任务锁 TTL（秒）
+JOB_LOCK_TTL=300
+# 任务最大重试次数
+JOB_MAX_RETRIES=3
+# 任务执行超时（秒）
+JOB_EXECUTION_TIMEOUT=300
+
+# =============================================================================
+# 外部服务配置（示例）
+# =============================================================================
+# OSS 配置
 # OSS_ENDPOINT=https://oss-cn-hangzhou.aliyuncs.com
 # OSS_ACCESS_KEY_ID=your_access_key
 # OSS_ACCESS_KEY_SECRET=your_secret_key
 # OSS_BUCKET_NAME=your_bucket
 
-# Database (if needed)
+# 数据库配置
-# DATABASE_URL=mysql://user:password@localhost:5432/dbname
+# DATABASE_URL=mysql://user:password@localhost:3306/dbname

AGENTS.md

@@ -0,0 +1,102 @@
+# Agent.md
+
+本文件为本仓库内各类智能体/助手提供工作指导，内容参考 `CLAUDE.md`，并针对日常开发与协作做了简化归纳。
+
+## 项目概述
+
+**FunctionalScaffold（函数式脚手架）** 是一个算法工程化 Serverless 解决方案的脚手架生成器。
+
+- 为了方便团队交流，项目自然语言使用中文，包括代码注释和文档
+- 核心目标：解决算力弹性、算法工程化门槛与后端集成复杂度问题
+
+## 技术与架构
+
+采用 **Docker 封装的 Serverless API 服务**方案：
+
+- 算法代码 + 运行环境打包为 Docker 镜像
+- 部署到云厂商 Serverless 平台实现自动扩缩容
+- FastAPI 作为 HTTP 接口层
+- 算法逻辑保持独立和专注
+
+架构流程概览：
+
+```
+用户请求 → API网关 → 容器实例（冷/热启动）→ FastAPI → 算法程序 → 返回结果
+                                              ↓
+                                        外部服务（OSS/数据库）
+```
+
+## 代码结构（src layout）
+
+```
+src/functional_scaffold/
+├── algorithms/          # 算法层 - 所有算法必须继承 BaseAlgorithm
+│   ├── base.py         # execute() 包装器（埋点、错误处理）
+│   └── prime_checker.py # 示例：质数判断算法
+├── api/                # API 层 - FastAPI 路由和模型
+│   ├── models.py       # Pydantic 数据模型（ConfigDict）
+│   ├── routes.py       # 路由定义（/invoke, /healthz, /readyz, /jobs）
+│   └── dependencies.py # 依赖注入（request_id 生成）
+├── core/               # 核心功能 - 横切关注点
+│   ├── errors.py       # 异常类层次结构
+│   ├── logging.py      # 结构化日志（JSON）
+│   ├── metrics.py      # Prometheus 指标和装饰器
+│   └── tracing.py      # 分布式追踪（ContextVar）
+├── utils/              # 工具函数
+│   └── validators.py   # 输入验证
+├── config.py           # 配置管理（pydantic-settings）
+└── main.py             # FastAPI 应用入口
+```
+
+## 关键设计约定
+
+1. **算法抽象层**：所有算法继承 `BaseAlgorithm`，只实现 `process()`；`execute()` 负责埋点、日志和错误包装。
+2. **依赖注入**：FastAPI `Depends()` 注入 `request_id`，通过 `ContextVar` 透传。
+3. **配置管理**：`pydantic-settings` 读取环境变量或 `.env`，支持类型校验。
+4. **可观测性**：JSON 结构化日志、Prometheus 指标、Request ID 追踪。
+5. **Pydantic V2**：使用 `ConfigDict` 和 `model_config`，不使用 `class Config`。
+
+## 常用命令
+
+环境设置：
+
+```bash
+python -m venv venv
+source venv/bin/activate
+pip install -e ".[dev]"
+```
+
+运行服务：
+
+```bash
+./scripts/run_dev.sh
+uvicorn functional_scaffold.main:app --reload --port 8000
+```
+
+测试与质量：
+
+```bash
+pytest tests/ -v
+black src/ tests/
+ruff check src/ tests/
+```
+
+## 添加新算法（简版步骤）
+
+1. 在 `src/functional_scaffold/algorithms/` 新建算法类，继承 `BaseAlgorithm` 并实现 `process()`。
+2. 在 `algorithms/__init__.py` 导出新算法类。
+3. 在 `api/routes.py` 添加端点，在 `api/models.py` 添加请求/响应模型。
+4. 在 `tests/` 编写对应测试。
+
+## 交付标准
+
+必须包含以下组件与规范：
+
+- `/invoke`, `/jobs`, `/healthz`, `/readyz`, `/metrics` 端点
+- 统一的请求/响应 Schema 与错误格式
+- 可观测性支持（日志、指标、追踪）
+
+## 开发理念
+
+算法同学只需关注 `process()` 的核心逻辑，其余基础设施能力由脚手架提供。
+

CLAUDE.md

@@ -2,12 +2,12 @@
 
 本文件为 Claude Code (claude.ai/code) 在此代码仓库中工作时提供指导。
 
+为了方便团队交流，项目的自然语言使用中文，包括代码注释和文档等
+
 ## 项目概述
 
 **FunctionalScaffold（函数式脚手架）** 是一个算法工程化 Serverless 解决方案的脚手架生成器。
 
-- 为了方便团队交流，项目的自然语言使用中文，包括代码注释和文档等
-
 ### 核心目标
 
 解决三大痛点：
@@ -65,9 +65,10 @@ src/functional_scaffold/
 3. **配置管理**：使用 `pydantic-settings` 从环境变量或 `.env` 文件加载配置，支持类型验证。
 
 4. **可观测性**：
-   - 日志：结构化 JSON 日志（pythonjsonlogger）
+   - 日志：结构化 JSON 日志（pythonjsonlogger），自动包含 request_id
    - 指标：Prometheus 格式（request_counter, request_latency, algorithm_counter）
    - 追踪：request_id 关联所有日志和指标
+   - 日志收集：Loki + Promtail 自动收集和查询日志
 
 ## 开发命令
 
@@ -87,10 +88,10 @@ pip install -e ".[dev]"
 ./scripts/run_dev.sh
 
 # 方式2：直接运行（开发模式，自动重载）
-uvicorn src.functional_scaffold.main:app --reload --port 8000
+uvicorn functional_scaffold.main:app --reload --port 8000
 
 # 方式3：生产模式
-uvicorn src.functional_scaffold.main:app --host 0.0.0.0 --port 8000 --workers 4
+uvicorn functional_scaffold.main:app --host 0.0.0.0 --port 8000 --workers 4
 ```
 
 访问地址：
@@ -150,11 +151,12 @@ docker build -f deployment/Dockerfile -t functional-scaffold:latest .
 # 运行容器
 docker run -p 8000:8000 functional-scaffold:latest
 
-# 使用 docker-compose（包含 Prometheus + Grafana）
+# 使用 docker-compose（包含 Prometheus + Grafana + Loki）
 cd deployment
 docker-compose up
 # Grafana: http://localhost:3000 (admin/admin)
 # Prometheus: http://localhost:9090
+# Loki: http://localhost:3100
 ```
 
 ### 文档
@@ -273,10 +275,52 @@ from functional_scaffold.core.logging import setup_logging
 # 设置日志
 logger = setup_logging(level="INFO", format_type="json")
 
-# 记录日志
+# 记录日志（自动包含 request_id）
 logger.info("处理请求", extra={"user_id": "123"})
 ```
 
+**日志特性：**
+- 结构化 JSON 格式
+- 自动包含 request_id（从 ContextVar 中提取）
+- 支持文件日志（可选，通过环境变量启用）
+- 日志轮转（100MB，保留 5 个备份）
+
+### 日志收集（Loki）
+
+项目集成了 Grafana Loki 日志收集系统，支持两种收集模式：
+
+**模式 1: Docker stdio 收集（默认，推荐）**
+- 自动收集容器标准输出/错误
+- 无需修改应用代码
+- 性能影响极小
+
+**模式 2: 文件收集（备用）**
+- 日志持久化到文件
+- 支持日志轮转
+- 需要设置 `LOG_FILE_ENABLED=true`
+
+**查询日志：**
+
+```bash
+# 使用 Loki API
+curl -G -s "http://localhost:3100/loki/api/v1/query_range" \
+  --data-urlencode 'query={job="functional-scaffold-app"}'
+
+# 按 request_id 过滤
+curl -G -s "http://localhost:3100/loki/api/v1/query_range" \
+  --data-urlencode 'query={job="functional-scaffold-app"} |= "request-id-here"'
+```
+
+**Grafana 仪表板：**
+- 访问 http://localhost:3000
+- 进入 "日志监控" 仪表板
+- 使用 Request ID 输入框过滤特定请求的日志
+
+**相关文档：**
+- 完整文档：`docs/loki-integration.md`
+- 使用说明：`docs/grafana-dashboard-usage.md`
+- 快速参考：`docs/loki-quick-reference.md`
+
 ### 指标
 
 使用 `core/metrics.py` 的装饰器：
@@ -298,7 +342,7 @@ def my_function():
 
 ### 追踪
 
-Request ID 自动注入到所有请求：
+Request ID 自动注入到所有请求和日志：
 
 ```python
 from functional_scaffold.core.tracing import get_request_id
@@ -307,6 +351,13 @@ from functional_scaffold.core.tracing import get_request_id
 request_id = get_request_id()
 ```
 
+**Request ID 特性：**
+- 自动生成或从请求头 `X-Request-ID` 获取
+- 通过 ContextVar 在异步上下文中传递
+- 自动添加到所有日志记录中
+- 可用于追踪单个请求的完整生命周期
+- 在 Grafana 仪表板中可按 request_id 过滤日志
+
 ## 部署
 
 ### Kubernetes
@@ -321,10 +372,23 @@ kubectl apply -f deployment/kubernetes/service.yaml
 - 资源限制：256Mi-512Mi 内存，250m-500m CPU
 - 健康检查：存活探针 (/healthz)，就绪探针 (/readyz)
 
-### 阿里云函数计算
+### 阿里云函数计算（FC 3.0）
 
 ```bash
-fun deploy -t deployment/serverless/aliyun-fc.yaml
+# 安装 Serverless Devs（如未安装）
+npm install -g @serverless-devs/s
+
+# 配置阿里云凭证（首次使用）
+s config add
+
+# 部署到阿里云函数计算
+cd deployment/serverless && s deploy
+
+# 验证配置语法
+cd deployment/serverless && s plan
+
+# 查看函数日志
+cd deployment/serverless && s logs --tail
 ```
 
 ### AWS Lambda
@@ -355,7 +419,8 @@ sam deploy --template-file deployment/serverless/aws-lambda.yaml
 - ✅ 参数校验（Pydantic + utils/validators.py）
 - ✅ 错误包装和标准化（core/errors.py）
 - ✅ 埋点（core/metrics.py - 延迟、失败率）
-- ✅ 分布式追踪的关联 ID（core/tracing.py）
+- ✅ 分布式追踪的关联 ID（core/tracing.py + RequestIdFilter）
+- ✅ 日志收集和查询（Loki + Promtail）
 - ⏳ Worker 运行时（重试、超时、DLQ - 待实现）
 
 ### 3. 脚手架生成器
@@ -365,8 +430,9 @@ sam deploy --template-file deployment/serverless/aws-lambda.yaml
 - ✅ Dockerfile（deployment/Dockerfile）
 - ✅ CI/CD 流水线配置（.github/workflows/）
 - ✅ Serverless 平台部署 YAML（deployment/serverless/）
-- ✅ Grafana 仪表板模板（monitoring/grafana/dashboard.json）
+- ✅ Grafana 仪表板模板（monitoring/grafana/dashboards/）
 - ✅ 告警规则配置（monitoring/alerts/rules.yaml）
+- ✅ Loki 日志收集配置（monitoring/loki.yaml, monitoring/promtail.yaml）
 
 ## 开发理念
 
@@ -397,3 +463,180 @@ sam deploy --template-file deployment/serverless/aws-lambda.yaml
 4. **Docker 构建**：Dockerfile 使用非 root 用户（appuser），包含健康检查。
 
 5. **配置优先级**：环境变量 > .env 文件 > 默认值。
+
+6. **Promtail 版本**：使用 Promtail 3.0.0 或更高版本，以支持较新的 Docker API（1.44+）。如果遇到 "client version too old" 错误，需要升级 Promtail 版本。
+
+## 日志收集系统（Loki）
+
+项目集成了 Grafana Loki 日志收集系统，提供强大的日志查询和分析能力。
+
+### 架构
+
+```
+应用容器 (stdout/stderr)
+    ↓
+Docker Engine
+    ↓
+Promtail (日志采集器)
+    ↓
+Loki (日志存储)
+    ↓
+Grafana (可视化)
+```
+
+### 服务组件
+
+**docker-compose 包含以下服务：**
+- **app**: 应用服务（端口 8111）
+- **loki**: 日志存储服务（端口 3100）
+- **promtail**: 日志采集服务（端口 9080）
+- **grafana**: 可视化服务（端口 3000）
+- **prometheus**: 指标收集服务（端口 9090）
+- **redis**: 缓存服务（端口 6380）
+
+### 日志收集模式
+
+#### 模式 1: Docker stdio 收集（默认）
+
+**特点：**
+- ✅ 无需修改应用代码
+- ✅ 自动收集容器标准输出/错误
+- ✅ 性能影响极小
+- ✅ 推荐用于生产环境
+
+**配置：**
+应用容器需要添加标签（已配置）：
+```yaml
+labels:
+  logging: "promtail"
+  logging_jobname: "functional-scaffold-app"
+```
+
+#### 模式 2: 文件收集（备用）
+
+**特点：**
+- ✅ 日志持久化到文件
+- ✅ 支持日志轮转（100MB，5个备份）
+- ✅ 适合需要本地日志文件的场景
+
+**启用方式：**
+```yaml
+# docker-compose.yml
+environment:
+  - LOG_FILE_ENABLED=true
+  - LOG_FILE_PATH=/var/log/app/app.log
+```
+
+### 日志格式
+
+所有日志使用 JSON 格式，自动包含以下字段：
+- `asctime`: 时间戳
+- `name`: 日志器名称
+- `levelname`: 日志级别（INFO, WARNING, ERROR）
+- `message`: 日志消息
+- `request_id`: 请求 ID（自动添加）
+- `timestamp`: ISO 格式时间戳
+
+### 查询日志
+
+#### 使用 Loki API
+
+```bash
+# 查询所有日志
+curl -G -s "http://localhost:3100/loki/api/v1/query_range" \
+  --data-urlencode 'query={job="functional-scaffold-app"}'
+
+# 按 request_id 过滤
+curl -G -s "http://localhost:3100/loki/api/v1/query_range" \
+  --data-urlencode 'query={job="functional-scaffold-app"} |= "request-id-here"'
+
+# 查询错误日志
+curl -G -s "http://localhost:3100/loki/api/v1/query_range" \
+  --data-urlencode 'query={job="functional-scaffold-app", level="ERROR"}'
+```
+
+#### 使用 Grafana 仪表板
+
+1. 访问 http://localhost:3000（admin/admin）
+2. 进入 "日志监控" 仪表板
+3. 使用 Request ID 输入框过滤特定请求的日志
+
+**仪表板面板：**
+- **日志流（实时）**: 实时日志流
+- **日志量趋势**: 按时间和级别统计
+- **日志级别分布**: INFO/WARNING/ERROR 分布
+- **错误日志**: 只显示 ERROR 级别
+
+#### 使用 Grafana Explore
+
+1. 访问 http://localhost:3000/explore
+2. 选择 Loki 数据源
+3. 使用 LogQL 查询语言
+
+**常用查询：**
+```logql
+# 查询所有日志
+{job="functional-scaffold-app"}
+
+# 查询错误日志
+{job="functional-scaffold-app", level="ERROR"}
+
+# 按 request_id 过滤
+{job="functional-scaffold-app"} |= "request-id-here"
+
+# 使用 JSON 解析
+{job="functional-scaffold-app"} | json | request_id="request-id-here"
+
+# 统计日志量
+sum by (level) (count_over_time({job="functional-scaffold-app"}[5m]))
+```
+
+### 验证和测试
+
+```bash
+# 验证 Loki 集成
+./scripts/verify_loki.sh
+
+# 测试 Request ID 过滤
+./scripts/test_request_id_filter.sh
+```
+
+### 配置文件
+
+- **Loki 配置**: `monitoring/loki.yaml`
+  - 日志保留期: 7 天
+  - 摄入速率限制: 10MB/s
+  - 自动压缩和清理
+
+- **Promtail 配置**: `monitoring/promtail.yaml`
+  - Docker stdio 收集配置
+  - 文件收集配置
+  - JSON 日志解析规则
+
+- **Grafana Provisioning**: `monitoring/grafana/`
+  - 数据源自动配置（datasources/）
+  - 仪表板自动加载（dashboards/）
+
+### 故障排查
+
+**看不到日志：**
+1. 检查服务状态: `docker-compose ps`
+2. 查看 Promtail 日志: `docker-compose logs promtail`
+3. 验证容器标签: `docker inspect <container> | grep Labels`
+
+**Docker socket 权限问题：**
+```bash
+sudo chmod 666 /var/run/docker.sock
+```
+
+**日志延迟：**
+- Promtail 每 5 秒刷新一次
+- 建议等待 5-10 秒后再查询
+
+### 相关文档
+
+- **完整文档**: `docs/loki-integration.md` - 包含查询示例、故障排查、性能优化
+- **快速参考**: `docs/loki-quick-reference.md` - 常用命令和 LogQL 查询
+- **仪表板使用**: `docs/grafana-dashboard-usage.md` - Grafana 仪表板使用说明
+- **实施总结**: `docs/loki-implementation-summary.md` - 架构和实施细节
+- **监控目录**: `monitoring/README.md` - 配置文件说明

LICENSE

@@ -0,0 +1,22 @@
+MIT License
+
+Copyright (c) 2026 Roog
+Copyright (c) 2026 Guxinpei
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

README.md

@@ -15,17 +15,19 @@
 - ✅ **容器化部署** - Docker 和 Kubernetes 支持
 - ✅ **Serverless 就绪** - 支持阿里云函数计算和 AWS Lambda
 - ✅ **完整测试** - 单元测试和集成测试
-- ✅ **CI/CD** - GitHub Actions 工作流
 
 ## 文档
 
-| 文档 | 描述 |
+| 文档                                             | 描述           |
-|------|------|
+|------------------------------------------------|--------------|
-| [快速入门](docs/getting-started.md) | 10 分钟上手指南 |
+| [快速入门](docs/getting-started.md)                | 10 分钟上手指南    |
-| [算法开发指南](docs/algorithm-development.md) | 详细的算法开发教程 |
+| [算法开发指南](docs/algorithm-development.md)        | 详细的算法开发教程    |
-| [API 参考](docs/api-reference.md) | 完整的 API 文档 |
+| [API 参考](docs/api-reference.md)                | 完整的 API 文档   |
-| [监控指南](docs/monitoring.md) | 监控和告警配置 |
+| [监控指南](docs/monitoring.md)                     | 监控和告警配置      |
-| [API 规范](docs/api/README.md) | OpenAPI 规范说明 |
+| [API 规范](docs/api/README.md)                   | OpenAPI 规范说明 |
+| [Kubernetes 部署](docs/kubernetes-deployment.md) | K8s 集群部署指南   |
+| [日志集成(Loki)](docs/loki-quick-reference.md)     | 日志收集部署说明     |
+| [阿里云函数运算FC部署入门](docs/fc-deploy.md)             | 阿里云FC部署入门    |
 
 ## 快速开始
 
@@ -58,7 +60,7 @@ pip install -e ".[dev]"
 ./scripts/run_dev.sh
 
 # 方式2：直接运行
-uvicorn src.functional_scaffold.main:app --reload --port 8000
+uvicorn functional_scaffold.main:app --reload --port 8000
 ```
 
 4. 访问 API 文档
@@ -80,6 +82,12 @@ docker run -p 8000:8000 functional-scaffold:latest
 # 或使用 docker-compose
 cd deployment
 docker-compose up
+
+# 如果阿里FC无法识别 Platform:unknown/unknown 的情况时，请按下列执行打包:
+export DOCKER_DEFAULT_PLATFORM=linux/amd64
+export BUILDX_NO_DEFAULT_ATTESTATIONS=1
+docker compose build
+docker compose push 
 ```
 
 ## API 端点

config/metrics.yaml

@@ -94,6 +94,26 @@ custom_metrics:
     type: counter
     description: "Webhook 回调发送总数"
     labels: [status]
+
+  # 队列监控指标
+  job_queue_length:
+    name: "job_queue_length"
+    type: gauge
+    description: "待处理任务队列长度"
+    labels: [queue]
+
+  job_oldest_waiting_seconds:
+    name: "job_oldest_waiting_seconds"
+    type: gauge
+    description: "最长任务等待时间（秒）"
+    labels: []
+
+  job_recovered_total:
+    name: "job_recovered_total"
+    type: counter
+    description: "回收的超时任务总数"
+    labels: []
+
   prime_check_total:
     name: "prime_check"
     type: counter

deployment/Dockerfile

@@ -1,4 +1,4 @@
-FROM python:3.11-slim
+FROM --platform=linux/amd64 python:3.11-slim
 
 WORKDIR /app
 
@@ -9,15 +9,21 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
 
 # 复制依赖文件
 COPY requirements.txt .
-COPY requirements-dev.txt .
 
 # 安装 Python 依赖
 RUN pip install --no-cache-dir -r requirements.txt
-RUN pip install --no-cache-dir -r requirements-dev.txt
+
+# 安装dev依赖
+#COPY requirements-dev.txt .
+#RUN pip install --no-cache-dir -r requirements-dev.txt
 
 # 复制应用代码和配置
 COPY src/ ./src/
 COPY config/ ./config/
+COPY pyproject.toml .
+
+# 安装包（使用 editable 模式）
+RUN pip install --no-cache-dir -e .
 
 # 创建非 root 用户
 RUN useradd -m -u 1000 appuser && chown -R appuser:appuser /app
@@ -26,9 +32,15 @@ USER appuser
 # 暴露端口
 EXPOSE 8000
 
-# 健康检查
+# 运行模式：api（默认）或 worker
-HEALTHCHECK --interval=30s --timeout=3s --start-period=5s --retries=3 \
+ENV RUN_MODE=api
-    CMD python -c "import urllib.request; urllib.request.urlopen('http://localhost:8000/healthz')"
 
-# 启动命令
+# 健康检查（仅对 API 模式有效）
-CMD ["uvicorn", "src.functional_scaffold.main:app", "--host", "0.0.0.0", "--port", "8000"]
+HEALTHCHECK --interval=30s --timeout=3s --start-period=5s --retries=3 \
+    CMD if [ "$RUN_MODE" = "api" ]; then python -c "import urllib.request; urllib.request.urlopen('http://localhost:8000/healthz')"; else exit 0; fi
+
+# 启动脚本
+COPY --chown=appuser:appuser deployment/entrypoint.sh /app/entrypoint.sh
+RUN chmod +x /app/entrypoint.sh
+
+CMD ["/app/entrypoint.sh"]

deployment/docker-compose.yml

@@ -5,21 +5,30 @@ services:
     build:
       context: ..
       dockerfile: deployment/Dockerfile
+    platform: linux/amd64
     ports:
       - "8111:8000"
     environment:
       - APP_ENV=development
       - LOG_LEVEL=INFO
       - METRICS_ENABLED=true
+      - RUN_MODE=api
       # Redis 指标存储配置
       - REDIS_HOST=redis
       - REDIS_PORT=6379
       - REDIS_DB=0
       # 指标配置文件路径
       - METRICS_CONFIG_PATH=config/metrics.yaml
+      # 日志文件配置
+      - LOG_FILE_ENABLED=false
+      - LOG_FILE_PATH=/var/log/app/app.log
     volumes:
       - ../src:/app/src
       - ../config:/app/config
+      - app_logs:/var/log/app
+    labels:
+      logging: "promtail"
+      logging_jobname: "functional-scaffold-app"
     restart: unless-stopped
     depends_on:
       redis:
@@ -31,6 +40,47 @@ services:
       retries: 3
       start_period: 5s
 
+  # Worker 服务 - 处理异步任务
+  worker:
+    build:
+      context: ..
+      dockerfile: deployment/Dockerfile
+    platform: linux/amd64
+    ports:
+      - "8112:8000"
+    environment:
+      - APP_ENV=development
+      - LOG_LEVEL=INFO
+      - METRICS_ENABLED=true
+      - RUN_MODE=worker
+      # Redis 配置
+      - REDIS_HOST=redis
+      - REDIS_PORT=6379
+      - REDIS_DB=0
+      # Worker 配置
+      - WORKER_POLL_INTERVAL=1.0
+      - MAX_CONCURRENT_JOBS=10
+      - JOB_MAX_RETRIES=3
+      - JOB_EXECUTION_TIMEOUT=300
+    volumes:
+      - ../src:/app/src
+      - ../config:/app/config
+    labels:
+      logging: "promtail"
+      logging_jobname: "functional-scaffold-worker"
+    restart: unless-stopped
+    depends_on:
+      redis:
+        condition: service_healthy
+    healthcheck:
+      test: ["CMD", "python", "-c", "import urllib.request; urllib.request.urlopen('http://localhost:8000/healthz')"]
+      interval: 30s
+      timeout: 3s
+      retries: 3
+      start_period: 10s
+    deploy:
+      replicas: 2
+
   # Redis - 用于集中式指标存储
   redis:
     image: redis:7-alpine
@@ -69,12 +119,47 @@ services:
       - GF_SECURITY_ADMIN_PASSWORD=admin
     volumes:
       - grafana_data:/var/lib/grafana
-      - ../monitoring/grafana:/etc/grafana/provisioning
+      - ../monitoring/grafana/datasources:/etc/grafana/provisioning/datasources
+      - ../monitoring/grafana/dashboards:/etc/grafana/provisioning/dashboards
     restart: unless-stopped
     depends_on:
       - prometheus
+      - loki
+
+  loki:
+    image: grafana/loki:2.9.3
+    ports:
+      - "3100:3100"
+    volumes:
+      - ../monitoring/loki.yaml:/etc/loki/local-config.yaml
+      - loki_data:/loki
+    command: -config.file=/etc/loki/local-config.yaml
+    restart: unless-stopped
+    healthcheck:
+      test: ["CMD", "wget", "--spider", "-q", "http://localhost:3100/ready"]
+      interval: 10s
+      timeout: 3s
+      retries: 3
+
+  promtail:
+    ports:
+      - "9080:9080"
+    image: grafana/promtail:3.0.0
+    volumes:
+      - ../monitoring/promtail.yaml:/etc/promtail/config.yml
+      # Docker stdio 收集
+      - /var/lib/docker/containers:/var/lib/docker/containers:ro
+      - /var/run/docker.sock:/var/run/docker.sock:ro
+      # Log 文件收集（备用）
+      - app_logs:/var/log/app:ro
+    command: -config.file=/etc/promtail/config.yml
+    restart: unless-stopped
+    depends_on:
+      - loki
 
 volumes:
   prometheus_data:
   grafana_data:
   redis_data:
+  loki_data:
+  app_logs:

deployment/entrypoint.sh

@@ -0,0 +1,12 @@
+#!/bin/bash
+# 启动脚本：根据 RUN_MODE 环境变量选择启动 API 或 Worker
+
+set -e
+
+if [ "$RUN_MODE" = "worker" ]; then
+    echo "启动 Worker 模式..."
+    exec python -m functional_scaffold.worker
+else
+    echo "启动 API 模式..."
+    exec uvicorn functional_scaffold.main:app --host 0.0.0.0 --port 8000
+fi

deployment/kubernetes/deployment.yaml

@@ -1,33 +1,70 @@
+# Kubernetes 部署配置
+# 包含：ConfigMap、API Deployment、Worker Deployment、Redis Deployment
+
+---
+# ConfigMap - 共享配置
+apiVersion: v1
+kind: ConfigMap
+metadata:
+  name: functional-scaffold-config
+  labels:
+    app: functional-scaffold
+data:
+  APP_ENV: "production"
+  LOG_LEVEL: "INFO"
+  LOG_FORMAT: "json"
+  METRICS_ENABLED: "true"
+  # Redis 配置（指向集群内 Redis 服务）
+  REDIS_HOST: "functional-scaffold-redis"
+  REDIS_PORT: "6379"
+  REDIS_DB: "0"
+  # 异步任务配置
+  MAX_CONCURRENT_JOBS: "10"
+  JOB_RESULT_TTL: "1800"
+  WEBHOOK_MAX_RETRIES: "3"
+  WEBHOOK_TIMEOUT: "10"
+  # Worker 配置
+  WORKER_POLL_INTERVAL: "1.0"
+  JOB_QUEUE_KEY: "job:queue"
+  JOB_CONCURRENCY_KEY: "job:concurrency"
+  JOB_LOCK_TTL: "300"
+  JOB_MAX_RETRIES: "3"
+  JOB_EXECUTION_TIMEOUT: "300"
+
+---
+# API Deployment - HTTP 服务
 apiVersion: apps/v1
 kind: Deployment
 metadata:
-  name: functional-scaffold
+  name: functional-scaffold-api
   labels:
     app: functional-scaffold
+    component: api
 spec:
   replicas: 3
   selector:
     matchLabels:
       app: functional-scaffold
+      component: api
   template:
     metadata:
       labels:
         app: functional-scaffold
+        component: api
     spec:
       containers:
-      - name: functional-scaffold
+      - name: api
         image: functional-scaffold:latest
         imagePullPolicy: IfNotPresent
         ports:
         - containerPort: 8000
           name: http
         env:
-        - name: APP_ENV
+        - name: RUN_MODE
-          value: "production"
+          value: "api"
-        - name: LOG_LEVEL
+        envFrom:
-          value: "INFO"
+        - configMapRef:
-        - name: METRICS_ENABLED
+            name: functional-scaffold-config
-          value: "true"
         resources:
           requests:
             memory: "256Mi"
@@ -51,3 +88,125 @@ spec:
           periodSeconds: 10
           timeoutSeconds: 3
           failureThreshold: 3
+
+---
+# Worker Deployment - 异步任务处理
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: functional-scaffold-worker
+  labels:
+    app: functional-scaffold
+    component: worker
+spec:
+  replicas: 2
+  selector:
+    matchLabels:
+      app: functional-scaffold
+      component: worker
+  template:
+    metadata:
+      labels:
+        app: functional-scaffold
+        component: worker
+    spec:
+      containers:
+      - name: worker
+        image: functional-scaffold:latest
+        imagePullPolicy: IfNotPresent
+        env:
+        - name: RUN_MODE
+          value: "worker"
+        envFrom:
+        - configMapRef:
+            name: functional-scaffold-config
+        resources:
+          requests:
+            memory: "256Mi"
+            cpu: "250m"
+          limits:
+            memory: "512Mi"
+            cpu: "500m"
+        # Worker 现在有 HTTP 健康检查端点
+        ports:
+        - containerPort: 8000
+          name: http
+        livenessProbe:
+          httpGet:
+            path: /healthz
+            port: 8000
+          initialDelaySeconds: 10
+          periodSeconds: 30
+          timeoutSeconds: 3
+          failureThreshold: 3
+        readinessProbe:
+          httpGet:
+            path: /readyz
+            port: 8000
+          initialDelaySeconds: 5
+          periodSeconds: 10
+          timeoutSeconds: 3
+          failureThreshold: 3
+
+---
+# Redis Deployment - 任务队列和状态存储
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: functional-scaffold-redis
+  labels:
+    app: functional-scaffold
+    component: redis
+spec:
+  replicas: 1
+  selector:
+    matchLabels:
+      app: functional-scaffold
+      component: redis
+  template:
+    metadata:
+      labels:
+        app: functional-scaffold
+        component: redis
+    spec:
+      containers:
+      - name: redis
+        image: redis:7-alpine
+        ports:
+        - containerPort: 6379
+          name: redis
+        command:
+        - redis-server
+        - --appendonly
+        - "yes"
+        resources:
+          requests:
+            memory: "128Mi"
+            cpu: "100m"
+          limits:
+            memory: "256Mi"
+            cpu: "200m"
+        livenessProbe:
+          exec:
+            command:
+            - redis-cli
+            - ping
+          initialDelaySeconds: 5
+          periodSeconds: 10
+          timeoutSeconds: 3
+          failureThreshold: 3
+        readinessProbe:
+          exec:
+            command:
+            - redis-cli
+            - ping
+          initialDelaySeconds: 5
+          periodSeconds: 5
+          timeoutSeconds: 3
+          failureThreshold: 3
+        volumeMounts:
+        - name: redis-data
+          mountPath: /data
+      volumes:
+      - name: redis-data
+        emptyDir: {}

deployment/kubernetes/service.yaml

@@ -1,9 +1,15 @@
+# Kubernetes Service 配置
+# 包含：API Service、Metrics Service、Redis Service
+
+---
+# API Service - 对外暴露 HTTP 服务
 apiVersion: v1
 kind: Service
 metadata:
-  name: functional-scaffold
+  name: functional-scaffold-api
   labels:
     app: functional-scaffold
+    component: api
 spec:
   type: ClusterIP
   ports:
@@ -13,13 +19,21 @@ spec:
     name: http
   selector:
     app: functional-scaffold
+    component: api
+
 ---
+# Metrics Service - Prometheus 抓取指标
 apiVersion: v1
 kind: Service
 metadata:
   name: functional-scaffold-metrics
   labels:
     app: functional-scaffold
+    component: api
+  annotations:
+    prometheus.io/scrape: "true"
+    prometheus.io/port: "8000"
+    prometheus.io/path: "/metrics"
 spec:
   type: ClusterIP
   ports:
@@ -29,3 +43,24 @@ spec:
     name: metrics
   selector:
     app: functional-scaffold
+    component: api
+
+---
+# Redis Service - 内部 Redis 服务
+apiVersion: v1
+kind: Service
+metadata:
+  name: functional-scaffold-redis
+  labels:
+    app: functional-scaffold
+    component: redis
+spec:
+  type: ClusterIP
+  ports:
+  - port: 6379
+    targetPort: 6379
+    protocol: TCP
+    name: redis
+  selector:
+    app: functional-scaffold
+    component: redis

deployment/serverless/aliyun-fc.yaml

@@ -1,40 +0,0 @@
-# 阿里云函数计算配置
-ROSTemplateFormatVersion: '2015-09-01'
-Transform: 'Aliyun::Serverless-2018-04-03'
-Resources:
-  functional-scaffold:
-    Type: 'Aliyun::Serverless::Service'
-    Properties:
-      Description: '算法工程化 Serverless 脚手架'
-      LogConfig:
-        Project: functional-scaffold-logs
-        Logstore: function-logs
-      VpcConfig:
-        VpcId: 'vpc-xxxxx'
-        VSwitchIds:
-          - 'vsw-xxxxx'
-        SecurityGroupId: 'sg-xxxxx'
-    prime-checker:
-      Type: 'Aliyun::Serverless::Function'
-      Properties:
-        Description: '质数判断算法服务'
-        Runtime: custom-container
-        MemorySize: 512
-        Timeout: 60
-        InstanceConcurrency: 10
-        CAPort: 8000
-        CustomContainerConfig:
-          Image: 'registry.cn-hangzhou.aliyuncs.com/your-namespace/functional-scaffold:latest'
-          Command: '["uvicorn", "src.functional_scaffold.main:app", "--host", "0.0.0.0", "--port", "8000"]'
-        EnvironmentVariables:
-          APP_ENV: production
-          LOG_LEVEL: INFO
-          METRICS_ENABLED: 'true'
-      Events:
-        httpTrigger:
-          Type: HTTP
-          Properties:
-            AuthType: ANONYMOUS
-            Methods:
-              - GET
-              - POST

deployment/serverless/s.yaml

@@ -0,0 +1,108 @@
+# 阿里云函数计算 FC 3.0 配置
+# 使用 Serverless Devs 部署: cd deployment/serverless && s deploy
+edition: 3.0.0
+name: functional-scaffold
+access: default
+
+vars:
+  region: cn-beijing
+  image: crpi-om2xd9y8cmaizszf-vpc.cn-beijing.personal.cr.aliyuncs.com/your-namespace/fc-test:test-v1
+  redis_host: 127.31.1.1
+  redis_port: "6379"
+  redis_password: "your-password"
+
+resources:
+  # API 服务函数
+  prime-checker-api:
+    component: fc3
+    props:
+      region: ${vars.region}
+      functionName: prime-checker-api
+      description: 质数判断算法服务（API）
+      runtime: custom-container
+      cpu: 0.35
+      memorySize: 512
+      diskSize: 512
+      timeout: 60
+      instanceConcurrency: 10
+      handler: not-used
+      customContainerConfig:
+        image: ${vars.image}
+        port: 8000
+        command:
+          - /app/entrypoint.sh
+        healthCheckConfig:
+          httpGetUrl: /healthz
+          initialDelaySeconds: 3
+          periodSeconds: 5
+          timeoutSeconds: 3
+          failureThreshold: 3
+          successThreshold: 1
+      environmentVariables:
+        APP_ENV: production
+        LOG_LEVEL: INFO
+        METRICS_ENABLED: "true"
+        RUN_MODE: api
+        REDIS_HOST: ${vars.redis_host}
+        REDIS_PORT: ${vars.redis_port}
+        REDIS_PASSWORD: ${vars.redis_password}
+      vpcConfig: auto
+      logConfig: auto
+      triggers:
+        - triggerName: http-trigger
+          triggerType: http
+          triggerConfig:
+            authType: anonymous
+            methods:
+              - GET
+              - POST
+              - PUT
+              - DELETE
+
+  # 异步任务 Worker 函数
+  job-worker:
+    component: fc3
+    props:
+      region: ${vars.region}
+      functionName: job-worker
+      description: 异步任务 Worker
+      runtime: custom-container
+      cpu: 0.35
+      memorySize: 512
+      diskSize: 512
+      timeout: 900
+      instanceConcurrency: 1
+      handler: not-used
+      customContainerConfig:
+        image: ${vars.image}
+        port: 8000
+        command:
+          - /app/entrypoint.sh
+        healthCheckConfig:
+          httpGetUrl: /healthz
+          initialDelaySeconds: 5
+          periodSeconds: 10
+          timeoutSeconds: 3
+          failureThreshold: 3
+          successThreshold: 1
+      environmentVariables:
+        APP_ENV: production
+        LOG_LEVEL: INFO
+        METRICS_ENABLED: "true"
+        RUN_MODE: worker
+        REDIS_HOST: ${vars.redis_host}
+        REDIS_PORT: ${vars.redis_port}
+        REDIS_PASSWORD: ${vars.redis_password}
+        WORKER_POLL_INTERVAL: "1.0"
+        MAX_CONCURRENT_JOBS: "5"
+        JOB_MAX_RETRIES: "3"
+        JOB_EXECUTION_TIMEOUT: "300"
+      vpcConfig: auto
+      logConfig: auto
+      triggers:
+        - triggerName: timer-trigger
+          triggerType: timer
+          triggerConfig:
+            cronExpression: "0 */1 * * * *"
+            enable: true
+            payload: "{}"

docs/algorithm-development.md

@@ -412,7 +412,7 @@ class MLPredictor(BaseAlgorithm):
 ```python
 # tests/test_text_processor.py
 import pytest
-from src.functional_scaffold.algorithms.text_processor import TextProcessor
+from functional_scaffold.algorithms.text_processor import TextProcessor
 
 class TestTextProcessor:
     """文本处理算法测试"""

docs/concurrency-control-changelog.md

@@ -0,0 +1,204 @@
+# 异步任务并发控制实现总结
+
+## 变更概述
+
+为异步任务管理器添加了并发控制功能，使用 `asyncio.Semaphore` 限制同时运行的任务数量，防止系统资源耗尽。
+
+## 修改的文件
+
+### 1. `src/functional_scaffold/config.py`
+
+**新增配置项：**
+```python
+max_concurrent_jobs: int = 10  # 最大并发任务数
+```
+
+### 2. `src/functional_scaffold/core/job_manager.py`
+
+**新增属性：**
+- `_semaphore: Optional[asyncio.Semaphore]` - 并发控制信号量
+- `_max_concurrent_jobs: int` - 最大并发数（存储配置值）
+
+**修改方法：**
+- `__init__()` - 初始化 semaphore 和 max_concurrent_jobs 属性
+- `initialize()` - 创建 Semaphore 实例
+- `execute_job()` - 使用 `async with self._semaphore` 包裹执行逻辑
+
+**新增方法：**
+- `get_concurrency_status()` - 返回并发状态（最大并发数、可用槽位、运行中任务数）
+
+### 3. `src/functional_scaffold/api/models.py`
+
+**新增模型：**
+```python
+class ConcurrencyStatusResponse(BaseModel):
+    """并发状态响应"""
+    max_concurrent: int
+    available_slots: int
+    running_jobs: int
+```
+
+### 4. `src/functional_scaffold/api/routes.py`
+
+**新增端点：**
+```python
+GET /jobs/concurrency/status
+```
+
+返回当前并发执行状态。
+
+### 5. `tests/test_job_manager.py`
+
+**新增测试类：**
+```python
+class TestConcurrencyControl:
+    - test_get_concurrency_status()
+    - test_get_concurrency_status_without_semaphore()
+    - test_concurrency_limit()
+    - test_concurrency_status_api()
+```
+
+**修改测试：**
+- `test_execute_job()` - 添加 semaphore 初始化
+
+## 工作原理
+
+### 并发控制流程
+
+```
+创建任务 (POST /jobs)
+       │
+       ▼
+asyncio.create_task(execute_job)
+       │
+       ▼
+检查 Redis 和 semaphore 可用性
+       │
+       ▼
+async with self._semaphore:  ← 获取槽位（阻塞直到有可用槽位）
+    │
+    ├─ 更新状态为 running
+    ├─ 执行算法
+    ├─ 更新状态为 completed/failed
+    └─ 发送 webhook
+       │
+       ▼
+自动释放槽位
+```
+
+### 关键设计决策
+
+1. **使用 asyncio.Semaphore**
+   - 简单、高效、无需外部依赖
+   - 自动管理槽位获取和释放
+   - 支持异步等待
+
+2. **在 execute_job 内部使用 semaphore**
+   - 快速失败的检查（Redis 可用性、任务存在性）在 semaphore 外部
+   - 只有真正要执行的任务才占用槽位
+   - 任务完成后自动释放（即使发生异常）
+
+3. **存储 _max_concurrent_jobs**
+   - Semaphore 不暴露最大值属性
+   - 需要单独存储以便 `get_concurrency_status()` 使用
+
+## 测试覆盖
+
+- ✅ 获取并发状态
+- ✅ 未初始化时的并发状态
+- ✅ 并发限制生效（创建超过限制的任务，验证只有限定数量在运行）
+- ✅ API 端点测试
+- ✅ 所有现有测试继续通过（60/60）
+
+## 使用示例
+
+### 配置并发限制
+
+```bash
+# 环境变量
+export MAX_CONCURRENT_JOBS=20
+
+# 或在 .env 文件
+MAX_CONCURRENT_JOBS=20
+```
+
+### 查询并发状态
+
+```bash
+curl http://localhost:8000/jobs/concurrency/status
+```
+
+响应：
+```json
+{
+  "max_concurrent": 10,
+  "available_slots": 7,
+  "running_jobs": 3
+}
+```
+
+### 测试并发控制
+
+```bash
+# 运行测试脚本
+./scripts/test_concurrency.sh
+```
+
+## 性能影响
+
+### 优点
+
+1. **防止资源耗尽**：限制同时运行的任务数
+2. **可预测的负载**：系统负载不会超过配置的限制
+3. **自动排队**：超过限制的任务自动等待
+4. **零开销**：未达到限制时，semaphore 几乎无性能开销
+
+### 注意事项
+
+1. **任务等待**：超过限制的任务会等待，可能导致响应延迟
+2. **内存占用**：等待中的任务仍占用内存（协程对象）
+3. **配置调优**：需要根据实际负载调整并发数
+
+## 监控建议
+
+### Prometheus 查询
+
+```promql
+# 任务创建速率
+rate(jobs_created_total[5m])
+
+# 任务完成速率
+rate(jobs_completed_total[5m])
+
+# 任务积压（创建 - 完成）
+rate(jobs_created_total[5m]) - rate(jobs_completed_total[5m])
+```
+
+### Grafana 面板
+
+建议添加以下面板：
+1. 并发状态时间序列（max_concurrent, available_slots, running_jobs）
+2. 任务创建/完成速率
+3. 任务执行时间分布（P50, P95, P99）
+
+## 未来改进
+
+1. **任务超时机制**：为长时间运行的任务设置超时
+2. **优先级队列**：支持高优先级任务优先执行
+3. **动态调整**：根据系统负载动态调整并发数
+4. **任务取消**：支持取消等待中或运行中的任务
+5. **资源限制**：更细粒度的 CPU、内存限制
+
+## 相关文档
+
+- [并发控制详细文档](./concurrency-control.md)
+- [异步任务接口实现计划](../plans/giggly-hatching-kite.md)
+- [监控指南](./monitoring.md)
+
+## 测试结果
+
+```
+======================== 60 passed, 7 warnings in 1.53s ========================
+```
+
+所有测试通过，包括 4 个新增的并发控制测试。

docs/concurrency-control-quickref.md

@@ -0,0 +1,102 @@
+# 并发控制快速参考
+
+## 配置
+
+```bash
+# 设置最大并发数（默认 10）
+export MAX_CONCURRENT_JOBS=20
+```
+
+## API
+
+### 查询并发状态
+
+```bash
+GET /jobs/concurrency/status
+```
+
+**响应：**
+```json
+{
+  "max_concurrent": 10,      // 最大并发数
+  "available_slots": 7,      // 可用槽位
+  "running_jobs": 3          // 运行中任务数
+}
+```
+
+## 代码示例
+
+### 在 JobManager 中使用
+
+```python
+# 并发控制自动生效，无需额外代码
+job_manager = await get_job_manager()
+job_id = await job_manager.create_job(...)
+
+# 任务会自动排队，等待可用槽位
+asyncio.create_task(job_manager.execute_job(job_id))
+```
+
+### 查询并发状态
+
+```python
+job_manager = await get_job_manager()
+status = job_manager.get_concurrency_status()
+
+print(f"运行中: {status['running_jobs']}/{status['max_concurrent']}")
+print(f"可用槽位: {status['available_slots']}")
+```
+
+## 监控
+
+### 实时监控
+
+```bash
+# 持续监控并发状态
+watch -n 1 'curl -s http://localhost:8000/jobs/concurrency/status | jq'
+```
+
+### 测试脚本
+
+```bash
+# 运行并发控制测试
+./scripts/test_concurrency.sh
+```
+
+## 推荐配置
+
+| 任务类型 | 推荐并发数 |
+|---------|-----------|
+| CPU 密集型 | 核心数 × 1.5 |
+| I/O 密集型 | 核心数 × 5-10 |
+| 混合型 | 核心数 × 2-3 |
+
+## 故障排查
+
+### 任务一直 pending
+
+```bash
+# 检查并发状态
+curl http://localhost:8000/jobs/concurrency/status
+
+# 如果 available_slots = 0，说明所有槽位被占用
+# 解决方案：
+# 1. 等待当前任务完成
+# 2. 增加并发限制
+# 3. 优化算法性能
+```
+
+### 系统资源耗尽
+
+```bash
+# 降低并发限制
+export MAX_CONCURRENT_JOBS=5
+
+# 重启服务
+./scripts/run_dev.sh
+```
+
+## 相关文档
+
+- [详细文档](./concurrency-control.md)
+- [实现总结](./concurrency-control-changelog.md)

docs/concurrency-control.md

@@ -0,0 +1,204 @@
+# 异步任务并发控制
+
+## 概述
+
+为了防止系统资源耗尽和控制负载，任务管理器实现了基于 `asyncio.Semaphore` 的并发控制机制。
+
+## 配置
+
+在 `config.py` 或环境变量中设置最大并发任务数：
+
+```python
+# config.py
+max_concurrent_jobs: int = 10  # 默认值
+```
+
+或通过环境变量：
+
+```bash
+export MAX_CONCURRENT_JOBS=20
+```
+
+## 工作原理
+
+1. **信号量机制**：使用 `asyncio.Semaphore` 限制同时运行的任务数
+2. **自动管理**：任务开始时获取槽位，完成后自动释放
+3. **队列等待**：超过限制的任务会自动等待，直到有可用槽位
+
+### 执行流程
+
+```
+POST /jobs 创建任务
+       │
+       ▼
+asyncio.create_task(execute_job)
+       │
+       ▼
+等待获取 semaphore 槽位
+       │
+       ▼
+async with semaphore:  ← 获取槽位
+    执行算法
+    更新状态
+    发送 webhook
+       │
+       ▼
+自动释放槽位
+```
+
+## API 端点
+
+### 查询并发状态
+
+```bash
+GET /jobs/concurrency/status
+```
+
+**响应示例：**
+
+```json
+{
+  "max_concurrent": 10,
+  "available_slots": 7,
+  "running_jobs": 3
+}
+```
+
+**字段说明：**
+
+- `max_concurrent`: 最大并发任务数（配置值）
+- `available_slots`: 当前可用槽位数
+- `running_jobs`: 当前正在运行的任务数
+
+## 使用示例
+
+### 1. 创建多个任务
+
+```bash
+# 创建 20 个任务
+for i in {1..20}; do
+  curl -X POST http://localhost:8000/jobs \
+    -H "Content-Type: application/json" \
+    -d "{\"algorithm\": \"PrimeChecker\", \"params\": {\"number\": $i}}"
+done
+```
+
+### 2. 监控并发状态
+
+```bash
+# 持续监控并发状态
+watch -n 1 'curl -s http://localhost:8000/jobs/concurrency/status | jq'
+```
+
+输出示例：
+
+```json
+{
+  "max_concurrent": 10,
+  "available_slots": 0,
+  "running_jobs": 10
+}
+```
+
+### 3. 调整并发限制
+
+```bash
+# 重启服务前设置环境变量
+export MAX_CONCURRENT_JOBS=20
+./scripts/run_dev.sh
+```
+
+## 性能考虑
+
+### 选择合适的并发数
+
+并发数应根据以下因素确定：
+
+1. **CPU 核心数**：CPU 密集型任务建议设置为核心数的 1-2 倍
+2. **内存限制**：每个任务的内存占用 × 并发数 < 可用内存
+3. **外部服务限制**：如果调用外部 API，考虑其速率限制
+4. **Redis 连接池**：确保 Redis 连接池大小 ≥ 并发数
+
+### 推荐配置
+
+| 场景 | 推荐并发数 | 说明 |
+|------|-----------|------|
+| CPU 密集型（如质数判断） | 核心数 × 1.5 | 充分利用 CPU |
+| I/O 密集型（如网络请求） | 核心数 × 5-10 | 等待 I/O 时可切换 |
+| 混合型 | 核心数 × 2-3 | 平衡 CPU 和 I/O |
+| 内存受限 | 根据内存计算 | 避免 OOM |
+
+### 示例计算
+
+假设：
+- 服务器：4 核 8GB 内存
+- 任务类型：I/O 密集型（网络请求）
+- 单任务内存：50MB
+
+```
+最大并发数 = min(
+    核心数 × 8 = 32,
+    可用内存 / 单任务内存 = 8000MB / 50MB = 160
+) = 32
+```
+
+## 监控指标
+
+相关 Prometheus 指标：
+
+```promql
+# 任务创建速率
+rate(jobs_created_total[5m])
+
+# 任务完成速率
+rate(jobs_completed_total[5m])
+
+# 任务执行时间分布
+histogram_quantile(0.95, job_execution_duration_seconds_bucket)
+```
+
+## 故障排查
+
+### 问题：任务一直处于 pending 状态
+
+**可能原因：**
+1. 所有槽位都被占用
+2. 某些任务执行时间过长
+
+**解决方案：**
+```bash
+# 1. 检查并发状态
+curl http://localhost:8000/jobs/concurrency/status
+
+# 2. 如果 available_slots = 0，说明所有槽位被占用
+# 3. 检查是否有长时间运行的任务
+# 4. 考虑增加并发限制或优化算法性能
+```
+
+### 问题：系统资源耗尽
+
+**可能原因：**
+并发数设置过高
+
+**解决方案：**
+```bash
+# 降低并发限制
+export MAX_CONCURRENT_JOBS=5
+# 重启服务
+```
+
+## 最佳实践
+
+1. **监控优先**：部署后持续监控并发状态和系统资源
+2. **逐步调整**：从保守值开始，逐步增加并发数
+3. **压力测试**：在生产环境前进行充分的压力测试
+4. **设置告警**：当 `available_slots = 0` 持续时间过长时告警
+5. **任务超时**：为长时间运行的任务设置超时机制（待实现）
+
+## 未来改进
+
+- [ ] 任务超时机制
+- [ ] 优先级队列
+- [ ] 动态调整并发数
+- [ ] 任务取消功能
+- [ ] 更细粒度的资源控制（CPU、内存限制）

docs/fc-deploy.md

@@ -0,0 +1,58 @@
+# 阿里云 函数运算FC 部署入门
+
+本指南帮助快速上手 FunctionalScaffold 脚手架，在 10 分钟内完成第一个算法服务的开发和部署。
+
+## 环境准备
+
+- 安装 [Serverless Devs CLI](https://serverless-devs.com/docs/overview)
+
+1. 首先安装Node 环境，在Node官网下载
+   - [Node.js 下载地址](https://nodejs.org/en/download/)
+2. 安装 Serverless Devs CLI
+
+```bash
+npm install @serverless-devs/s -g
+```
+
+## 初始化 serverless dev cli 配置
+
+执行以下命令初始化 serverless dev cli 配置
+
+```bash
+s config add
+```
+
+根据引导进行操作，填入你的access key id 和 access key secret
+
+## 部署算法服务
+
+部署算法服务前，请确保已经完成环境准备和配置。
+
+修改 `s.yaml` 文件中的 vars 部分
+
+```yaml
+# 阿里云函数计算 FC 3.0 配置
+# 使用 Serverless Devs 部署: cd deployment/serverless && s deploy
+edition: 3.0.0
+name: functional-scaffold
+access: default
+
+vars:
+  region: cn-hangzhou # 换成你的区域
+  image: registry.cn-hangzhou.aliyuncs.com/your-namespace/functional-scaffold:latest # 换成你的docker 镜像
+  redis_host: r-xxxxx.redis.rds.aliyuncs.com # 换成你的redis连接
+  redis_port: "6379" # redis 端口号
+  redis_password: "your-password" #redis 密码，如果没有可留空
+```
+
+```bash
+cd deployment && s deploy
+```
+
+部署完成后，可以在控制台查看服务的运行状态和日志。
+
+## 删除算法服务
+
+```bash
+cd deployment && s remove
+```

docs/getting-started.md

@@ -34,7 +34,9 @@ pip install -e ".[dev]"
 
 ```bash
 # 开发模式（自动重载）
-uvicorn src.functional_scaffold.main:app --reload --port 8000
+uvicorn functional_scaffold.main:app --reload --port 8000
+# docker 开发者模式
+cd deployment && docker compose up -d
 ```
 
 ### 3. 验证服务

docs/grafana-dashboard-usage.md

@@ -0,0 +1,182 @@
+# Grafana 日志仪表板使用说明
+
+## Request ID 过滤功能
+
+日志监控仪表板现在支持按 request_id 过滤日志，可以追踪单个请求的完整生命周期。
+
+### 如何使用
+
+1. **访问仪表板**
+   - 打开 Grafana: http://localhost:3000
+   - 登录（admin/admin）
+   - 进入 "日志监控" 仪表板
+
+2. **使用 Request ID 过滤**
+   - 在仪表板顶部找到 "Request ID" 输入框
+   - 输入完整的 request_id（例如：`59017bdd-5963-40b1-a325-5088593382c0`）
+   - 所有面板会自动更新，只显示该 request_id 的日志
+
+3. **查看所有日志**
+   - 清空 "Request ID" 输入框
+   - 所有面板会显示所有日志
+
+### 示例
+
+#### 获取 Request ID
+
+从 API 响应中获取：
+
+```bash
+curl -X POST http://localhost:8111/invoke \
+  -H "Content-Type: application/json" \
+  -d '{"number": 17}' | jq -r '.request_id'
+```
+
+输出示例：
+```
+59017bdd-5963-40b1-a325-5088593382c0
+```
+
+#### 在仪表板中过滤
+
+1. 复制上面的 request_id
+2. 在 Grafana 仪表板顶部的 "Request ID" 输入框中粘贴
+3. 按回车或点击刷新
+
+#### 查看结果
+
+过滤后，你会看到该请求的所有日志：
+
+- **日志流面板**：显示该请求的所有日志条目
+- **日志量趋势**：显示该请求的日志分布
+- **日志级别分布**：显示该请求的日志级别统计
+- **错误日志**：如果该请求有错误，会显示在这里
+
+### 典型的请求日志流
+
+一个成功的请求通常包含以下日志：
+
+```
+1. Request: POST /invoke
+2. Processing request {request_id} with number=17
+3. Starting algorithm: PrimeChecker
+4. Algorithm PrimeChecker completed successfully in 0.001s
+5. Response: 200
+```
+
+所有这些日志都有相同的 request_id，可以通过过滤功能一起查看。
+
+### 高级用法
+
+#### 在 Explore 中使用
+
+1. 进入 Grafana Explore: http://localhost:3000/explore
+2. 选择 Loki 数据源
+3. 使用以下查询：
+
+```logql
+# 查询特定 request_id
+{job="functional-scaffold-app"} |= "59017bdd-5963-40b1-a325-5088593382c0"
+
+# 使用 JSON 解析（更精确）
+{job="functional-scaffold-app"} | json | request_id="59017bdd-5963-40b1-a325-5088593382c0"
+
+# 查询特定 request_id 的错误日志
+{job="functional-scaffold-app", level="ERROR"} |= "59017bdd-5963-40b1-a325-5088593382c0"
+```
+
+#### 组合过滤
+
+可以结合其他过滤条件：
+
+```logql
+# 特定 request_id 的 ERROR 日志
+{job="functional-scaffold-app", level="ERROR"} |= "59017bdd-5963-40b1-a325-5088593382c0"
+
+# 特定 request_id 的特定 logger
+{job="functional-scaffold-app", logger="functional_scaffold.algorithms.base"} |= "59017bdd-5963-40b1-a325-5088593382c0"
+```
+
+### 故障排查
+
+#### Request ID 过滤不生效
+
+1. **检查 request_id 格式**
+   - 确保输入的是完整的 UUID 格式
+   - 不要包含额外的空格或引号
+
+2. **检查时间范围**
+   - 确保仪表板的时间范围包含该请求的时间
+   - 可以调整为 "Last 15 minutes" 或更长
+
+3. **刷新仪表板**
+   - 点击右上角的刷新按钮
+   - 或者按 Ctrl+R (Cmd+R on Mac)
+
+4. **验证日志是否存在**
+   - 在 Explore 中手动查询：
+     ```logql
+     {job="functional-scaffold-app"} |= "your-request-id"
+     ```
+   - 如果没有结果，说明日志还没有被收集
+
+#### 日志延迟
+
+- Promtail 每 5 秒刷新一次
+- Loki 可能有几秒的延迟
+- 建议等待 5-10 秒后再查询
+
+### 最佳实践
+
+1. **调试单个请求**
+   - 发送请求并记录 request_id
+   - 在仪表板中输入 request_id
+   - 查看完整的请求处理流程
+
+2. **追踪错误**
+   - 当发现错误时，从错误日志中提取 request_id
+   - 使用 request_id 过滤查看完整的请求上下文
+   - 分析错误发生前后的日志
+
+3. **性能分析**
+   - 使用 request_id 过滤慢请求
+   - 查看算法执行时间
+   - 分析性能瓶颈
+
+4. **用户问题排查**
+   - 从用户报告中获取 request_id（如果有）
+   - 使用 request_id 重现问题场景
+   - 查看完整的请求处理过程
+
+### 技术细节
+
+#### 过滤实现
+
+仪表板使用 LogQL 的文本匹配操作符 `|=`：
+
+```logql
+{job="functional-scaffold-app"} |= "$request_id"
+```
+
+- 当 `$request_id` 为空时，`|= ""` 匹配所有日志
+- 当 `$request_id` 有值时，只匹配包含该字符串的日志
+
+#### 性能考虑
+
+- 文本匹配 (`|=`) 比 JSON 解析更快
+- 适合实时查询和仪表板
+- 对于精确匹配，可以在 Explore 中使用 JSON 解析
+
+#### 变量配置
+
+Request ID 变量配置：
+- 类型：textbox（文本输入框）
+- 名称：request_id
+- 标签：Request ID
+- 默认值：空字符串
+
+## 相关文档
+
+- [Loki 集成文档](loki-integration.md)
+- [Loki 快速参考](loki-quick-reference.md)
+- [LogQL 查询语言](https://grafana.com/docs/loki/latest/logql/)

docs/kubernetes-deployment.md

@@ -0,0 +1,307 @@
+# Kubernetes 部署指南
+
+本文档介绍如何在 Kubernetes 集群中部署 FunctionalScaffold 服务。
+
+## 架构概览
+
+```
+                    ┌─────────────────┐
+                    │   Ingress/LB    │
+                    └────────┬────────┘
+                             │
+                    ┌────────▼────────┐
+                    │   API Service   │
+                    │   (ClusterIP)   │
+                    └────────┬────────┘
+                             │
+              ┌──────────────┼──────────────┐
+              │              │              │
+       ┌──────▼──────┐ ┌─────▼─────┐ ┌─────▼─────┐
+       │  API Pod 1  │ │ API Pod 2 │ │ API Pod 3 │
+       └─────────────┘ └───────────┘ └───────────┘
+                             │
+                    ┌────────▼────────┐
+                    │  Redis Service  │
+                    └────────┬────────┘
+                             │
+              ┌──────────────┼──────────────┐
+              │              │              │
+       ┌──────▼──────┐ ┌─────▼─────┐        │
+       │ Worker Pod 1│ │Worker Pod2│        │
+       └─────────────┘ └───────────┘        │
+                                     ┌──────▼──────┐
+                                     │  Redis Pod  │
+                                     └─────────────┘
+```
+
+## 组件说明
+
+| 组件 | 副本数 | 说明 |
+|------|--------|------|
+| **API Deployment** | 3 | HTTP 服务，处理同步请求和任务创建 |
+| **Worker Deployment** | 2 | 异步任务处理，从 Redis 队列消费任务 |
+| **Redis Deployment** | 1 | 任务队列和状态存储 |
+| **ConfigMap** | - | 共享配置管理 |
+
+## 快速部署
+
+```bash
+# 部署所有资源
+kubectl apply -f deployment/kubernetes/deployment.yaml
+kubectl apply -f deployment/kubernetes/service.yaml
+
+# 查看部署状态
+kubectl get pods -l app=functional-scaffold
+kubectl get svc -l app=functional-scaffold
+```
+
+## 配置文件说明
+
+### deployment.yaml
+
+包含以下资源：
+
+#### ConfigMap
+
+```yaml
+apiVersion: v1
+kind: ConfigMap
+metadata:
+  name: functional-scaffold-config
+data:
+  APP_ENV: "production"
+  LOG_LEVEL: "INFO"
+  REDIS_HOST: "functional-scaffold-redis"
+  # ... 更多配置
+```
+
+主要配置项：
+
+| 配置项 | 默认值 | 说明 |
+|--------|--------|------|
+| `APP_ENV` | production | 运行环境 |
+| `LOG_LEVEL` | INFO | 日志级别 |
+| `REDIS_HOST` | functional-scaffold-redis | Redis 服务地址 |
+| `MAX_CONCURRENT_JOBS` | 10 | 最大并发任务数 |
+| `JOB_EXECUTION_TIMEOUT` | 300 | 任务执行超时（秒） |
+
+#### API Deployment
+
+- **副本数**: 3
+- **资源限制**: 256Mi-512Mi 内存，250m-500m CPU
+- **健康检查**: `/healthz`（存活）、`/readyz`（就绪）
+- **环境变量**: `RUN_MODE=api`
+
+#### Worker Deployment
+
+- **副本数**: 2
+- **资源限制**: 256Mi-512Mi 内存，250m-500m CPU
+- **健康检查**: exec 探针检查 Redis 连接
+- **环境变量**: `RUN_MODE=worker`
+
+#### Redis Deployment
+
+- **副本数**: 1
+- **资源限制**: 128Mi-256Mi 内存，100m-200m CPU
+- **持久化**: AOF 模式（appendonly yes）
+- **存储**: emptyDir（开发环境）
+
+### service.yaml
+
+| Service | 类型 | 端口 | 说明 |
+|---------|------|------|------|
+| `functional-scaffold-api` | ClusterIP | 80 → 8000 | API 服务 |
+| `functional-scaffold-metrics` | ClusterIP | 8000 | Prometheus 指标 |
+| `functional-scaffold-redis` | ClusterIP | 6379 | Redis 服务 |
+
+## 生产环境建议
+
+### 1. 使用外部 Redis
+
+生产环境建议使用托管 Redis 服务（如阿里云 Redis、AWS ElastiCache）：
+
+```yaml
+# 修改 ConfigMap
+data:
+  REDIS_HOST: "r-xxxxx.redis.rds.aliyuncs.com"
+  REDIS_PORT: "6379"
+  REDIS_PASSWORD: ""  # 使用 Secret 管理
+```
+
+### 2. 使用 Secret 管理敏感信息
+
+```yaml
+apiVersion: v1
+kind: Secret
+metadata:
+  name: functional-scaffold-secrets
+type: Opaque
+stringData:
+  REDIS_PASSWORD: "your-password"
+  DATABASE_URL: "postgresql://..."
+```
+
+在 Deployment 中引用：
+
+```yaml
+envFrom:
+- configMapRef:
+    name: functional-scaffold-config
+- secretRef:
+    name: functional-scaffold-secrets
+```
+
+### 3. 配置 HPA 自动扩缩容
+
+```yaml
+apiVersion: autoscaling/v2
+kind: HorizontalPodAutoscaler
+metadata:
+  name: functional-scaffold-api-hpa
+spec:
+  scaleTargetRef:
+    apiVersion: apps/v1
+    kind: Deployment
+    name: functional-scaffold-api
+  minReplicas: 2
+  maxReplicas: 10
+  metrics:
+  - type: Resource
+    resource:
+      name: cpu
+      target:
+        type: Utilization
+        averageUtilization: 70
+```
+
+### 4. 配置 PDB 保证可用性
+
+```yaml
+apiVersion: policy/v1
+kind: PodDisruptionBudget
+metadata:
+  name: functional-scaffold-api-pdb
+spec:
+  minAvailable: 2
+  selector:
+    matchLabels:
+      app: functional-scaffold
+      component: api
+```
+
+### 5. 使用 PVC 持久化 Redis 数据
+
+```yaml
+apiVersion: v1
+kind: PersistentVolumeClaim
+metadata:
+  name: redis-data-pvc
+spec:
+  accessModes:
+    - ReadWriteOnce
+  resources:
+    requests:
+      storage: 10Gi
+```
+
+## 监控集成
+
+### Prometheus 抓取配置
+
+`functional-scaffold-metrics` Service 已添加 Prometheus 注解：
+
+```yaml
+annotations:
+  prometheus.io/scrape: "true"
+  prometheus.io/port: "8000"
+  prometheus.io/path: "/metrics"
+```
+
+### ServiceMonitor（如使用 Prometheus Operator）
+
+```yaml
+apiVersion: monitoring.coreos.com/v1
+kind: ServiceMonitor
+metadata:
+  name: functional-scaffold
+spec:
+  selector:
+    matchLabels:
+      app: functional-scaffold
+      component: api
+  endpoints:
+  - port: metrics
+    path: /metrics
+    interval: 30s
+```
+
+## 常用命令
+
+```bash
+# 查看所有资源
+kubectl get all -l app=functional-scaffold
+
+# 查看 Pod 日志
+kubectl logs -l app=functional-scaffold,component=api -f
+kubectl logs -l app=functional-scaffold,component=worker -f
+
+# 扩缩容
+kubectl scale deployment functional-scaffold-api --replicas=5
+kubectl scale deployment functional-scaffold-worker --replicas=3
+
+# 滚动更新
+kubectl set image deployment/functional-scaffold-api \
+  api=functional-scaffold:v2.0.0
+
+# 回滚
+kubectl rollout undo deployment/functional-scaffold-api
+
+# 查看部署历史
+kubectl rollout history deployment/functional-scaffold-api
+
+# 进入 Pod 调试
+kubectl exec -it <pod-name> -- /bin/sh
+
+# 端口转发（本地调试）
+kubectl port-forward svc/functional-scaffold-api 8000:80
+```
+
+## 故障排查
+
+### Pod 启动失败
+
+```bash
+# 查看 Pod 事件
+kubectl describe pod <pod-name>
+
+# 查看 Pod 日志
+kubectl logs <pod-name> --previous
+```
+
+### Redis 连接失败
+
+```bash
+# 检查 Redis Service
+kubectl get svc functional-scaffold-redis
+
+# 测试 Redis 连接
+kubectl run redis-test --rm -it --image=redis:7-alpine -- \
+  redis-cli -h functional-scaffold-redis ping
+```
+
+### Worker 不消费任务
+
+```bash
+# 检查 Worker 日志
+kubectl logs -l component=worker -f
+
+# 检查 Redis 队列
+kubectl exec -it <redis-pod> -- redis-cli LLEN job:queue
+```
+
+## 相关文档
+
+- [快速入门](getting-started.md)
+- [监控指南](monitoring.md)
+- [并发控制](concurrency-control.md)
+- [日志集成](loki-quick-reference.md)

docs/loki-integration.md

@@ -0,0 +1,564 @@
+# Loki 日志收集系统集成文档
+
+## 概述
+
+本项目已集成 Grafana Loki 日志收集系统，支持两种日志收集模式：
+
+1. **Docker stdio 收集**（推荐）- 从容器标准输出/错误收集日志
+2. **Log 文件收集**（备用）- 从日志文件收集日志
+
+## 架构
+
+```
+应用容器 (stdout/stderr)
+    ↓
+Docker Engine
+    ↓
+Promtail (日志采集器)
+    ↓
+Loki (日志存储)
+    ↓
+Grafana (可视化)
+```
+
+## 快速开始
+
+### 1. 启动服务
+
+```bash
+cd deployment
+docker-compose up -d
+```
+
+这将启动以下服务：
+- **app**: 应用服务 (端口 8111)
+- **loki**: 日志存储服务 (端口 3100)
+- **promtail**: 日志采集服务 (端口 9080)
+- **grafana**: 可视化服务 (端口 3000)
+- **prometheus**: 指标收集服务 (端口 9090)
+- **redis**: 缓存服务 (端口 6380)
+
+### 2. 访问 Grafana
+
+1. 打开浏览器访问 http://localhost:3000
+2. 使用默认凭据登录：
+   - 用户名: `admin`
+   - 密码: `admin`
+3. 首次登录后建议修改密码
+
+### 3. 查看日志
+
+#### 方式 1: 使用预配置的日志仪表板
+
+1. 在 Grafana 左侧菜单点击 **Dashboards**
+2. 选择 **日志监控** 仪表板
+3. 查看以下面板：
+   - **日志流 (实时)**: 实时日志流
+   - **日志量趋势（按级别）**: 时间序列图表
+   - **日志级别分布**: 按级别统计
+   - **错误日志**: 只显示 ERROR 级别日志
+
+#### 方式 2: 使用 Explore 功能
+
+1. 在 Grafana 左侧菜单点击 **Explore** (指南针图标)
+2. 选择 **Loki** 数据源
+3. 输入 LogQL 查询语句（见下文）
+
+## LogQL 查询示例
+
+### 基础查询
+
+```logql
+# 查询所有应用日志
+{job="functional-scaffold-app"}
+
+# 查询特定级别的日志
+{job="functional-scaffold-app", level="ERROR"}
+{job="functional-scaffold-app", level="INFO"}
+
+# 查询特定容器的日志
+{container="functional-scaffold-app-1"}
+```
+
+### 文本过滤
+
+```logql
+# 包含特定文本
+{job="functional-scaffold-app"} |= "request_id"
+
+# 不包含特定文本
+{job="functional-scaffold-app"} != "healthz"
+
+# 正则表达式匹配
+{job="functional-scaffold-app"} |~ "error|exception"
+
+# 正则表达式不匹配
+{job="functional-scaffold-app"} !~ "debug|trace"
+```
+
+### JSON 字段提取
+
+```logql
+# 提取 request_id 字段
+{job="functional-scaffold-app"} | json | request_id != ""
+
+# 提取并过滤特定 request_id
+{job="functional-scaffold-app"} | json | request_id = "abc123"
+
+# 提取 logger 字段
+{job="functional-scaffold-app"} | json | logger = "functional_scaffold.api.routes"
+```
+
+### 聚合查询
+
+```logql
+# 统计日志数量
+count_over_time({job="functional-scaffold-app"}[5m])
+
+# 按级别统计
+sum by (level) (count_over_time({job="functional-scaffold-app"}[5m]))
+
+# 计算错误率
+sum(rate({job="functional-scaffold-app", level="ERROR"}[5m]))
+/
+sum(rate({job="functional-scaffold-app"}[5m]))
+```
+
+## 日志收集模式
+
+### 模式 1: Docker stdio 收集（默认，推荐）
+
+**特点：**
+- 无需修改应用代码
+- 自动收集容器标准输出/错误
+- 性能影响极小
+- 配置简单
+
+**工作原理：**
+1. 应用将日志输出到 stdout/stderr
+2. Docker Engine 捕获日志
+3. Promtail 通过 Docker API 读取日志
+4. 日志发送到 Loki 存储
+
+**配置：**
+- 应用容器需要添加标签：
+  ```yaml
+  labels:
+    logging: "promtail"
+    logging_jobname: "functional-scaffold-app"
+  ```
+
+### 模式 2: Log 文件收集（备用）
+
+**特点：**
+- 日志持久化到文件
+- 支持日志轮转
+- 适合需要本地日志文件的场景
+
+**启用方式：**
+
+1. 修改 `deployment/docker-compose.yml`：
+   ```yaml
+   environment:
+     - LOG_FILE_ENABLED=true
+     - LOG_FILE_PATH=/var/log/app/app.log
+   ```
+
+2. 重启服务：
+   ```bash
+   docker-compose up -d app
+   ```
+
+**日志文件配置：**
+- 最大文件大小: 100MB
+- 保留备份数: 5 个
+- 总存储空间: 最多 500MB
+
+## 配置说明
+
+### Loki 配置 (monitoring/loki.yaml)
+
+```yaml
+limits_config:
+  retention_period: 168h  # 日志保留 7 天
+  ingestion_rate_mb: 10   # 摄入速率限制 10MB/s
+  ingestion_burst_size_mb: 20  # 突发大小 20MB
+```
+
+**可调整参数：**
+- `retention_period`: 日志保留时间（默认 7 天）
+- `ingestion_rate_mb`: 每秒摄入速率限制
+- `ingestion_burst_size_mb`: 突发流量大小
+
+### Promtail 配置 (monitoring/promtail.yaml)
+
+**Docker stdio 收集配置：**
+```yaml
+scrape_configs:
+  - job_name: docker
+    docker_sd_configs:
+      - host: unix:///var/run/docker.sock
+        filters:
+          - name: label
+            values: ["logging=promtail"]
+```
+
+**文件收集配置：**
+```yaml
+scrape_configs:
+  - job_name: app_files
+    static_configs:
+      - targets:
+          - localhost
+        labels:
+          job: functional-scaffold-app-files
+          __path__: /var/log/app/*.log
+```
+
+## 验证和测试
+
+### 1. 检查服务状态
+
+```bash
+# 查看所有服务
+docker-compose ps
+
+# 检查 Loki 健康状态
+curl http://localhost:3100/ready
+
+# 检查 Promtail 健康状态
+curl http://localhost:9080/ready
+```
+
+### 2. 生成测试日志
+
+```bash
+# 发送测试请求
+curl -X POST http://localhost:8111/invoke \
+  -H "Content-Type: application/json" \
+  -d '{"algorithm": "PrimeChecker", "params": {"number": 17}}'
+```
+
+### 3. 查询日志
+
+```bash
+# 使用 Loki API 查询
+curl -G -s "http://localhost:3100/loki/api/v1/query_range" \
+  --data-urlencode 'query={job="functional-scaffold-app"}' \
+  --data-urlencode 'limit=10' \
+  | jq '.data.result'
+```
+
+### 4. 在 Grafana 中验证
+
+1. 访问 http://localhost:3000/explore
+2. 选择 Loki 数据源
+3. 输入查询: `{job="functional-scaffold-app"}`
+4. 应该能看到应用日志
+
+## 故障排查
+
+### 问题 1: 看不到日志
+
+**检查步骤：**
+
+1. 确认 Promtail 正在运行：
+   ```bash
+   docker-compose ps promtail
+   ```
+
+2. 检查 Promtail 日志：
+   ```bash
+   docker-compose logs promtail
+   ```
+
+3. 确认应用容器有正确的标签：
+   ```bash
+   docker inspect functional-scaffold-app-1 | grep -A 5 Labels
+   ```
+
+4. 检查 Loki 是否接收到日志：
+   ```bash
+   curl -G -s "http://localhost:3100/loki/api/v1/label/job/values" | jq
+   ```
+
+### 问题 2: Promtail 无法访问 Docker socket
+
+**错误信息：**
+```
+permission denied while trying to connect to the Docker daemon socket
+```
+
+**解决方案：**
+
+在 macOS/Linux 上，确保 Docker socket 权限正确：
+```bash
+sudo chmod 666 /var/run/docker.sock
+```
+
+或者将 Promtail 容器添加到 docker 组（Linux）：
+```yaml
+promtail:
+  user: root
+  group_add:
+    - docker
+```
+
+### 问题 3: 日志量过大
+
+**症状：**
+- Loki 响应缓慢
+- 磁盘空间不足
+
+**解决方案：**
+
+1. 调整日志保留期：
+   ```yaml
+   # monitoring/loki.yaml
+   limits_config:
+     retention_period: 72h  # 改为 3 天
+   ```
+
+2. 增加摄入速率限制：
+   ```yaml
+   limits_config:
+     ingestion_rate_mb: 5  # 降低到 5MB/s
+   ```
+
+3. 添加日志过滤：
+   ```yaml
+   # monitoring/promtail.yaml
+   pipeline_stages:
+     - match:
+         selector: '{job="functional-scaffold-app"}'
+         stages:
+           - drop:
+               expression: ".*healthz.*"  # 丢弃健康检查日志
+   ```
+
+### 问题 4: 文件模式下看不到日志
+
+**检查步骤：**
+
+1. 确认文件日志已启用：
+   ```bash
+   docker-compose exec app env | grep LOG_FILE
+   ```
+
+2. 检查日志文件是否存在：
+   ```bash
+   docker-compose exec app ls -lh /var/log/app/
+   ```
+
+3. 检查 Promtail 是否能访问日志文件：
+   ```bash
+   docker-compose exec promtail ls -lh /var/log/app/
+   ```
+
+## 性能优化
+
+### 1. 减少日志量
+
+**在应用层面：**
+- 调整日志级别为 WARNING 或 ERROR
+- 过滤掉不必要的日志（如健康检查）
+
+```yaml
+# docker-compose.yml
+environment:
+  - LOG_LEVEL=WARNING
+```
+
+**在 Promtail 层面：**
+```yaml
+# monitoring/promtail.yaml
+pipeline_stages:
+  - drop:
+      expression: ".*healthz.*"
+      drop_counter_reason: "healthcheck"
+```
+
+### 2. 优化查询性能
+
+**使用标签过滤：**
+```logql
+# 好：使用标签过滤（快）
+{job="functional-scaffold-app", level="ERROR"}
+
+# 差：使用文本过滤（慢）
+{job="functional-scaffold-app"} |= "ERROR"
+```
+
+**限制时间范围：**
+```logql
+# 查询最近 5 分钟
+{job="functional-scaffold-app"}[5m]
+
+# 避免查询过长时间范围
+{job="functional-scaffold-app"}[7d]  # 慢
+```
+
+### 3. 存储优化
+
+**定期清理旧数据：**
+```bash
+# Loki 会自动根据 retention_period 清理
+# 也可以手动清理
+docker-compose exec loki rm -rf /loki/chunks/*
+```
+
+**监控磁盘使用：**
+```bash
+docker-compose exec loki du -sh /loki/chunks
+```
+
+## 高级功能
+
+### 1. 告警规则
+
+在 Loki 中配置告警规则（需要 Loki Ruler）：
+
+```yaml
+# monitoring/loki-rules.yaml
+groups:
+  - name: error_alerts
+    interval: 1m
+    rules:
+      - alert: HighErrorRate
+        expr: |
+          sum(rate({job="functional-scaffold-app", level="ERROR"}[5m]))
+          /
+          sum(rate({job="functional-scaffold-app"}[5m]))
+          > 0.05
+        for: 5m
+        labels:
+          severity: warning
+        annotations:
+          summary: "错误率过高"
+          description: "应用错误率超过 5%"
+```
+
+### 2. 日志导出
+
+**导出为 JSON：**
+```bash
+curl -G -s "http://localhost:3100/loki/api/v1/query_range" \
+  --data-urlencode 'query={job="functional-scaffold-app"}' \
+  --data-urlencode 'start=2024-01-01T00:00:00Z' \
+  --data-urlencode 'end=2024-01-02T00:00:00Z' \
+  | jq '.data.result' > logs.json
+```
+
+**导出为文本：**
+```bash
+curl -G -s "http://localhost:3100/loki/api/v1/query_range" \
+  --data-urlencode 'query={job="functional-scaffold-app"}' \
+  | jq -r '.data.result[].values[][1]' > logs.txt
+```
+
+### 3. 与 Prometheus 集成
+
+在 Grafana 仪表板中同时显示日志和指标：
+
+```json
+{
+  "panels": [
+    {
+      "title": "错误率和错误日志",
+      "targets": [
+        {
+          "datasource": "Prometheus",
+          "expr": "rate(http_requests_total{status=\"error\"}[5m])"
+        },
+        {
+          "datasource": "Loki",
+          "expr": "{job=\"functional-scaffold-app\", level=\"ERROR\"}"
+        }
+      ]
+    }
+  ]
+}
+```
+
+## 最佳实践
+
+### 1. 日志格式
+
+**使用结构化日志（JSON）：**
+```python
+logger.info("处理请求", extra={
+    "request_id": "abc123",
+    "user_id": "user456",
+    "duration": 0.123
+})
+```
+
+**输出：**
+```json
+{
+  "asctime": "2024-01-01 12:00:00,000",
+  "name": "functional_scaffold.api.routes",
+  "levelname": "INFO",
+  "message": "处理请求",
+  "request_id": "abc123",
+  "user_id": "user456",
+  "duration": 0.123
+}
+```
+
+### 2. 标签策略
+
+**好的标签：**
+- 低基数（值的种类少）
+- 用于过滤和分组
+- 例如：`level`, `logger`, `container`
+
+**不好的标签：**
+- 高基数（值的种类多）
+- 例如：`request_id`, `user_id`, `timestamp`
+
+**正确做法：**
+```logql
+# 使用标签过滤
+{job="functional-scaffold-app", level="ERROR"}
+
+# 使用 JSON 提取高基数字段
+{job="functional-scaffold-app"} | json | request_id = "abc123"
+```
+
+### 3. 查询优化
+
+**使用时间范围：**
+```logql
+{job="functional-scaffold-app"}[5m]  # 最近 5 分钟
+```
+
+**限制返回行数：**
+```logql
+{job="functional-scaffold-app"} | limit 100
+```
+
+**使用聚合减少数据量：**
+```logql
+sum by (level) (count_over_time({job="functional-scaffold-app"}[5m]))
+```
+
+## 参考资料
+
+- [Loki 官方文档](https://grafana.com/docs/loki/latest/)
+- [LogQL 查询语言](https://grafana.com/docs/loki/latest/logql/)
+- [Promtail 配置](https://grafana.com/docs/loki/latest/clients/promtail/configuration/)
+- [Grafana Explore](https://grafana.com/docs/grafana/latest/explore/)
+
+## 总结
+
+本项目的 Loki 集成提供了：
+
+✅ **开箱即用** - 无需额外配置即可收集日志
+✅ **双模式支持** - Docker stdio（默认）和文件收集
+✅ **自动化配置** - 数据源和仪表板自动加载
+✅ **结构化日志** - JSON 格式，支持字段提取
+✅ **高性能** - 低资源占用，快速查询
+✅ **易于扩展** - 支持自定义标签和过滤规则
+
+如有问题，请参考故障排查章节或查阅官方文档。

docs/loki-quick-reference.md

@@ -0,0 +1,237 @@
+# Loki 快速参考
+
+## 常用命令
+
+### 服务管理
+
+```bash
+# 启动所有服务
+cd deployment && docker-compose up -d
+
+# 查看服务状态
+docker-compose ps
+
+# 查看日志
+docker-compose logs -f loki
+docker-compose logs -f promtail
+
+# 重启服务
+docker-compose restart loki promtail
+
+# 停止服务
+docker-compose down
+```
+
+### 健康检查
+
+```bash
+# Loki
+curl http://localhost:3100/ready
+
+# Promtail
+curl http://localhost:9080/ready
+
+# 验证脚本
+./scripts/verify_loki.sh
+```
+
+## 常用 LogQL 查询
+
+### 基础查询
+
+```logql
+# 所有日志
+{job="functional-scaffold-app"}
+
+# 错误日志
+{job="functional-scaffold-app", level="ERROR"}
+
+# 特定时间范围
+{job="functional-scaffold-app"}[5m]
+```
+
+### 文本过滤
+
+```logql
+# 包含文本
+{job="functional-scaffold-app"} |= "error"
+
+# 不包含文本
+{job="functional-scaffold-app"} != "healthz"
+
+# 正则匹配
+{job="functional-scaffold-app"} |~ "error|exception"
+```
+
+### JSON 提取
+
+```logql
+# 提取 request_id
+{job="functional-scaffold-app"} | json | request_id != ""
+
+# 按 request_id 过滤
+{job="functional-scaffold-app"} | json | request_id = "abc123"
+```
+
+### 聚合统计
+
+```logql
+# 日志数量
+count_over_time({job="functional-scaffold-app"}[5m])
+
+# 按级别统计
+sum by (level) (count_over_time({job="functional-scaffold-app"}[5m]))
+
+# 错误率
+sum(rate({job="functional-scaffold-app", level="ERROR"}[5m]))
+/
+sum(rate({job="functional-scaffold-app"}[5m]))
+```
+
+## API 查询
+
+### 查询日志
+
+```bash
+# 查询最近的日志
+curl -G -s "http://localhost:3100/loki/api/v1/query_range" \
+  --data-urlencode 'query={job="functional-scaffold-app"}' \
+  --data-urlencode 'limit=10' \
+  | jq '.data.result'
+
+# 查询错误日志
+curl -G -s "http://localhost:3100/loki/api/v1/query_range" \
+  --data-urlencode 'query={job="functional-scaffold-app", level="ERROR"}' \
+  | jq '.data.result'
+```
+
+### 查询标签
+
+```bash
+# 查询所有 job 标签值
+curl -s "http://localhost:3100/loki/api/v1/label/job/values" | jq
+
+# 查询所有 level 标签值
+curl -s "http://localhost:3100/loki/api/v1/label/level/values" | jq
+```
+
+## 配置切换
+
+### 启用文件日志
+
+编辑 `deployment/docker-compose.yml`:
+
+```yaml
+environment:
+  - LOG_FILE_ENABLED=true
+```
+
+重启服务:
+
+```bash
+docker-compose up -d app
+```
+
+### 调整日志级别
+
+编辑 `deployment/docker-compose.yml`:
+
+```yaml
+environment:
+  - LOG_LEVEL=WARNING  # DEBUG, INFO, WARNING, ERROR, CRITICAL
+```
+
+### 修改保留期
+
+编辑 `monitoring/loki.yaml`:
+
+```yaml
+limits_config:
+  retention_period: 72h  # 改为 3 天
+```
+
+重启 Loki:
+
+```bash
+docker-compose restart loki
+```
+
+## 访问地址
+
+| 服务 | 地址 | 凭据 |
+|------|------|------|
+| Grafana | http://localhost:3000 | admin/admin |
+| Loki API | http://localhost:3100 | - |
+| Promtail | http://localhost:9080 | - |
+| Prometheus | http://localhost:9090 | - |
+| App | http://localhost:8111 | - |
+
+## 故障排查
+
+### 看不到日志
+
+```bash
+# 1. 检查 Promtail 日志
+docker-compose logs promtail | tail -50
+
+# 2. 检查容器标签
+docker inspect deployment-app-1 | grep -A 5 Labels
+
+# 3. 查询 Loki
+curl -s "http://localhost:3100/loki/api/v1/label/job/values" | jq
+```
+
+### Docker socket 权限
+
+```bash
+sudo chmod 666 /var/run/docker.sock
+```
+
+### 清理日志数据
+
+```bash
+# 停止 Loki
+docker-compose stop loki
+
+# 清理数据
+docker-compose exec loki rm -rf /loki/chunks/*
+
+# 重启 Loki
+docker-compose start loki
+```
+
+## 性能优化
+
+### 减少日志量
+
+```yaml
+# docker-compose.yml
+environment:
+  - LOG_LEVEL=WARNING  # 只记录警告和错误
+```
+
+### 过滤健康检查日志
+
+编辑 `monitoring/promtail.yaml`:
+
+```yaml
+pipeline_stages:
+  - drop:
+      expression: ".*healthz.*"
+```
+
+### 限制查询范围
+
+```logql
+# 好：限制时间范围
+{job="functional-scaffold-app"}[5m]
+
+# 差：查询所有时间
+{job="functional-scaffold-app"}
+```
+
+## 文档链接
+
+- 完整文档: `docs/loki-integration.md`
+- 实施总结: `docs/loki-implementation-summary.md`
+- 验证脚本: `scripts/verify_loki.sh`

docs/metrics-filtering-changelog.md

@@ -0,0 +1,126 @@
+# 指标过滤和路径规范化
+
+## 变更说明
+
+本次修改优化了 HTTP 请求指标的记录逻辑，主要包括两个方面：
+
+### 1. 跳过健康检查端点
+
+以下端点不再记录到 Prometheus 指标中：
+- `/metrics` - 指标端点本身
+- `/healthz` - 存活检查
+- `/readyz` - 就绪检查
+
+**原因**：这些端点通常被频繁调用（如 Kubernetes 健康检查、Prometheus 抓取），但对业务监控意义不大，会产生大量噪音数据。
+
+### 2. 路径参数规范化
+
+带有路径参数的端点会被规范化为模板形式：
+
+| 原始路径 | 规范化后 |
+|---------|---------|
+| `GET /jobs/a1b2c3d4e5f6` | `GET /jobs/{job_id}` |
+| `GET /jobs/xyz123456789` | `GET /jobs/{job_id}` |
+
+**原因**：避免因为不同的路径参数值产生过多的指标标签，导致指标基数爆炸（cardinality explosion），影响 Prometheus 性能。
+
+## 实现细节
+
+### 代码修改
+
+**文件：`src/functional_scaffold/main.py`**
+
+1. 添加 `normalize_path()` 函数：
+```python
+def normalize_path(path: str) -> str:
+    """规范化路径，将路径参数替换为模板形式"""
+    if path.startswith("/jobs/") and len(path) > 6:
+        return "/jobs/{job_id}"
+    return path
+```
+
+2. 修改 `track_metrics` 中间件：
+```python
+# 跳过不需要记录指标的端点
+skip_paths = {"/metrics", "/readyz", "/healthz"}
+if request.url.path in skip_paths:
+    return await call_next(request)
+
+# 使用规范化后的路径记录指标
+normalized_path = normalize_path(request.url.path)
+incr("http_requests_total",
+     {"method": request.method, "endpoint": normalized_path, "status": status})
+```
+
+### 测试覆盖
+
+**文件：`tests/test_middleware.py`**
+
+新增 6 个测试用例：
+- `test_normalize_jobs_path` - 测试任务路径规范化
+- `test_normalize_other_paths` - 测试其他路径保持不变
+- `test_normalize_jobs_root` - 测试 /jobs 根路径
+- `test_skip_health_endpoints` - 测试跳过健康检查端点
+- `test_record_normal_endpoints` - 测试记录普通端点
+- `test_normalize_job_path` - 测试规范化任务路径的集成测试
+
+所有测试通过：✅ 56/56 passed
+
+## 验证方法
+
+### 手动测试
+
+使用提供的测试脚本：
+```bash
+./scripts/test_metrics_filtering.sh
+```
+
+### 预期结果
+
+访问 `/metrics` 端点后，应该看到：
+
+✅ **应该出现的指标：**
+```
+http_requests_total{method="POST",endpoint="/invoke",status="success"} 1
+http_requests_total{method="GET",endpoint="/jobs/{job_id}",status="error"} 2
+```
+
+❌ **不应该出现的指标：**
+```
+http_requests_total{method="GET",endpoint="/healthz",...}
+http_requests_total{method="GET",endpoint="/readyz",...}
+http_requests_total{method="GET",endpoint="/metrics",...}
+http_requests_total{method="GET",endpoint="/jobs/a1b2c3d4e5f6",...}
+```
+
+## 扩展性
+
+如果需要添加更多路径规范化规则，只需修改 `normalize_path()` 函数：
+
+```python
+def normalize_path(path: str) -> str:
+    """规范化路径，将路径参数替换为模板形式"""
+    # 任务路径
+    if path.startswith("/jobs/") and len(path) > 6:
+        return "/jobs/{job_id}"
+
+    # 用户路径（示例）
+    if path.startswith("/users/") and len(path) > 7:
+        return "/users/{user_id}"
+
+    # 其他路径保持不变
+    return path
+```
+
+## 影响范围
+
+- ✅ 不影响现有功能
+- ✅ 不影响 API 行为
+- ✅ 仅影响指标记录逻辑
+- ✅ 向后兼容
+- ✅ 所有测试通过
+
+## 相关文档
+
+- [监控指南](../docs/monitoring.md) - 已更新指标说明
+- [测试脚本](../scripts/test_metrics_filtering.sh) - 手动验证脚本

docs/monitoring.md

@@ -61,6 +61,19 @@ docker-compose up -d redis prometheus grafana
 | `http_request_duration_seconds` | Histogram | method, endpoint | HTTP 请求延迟分布 |
 | `http_requests_in_progress` | Gauge | - | 当前进行中的请求数 |
 
+**注意事项：**
+
+1. **跳过的端点**：以下端点不会被记录到指标中，以减少噪音：
+   - `/metrics` - 指标端点本身
+   - `/healthz` - 存活检查
+   - `/readyz` - 就绪检查
+
+2. **路径规范化**：带有路径参数的端点会被规范化为模板形式：
+   - `GET /jobs/a1b2c3d4e5f6` → `GET /jobs/{job_id}`
+   - `GET /jobs/xyz123456789` → `GET /jobs/{job_id}`
+
+   这样可以避免因为不同的路径参数值产生过多的指标标签，导致指标基数爆炸。
+
 ### 算法执行指标
 
 | 指标 | 类型 | 标签 | 描述 |

main.py

@@ -1,16 +0,0 @@
-# 这是一个示例 Python 脚本。
-
-# 按 ⌃R 执行或将其替换为您的代码。
-# 按 双击 ⇧ 在所有地方搜索类、文件、工具窗口、操作和设置。
-
-
-def print_hi(name):
-    # 在下面的代码行中使用断点来调试脚本。
-    print(f'Hi, {name}')  # 按 ⌘F8 切换断点。
-
-
-# 按装订区域中的绿色按钮以运行脚本。
-if __name__ == '__main__':
-    print_hi('PyCharm')
-
-# 访问 https://www.jetbrains.com/help/pycharm/ 获取 PyCharm 帮助

monitoring/README.md

@@ -0,0 +1,258 @@
+# Monitoring 目录说明
+
+本目录包含所有监控和日志收集相关的配置文件。
+
+## 目录结构
+
+```
+monitoring/
+├── alerts/                          # Prometheus 告警规则
+│   └── rules.yaml                   # 告警规则配置
+├── grafana/                         # Grafana 配置
+│   ├── datasources/                 # 数据源自动配置
+│   │   ├── prometheus.yaml          # Prometheus 数据源
+│   │   └── loki.yaml                # Loki 数据源
+│   └── dashboards/                  # 仪表板自动加载
+│       ├── provider.yaml            # Dashboard provider 配置
+│       ├── dashboard.json           # 指标监控仪表板
+│       └── logs-dashboard.json      # 日志监控仪表板
+├── loki.yaml                        # Loki 日志存储配置
+├── promtail.yaml                    # Promtail 日志采集配置
+└── prometheus.yml                   # Prometheus 指标收集配置
+```
+
+## 配置文件说明
+
+### Prometheus 配置
+
+**文件**: `prometheus.yml`
+
+Prometheus 指标收集配置，包括：
+- 抓取间隔: 5 秒
+- 目标: app 服务的 `/metrics` 端点
+- 告警规则: 从 `alerts/` 目录加载
+
+### Loki 配置
+
+**文件**: `loki.yaml`
+
+Loki 日志存储配置，包括：
+- 存储方式: 本地文件系统
+- 日志保留期: 7 天
+- 摄入速率限制: 10MB/s
+- 自动压缩和清理
+
+**关键配置**:
+```yaml
+limits_config:
+  retention_period: 168h  # 7 天
+  ingestion_rate_mb: 10   # 10MB/s
+```
+
+### Promtail 配置
+
+**文件**: `promtail.yaml`
+
+Promtail 日志采集配置，支持两种模式：
+
+**模式 1: Docker stdio 收集（默认）**
+- 通过 Docker API 自动发现容器
+- 过滤带有 `logging=promtail` 标签的容器
+- 自动解析 JSON 日志
+
+**模式 2: 文件收集（备用）**
+- 从 `/var/log/app/*.log` 读取日志文件
+- 支持日志轮转
+- 需要设置 `LOG_FILE_ENABLED=true`
+
+### Grafana Provisioning
+
+**数据源** (`grafana/datasources/`)
+
+自动配置 Grafana 数据源：
+- `prometheus.yaml`: Prometheus 数据源（默认）
+- `loki.yaml`: Loki 数据源
+
+**仪表板** (`grafana/dashboards/`)
+
+自动加载 Grafana 仪表板：
+- `provider.yaml`: Dashboard provider 配置
+- `dashboard.json`: 指标监控仪表板（HTTP 请求、算法执行等）
+- `logs-dashboard.json`: 日志监控仪表板（日志流、错误日志等）
+
+### 告警规则
+
+**文件**: `alerts/rules.yaml`
+
+Prometheus 告警规则，包括：
+- 高错误率告警
+- 高延迟告警
+- 服务不可用告警
+
+## 修改配置
+
+### 调整日志保留期
+
+编辑 `loki.yaml`:
+
+```yaml
+limits_config:
+  retention_period: 72h  # 改为 3 天
+```
+
+重启 Loki:
+
+```bash
+cd deployment
+docker-compose restart loki
+```
+
+### 调整指标抓取间隔
+
+编辑 `prometheus.yml`:
+
+```yaml
+global:
+  scrape_interval: 10s  # 改为 10 秒
+```
+
+重启 Prometheus:
+
+```bash
+cd deployment
+docker-compose restart prometheus
+```
+
+### 添加新的告警规则
+
+编辑 `alerts/rules.yaml`，添加新规则：
+
+```yaml
+groups:
+  - name: my_alerts
+    rules:
+      - alert: MyAlert
+        expr: my_metric > 100
+        for: 5m
+        labels:
+          severity: warning
+        annotations:
+          summary: "我的告警"
+```
+
+重启 Prometheus:
+
+```bash
+cd deployment
+docker-compose restart prometheus
+```
+
+### 添加新的仪表板
+
+1. 在 Grafana UI 中创建仪表板
+2. 导出为 JSON
+3. 保存到 `grafana/dashboards/my-dashboard.json`
+4. 重启 Grafana（或等待自动重载）
+
+```bash
+cd deployment
+docker-compose restart grafana
+```
+
+## 验证配置
+
+### 检查 Prometheus 配置
+
+```bash
+# 访问 Prometheus UI
+open http://localhost:9090
+
+# 检查目标状态
+open http://localhost:9090/targets
+
+# 检查告警规则
+open http://localhost:9090/alerts
+```
+
+### 检查 Loki 配置
+
+```bash
+# 检查 Loki 健康状态
+curl http://localhost:3100/ready
+
+# 查询标签
+curl -s "http://localhost:3100/loki/api/v1/label/job/values" | jq
+```
+
+### 检查 Grafana 配置
+
+```bash
+# 访问 Grafana UI
+open http://localhost:3000
+
+# 检查数据源
+curl -s -u admin:admin http://localhost:3000/api/datasources | jq
+
+# 检查仪表板
+curl -s -u admin:admin http://localhost:3000/api/search | jq
+```
+
+## 故障排查
+
+### Prometheus 无法抓取指标
+
+1. 检查 app 服务是否运行: `docker-compose ps app`
+2. 检查 metrics 端点: `curl http://localhost:8111/metrics`
+3. 查看 Prometheus 日志: `docker-compose logs prometheus`
+
+### Loki 无法接收日志
+
+1. 检查 Promtail 是否运行: `docker-compose ps promtail`
+2. 查看 Promtail 日志: `docker-compose logs promtail`
+3. 检查容器标签: `docker inspect <container> | grep Labels`
+
+### Grafana 数据源未加载
+
+1. 检查 provisioning 目录挂载: `docker-compose config | grep grafana -A 10`
+2. 查看 Grafana 日志: `docker-compose logs grafana`
+3. 手动重启 Grafana: `docker-compose restart grafana`
+
+## 相关文档
+
+- [Loki 集成文档](../docs/loki-integration.md) - 完整的 Loki 使用文档
+- [Loki 快速参考](../docs/loki-quick-reference.md) - 常用命令和查询
+- [Loki 实施总结](../docs/loki-implementation-summary.md) - 实施细节和架构说明
+- [Prometheus 官方文档](https://prometheus.io/docs/)
+- [Loki 官方文档](https://grafana.com/docs/loki/latest/)
+- [Grafana 官方文档](https://grafana.com/docs/grafana/latest/)
+
+## 性能建议
+
+### 日志量控制
+
+- 调整日志级别为 WARNING 或 ERROR
+- 过滤掉不必要的日志（如健康检查）
+- 减少日志保留期
+
+### 指标优化
+
+- 增加抓取间隔（如 15s 或 30s）
+- 减少指标基数（避免高基数标签）
+- 定期清理旧数据
+
+### 存储优化
+
+- 监控磁盘使用: `docker-compose exec loki du -sh /loki`
+- 定期备份重要数据
+- 考虑使用对象存储（S3/OSS）作为后端
+
+## 总结
+
+本目录包含完整的监控和日志收集配置：
+
+✅ **Prometheus** - 指标收集和告警
+✅ **Loki** - 日志存储和查询
+✅ **Promtail** - 日志采集
+✅ **Grafana** - 可视化和仪表板
+
+所有配置都支持自动加载，无需手动配置。

monitoring/grafana/dashboards/dashboard.json

@@ -1395,6 +1395,504 @@
       ],
       "title": "Webhook 发送状态",
       "type": "piechart"
+    },
+    {
+      "collapsed": false,
+      "gridPos": {
+        "h": 1,
+        "w": 24,
+        "x": 0,
+        "y": 53
+      },
+      "id": 200,
+      "panels": [],
+      "title": "队列监控",
+      "type": "row"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "${DS_PROMETHEUS}"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "palette-classic"
+          },
+          "custom": {
+            "axisCenteredZero": false,
+            "axisColorMode": "text",
+            "axisLabel": "任务数",
+            "axisPlacement": "auto",
+            "barAlignment": 0,
+            "drawStyle": "line",
+            "fillOpacity": 20,
+            "gradientMode": "opacity",
+            "hideFrom": {
+              "tooltip": false,
+              "viz": false,
+              "legend": false
+            },
+            "lineInterpolation": "smooth",
+            "lineWidth": 2,
+            "pointSize": 5,
+            "scaleDistribution": {
+              "type": "linear"
+            },
+            "showPoints": "never",
+            "spanNulls": true,
+            "stacking": {
+              "group": "A",
+              "mode": "none"
+            },
+            "thresholdsStyle": {
+              "mode": "off"
+            }
+          },
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              },
+              {
+                "color": "yellow",
+                "value": 50
+              },
+              {
+                "color": "red",
+                "value": 100
+              }
+            ]
+          },
+          "unit": "short"
+        },
+        "overrides": [
+          {
+            "matcher": {
+              "id": "byName",
+              "options": "pending"
+            },
+            "properties": [
+              {
+                "id": "color",
+                "value": {
+                  "fixedColor": "blue",
+                  "mode": "fixed"
+                }
+              }
+            ]
+          },
+          {
+            "matcher": {
+              "id": "byName",
+              "options": "processing"
+            },
+            "properties": [
+              {
+                "id": "color",
+                "value": {
+                  "fixedColor": "orange",
+                  "mode": "fixed"
+                }
+              }
+            ]
+          },
+          {
+            "matcher": {
+              "id": "byName",
+              "options": "dlq"
+            },
+            "properties": [
+              {
+                "id": "color",
+                "value": {
+                  "fixedColor": "red",
+                  "mode": "fixed"
+                }
+              }
+            ]
+          }
+        ]
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 12,
+        "x": 0,
+        "y": 54
+      },
+      "id": 19,
+      "options": {
+        "legend": {
+          "calcs": ["mean", "last", "max"],
+          "displayMode": "table",
+          "placement": "bottom",
+          "showLegend": true
+        },
+        "tooltip": {
+          "mode": "multi",
+          "sort": "desc"
+        }
+      },
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "${DS_PROMETHEUS}"
+          },
+          "expr": "job_queue_length",
+          "legendFormat": "{{queue}}",
+          "refId": "A"
+        }
+      ],
+      "title": "队列长度趋势",
+      "type": "timeseries"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "${DS_PROMETHEUS}"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "palette-classic"
+          },
+          "custom": {
+            "axisCenteredZero": false,
+            "axisColorMode": "text",
+            "axisLabel": "秒",
+            "axisPlacement": "auto",
+            "barAlignment": 0,
+            "drawStyle": "line",
+            "fillOpacity": 20,
+            "gradientMode": "opacity",
+            "hideFrom": {
+              "tooltip": false,
+              "viz": false,
+              "legend": false
+            },
+            "lineInterpolation": "smooth",
+            "lineWidth": 2,
+            "pointSize": 5,
+            "scaleDistribution": {
+              "type": "linear"
+            },
+            "showPoints": "never",
+            "spanNulls": true,
+            "stacking": {
+              "group": "A",
+              "mode": "none"
+            },
+            "thresholdsStyle": {
+              "mode": "line"
+            }
+          },
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              },
+              {
+                "color": "yellow",
+                "value": 60
+              },
+              {
+                "color": "red",
+                "value": 300
+              }
+            ]
+          },
+          "unit": "s"
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 12,
+        "x": 12,
+        "y": 54
+      },
+      "id": 20,
+      "options": {
+        "legend": {
+          "calcs": ["mean", "last", "max"],
+          "displayMode": "table",
+          "placement": "bottom",
+          "showLegend": true
+        },
+        "tooltip": {
+          "mode": "multi",
+          "sort": "desc"
+        }
+      },
+      "pluginVersion": "9.0.0",
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "${DS_PROMETHEUS}"
+          },
+          "expr": "job_oldest_waiting_seconds",
+          "legendFormat": "最长等待时间",
+          "refId": "A"
+        }
+      ],
+      "title": "最长任务等待时间",
+      "type": "timeseries"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "${DS_PROMETHEUS}"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "thresholds"
+          },
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              },
+              {
+                "color": "yellow",
+                "value": 10
+              },
+              {
+                "color": "red",
+                "value": 50
+              }
+            ]
+          },
+          "unit": "short"
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 4,
+        "w": 6,
+        "x": 0,
+        "y": 62
+      },
+      "id": 21,
+      "options": {
+        "colorMode": "value",
+        "graphMode": "area",
+        "justifyMode": "auto",
+        "orientation": "auto",
+        "reduceOptions": {
+          "values": false,
+          "calcs": ["last"],
+          "fields": ""
+        },
+        "textMode": "auto"
+      },
+      "pluginVersion": "9.0.0",
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "${DS_PROMETHEUS}"
+          },
+          "expr": "job_queue_length{queue=\"pending\"}",
+          "refId": "A"
+        }
+      ],
+      "title": "待处理队列",
+      "type": "stat"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "${DS_PROMETHEUS}"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "thresholds"
+          },
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              },
+              {
+                "color": "yellow",
+                "value": 5
+              },
+              {
+                "color": "red",
+                "value": 10
+              }
+            ]
+          },
+          "unit": "short"
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 4,
+        "w": 6,
+        "x": 6,
+        "y": 62
+      },
+      "id": 22,
+      "options": {
+        "colorMode": "value",
+        "graphMode": "area",
+        "justifyMode": "auto",
+        "orientation": "auto",
+        "reduceOptions": {
+          "values": false,
+          "calcs": ["last"],
+          "fields": ""
+        },
+        "textMode": "auto"
+      },
+      "pluginVersion": "9.0.0",
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "${DS_PROMETHEUS}"
+          },
+          "expr": "job_queue_length{queue=\"processing\"}",
+          "refId": "A"
+        }
+      ],
+      "title": "处理中队列",
+      "type": "stat"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "${DS_PROMETHEUS}"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "thresholds"
+          },
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              },
+              {
+                "color": "red",
+                "value": 1
+              }
+            ]
+          },
+          "unit": "short"
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 4,
+        "w": 6,
+        "x": 12,
+        "y": 62
+      },
+      "id": 23,
+      "options": {
+        "colorMode": "value",
+        "graphMode": "area",
+        "justifyMode": "auto",
+        "orientation": "auto",
+        "reduceOptions": {
+          "values": false,
+          "calcs": ["last"],
+          "fields": ""
+        },
+        "textMode": "auto"
+      },
+      "pluginVersion": "9.0.0",
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "${DS_PROMETHEUS}"
+          },
+          "expr": "job_queue_length{queue=\"dlq\"}",
+          "refId": "A"
+        }
+      ],
+      "title": "死信队列",
+      "type": "stat"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "${DS_PROMETHEUS}"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "thresholds"
+          },
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              }
+            ]
+          },
+          "unit": "short"
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 4,
+        "w": 6,
+        "x": 18,
+        "y": 62
+      },
+      "id": 24,
+      "options": {
+        "colorMode": "value",
+        "graphMode": "area",
+        "justifyMode": "auto",
+        "orientation": "auto",
+        "reduceOptions": {
+          "values": false,
+          "calcs": ["last"],
+          "fields": ""
+        },
+        "textMode": "auto"
+      },
+      "pluginVersion": "9.0.0",
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "${DS_PROMETHEUS}"
+          },
+          "expr": "sum(job_recovered_total) or vector(0)",
+          "refId": "A"
+        }
+      ],
+      "title": "回收任务总数",
+      "type": "stat"
     }
   ],
   "refresh": "5s",

monitoring/grafana/dashboards/logs-dashboard.json

@@ -0,0 +1,292 @@
+{
+  "annotations": {
+    "list": [
+      {
+        "builtIn": 1,
+        "datasource": {
+          "type": "grafana",
+          "uid": "-- Grafana --"
+        },
+        "enable": true,
+        "hide": true,
+        "iconColor": "rgba(0, 211, 255, 1)",
+        "name": "Annotations & Alerts",
+        "type": "dashboard"
+      }
+    ]
+  },
+  "editable": true,
+  "fiscalYearStartMonth": 0,
+  "graphTooltip": 0,
+  "id": null,
+  "links": [],
+  "liveNow": false,
+  "panels": [
+    {
+      "datasource": {
+        "type": "loki",
+        "uid": "Loki"
+      },
+      "gridPos": {
+        "h": 10,
+        "w": 24,
+        "x": 0,
+        "y": 0
+      },
+      "id": 1,
+      "options": {
+        "dedupStrategy": "none",
+        "enableLogDetails": true,
+        "prettifyLogMessage": false,
+        "showCommonLabels": false,
+        "showLabels": false,
+        "showTime": true,
+        "sortOrder": "Descending",
+        "wrapLogMessage": false
+      },
+      "targets": [
+        {
+          "datasource": {
+            "type": "loki",
+            "uid": "Loki"
+          },
+          "editorMode": "code",
+          "expr": "{job=\"functional-scaffold-app\"} |= \"$request_id\"",
+          "queryType": "range",
+          "refId": "A"
+        }
+      ],
+      "title": "日志流 (实时)",
+      "type": "logs"
+    },
+    {
+      "datasource": {
+        "type": "loki",
+        "uid": "Loki"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "palette-classic"
+          },
+          "custom": {
+            "axisCenteredZero": false,
+            "axisColorMode": "text",
+            "axisLabel": "",
+            "axisPlacement": "auto",
+            "barAlignment": 0,
+            "drawStyle": "line",
+            "fillOpacity": 10,
+            "gradientMode": "none",
+            "hideFrom": {
+              "tooltip": false,
+              "viz": false,
+              "legend": false
+            },
+            "lineInterpolation": "linear",
+            "lineWidth": 1,
+            "pointSize": 5,
+            "scaleDistribution": {
+              "type": "linear"
+            },
+            "showPoints": "never",
+            "spanNulls": false,
+            "stacking": {
+              "group": "A",
+              "mode": "none"
+            },
+            "thresholdsStyle": {
+              "mode": "off"
+            }
+          },
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              }
+            ]
+          },
+          "unit": "short"
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 12,
+        "x": 0,
+        "y": 10
+      },
+      "id": 2,
+      "options": {
+        "legend": {
+          "calcs": [],
+          "displayMode": "list",
+          "placement": "bottom",
+          "showLegend": true
+        },
+        "tooltip": {
+          "mode": "single",
+          "sort": "none"
+        }
+      },
+      "targets": [
+        {
+          "datasource": {
+            "type": "loki",
+            "uid": "Loki"
+          },
+          "editorMode": "code",
+          "expr": "sum by (level) (count_over_time({job=\"functional-scaffold-app\"} |= \"$request_id\" [1m]))",
+          "queryType": "range",
+          "refId": "A"
+        }
+      ],
+      "title": "日志量趋势（按级别）",
+      "type": "timeseries"
+    },
+    {
+      "datasource": {
+        "type": "loki",
+        "uid": "Loki"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "thresholds"
+          },
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              },
+              {
+                "color": "yellow",
+                "value": 10
+              },
+              {
+                "color": "red",
+                "value": 50
+              }
+            ]
+          }
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 12,
+        "x": 12,
+        "y": 10
+      },
+      "id": 3,
+      "options": {
+        "orientation": "auto",
+        "reduceOptions": {
+          "values": false,
+          "calcs": [
+            "lastNotNull"
+          ],
+          "fields": ""
+        },
+        "showThresholdLabels": false,
+        "showThresholdMarkers": true
+      },
+      "pluginVersion": "9.5.3",
+      "targets": [
+        {
+          "datasource": {
+            "type": "loki",
+            "uid": "Loki"
+          },
+          "editorMode": "code",
+          "expr": "sum by (level) (count_over_time({job=\"functional-scaffold-app\"} |= \"$request_id\" [$__range]))",
+          "queryType": "range",
+          "refId": "A"
+        }
+      ],
+      "title": "日志级别分布",
+      "type": "gauge"
+    },
+    {
+      "datasource": {
+        "type": "loki",
+        "uid": "Loki"
+      },
+      "gridPos": {
+        "h": 10,
+        "w": 24,
+        "x": 0,
+        "y": 18
+      },
+      "id": 4,
+      "options": {
+        "dedupStrategy": "none",
+        "enableLogDetails": true,
+        "prettifyLogMessage": false,
+        "showCommonLabels": false,
+        "showLabels": false,
+        "showTime": true,
+        "sortOrder": "Descending",
+        "wrapLogMessage": false
+      },
+      "targets": [
+        {
+          "datasource": {
+            "type": "loki",
+            "uid": "Loki"
+          },
+          "editorMode": "code",
+          "expr": "{job=\"functional-scaffold-app\", level=\"ERROR\"} |= \"$request_id\"",
+          "queryType": "range",
+          "refId": "A"
+        }
+      ],
+      "title": "错误日志",
+      "type": "logs"
+    }
+  ],
+  "refresh": "5s",
+  "schemaVersion": 38,
+  "style": "dark",
+  "tags": ["logs", "loki"],
+  "templating": {
+    "list": [
+      {
+        "current": {
+          "selected": false,
+          "text": "",
+          "value": ""
+        },
+        "hide": 0,
+        "label": "Request ID",
+        "name": "request_id",
+        "options": [
+          {
+            "selected": true,
+            "text": "",
+            "value": ""
+          }
+        ],
+        "query": "",
+        "skipUrlSync": false,
+        "type": "textbox"
+      }
+    ]
+  },
+  "time": {
+    "from": "now-15m",
+    "to": "now"
+  },
+  "timepicker": {},
+  "timezone": "",
+  "title": "日志监控",
+  "uid": "logs-dashboard",
+  "version": 0,
+  "weekStart": ""
+}

monitoring/grafana/dashboards/provider.yaml

@@ -0,0 +1,13 @@
+apiVersion: 1
+
+providers:
+  - name: 'default'
+    orgId: 1
+    folder: ''
+    type: file
+    disableDeletion: false
+    updateIntervalSeconds: 10
+    allowUiUpdates: true
+    options:
+      path: /etc/grafana/provisioning/dashboards
+      foldersFromFilesStructure: true

monitoring/grafana/datasources/loki.yaml

@@ -0,0 +1,11 @@
+apiVersion: 1
+
+datasources:
+  - name: Loki
+    type: loki
+    access: proxy
+    url: http://loki:3100
+    isDefault: false
+    editable: false
+    jsonData:
+      maxLines: 1000

monitoring/grafana/datasources/prometheus.yaml

@@ -0,0 +1,11 @@
+apiVersion: 1
+
+datasources:
+  - name: Prometheus
+    type: prometheus
+    access: proxy
+    url: http://prometheus:9090
+    isDefault: true
+    editable: false
+    jsonData:
+      timeInterval: "5s"

monitoring/loki.yaml

@@ -0,0 +1,39 @@
+auth_enabled: false
+
+server:
+  http_listen_port: 3100
+  grpc_listen_port: 9096
+
+common:
+  path_prefix: /loki
+  storage:
+    filesystem:
+      chunks_directory: /loki/chunks
+      rules_directory: /loki/rules
+  replication_factor: 1
+  ring:
+    instance_addr: 127.0.0.1
+    kvstore:
+      store: inmemory
+
+schema_config:
+  configs:
+    - from: 2020-10-24
+      store: boltdb-shipper
+      object_store: filesystem
+      schema: v11
+      index:
+        prefix: index_
+        period: 24h
+
+limits_config:
+  retention_period: 168h  # 7 天
+  ingestion_rate_mb: 10
+  ingestion_burst_size_mb: 20
+
+compactor:
+  working_directory: /loki/compactor
+  shared_store: filesystem
+  compaction_interval: 10m
+  retention_enabled: true
+  retention_delete_delay: 2h

monitoring/promtail.yaml

@@ -0,0 +1,71 @@
+server:
+  http_listen_port: 9080
+  grpc_listen_port: 0
+
+positions:
+  filename: /tmp/positions.yaml
+
+clients:
+  - url: http://loki:3100/loki/api/v1/push
+
+scrape_configs:
+  # 场景 1: Docker stdio 收集（主要方式）
+  - job_name: docker
+    docker_sd_configs:
+      - host: unix:///var/run/docker.sock
+        refresh_interval: 5s
+        filters:
+          - name: label
+            values: ["logging=promtail"]
+    relabel_configs:
+      - source_labels: ['__meta_docker_container_name']
+        regex: '/(.*)'
+        target_label: 'container'
+      - source_labels: ['__meta_docker_container_label_logging_jobname']
+        target_label: 'job'
+      - source_labels: ['__meta_docker_container_id']
+        target_label: '__path__'
+        replacement: '/var/lib/docker/containers/$1/*.log'
+    pipeline_stages:
+      - json:
+          expressions:
+            log: log
+            stream: stream
+            time: time
+      - json:
+          source: log
+          expressions:
+            level: levelname
+            logger: name
+            message: message
+            request_id: request_id
+      - labels:
+          level:
+          logger:
+      - output:
+          source: log
+
+  # 场景 2: Log 文件收集（备用）
+  - job_name: app_files
+    static_configs:
+      - targets:
+          - localhost
+        labels:
+          job: functional-scaffold-app-files
+          __path__: /var/log/app/*.log
+    pipeline_stages:
+      - json:
+          expressions:
+            timestamp: asctime
+            level: levelname
+            logger: name
+            message: message
+            request_id: request_id
+      - timestamp:
+          source: timestamp
+          format: "2006-01-02 15:04:05,000"
+      - labels:
+          level:
+          logger:
+      - output:
+          source: message

pyproject.toml

@@ -19,6 +19,14 @@ dependencies = [
     "pydantic-settings>=2.0.0",
     "prometheus-client>=0.19.0",
     "python-json-logger>=2.0.7",
+    # Redis - 任务队列和指标存储
+    "redis>=5.0.0",
+    # YAML 配置解析
+    "pyyaml>=6.0.0",
+    # HTTP 客户端（Webhook 回调）
+    "httpx>=0.27.0",
+    # 轻量级 HTTP 服务器（Worker 健康检查）
+    "aiohttp>=3.9.0",
 ]
 
 [project.optional-dependencies]
@@ -26,7 +34,6 @@ dev = [
     "pytest>=7.4.0",
     "pytest-asyncio>=0.21.0",
     "pytest-cov>=4.1.0",
-    "httpx>=0.26.0",
     "black>=23.12.0",
     "ruff>=0.1.0",
 ]
@@ -48,3 +55,4 @@ python_files = ["test_*.py"]
 python_classes = ["Test*"]
 python_functions = ["test_*"]
 addopts = "-v --strict-markers"
+pythonpath = ["src"]

requirements.txt

@@ -1,16 +1,17 @@
+# 核心依赖 - 与 pyproject.toml 保持同步
 fastapi>=0.109.0
 uvicorn[standard]>=0.27.0
 pydantic>=2.5.0
 pydantic-settings>=2.0.0
 prometheus-client>=0.19.0
 python-json-logger>=2.0.7
+aiohttp>=3.9.0
 
-# 指标存储方案（可选，根据选择的方案安装）
+# Redis - 任务队列和指标存储
-# 方案2：Redis 方案需要
 redis>=5.0.0
 
 # YAML 配置解析
 pyyaml>=6.0.0
 
-# HTTP 客户端（用于 Webhook 回调）
+# HTTP 客户端（Webhook 回调）
 httpx>=0.27.0

scripts/run_dev.sh

@@ -21,4 +21,4 @@ pip install -e ".[dev]"
 # 启动服务
 echo "Starting server on http://localhost:8000"
 echo "API docs available at http://localhost:8000/docs"
-uvicorn src.functional_scaffold.main:app --reload --host 0.0.0.0 --port 8000
+uvicorn functional_scaffold.main:app --reload --host 0.0.0.0 --port 8000

scripts/start_metrics.sh

@@ -1,114 +0,0 @@
-#!/bin/bash
-# 指标方案快速启动脚本
-
-set -e
-
-# 颜色定义
-RED='\033[0;31m'
-GREEN='\033[0;32m'
-YELLOW='\033[1;33m'
-NC='\033[0m' # No Color
-
-echo "=========================================="
-echo "FunctionalScaffold 指标方案启动脚本"
-echo "=========================================="
-
-# 检查 docker-compose
-if ! command -v docker-compose &> /dev/null; then
-    echo -e "${RED}错误: docker-compose 未安装${NC}"
-    exit 1
-fi
-
-# 选择方案
-echo ""
-echo "请选择指标方案："
-echo "1. Pushgateway（推荐，适合 Serverless）"
-echo "2. Redis + Exporter（适合高并发）"
-echo "3. 两者都启动（用于对比测试）"
-echo ""
-read -p "输入选项 (1/2/3): " choice
-
-cd "$(dirname "$0")/../deployment"
-
-case $choice in
-    1)
-        echo -e "${GREEN}启动 Pushgateway 方案...${NC}"
-        docker-compose up -d redis pushgateway prometheus grafana
-        echo ""
-        echo -e "${GREEN}✓ Pushgateway 方案已启动${NC}"
-        echo ""
-        echo "服务地址："
-        echo "  - Pushgateway: http://localhost:9091"
-        echo "  - Prometheus:  http://localhost:9090"
-        echo "  - Grafana:     http://localhost:3000 (admin/admin)"
-        echo ""
-        echo "下一步："
-        echo "  1. 修改代码导入: from functional_scaffold.core.metrics_pushgateway import ..."
-        echo "  2. 配置环境变量: PUSHGATEWAY_URL=localhost:9091"
-        echo "  3. 启动应用: ./scripts/run_dev.sh"
-        echo "  4. 运行测试: python scripts/test_metrics.py pushgateway"
-        ;;
-    2)
-        echo -e "${GREEN}启动 Redis 方案...${NC}"
-
-        # 检查 redis 依赖
-        if ! python -c "import redis" 2>/dev/null; then
-            echo -e "${YELLOW}警告: redis 库未安装${NC}"
-            echo "正在安装 redis..."
-            pip install redis
-        fi
-
-        docker-compose up -d redis redis-exporter prometheus grafana
-        echo ""
-        echo -e "${GREEN}✓ Redis 方案已启动${NC}"
-        echo ""
-        echo "服务地址："
-        echo "  - Redis:        localhost:6379"
-        echo "  - Redis Exporter: http://localhost:8001/metrics"
-        echo "  - Prometheus:   http://localhost:9090"
-        echo "  - Grafana:      http://localhost:3000 (admin/admin)"
-        echo ""
-        echo "下一步："
-        echo "  1. 修改代码导入: from functional_scaffold.core.metrics_redis import ..."
-        echo "  2. 配置环境变量: REDIS_HOST=localhost REDIS_PORT=6379"
-        echo "  3. 启动应用: ./scripts/run_dev.sh"
-        echo "  4. 运行测试: python scripts/test_metrics.py redis"
-        ;;
-    3)
-        echo -e "${GREEN}启动所有服务...${NC}"
-
-        # 检查 redis 依赖
-        if ! python -c "import redis" 2>/dev/null; then
-            echo -e "${YELLOW}警告: redis 库未安装${NC}"
-            echo "正在安装 redis..."
-            pip install redis
-        fi
-
-        docker-compose up -d
-        echo ""
-        echo -e "${GREEN}✓ 所有服务已启动${NC}"
-        echo ""
-        echo "服务地址："
-        echo "  - 应用:         http://localhost:8000"
-        echo "  - Pushgateway:  http://localhost:9091"
-        echo "  - Redis:        localhost:6379"
-        echo "  - Redis Exporter: http://localhost:8001/metrics"
-        echo "  - Prometheus:   http://localhost:9090"
-        echo "  - Grafana:      http://localhost:3000 (admin/admin)"
-        echo ""
-        echo "下一步："
-        echo "  1. 查看文档: cat docs/metrics-guide.md"
-        echo "  2. 运行测试: python scripts/test_metrics.py"
-        ;;
-    *)
-        echo -e "${RED}无效的选项${NC}"
-        exit 1
-        ;;
-esac
-
-echo ""
-echo "=========================================="
-echo "查看日志: docker-compose logs -f"
-echo "停止服务: docker-compose down"
-echo "查看文档: cat ../docs/metrics-guide.md"
-echo "=========================================="

scripts/test_concurrency.sh

@@ -0,0 +1,104 @@
+#!/bin/bash
+# 并发控制测试脚本
+
+set -e
+
+BASE_URL="http://localhost:8000"
+
+echo "=== 异步任务并发控制测试 ==="
+echo ""
+
+# 1. 检查服务是否运行
+echo "1. 检查服务状态..."
+if ! curl -s "${BASE_URL}/healthz" > /dev/null; then
+    echo "❌ 服务未运行，请先启动服务"
+    exit 1
+fi
+echo "✅ 服务正常运行"
+echo ""
+
+# 2. 查询初始并发状态
+echo "2. 查询初始并发状态..."
+curl -s "${BASE_URL}/jobs/concurrency/status" | jq '.'
+echo ""
+
+# 3. 创建多个任务
+echo "3. 创建 15 个任务（测试并发限制）..."
+JOB_IDS=()
+for i in {1..15}; do
+    # 使用较大的质数，让任务执行时间更长
+    NUMBER=$((10000 + i * 1000))
+    RESPONSE=$(curl -s -X POST "${BASE_URL}/jobs" \
+        -H "Content-Type: application/json" \
+        -d "{\"algorithm\": \"PrimeChecker\", \"params\": {\"number\": ${NUMBER}}}")
+
+    JOB_ID=$(echo "$RESPONSE" | jq -r '.job_id')
+    JOB_IDS+=("$JOB_ID")
+    echo "  创建任务 ${i}/15: job_id=${JOB_ID}"
+
+    # 短暂延迟，避免请求过快
+    sleep 0.1
+done
+echo ""
+
+# 4. 立即查询并发状态（应该看到多个任务在运行）
+echo "4. 查询并发状态（任务执行中）..."
+for i in {1..5}; do
+    echo "  第 ${i} 次查询:"
+    STATUS=$(curl -s "${BASE_URL}/jobs/concurrency/status")
+    echo "    $(echo "$STATUS" | jq -c '.')"
+    sleep 1
+done
+echo ""
+
+# 5. 等待所有任务完成
+echo "5. 等待任务完成..."
+COMPLETED=0
+TOTAL=${#JOB_IDS[@]}
+
+while [ $COMPLETED -lt $TOTAL ]; do
+    COMPLETED=0
+    for JOB_ID in "${JOB_IDS[@]}"; do
+        STATUS=$(curl -s "${BASE_URL}/jobs/${JOB_ID}" | jq -r '.status')
+        if [ "$STATUS" = "completed" ] || [ "$STATUS" = "failed" ]; then
+            ((COMPLETED++))
+        fi
+    done
+
+    echo "  进度: ${COMPLETED}/${TOTAL} 任务完成"
+
+    # 显示当前并发状态
+    CONCURRENCY=$(curl -s "${BASE_URL}/jobs/concurrency/status")
+    echo "  并发状态: $(echo "$CONCURRENCY" | jq -c '.')"
+
+    if [ $COMPLETED -lt $TOTAL ]; then
+        sleep 2
+    fi
+done
+echo ""
+
+# 6. 查询最终并发状态
+echo "6. 查询最终并发状态..."
+curl -s "${BASE_URL}/jobs/concurrency/status" | jq '.'
+echo ""
+
+# 7. 显示任务结果统计
+echo "7. 任务结果统计..."
+COMPLETED_COUNT=0
+FAILED_COUNT=0
+
+for JOB_ID in "${JOB_IDS[@]}"; do
+    STATUS=$(curl -s "${BASE_URL}/jobs/${JOB_ID}" | jq -r '.status')
+    if [ "$STATUS" = "completed" ]; then
+        ((COMPLETED_COUNT++))
+    elif [ "$STATUS" = "failed" ]; then
+        ((FAILED_COUNT++))
+    fi
+done
+
+echo "  总任务数: ${TOTAL}"
+echo "  成功: ${COMPLETED_COUNT}"
+echo "  失败: ${FAILED_COUNT}"
+echo ""
+
+echo "=== 测试完成 ==="

scripts/test_metrics_filtering.sh

@@ -0,0 +1,39 @@
+#!/bin/bash
+# 测试指标过滤和路径规范化
+
+echo "=== 测试指标过滤和路径规范化 ==="
+echo ""
+
+# 启动服务（假设已经在运行）
+BASE_URL="http://localhost:8000"
+
+echo "1. 访问健康检查端点（应该被跳过，不记录指标）"
+curl -s "$BASE_URL/healthz" > /dev/null
+curl -s "$BASE_URL/readyz" > /dev/null
+echo "   ✓ 已访问 /healthz 和 /readyz"
+echo ""
+
+echo "2. 访问普通端点（应该记录指标）"
+curl -s -X POST "$BASE_URL/invoke" \
+  -H "Content-Type: application/json" \
+  -d '{"number": 17}' > /dev/null
+echo "   ✓ 已访问 POST /invoke"
+echo ""
+
+echo "3. 访问任务端点（应该规范化为 /jobs/{job_id}）"
+curl -s "$BASE_URL/jobs/a1b2c3d4e5f6" > /dev/null
+curl -s "$BASE_URL/jobs/xyz123456789" > /dev/null
+echo "   ✓ 已访问 GET /jobs/a1b2c3d4e5f6 和 GET /jobs/xyz123456789"
+echo ""
+
+echo "4. 查看指标输出"
+echo "   查找 http_requests_total 指标："
+curl -s "$BASE_URL/metrics" | grep 'http_requests_total{' | grep -v '#'
+echo ""
+echo "   预期结果："
+echo "   - 应该看到 endpoint=\"/invoke\" 的记录"
+echo "   - 应该看到 endpoint=\"/jobs/{job_id}\" 的记录（而不是具体的 job_id）"
+echo "   - 不应该看到 endpoint=\"/healthz\" 或 endpoint=\"/readyz\" 的记录"
+echo "   - 不应该看到 endpoint=\"/metrics\" 的记录"
+echo ""
+echo "=== 测试完成 ==="

scripts/test_request_id_filter.sh

@@ -0,0 +1,69 @@
+#!/bin/bash
+# Grafana Request ID 过滤功能测试脚本
+
+set -e
+
+echo "========================================="
+echo "Grafana Request ID 过滤功能测试"
+echo "========================================="
+echo ""
+
+# 颜色定义
+GREEN='\033[0;32m'
+BLUE='\033[0;34m'
+YELLOW='\033[1;33m'
+NC='\033[0m' # No Color
+
+echo "1. 生成测试请求..."
+echo "-------------------"
+RESPONSE=$(curl -X POST http://localhost:8111/invoke \
+  -H "Content-Type: application/json" \
+  -d '{"number": 43}' \
+  -s)
+
+REQUEST_ID=$(echo "$RESPONSE" | jq -r '.request_id')
+echo -e "${GREEN}✓ 请求成功${NC}"
+echo -e "${BLUE}Request ID: $REQUEST_ID${NC}"
+
+echo ""
+echo "2. 等待日志收集 (5秒)..."
+sleep 5
+
+echo ""
+echo "3. 测试 Loki 过滤..."
+echo "-------------------"
+
+# 测试过滤特定 request_id
+LOG_COUNT=$(curl -G -s "http://localhost:3100/loki/api/v1/query_range" \
+  --data-urlencode "query={job=\"functional-scaffold-app\"} |= \"$REQUEST_ID\"" \
+  | jq '.data.result[0].values | length')
+
+if [ "$LOG_COUNT" -gt 0 ]; then
+    echo -e "${GREEN}✓ 找到 $LOG_COUNT 条日志${NC}"
+else
+    echo -e "${YELLOW}⚠ 没有找到日志，可能需要等待更长时间${NC}"
+fi
+
+echo ""
+echo "4. 显示日志内容..."
+echo "-------------------"
+curl -G -s "http://localhost:3100/loki/api/v1/query_range" \
+  --data-urlencode "query={job=\"functional-scaffold-app\"} |= \"$REQUEST_ID\"" \
+  | jq -r '.data.result[0].values[].[-1]' \
+  | jq -r '.message' \
+  | nl
+
+echo ""
+echo "========================================="
+echo "测试完成！"
+echo "========================================="
+echo ""
+echo "在 Grafana 中测试："
+echo "  1. 访问: http://localhost:3000"
+echo "  2. 进入 '日志监控' 仪表板"
+echo "  3. 在顶部 'Request ID' 输入框中输入："
+echo -e "     ${BLUE}$REQUEST_ID${NC}"
+echo "  4. 按回车，查看过滤后的日志"
+echo ""
+echo "清空 Request ID 输入框可以查看所有日志"
+echo ""

scripts/verify_loki.sh

@@ -0,0 +1,100 @@
+#!/bin/bash
+# Loki 集成验证脚本
+
+set -e
+
+echo "========================================="
+echo "Loki 日志收集系统验证"
+echo "========================================="
+echo ""
+
+# 颜色定义
+GREEN='\033[0;32m'
+RED='\033[0;31m'
+YELLOW='\033[1;33m'
+NC='\033[0m' # No Color
+
+# 检查服务状态
+echo "1. 检查服务状态..."
+echo "-------------------"
+docker-compose ps
+
+echo ""
+echo "2. 检查 Loki 健康状态..."
+echo "-------------------"
+if curl -s http://localhost:3100/ready | grep -q "ready"; then
+    echo -e "${GREEN}✓ Loki 服务正常${NC}"
+else
+    echo -e "${RED}✗ Loki 服务异常${NC}"
+    exit 1
+fi
+
+echo ""
+echo "3. 检查 Promtail 健康状态..."
+echo "-------------------"
+if curl -s http://localhost:9080/ready | grep -q "ready"; then
+    echo -e "${GREEN}✓ Promtail 服务正常${NC}"
+else
+    echo -e "${RED}✗ Promtail 服务异常${NC}"
+    exit 1
+fi
+
+echo ""
+echo "4. 生成测试日志..."
+echo "-------------------"
+curl -X POST http://localhost:8111/invoke \
+  -H "Content-Type: application/json" \
+  -d '{"algorithm": "PrimeChecker", "params": {"number": 17}}' \
+  -s -o /dev/null -w "HTTP Status: %{http_code}\n"
+
+echo ""
+echo "5. 等待日志收集 (5秒)..."
+sleep 5
+
+echo ""
+echo "6. 查询 Loki 日志..."
+echo "-------------------"
+LOGS=$(curl -G -s "http://localhost:3100/loki/api/v1/query_range" \
+  --data-urlencode 'query={job="functional-scaffold-app"}' \
+  --data-urlencode 'limit=5')
+
+if echo "$LOGS" | jq -e '.data.result | length > 0' > /dev/null 2>&1; then
+    echo -e "${GREEN}✓ 成功查询到日志${NC}"
+    echo ""
+    echo "最近的日志条目："
+    echo "$LOGS" | jq -r '.data.result[0].values[-1][1]' | head -3
+else
+    echo -e "${YELLOW}⚠ 暂时没有查询到日志，可能需要等待更长时间${NC}"
+fi
+
+echo ""
+echo "7. 检查 Grafana 数据源..."
+echo "-------------------"
+DATASOURCES=$(curl -s -u admin:admin http://localhost:3000/api/datasources)
+if echo "$DATASOURCES" | jq -e '.[] | select(.name == "Loki")' > /dev/null 2>&1; then
+    echo -e "${GREEN}✓ Loki 数据源已配置${NC}"
+else
+    echo -e "${RED}✗ Loki 数据源未配置${NC}"
+fi
+
+if echo "$DATASOURCES" | jq -e '.[] | select(.name == "Prometheus")' > /dev/null 2>&1; then
+    echo -e "${GREEN}✓ Prometheus 数据源已配置${NC}"
+else
+    echo -e "${RED}✗ Prometheus 数据源未配置${NC}"
+fi
+
+echo ""
+echo "========================================="
+echo "验证完成！"
+echo "========================================="
+echo ""
+echo "访问地址："
+echo "  - Grafana: http://localhost:3000 (admin/admin)"
+echo "  - Loki: http://localhost:3100"
+echo "  - Promtail: http://localhost:9080"
+echo ""
+echo "查看日志："
+echo "  1. 访问 Grafana Explore: http://localhost:3000/explore"
+echo "  2. 选择 Loki 数据源"
+echo "  3. 输入查询: {job=\"functional-scaffold-app\"}"
+echo ""

src/functional_scaffold/algorithms/base.py

@@ -32,7 +32,7 @@ class BaseAlgorithm(ABC):
         Returns:
             Dict[str, Any]: 包含结果和元数据的字典
         """
-        from ..core.metrics_unified import incr, observe
+        from ..core.metrics_unified import incr_sync, observe_sync
 
         start_time = time.time()
         status = "success"
@@ -71,5 +71,7 @@ class BaseAlgorithm(ABC):
         finally:
             # 记录算法执行指标
             elapsed_time = time.time() - start_time
-            incr("algorithm_executions_total", {"algorithm": self.name, "status": status})
+            incr_sync("algorithm_executions_total", {"algorithm": self.name, "status": status})
-            observe("algorithm_execution_duration_seconds", {"algorithm": self.name}, elapsed_time)
+            observe_sync(
+                "algorithm_execution_duration_seconds", {"algorithm": self.name}, elapsed_time
+            )

src/functional_scaffold/algorithms/prime_checker.py

@@ -2,7 +2,7 @@
 
 from typing import Dict, Any, List
 from .base import BaseAlgorithm
-from ..core.metrics_unified import incr
+from ..core.metrics_unified import incr_sync
 
 
 class PrimeChecker(BaseAlgorithm):
@@ -31,12 +31,12 @@ class PrimeChecker(BaseAlgorithm):
             ValueError: 如果输入不是整数
         """
         if not isinstance(number, int):
-            incr('prime_check',{"status":"invalid_input"})
+            incr_sync('prime_check', {"status": "invalid_input"})
             raise ValueError(f"Input must be an integer, got {type(number).__name__}")
 
         # 小于2的数不是质数
         if number < 2:
-            incr('prime_check', {"status": "number_little_two"})
+            incr_sync('prime_check', {"status": "number_little_two"})
             return {
                 "number": number,
                 "is_prime": False,
@@ -50,7 +50,7 @@ class PrimeChecker(BaseAlgorithm):
 
         # 如果不是质数，计算因数
         factors = [] if is_prime else self._get_factors(number)
-        incr('prime_check', {"status": "success"})
+        incr_sync('prime_check', {"status": "success"})
         return {
             "number": number,
             "is_prime": is_prime,

src/functional_scaffold/api/dependencies.py

@@ -2,7 +2,7 @@
 
 from fastapi import Header, HTTPException
 from typing import Optional
-from ..core.tracing import set_request_id, generate_request_id
+from ..core.tracing import set_request_id, generate_request_id, get_request_id as get_current_request_id
 
 
 async def get_request_id(x_request_id: Optional[str] = Header(None)) -> str:
@@ -15,6 +15,12 @@ async def get_request_id(x_request_id: Optional[str] = Header(None)) -> str:
     Returns:
         str: 请求ID
     """
+    # 先检查 ContextVar 中是否已经有 request_id（由中间件设置）
+    existing_request_id = get_current_request_id()
+    if existing_request_id:
+        return existing_request_id
+
+    # 如果没有，则从请求头获取或生成新的
     request_id = x_request_id or generate_request_id()
     set_request_id(request_id)
     return request_id

src/functional_scaffold/api/models.py

@@ -152,3 +152,21 @@ class JobStatusResponse(BaseModel):
     result: Optional[Dict[str, Any]] = Field(None, description="执行结果（仅完成时返回）")
     error: Optional[str] = Field(None, description="错误信息（仅失败时返回）")
     metadata: Optional[Dict[str, Any]] = Field(None, description="元数据信息")
+
+
+class ConcurrencyStatusResponse(BaseModel):
+    """并发状态响应"""
+
+    model_config = ConfigDict(
+        json_schema_extra={
+            "example": {
+                "max_concurrent": 10,
+                "available_slots": 7,
+                "running_jobs": 3,
+            }
+        }
+    )
+
+    max_concurrent: int = Field(..., description="最大并发任务数")
+    available_slots: int = Field(..., description="当前可用槽位数")
+    running_jobs: int = Field(..., description="当前运行中的任务数")

src/functional_scaffold/api/routes.py

@@ -1,6 +1,5 @@
 """API 路由"""
 
-import asyncio
 from fastapi import APIRouter, HTTPException, Depends, status
 import time
 import logging
@@ -15,6 +14,7 @@ from .models import (
     JobCreateResponse,
     JobStatusResponse,
     JobStatus,
+    ConcurrencyStatusResponse,
 )
 from .dependencies import get_request_id
 from ..algorithms.prime_checker import PrimeChecker
@@ -199,10 +199,10 @@ async def create_job(
         # 获取任务信息
         job_data = await job_manager.get_job(job_id)
 
-        # 后台执行任务
+        # 任务入队，由 Worker 执行
-        asyncio.create_task(job_manager.execute_job(job_id))
+        await job_manager.enqueue_job(job_id)
 
-        logger.info(f"异步任务已创建: job_id={job_id}, request_id={request_id}")
+        logger.info(f"异步任务已创建并入队: job_id={job_id}, request_id={request_id}")
 
         return JobCreateResponse(
             job_id=job_id,
@@ -292,3 +292,57 @@ async def get_job_status(job_id: str):
                 "message": str(e),
             },
         )
+
+
+@router.get(
+    "/jobs/concurrency/status",
+    response_model=ConcurrencyStatusResponse,
+    summary="查询并发状态",
+    description="查询任务管理器的并发执行状态",
+    responses={
+        200: {"description": "成功", "model": ConcurrencyStatusResponse},
+        503: {"description": "服务不可用", "model": ErrorResponse},
+    },
+)
+async def get_concurrency_status():
+    """
+    查询并发状态
+
+    返回当前任务管理器的并发执行状态，包括：
+    - 最大并发任务数
+    - 当前可用槽位数
+    - 当前运行中的任务数
+    """
+    try:
+        job_manager = await get_job_manager()
+
+        # 检查任务管理器是否可用
+        if not job_manager.is_available():
+            raise HTTPException(
+                status_code=status.HTTP_503_SERVICE_UNAVAILABLE,
+                detail={
+                    "error": "SERVICE_UNAVAILABLE",
+                    "message": "任务管理器不可用",
+                },
+            )
+
+        concurrency_status = job_manager.get_concurrency_status()
+
+        return ConcurrencyStatusResponse(
+            max_concurrent=concurrency_status["max_concurrent"],
+            available_slots=concurrency_status["available_slots"],
+            running_jobs=concurrency_status["running_jobs"],
+        )
+
+    except HTTPException:
+        raise
+
+    except Exception as e:
+        logger.error(f"查询并发状态失败: {str(e)}", exc_info=True)
+        raise HTTPException(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            detail={
+                "error": "INTERNAL_ERROR",
+                "message": str(e),
+            },
+        )

src/functional_scaffold/config.py

@@ -23,6 +23,8 @@ class Settings(BaseSettings):
     # 日志配置
     log_level: str = "INFO"
     log_format: str = "json"
+    log_file_enabled: bool = False
+    log_file_path: str = "/var/log/app/app.log"
 
     # 指标配置
     metrics_enabled: bool = True
@@ -53,6 +55,27 @@ class Settings(BaseSettings):
     job_result_ttl: int = 1800  # 结果缓存时间（秒），默认 30 分钟
     webhook_max_retries: int = 3  # Webhook 最大重试次数
     webhook_timeout: int = 10  # Webhook 超时时间（秒）
+    max_concurrent_jobs: int = 10  # 最大并发任务数
+
+    # Worker 配置
+    worker_poll_interval: float = 0.1  # Worker 轮询间隔（秒）
+    job_queue_key: str = "job:queue"  # 任务队列 Redis Key
+    job_concurrency_key: str = "job:concurrency"  # 全局并发计数器 Redis Key
+    job_lock_ttl: int = 300  # 任务锁 TTL（秒）
+    job_max_retries: int = 3  # 任务最大重试次数
+    job_execution_timeout: int = 300  # 任务执行超时（秒）
+
+    # 处理队列配置
+    job_processing_key: str = "job:processing"  # 处理中队列
+    job_processing_ts_key: str = "job:processing:ts"  # 处理时间戳 ZSET
+    job_dlq_key: str = "job:dlq"  # 死信队列
+
+    # 锁配置扩展
+    job_lock_buffer: int = 60  # 锁 TTL 缓冲时间（秒）
+
+    # 回收器配置
+    job_sweeper_enabled: bool = True  # 启用回收器
+    job_sweeper_interval: int = 60  # 回收扫描间隔（秒）
 
 
 # 全局配置实例

src/functional_scaffold/core/job_manager.py

@@ -7,6 +7,7 @@ import asyncio
 import json
 import logging
 import secrets
+import time
 from datetime import datetime, timezone
 from typing import Any, Dict, List, Optional, Type
 
@@ -16,6 +17,7 @@ import redis.asyncio as aioredis
 from ..algorithms.base import BaseAlgorithm
 from ..config import settings
 from ..core.metrics_unified import incr, observe
+from ..core.tracing import set_request_id
 
 logger = logging.getLogger(__name__)
 
@@ -23,10 +25,30 @@ logger = logging.getLogger(__name__)
 class JobManager:
     """异步任务管理器"""
 
+    # Lua 脚本：安全释放锁（验证 token）
+    RELEASE_LOCK_SCRIPT = """
+local current = redis.call('GET', KEYS[1])
+if current == ARGV[1] then
+    return redis.call('DEL', KEYS[1])
+end
+return 0
+"""
+
+    # Lua 脚本：锁续租（验证 token 后延长 TTL）
+    RENEW_LOCK_SCRIPT = """
+local current = redis.call('GET', KEYS[1])
+if current == ARGV[1] then
+    return redis.call('EXPIRE', KEYS[1], ARGV[2])
+end
+return 0
+"""
+
     def __init__(self):
         self._redis_client: Optional[aioredis.Redis] = None
         self._algorithm_registry: Dict[str, Type[BaseAlgorithm]] = {}
         self._http_client: Optional[httpx.AsyncClient] = None
+        self._semaphore: Optional[asyncio.Semaphore] = None
+        self._max_concurrent_jobs: int = 0
 
     async def initialize(self) -> None:
         """初始化 Redis 连接和 HTTP 客户端"""
@@ -51,6 +73,11 @@ class JobManager:
         # 初始化 HTTP 客户端
         self._http_client = httpx.AsyncClient(timeout=settings.webhook_timeout)
 
+        # 初始化并发控制信号量
+        self._max_concurrent_jobs = settings.max_concurrent_jobs
+        self._semaphore = asyncio.Semaphore(self._max_concurrent_jobs)
+        logger.info(f"任务并发限制已设置: {self._max_concurrent_jobs}")
+
         # 注册算法
         self._register_algorithms()
 
@@ -141,7 +168,7 @@ class JobManager:
         await self._redis_client.hset(key, mapping=job_data)
 
         # 记录指标
-        incr("jobs_created_total", {"algorithm": algorithm})
+        await incr("jobs_created_total", {"algorithm": algorithm})
 
         logger.info(f"任务已创建: job_id={job_id}, algorithm={algorithm}")
         return job_id
@@ -169,6 +196,7 @@ class JobManager:
             "job_id": job_id,
             "status": job_data.get("status", ""),
             "algorithm": job_data.get("algorithm", ""),
+            "request_id": job_data.get("request_id") or None,
             "created_at": job_data.get("created_at", ""),
             "started_at": job_data.get("started_at") or None,
             "completed_at": job_data.get("completed_at") or None,
@@ -203,6 +231,10 @@ class JobManager:
             logger.error(f"Redis 不可用，无法执行任务: {job_id}")
             return
 
+        if not self._semaphore:
+            logger.error(f"并发控制未初始化，无法执行任务: {job_id}")
+            return
+
         key = f"job:{job_id}"
         job_data = await self._redis_client.hgetall(key)
 
@@ -212,6 +244,11 @@ class JobManager:
 
         algorithm_name = job_data.get("algorithm", "")
         webhook_url = job_data.get("webhook", "")
+        request_id = job_data.get("request_id", "")
+
+        # 设置 request_id 上下文，确保日志中包含 request_id
+        if request_id:
+            set_request_id(request_id)
 
         # 解析参数
         try:
@@ -219,74 +256,82 @@ class JobManager:
         except json.JSONDecodeError:
             params = {}
 
-        # 更新状态为 running
+        # 使用信号量控制并发
-        started_at = self._get_timestamp()
+        async with self._semaphore:
-        await self._redis_client.hset(key, mapping={"status": "running", "started_at": started_at})
+            # 更新状态为 running
+            started_at = self._get_timestamp()
+            await self._redis_client.hset(
+                key, mapping={"status": "running", "started_at": started_at}
+            )
 
-        logger.info(f"开始执行任务: job_id={job_id}, algorithm={algorithm_name}")
+            logger.info(f"开始执行任务: job_id={job_id}, algorithm={algorithm_name}")
 
-        import time
+            import time
 
-        start_time = time.time()
+            start_time = time.time()
-        status = "completed"
+            status = "completed"
-        result_data = None
+            result_data = None
-        error_msg = None
+            error_msg = None
-        metadata = None
+            metadata = None
 
-        try:
+            try:
-            # 获取算法类并执行
+                # 获取算法类并执行
-            algorithm_cls = self._algorithm_registry.get(algorithm_name)
+                algorithm_cls = self._algorithm_registry.get(algorithm_name)
-            if not algorithm_cls:
+                if not algorithm_cls:
-                raise ValueError(f"算法 '{algorithm_name}' 不存在")
+                    raise ValueError(f"算法 '{algorithm_name}' 不存在")
 
-            algorithm = algorithm_cls()
+                algorithm = algorithm_cls()
 
-            # 根据算法类型传递参数
+                # 根据算法类型传递参数
-            if algorithm_name == "PrimeChecker":
+                if algorithm_name == "PrimeChecker":
-                execution_result = algorithm.execute(params.get("number", 0))
+                    execution_result = algorithm.execute(params.get("number", 0))
-            else:
+                else:
-                # 通用参数传递
+                    # 通用参数传递
-                execution_result = algorithm.execute(**params)
+                    execution_result = algorithm.execute(**params)
 
-            if execution_result.get("success"):
+                if execution_result.get("success"):
-                result_data = execution_result.get("result", {})
+                    result_data = execution_result.get("result", {})
-                metadata = execution_result.get("metadata", {})
+                    metadata = execution_result.get("metadata", {})
-            else:
+                else:
+                    status = "failed"
+                    error_msg = execution_result.get("error", "算法执行失败")
+                    metadata = execution_result.get("metadata", {})
+
+            except Exception as e:
                 status = "failed"
-                error_msg = execution_result.get("error", "算法执行失败")
+                error_msg = str(e)
-                metadata = execution_result.get("metadata", {})
+                logger.error(f"任务执行失败: job_id={job_id}, error={e}", exc_info=True)
 
-        except Exception as e:
+            # 计算执行时间
-            status = "failed"
+            elapsed_time = time.time() - start_time
-            error_msg = str(e)
+            completed_at = self._get_timestamp()
-            logger.error(f"任务执行失败: job_id={job_id}, error={e}", exc_info=True)
 
-        # 计算执行时间
+            # 更新任务状态
-        elapsed_time = time.time() - start_time
+            update_data = {
-        completed_at = self._get_timestamp()
+                "status": status,
+                "completed_at": completed_at,
+                "result": json.dumps(result_data) if result_data else "",
+                "error": error_msg or "",
+                "metadata": json.dumps(metadata) if metadata else "",
+            }
+            await self._redis_client.hset(key, mapping=update_data)
 
-        # 更新任务状态
+            # 设置 TTL
-        update_data = {
+            await self._redis_client.expire(key, settings.job_result_ttl)
-            "status": status,
-            "completed_at": completed_at,
-            "result": json.dumps(result_data) if result_data else "",
-            "error": error_msg or "",
-            "metadata": json.dumps(metadata) if metadata else "",
-        }
-        await self._redis_client.hset(key, mapping=update_data)
 
-        # 设置 TTL
+            # 记录指标
-        await self._redis_client.expire(key, settings.job_result_ttl)
+            await incr("jobs_completed_total", {"algorithm": algorithm_name, "status": status})
+            await observe(
+                "job_execution_duration_seconds", {"algorithm": algorithm_name}, elapsed_time
+            )
 
-        # 记录指标
+            logger.info(
-        incr("jobs_completed_total", {"algorithm": algorithm_name, "status": status})
+                f"任务执行完成: job_id={job_id}, status={status}, elapsed={elapsed_time:.3f}s"
-        observe("job_execution_duration_seconds", {"algorithm": algorithm_name}, elapsed_time)
+            )
 
-        logger.info(f"任务执行完成: job_id={job_id}, status={status}, elapsed={elapsed_time:.3f}s")
+            # 发送 Webhook 回调
-
+            if webhook_url:
-        # 发送 Webhook 回调
+                await self._send_webhook(job_id, webhook_url)
-        if webhook_url:
-            await self._send_webhook(job_id, webhook_url)
 
     async def _send_webhook(self, job_id: str, webhook_url: str) -> None:
         """发送 Webhook 回调（带重试）
@@ -329,7 +374,7 @@ class JobManager:
                 )
 
                 if response.status_code < 400:
-                    incr("webhook_deliveries_total", {"status": "success"})
+                    await incr("webhook_deliveries_total", {"status": "success"})
                     logger.info(
                         f"Webhook 发送成功: job_id={job_id}, url={webhook_url}, "
                         f"status_code={response.status_code}"
@@ -352,13 +397,445 @@ class JobManager:
                 await asyncio.sleep(delay)
 
         # 所有重试都失败
-        incr("webhook_deliveries_total", {"status": "failed"})
+        await incr("webhook_deliveries_total", {"status": "failed"})
         logger.error(f"Webhook 发送最终失败: job_id={job_id}, url={webhook_url}")
 
     def is_available(self) -> bool:
         """检查任务管理器是否可用"""
         return self._redis_client is not None
 
+    async def enqueue_job(self, job_id: str) -> bool:
+        """将任务加入队列
+
+        Args:
+            job_id: 任务 ID
+
+        Returns:
+            bool: 是否成功入队
+        """
+        if not self._redis_client:
+            logger.error(f"Redis 不可用，无法入队任务: {job_id}")
+            return False
+
+        try:
+            await self._redis_client.lpush(settings.job_queue_key, job_id)
+            logger.info(f"任务已入队: job_id={job_id}")
+            return True
+        except Exception as e:
+            logger.error(f"任务入队失败: job_id={job_id}, error={e}")
+            return False
+
+    async def dequeue_job(self, timeout: int = 5) -> Optional[str]:
+        """从队列获取任务（阻塞式，转移式出队）
+
+        使用 BLMOVE 原子性地将任务从 job:queue 移动到 job:processing，
+        防止 Worker 崩溃时任务丢失。
+
+        Args:
+            timeout: 阻塞超时时间（秒）
+
+        Returns:
+            Optional[str]: 任务 ID，超时返回 None
+        """
+        if not self._redis_client:
+            return None
+
+        try:
+            # 使用 BLMOVE 原子性转移任务
+            job_id = await self._redis_client.blmove(
+                settings.job_queue_key,  # 源: job:queue
+                settings.job_processing_key,  # 目标: job:processing
+                timeout,
+                "RIGHT",
+                "LEFT",
+            )
+            if job_id:
+                # 记录出队时间戳到 ZSET
+                await self._redis_client.zadd(settings.job_processing_ts_key, {job_id: time.time()})
+                logger.debug(f"任务已转移到处理队列: {job_id}")
+            return job_id
+        except Exception as e:
+            logger.error(f"任务出队失败: error={e}")
+            return None
+
+    async def acquire_job_lock(self, job_id: str) -> Optional[str]:
+        """获取任务执行锁（分布式锁，带 Token）
+
+        Args:
+            job_id: 任务 ID
+
+        Returns:
+            Optional[str]: 成功时返回锁 token，失败返回 None
+        """
+        if not self._redis_client:
+            return None
+
+        lock_key = f"job:lock:{job_id}"
+        lock_token = secrets.token_hex(16)  # 随机 token
+        lock_ttl = settings.job_execution_timeout + settings.job_lock_buffer
+        try:
+            acquired = await self._redis_client.set(lock_key, lock_token, nx=True, ex=lock_ttl)
+            if acquired:
+                logger.debug(f"获取任务锁成功: job_id={job_id}")
+                return lock_token
+            return None
+        except Exception as e:
+            logger.error(f"获取任务锁失败: job_id={job_id}, error={e}")
+            return None
+
+    async def release_job_lock(self, job_id: str, lock_token: Optional[str] = None) -> bool:
+        """释放任务执行锁（使用 Lua 脚本验证 token）
+
+        Args:
+            job_id: 任务 ID
+            lock_token: 锁 token（用于验证所有权）
+
+        Returns:
+            bool: 是否成功释放锁
+        """
+        if not self._redis_client:
+            return False
+
+        lock_key = f"job:lock:{job_id}"
+        try:
+            if lock_token:
+                # 使用 Lua 脚本安全释放锁
+                result = await self._redis_client.eval(
+                    self.RELEASE_LOCK_SCRIPT, 1, lock_key, lock_token
+                )
+                if result == 1:
+                    logger.debug(f"释放任务锁成功: job_id={job_id}")
+                    return True
+                else:
+                    logger.warning(f"释放任务锁失败（token 不匹配）: job_id={job_id}")
+                    return False
+            else:
+                # 向后兼容：无 token 时直接删除
+                await self._redis_client.delete(lock_key)
+                logger.debug(f"释放任务锁成功（无 token 验证）: job_id={job_id}")
+                return True
+        except Exception as e:
+            logger.error(f"释放任务锁失败: job_id={job_id}, error={e}")
+            return False
+
+    async def increment_concurrency(self) -> int:
+        """增加全局并发计数
+
+        Returns:
+            int: 增加后的并发数
+        """
+        if not self._redis_client:
+            return 0
+
+        try:
+            count = await self._redis_client.incr(settings.job_concurrency_key)
+            return count
+        except Exception as e:
+            logger.error(f"增加并发计数失败: error={e}")
+            return 0
+
+    async def decrement_concurrency(self) -> int:
+        """减少全局并发计数
+
+        Returns:
+            int: 减少后的并发数
+        """
+        if not self._redis_client:
+            return 0
+
+        try:
+            count = await self._redis_client.decr(settings.job_concurrency_key)
+            # 防止计数变为负数
+            if count < 0:
+                await self._redis_client.set(settings.job_concurrency_key, 0)
+                return 0
+            return count
+        except Exception as e:
+            logger.error(f"减少并发计数失败: error={e}")
+            return 0
+
+    async def get_global_concurrency(self) -> int:
+        """获取当前全局并发数
+
+        Returns:
+            int: 当前并发数
+        """
+        if not self._redis_client:
+            return 0
+
+        try:
+            count = await self._redis_client.get(settings.job_concurrency_key)
+            return int(count) if count else 0
+        except Exception as e:
+            logger.error(f"获取并发计数失败: error={e}")
+            return 0
+
+    async def can_execute(self) -> bool:
+        """检查是否可以执行新任务（全局并发控制）
+
+        Returns:
+            bool: 是否可以执行
+        """
+        current = await self.get_global_concurrency()
+        return current < settings.max_concurrent_jobs
+
+    async def get_job_retry_count(self, job_id: str) -> int:
+        """获取任务重试次数
+
+        Args:
+            job_id: 任务 ID
+
+        Returns:
+            int: 重试次数
+        """
+        if not self._redis_client:
+            return 0
+
+        key = f"job:{job_id}"
+        try:
+            retry_count = await self._redis_client.hget(key, "retry_count")
+            return int(retry_count) if retry_count else 0
+        except Exception:
+            return 0
+
+    async def increment_job_retry(self, job_id: str) -> int:
+        """增加任务重试次数
+
+        Args:
+            job_id: 任务 ID
+
+        Returns:
+            int: 增加后的重试次数
+        """
+        if not self._redis_client:
+            return 0
+
+        key = f"job:{job_id}"
+        try:
+            await self._redis_client.hincrby(key, "retry_count", 1)
+            retry_count = await self._redis_client.hget(key, "retry_count")
+            return int(retry_count) if retry_count else 1
+        except Exception as e:
+            logger.error(f"增加重试次数失败: job_id={job_id}, error={e}")
+            return 0
+
+    async def ack_job(self, job_id: str) -> bool:
+        """确认任务完成（从处理队列移除）
+
+        Args:
+            job_id: 任务 ID
+
+        Returns:
+            bool: 是否成功确认
+        """
+        if not self._redis_client:
+            return False
+
+        try:
+            async with self._redis_client.pipeline(transaction=True) as pipe:
+                pipe.lrem(settings.job_processing_key, 1, job_id)
+                pipe.zrem(settings.job_processing_ts_key, job_id)
+                await pipe.execute()
+            logger.debug(f"任务已确认完成: job_id={job_id}")
+            return True
+        except Exception as e:
+            logger.error(f"确认任务失败: job_id={job_id}, error={e}")
+            return False
+
+    async def nack_job(self, job_id: str, requeue: bool = True) -> bool:
+        """拒绝任务（从处理队列移除，根据重试次数决定重新入队或进死信队列）
+
+        Args:
+            job_id: 任务 ID
+            requeue: 是否尝试重新入队
+
+        Returns:
+            bool: 是否成功处理
+        """
+        if not self._redis_client:
+            return False
+
+        try:
+            retry_count = await self.get_job_retry_count(job_id)
+            async with self._redis_client.pipeline(transaction=True) as pipe:
+                pipe.lrem(settings.job_processing_key, 1, job_id)
+                pipe.zrem(settings.job_processing_ts_key, job_id)
+                if requeue and retry_count < settings.job_max_retries:
+                    pipe.lpush(settings.job_queue_key, job_id)
+                    logger.info(f"任务重新入队: job_id={job_id}, retry_count={retry_count}")
+                else:
+                    pipe.lpush(settings.job_dlq_key, job_id)
+                    logger.warning(f"任务进入死信队列: job_id={job_id}, retry_count={retry_count}")
+                await pipe.execute()
+            return True
+        except Exception as e:
+            logger.error(f"拒绝任务失败: job_id={job_id}, error={e}")
+            return False
+
+    async def renew_job_lock(self, job_id: str, lock_token: str) -> bool:
+        """续租任务锁（延长 TTL）
+
+        Args:
+            job_id: 任务 ID
+            lock_token: 锁 token
+
+        Returns:
+            bool: 是否成功续租
+        """
+        if not self._redis_client:
+            return False
+
+        lock_key = f"job:lock:{job_id}"
+        lock_ttl = settings.job_execution_timeout + settings.job_lock_buffer
+        try:
+            result = await self._redis_client.eval(
+                self.RENEW_LOCK_SCRIPT, 1, lock_key, lock_token, lock_ttl
+            )
+            if result == 1:
+                logger.debug(f"锁续租成功: job_id={job_id}")
+                return True
+            else:
+                logger.warning(f"锁续租失败（token 不匹配或锁已过期）: job_id={job_id}")
+                return False
+        except Exception as e:
+            logger.error(f"锁续租失败: job_id={job_id}, error={e}")
+            return False
+
+    async def recover_stale_jobs(self) -> int:
+        """回收超时任务
+
+        扫描 job:processing:ts ZSET，找出超时的任务，
+        根据重试次数决定重新入队或进死信队列。
+
+        Returns:
+            int: 回收的任务数量
+        """
+        if not self._redis_client:
+            return 0
+
+        timeout = settings.job_execution_timeout + settings.job_lock_buffer
+        cutoff = time.time() - timeout
+
+        try:
+            # 获取超时任务列表
+            stale_jobs = await self._redis_client.zrangebyscore(
+                settings.job_processing_ts_key, "-inf", cutoff
+            )
+
+            recovered = 0
+            for job_id in stale_jobs:
+                # 增加重试次数
+                await self.increment_job_retry(job_id)
+                retry_count = await self.get_job_retry_count(job_id)
+
+                async with self._redis_client.pipeline(transaction=True) as pipe:
+                    pipe.lrem(settings.job_processing_key, 1, job_id)
+                    pipe.zrem(settings.job_processing_ts_key, job_id)
+                    if retry_count < settings.job_max_retries:
+                        pipe.lpush(settings.job_queue_key, job_id)
+                        logger.info(f"超时任务重新入队: job_id={job_id}, retry_count={retry_count}")
+                    else:
+                        pipe.lpush(settings.job_dlq_key, job_id)
+                        logger.warning(
+                            f"超时任务进入死信队列: job_id={job_id}, retry_count={retry_count}"
+                        )
+                    await pipe.execute()
+                recovered += 1
+
+            if recovered > 0:
+                logger.info(f"回收超时任务完成: 共 {recovered} 个")
+            return recovered
+        except Exception as e:
+            logger.error(f"回收超时任务失败: error={e}")
+            return 0
+
+    def get_concurrency_status(self) -> Dict[str, int]:
+        """获取并发状态
+
+        Returns:
+            Dict[str, int]: 包含以下键的字典
+                - max_concurrent: 最大并发数
+                - available_slots: 可用槽位数
+                - running_jobs: 当前运行中的任务数
+        """
+        if not self._semaphore:
+            return {
+                "max_concurrent": 0,
+                "available_slots": 0,
+                "running_jobs": 0,
+            }
+
+        max_concurrent = self._max_concurrent_jobs
+        available_slots = self._semaphore._value
+        running_jobs = max_concurrent - available_slots
+
+        return {
+            "max_concurrent": max_concurrent,
+            "available_slots": available_slots,
+            "running_jobs": running_jobs,
+        }
+
+    async def collect_queue_metrics(self) -> Dict[str, Any]:
+        """收集队列监控指标
+
+        Returns:
+            Dict[str, Any]: 包含以下键的字典
+                - queue_length: 待处理队列长度
+                - processing_length: 处理中队列长度
+                - dlq_length: 死信队列长度
+                - oldest_waiting_seconds: 最长等待时间（秒）
+        """
+        if not self._redis_client:
+            return {
+                "queue_length": 0,
+                "processing_length": 0,
+                "dlq_length": 0,
+                "oldest_waiting_seconds": 0,
+            }
+
+        try:
+            # 使用 pipeline 批量获取队列长度
+            async with self._redis_client.pipeline(transaction=False) as pipe:
+                pipe.llen(settings.job_queue_key)
+                pipe.llen(settings.job_processing_key)
+                pipe.llen(settings.job_dlq_key)
+                pipe.zrange(settings.job_processing_ts_key, 0, 0, withscores=True)
+                results = await pipe.execute()
+
+            queue_length = results[0] or 0
+            processing_length = results[1] or 0
+            dlq_length = results[2] or 0
+
+            # 计算最长等待时间
+            oldest_waiting_seconds = 0
+            if results[3]:
+                # results[3] 是 [(job_id, timestamp), ...] 格式
+                oldest_ts = results[3][0][1]
+                oldest_waiting_seconds = time.time() - oldest_ts
+
+            # 更新指标
+            from .metrics_unified import set as metrics_set
+
+            await metrics_set("job_queue_length", {"queue": "pending"}, queue_length)
+            await metrics_set("job_queue_length", {"queue": "processing"}, processing_length)
+            await metrics_set("job_queue_length", {"queue": "dlq"}, dlq_length)
+            await metrics_set("job_oldest_waiting_seconds", None, oldest_waiting_seconds)
+
+            return {
+                "queue_length": queue_length,
+                "processing_length": processing_length,
+                "dlq_length": dlq_length,
+                "oldest_waiting_seconds": oldest_waiting_seconds,
+            }
+        except Exception as e:
+            logger.error(f"收集队列指标失败: error={e}")
+            return {
+                "queue_length": 0,
+                "processing_length": 0,
+                "dlq_length": 0,
+                "oldest_waiting_seconds": 0,
+            }
+
 
 # 全局单例
 _job_manager: Optional[JobManager] = None

src/functional_scaffold/core/logging.py

@@ -2,14 +2,39 @@
 
 import logging
 import sys
+from pathlib import Path
 from typing import Optional
+from logging.handlers import RotatingFileHandler
 from pythonjsonlogger.json import JsonFormatter
 
+from .tracing import get_request_id
+
+
+class RequestIdFilter(logging.Filter):
+    """自动添加 request_id 到日志记录的过滤器"""
+
+    def filter(self, record: logging.LogRecord) -> bool:
+        """
+        为日志记录添加 request_id 字段
+
+        Args:
+            record: 日志记录
+
+        Returns:
+            bool: 总是返回 True（不过滤任何日志）
+        """
+        # 从 ContextVar 中获取 request_id
+        request_id = get_request_id()
+        # 添加到日志记录中，如果没有则设置为 None
+        record.request_id = request_id if request_id else "-"
+        return True
+
 
 def setup_logging(
     level: str = "INFO",
     format_type: str = "json",
     logger_name: Optional[str] = None,
+    file_path: Optional[str] = None,
 ) -> logging.Logger:
     """
     配置日志系统
@@ -18,6 +43,7 @@ def setup_logging(
         level: 日志级别 (DEBUG, INFO, WARNING, ERROR, CRITICAL)
         format_type: 日志格式 ('json' 或 'text')
         logger_name: 日志器名称，None表示根日志器
+        file_path: 日志文件路径，None表示不写入文件
 
     Returns:
         logging.Logger: 配置好的日志器
@@ -28,23 +54,45 @@ def setup_logging(
     # 清除现有处理器
     logger.handlers.clear()
 
-    # 创建控制台处理器
-    handler = logging.StreamHandler(sys.stdout)
-    handler.setLevel(getattr(logging, level.upper()))
-
     # 设置格式
     if format_type == "json":
         formatter = JsonFormatter(
-            "%(asctime)s %(name)s %(levelname)s %(message)s",
+            "%(asctime)s %(name)s %(levelname)s %(message)s %(request_id)s",
             timestamp=True,
         )
     else:
         formatter = logging.Formatter(
-            "%(asctime)s - %(name)s - %(levelname)s - %(message)s",
+            "%(asctime)s - %(name)s - %(levelname)s - [%(request_id)s] - %(message)s",
             datefmt="%Y-%m-%d %H:%M:%S",
         )
 
-    handler.setFormatter(formatter)
+    # 创建 RequestIdFilter
-    logger.addHandler(handler)
+    request_id_filter = RequestIdFilter()
+
+    # 创建控制台处理器
+    console_handler = logging.StreamHandler(sys.stdout)
+    console_handler.setLevel(getattr(logging, level.upper()))
+    console_handler.setFormatter(formatter)
+    console_handler.addFilter(request_id_filter)
+    logger.addHandler(console_handler)
+
+    # 创建文件处理器（如果指定了文件路径）
+    if file_path:
+        # 确保日志目录存在
+        log_dir = Path(file_path).parent
+        log_dir.mkdir(parents=True, exist_ok=True)
+
+        # 创建 RotatingFileHandler
+        # 最大 100MB，保留 5 个备份
+        file_handler = RotatingFileHandler(
+            file_path,
+            maxBytes=100 * 1024 * 1024,  # 100MB
+            backupCount=5,
+            encoding="utf-8",
+        )
+        file_handler.setLevel(getattr(logging, level.upper()))
+        file_handler.setFormatter(formatter)
+        file_handler.addFilter(request_id_filter)
+        logger.addHandler(file_handler)
 
     return logger

src/functional_scaffold/core/metrics_unified.py

@@ -1,19 +1,21 @@
 """统一指标管理模块
 
 基于 Redis 的指标收集方案，支持多实例部署和 YAML 配置。
+使用异步 Redis 客户端，避免在异步请求路径中阻塞事件循环。
 """
 
 import os
 import re
 import socket
 import logging
+import asyncio
 from pathlib import Path
 from typing import Any, Dict, List, Optional
 from functools import wraps
 import time
 
 import yaml
-import redis
+import redis.asyncio as aioredis
 
 logger = logging.getLogger(__name__)
 
@@ -22,7 +24,7 @@ class MetricsManager:
     """统一指标管理器
 
     支持从 YAML 配置文件加载指标定义，使用 Redis 存储指标数据，
-    并导出 Prometheus 格式的指标。
+    并导出 Prometheus 格式的指标。使用异步 Redis 客户端。
     """
 
     def __init__(self, config_path: Optional[str] = None):
@@ -37,16 +39,22 @@ class MetricsManager:
         self.instance_id = settings.metrics_instance_id or socket.gethostname()
         self.config: Dict[str, Any] = {}
         self.metrics_definitions: Dict[str, Dict[str, Any]] = {}
-        self._redis_client: Optional[redis.Redis] = None
+        self._redis_client: Optional[aioredis.Redis] = None
         self._redis_available = False
+        self._initialized = False
 
-        # 加载配置
+        # 加载配置（同步操作）
         self._load_config()
-        # 初始化 Redis 连接
+        # 注册指标定义（同步操作）
-        self._init_redis()
-        # 注册指标定义
         self._register_metrics()
 
+    async def initialize(self) -> None:
+        """异步初始化 Redis 连接"""
+        if self._initialized:
+            return
+        await self._init_redis()
+        self._initialized = True
+
     def _load_config(self) -> None:
         """加载 YAML 配置文件"""
         # 尝试多个路径
@@ -138,8 +146,8 @@ class MetricsManager:
             "custom_metrics": {},
         }
 
-    def _init_redis(self) -> None:
+    async def _init_redis(self) -> None:
-        """初始化 Redis 连接"""
+        """异步初始化 Redis 连接"""
         from ..config import settings
 
         redis_config = self.config.get("redis", {})
@@ -149,7 +157,7 @@ class MetricsManager:
         password = redis_config.get("password") or settings.redis_password
 
         try:
-            self._redis_client = redis.Redis(
+            self._redis_client = aioredis.Redis(
                 host=host,
                 port=port,
                 db=db,
@@ -159,10 +167,10 @@ class MetricsManager:
                 socket_timeout=5,
             )
             # 测试连接
-            self._redis_client.ping()
+            await self._redis_client.ping()
             self._redis_available = True
             logger.info(f"Redis 连接成功: {host}:{port}/{db}")
-        except redis.ConnectionError as e:
+        except aioredis.ConnectionError as e:
             logger.warning(f"Redis 连接失败: {e}，指标将不会被收集")
             self._redis_available = False
         except Exception as e:
@@ -235,7 +243,9 @@ class MetricsManager:
 
     # === 简单 API（业务代码使用）===
 
-    def incr(self, name: str, labels: Optional[Dict[str, str]] = None, value: int = 1) -> None:
+    async def incr(
+        self, name: str, labels: Optional[Dict[str, str]] = None, value: int = 1
+    ) -> None:
         """增加计数器
 
         Args:
@@ -252,11 +262,13 @@ class MetricsManager:
         try:
             key = f"metrics:counter:{name}"
             field = self._labels_to_key(labels) or "_default_"
-            self._redis_client.hincrbyfloat(key, field, value)
+            await self._redis_client.hincrbyfloat(key, field, value)
         except Exception as e:
             logger.error(f"增加计数器失败: {e}")
 
-    def set(self, name: str, labels: Optional[Dict[str, str]] = None, value: float = 0) -> None:
+    async def set(
+        self, name: str, labels: Optional[Dict[str, str]] = None, value: float = 0
+    ) -> None:
         """设置仪表盘值
 
         Args:
@@ -273,11 +285,11 @@ class MetricsManager:
         try:
             key = f"metrics:gauge:{name}"
             field = self._labels_to_key(labels) or "_default_"
-            self._redis_client.hset(key, field, value)
+            await self._redis_client.hset(key, field, value)
         except Exception as e:
             logger.error(f"设置仪表盘失败: {e}")
 
-    def gauge_incr(
+    async def gauge_incr(
         self, name: str, labels: Optional[Dict[str, str]] = None, value: float = 1
     ) -> None:
         """增加仪表盘值
@@ -296,11 +308,11 @@ class MetricsManager:
         try:
             key = f"metrics:gauge:{name}"
             field = self._labels_to_key(labels) or "_default_"
-            self._redis_client.hincrbyfloat(key, field, value)
+            await self._redis_client.hincrbyfloat(key, field, value)
         except Exception as e:
             logger.error(f"增加仪表盘失败: {e}")
 
-    def gauge_decr(
+    async def gauge_decr(
         self, name: str, labels: Optional[Dict[str, str]] = None, value: float = 1
     ) -> None:
         """减少仪表盘值
@@ -310,9 +322,11 @@ class MetricsManager:
             labels: 标签字典
             value: 减少的值
         """
-        self.gauge_incr(name, labels, -value)
+        await self.gauge_incr(name, labels, -value)
 
-    def observe(self, name: str, labels: Optional[Dict[str, str]] = None, value: float = 0) -> None:
+    async def observe(
+        self, name: str, labels: Optional[Dict[str, str]] = None, value: float = 0
+    ) -> None:
         """记录直方图观测值
 
         Args:
@@ -348,13 +362,13 @@ class MetricsManager:
             # +Inf 桶总是增加
             pipe.hincrbyfloat(f"metrics:histogram:{name}:bucket:+Inf", label_key, 1)
 
-            pipe.execute()
+            await pipe.execute()
         except Exception as e:
             logger.error(f"记录直方图失败: {e}")
 
     # === 导出方法 ===
 
-    def export(self) -> str:
+    async def export(self) -> str:
         """导出 Prometheus 格式指标
 
         Returns:
@@ -375,11 +389,11 @@ class MetricsManager:
                 lines.append(f"# TYPE {name} {metric_type}")
 
                 if metric_type == "counter":
-                    lines.extend(self._export_counter(name))
+                    lines.extend(await self._export_counter(name))
                 elif metric_type == "gauge":
-                    lines.extend(self._export_gauge(name))
+                    lines.extend(await self._export_gauge(name))
                 elif metric_type == "histogram":
-                    lines.extend(self._export_histogram(name, definition))
+                    lines.extend(await self._export_histogram(name, definition))
 
                 lines.append("")  # 空行分隔
 
@@ -389,12 +403,12 @@ class MetricsManager:
 
         return "\n".join(lines)
 
-    def _export_counter(self, name: str) -> List[str]:
+    async def _export_counter(self, name: str) -> List[str]:
         """导出计数器指标"""
         lines = []
         key = f"metrics:counter:{name}"
 
-        data = self._redis_client.hgetall(key)
+        data = await self._redis_client.hgetall(key)
         for field, value in data.items():
             if field == "_default_":
                 lines.append(f"{name} {value}")
@@ -404,12 +418,12 @@ class MetricsManager:
 
         return lines
 
-    def _export_gauge(self, name: str) -> List[str]:
+    async def _export_gauge(self, name: str) -> List[str]:
         """导出仪表盘指标"""
         lines = []
         key = f"metrics:gauge:{name}"
 
-        data = self._redis_client.hgetall(key)
+        data = await self._redis_client.hgetall(key)
         for field, value in data.items():
             if field == "_default_":
                 lines.append(f"{name} {value}")
@@ -419,14 +433,14 @@ class MetricsManager:
 
         return lines
 
-    def _export_histogram(self, name: str, definition: Dict[str, Any]) -> List[str]:
+    async def _export_histogram(self, name: str, definition: Dict[str, Any]) -> List[str]:
         """导出直方图指标"""
         lines = []
         buckets = definition.get("buckets", [])
 
         # 获取所有标签组合
-        count_data = self._redis_client.hgetall(f"metrics:histogram:{name}:count")
+        count_data = await self._redis_client.hgetall(f"metrics:histogram:{name}:count")
-        sum_data = self._redis_client.hgetall(f"metrics:histogram:{name}:sum")
+        sum_data = await self._redis_client.hgetall(f"metrics:histogram:{name}:sum")
 
         for label_key in count_data.keys():
             prom_labels = self._key_to_prometheus_labels(label_key)
@@ -434,7 +448,7 @@ class MetricsManager:
             # 导出各个桶
             for bucket in buckets:
                 bucket_key = f"metrics:histogram:{name}:bucket:{bucket}"
-                bucket_value = self._redis_client.hget(bucket_key, label_key) or "0"
+                bucket_value = await self._redis_client.hget(bucket_key, label_key) or "0"
                 if label_key == "_default_":
                     lines.append(f'{name}_bucket{{le="{bucket}"}} {bucket_value}')
                 else:
@@ -442,7 +456,7 @@ class MetricsManager:
 
             # +Inf 桶
             inf_key = f"metrics:histogram:{name}:bucket:+Inf"
-            inf_value = self._redis_client.hget(inf_key, label_key) or "0"
+            inf_value = await self._redis_client.hget(inf_key, label_key) or "0"
             if label_key == "_default_":
                 lines.append(f'{name}_bucket{{le="+Inf"}} {inf_value}')
             else:
@@ -464,43 +478,79 @@ class MetricsManager:
         """检查 Redis 是否可用"""
         return self._redis_available
 
-    def reset(self) -> None:
+    async def reset(self) -> None:
         """重置所有指标（主要用于测试）"""
         if not self._redis_available:
             return
 
         try:
             # 删除所有指标相关的 key
-            keys = self._redis_client.keys("metrics:*")
+            keys = await self._redis_client.keys("metrics:*")
             if keys:
-                self._redis_client.delete(*keys)
+                await self._redis_client.delete(*keys)
             logger.info("已重置所有指标")
         except Exception as e:
             logger.error(f"重置指标失败: {e}")
 
+    async def close(self) -> None:
+        """关闭 Redis 连接"""
+        if self._redis_client:
+            await self._redis_client.close()
+            self._redis_client = None
+            self._redis_available = False
+            self._initialized = False
+
 
 # 全局单例
 _manager: Optional[MetricsManager] = None
+_manager_lock = asyncio.Lock()
 
 
-def get_metrics_manager() -> MetricsManager:
+async def get_metrics_manager() -> MetricsManager:
-    """获取指标管理器单例"""
+    """获取指标管理器单例（异步）"""
+    global _manager
+    if _manager is None:
+        async with _manager_lock:
+            if _manager is None:
+                _manager = MetricsManager()
+                await _manager.initialize()
+    elif not _manager._initialized:
+        await _manager.initialize()
+    return _manager
+
+
+def get_metrics_manager_sync() -> MetricsManager:
+    """获取指标管理器单例（同步，仅用于非异步上下文）
+
+    注意：此方法不会初始化 Redis 连接，需要在异步上下文中调用 initialize()
+    """
     global _manager
     if _manager is None:
         _manager = MetricsManager()
     return _manager
 
 
-def reset_metrics_manager() -> None:
+async def reset_metrics_manager() -> None:
     """重置指标管理器单例（主要用于测试）"""
     global _manager
+    if _manager is not None:
+        await _manager.close()
+    _manager = None
+
+
+def reset_metrics_manager_sync() -> None:
+    """同步重置指标管理器单例（主要用于测试）
+
+    注意：此方法不会关闭 Redis 连接，仅重置单例引用
+    """
+    global _manager
     _manager = None
 
 
 # === 便捷函数（业务代码直接调用）===
 
 
-def incr(name: str, labels: Optional[Dict[str, str]] = None, value: int = 1) -> None:
+async def incr(name: str, labels: Optional[Dict[str, str]] = None, value: int = 1) -> None:
     """增加计数器 - 便捷函数
 
     Args:
@@ -508,10 +558,11 @@ def incr(name: str, labels: Optional[Dict[str, str]] = None, value: int = 1) ->
         labels: 标签字典
         value: 增加的值，默认为 1
     """
-    get_metrics_manager().incr(name, labels, value)
+    manager = await get_metrics_manager()
+    await manager.incr(name, labels, value)
 
 
-def set(name: str, labels: Optional[Dict[str, str]] = None, value: float = 0) -> None:
+async def set(name: str, labels: Optional[Dict[str, str]] = None, value: float = 0) -> None:
     """设置仪表盘 - 便捷函数
 
     Args:
@@ -519,10 +570,13 @@ def set(name: str, labels: Optional[Dict[str, str]] = None, value: float = 0) ->
         labels: 标签字典
         value: 设置的值
     """
-    get_metrics_manager().set(name, labels, value)
+    manager = await get_metrics_manager()
+    await manager.set(name, labels, value)
 
 
-def gauge_incr(name: str, labels: Optional[Dict[str, str]] = None, value: float = 1) -> None:
+async def gauge_incr(
+    name: str, labels: Optional[Dict[str, str]] = None, value: float = 1
+) -> None:
     """增加仪表盘 - 便捷函数
 
     Args:
@@ -530,10 +584,13 @@ def gauge_incr(name: str, labels: Optional[Dict[str, str]] = None, value: float
         labels: 标签字典
         value: 增加的值
     """
-    get_metrics_manager().gauge_incr(name, labels, value)
+    manager = await get_metrics_manager()
+    await manager.gauge_incr(name, labels, value)
 
 
-def gauge_decr(name: str, labels: Optional[Dict[str, str]] = None, value: float = 1) -> None:
+async def gauge_decr(
+    name: str, labels: Optional[Dict[str, str]] = None, value: float = 1
+) -> None:
     """减少仪表盘 - 便捷函数
 
     Args:
@@ -541,10 +598,13 @@ def gauge_decr(name: str, labels: Optional[Dict[str, str]] = None, value: float
         labels: 标签字典
         value: 减少的值
     """
-    get_metrics_manager().gauge_decr(name, labels, value)
+    manager = await get_metrics_manager()
+    await manager.gauge_decr(name, labels, value)
 
 
-def observe(name: str, labels: Optional[Dict[str, str]] = None, value: float = 0) -> None:
+async def observe(
+    name: str, labels: Optional[Dict[str, str]] = None, value: float = 0
+) -> None:
     """记录直方图 - 便捷函数
 
     Args:
@@ -552,21 +612,105 @@ def observe(name: str, labels: Optional[Dict[str, str]] = None, value: float = 0
         labels: 标签字典
         value: 观测值
     """
-    get_metrics_manager().observe(name, labels, value)
+    manager = await get_metrics_manager()
+    await manager.observe(name, labels, value)
 
 
-def export() -> str:
+async def export() -> str:
     """导出指标 - 便捷函数
 
     Returns:
         Prometheus 文本格式的指标字符串
     """
-    return get_metrics_manager().export()
+    manager = await get_metrics_manager()
+    return await manager.export()
 
 
-def is_available() -> bool:
+async def is_available() -> bool:
     """检查 Redis 是否可用 - 便捷函数"""
-    return get_metrics_manager().is_available()
+    manager = await get_metrics_manager()
+    return manager.is_available()
+
+
+# === 同步便捷函数（用于同步代码中的 fire-and-forget 模式）===
+
+
+def _schedule_async(coro) -> None:
+    """在后台调度异步协程（fire-and-forget 模式）
+
+    如果当前没有运行的事件循环，则静默忽略。
+    """
+    try:
+        loop = asyncio.get_running_loop()
+        loop.create_task(coro)
+    except RuntimeError:
+        # 没有运行的事件循环，静默忽略
+        pass
+
+
+def incr_sync(
+    name: str, labels: Optional[Dict[str, str]] = None, value: int = 1
+) -> None:
+    """增加计数器 - 同步便捷函数（fire-and-forget）
+
+    Args:
+        name: 指标名称
+        labels: 标签字典
+        value: 增加的值，默认为 1
+    """
+    _schedule_async(incr(name, labels, value))
+
+
+def set_sync(
+    name: str, labels: Optional[Dict[str, str]] = None, value: float = 0
+) -> None:
+    """设置仪表盘 - 同步便捷函数（fire-and-forget）
+
+    Args:
+        name: 指标名称
+        labels: 标签字典
+        value: 设置的值
+    """
+    _schedule_async(set(name, labels, value))
+
+
+def gauge_incr_sync(
+    name: str, labels: Optional[Dict[str, str]] = None, value: float = 1
+) -> None:
+    """增加仪表盘 - 同步便捷函数（fire-and-forget）
+
+    Args:
+        name: 指标名称
+        labels: 标签字典
+        value: 增加的值
+    """
+    _schedule_async(gauge_incr(name, labels, value))
+
+
+def gauge_decr_sync(
+    name: str, labels: Optional[Dict[str, str]] = None, value: float = 1
+) -> None:
+    """减少仪表盘 - 同步便捷函数（fire-and-forget）
+
+    Args:
+        name: 指标名称
+        labels: 标签字典
+        value: 减少的值
+    """
+    _schedule_async(gauge_decr(name, labels, value))
+
+
+def observe_sync(
+    name: str, labels: Optional[Dict[str, str]] = None, value: float = 0
+) -> None:
+    """记录直方图 - 同步便捷函数（fire-and-forget）
+
+    Args:
+        name: 指标名称
+        labels: 标签字典
+        value: 观测值
+    """
+    _schedule_async(observe(name, labels, value))
 
 
 # === 装饰器（兼容旧 API）===
@@ -593,8 +737,11 @@ def track_algorithm_execution(algorithm_name: str):
                 raise e
             finally:
                 elapsed = time.time() - start_time
-                incr("algorithm_executions_total", {"algorithm": algorithm_name, "status": status})
+                incr_sync(
-                observe(
+                    "algorithm_executions_total",
+                    {"algorithm": algorithm_name, "status": status},
+                )
+                observe_sync(
                     "algorithm_execution_duration_seconds",
                     {"algorithm": algorithm_name},
                     elapsed,

src/functional_scaffold/main.py

@@ -9,6 +9,7 @@ import time
 from .api import router
 from .config import settings
 from .core.logging import setup_logging
+from .core.tracing import generate_request_id, set_request_id, get_request_id
 from .core.metrics_unified import (
     get_metrics_manager,
     incr,
@@ -20,7 +21,11 @@ from .core.metrics_unified import (
 from .core.job_manager import get_job_manager, shutdown_job_manager
 
 # 设置日志
-setup_logging(level=settings.log_level, format_type=settings.log_format)
+setup_logging(
+    level=settings.log_level,
+    format_type=settings.log_format,
+    file_path=settings.log_file_path if settings.log_file_enabled else None,
+)
 logger = logging.getLogger(__name__)
 
 # 创建 FastAPI 应用
@@ -47,12 +52,37 @@ app.add_middleware(
 @app.middleware("http")
 async def log_requests(request: Request, call_next):
     """记录所有HTTP请求"""
+    # 从请求头获取或生成 request_id
+    request_id = request.headers.get("x-request-id") or generate_request_id()
+    set_request_id(request_id)
+
     logger.info(f"Request: {request.method} {request.url.path}")
     response = await call_next(request)
     logger.info(f"Response: {response.status_code}")
     return response
 
 
+def normalize_path(path: str) -> str:
+    """
+    规范化路径，将路径参数替换为模板形式
+
+    Args:
+        path: 原始路径
+
+    Returns:
+        规范化后的路径
+
+    Examples:
+        /jobs/a1b2c3d4e5f6 -> /jobs/{job_id}
+        /invoke -> /invoke
+    """
+    # 匹配 /jobs/{任意字符串} 模式
+    if path.startswith("/jobs/") and len(path) > 6:
+        return "/jobs/{job_id}"
+
+    return path
+
+
 # 指标跟踪中间件
 @app.middleware("http")
 async def track_metrics(request: Request, call_next):
@@ -60,11 +90,12 @@ async def track_metrics(request: Request, call_next):
     if not settings.metrics_enabled:
         return await call_next(request)
 
-    # 跳过 /metrics 端点本身，避免循环记录
+    # 跳过不需要记录指标的端点
-    if request.url.path == "/metrics":
+    skip_paths = {"/metrics", "/readyz", "/healthz"}
+    if request.url.path in skip_paths:
         return await call_next(request)
 
-    gauge_incr("http_requests_in_progress")
+    await gauge_incr("http_requests_in_progress")
     start_time = time.time()
     status = "success"
 
@@ -79,16 +110,18 @@ async def track_metrics(request: Request, call_next):
         raise e
     finally:
         elapsed = time.time() - start_time
-        incr(
+        # 使用规范化后的路径记录指标
+        normalized_path = normalize_path(request.url.path)
+        await incr(
             "http_requests_total",
-            {"method": request.method, "endpoint": request.url.path, "status": status},
+            {"method": request.method, "endpoint": normalized_path, "status": status},
         )
-        observe(
+        await observe(
             "http_request_duration_seconds",
-            {"method": request.method, "endpoint": request.url.path},
+            {"method": request.method, "endpoint": normalized_path},
             elapsed,
         )
-        gauge_decr("http_requests_in_progress")
+        await gauge_decr("http_requests_in_progress")
 
 
 # 注册路由
@@ -112,7 +145,7 @@ async def metrics():
         return Response(content="Metrics disabled", status_code=404)
 
     return Response(
-        content=export(),
+        content=await export(),
         media_type="text/plain; version=0.0.4; charset=utf-8",
     )
 
@@ -127,7 +160,7 @@ async def startup_event():
 
     # 初始化指标管理器
     if settings.metrics_enabled:
-        manager = get_metrics_manager()
+        manager = await get_metrics_manager()
         if manager.is_available():
             logger.info("Redis 指标收集已启用")
         else:

src/functional_scaffold/worker.py

@@ -0,0 +1,373 @@
+"""Worker 进程模块
+
+基于 Redis 队列的任务 Worker，支持分布式锁和全局并发控制。
+"""
+
+import asyncio
+import logging
+import signal
+import sys
+from typing import Optional
+
+from aiohttp import web
+
+from .config import settings
+from .core.job_manager import JobManager
+from .core.logging import setup_logging
+from .core.tracing import set_request_id
+
+logger = logging.getLogger(__name__)
+
+
+class HealthCheckServer:
+    """轻量级健康检查 HTTP 服务器
+
+    为 Worker 模式提供健康检查端点，满足 FC 3.0 容器健康检查要求。
+    """
+
+    def __init__(self, host: str = "0.0.0.0", port: int = 8000):
+        self._host = host
+        self._port = port
+        self._app: Optional[web.Application] = None
+        self._runner: Optional[web.AppRunner] = None
+        self._site: Optional[web.TCPSite] = None
+        self._healthy = True
+
+    async def start(self) -> None:
+        """启动健康检查服务器"""
+        self._app = web.Application()
+        self._app.router.add_get("/healthz", self._healthz_handler)
+        self._app.router.add_get("/readyz", self._readyz_handler)
+
+        self._runner = web.AppRunner(self._app)
+        await self._runner.setup()
+        self._site = web.TCPSite(self._runner, self._host, self._port)
+        await self._site.start()
+        logger.info(f"健康检查服务器已启动: http://{self._host}:{self._port}")
+
+    async def stop(self) -> None:
+        """停止健康检查服务器"""
+        if self._runner:
+            await self._runner.cleanup()
+            logger.info("健康检查服务器已停止")
+
+    def set_healthy(self, healthy: bool) -> None:
+        """设置健康状态"""
+        self._healthy = healthy
+
+    async def _healthz_handler(self, request: web.Request) -> web.Response:
+        """存活检查端点"""
+        return web.json_response({"status": "healthy", "mode": "worker"})
+
+    async def _readyz_handler(self, request: web.Request) -> web.Response:
+        """就绪检查端点"""
+        if self._healthy:
+            return web.json_response({"status": "ready", "mode": "worker"})
+        return web.json_response({"status": "not ready"}, status=503)
+
+
+class JobWorker:
+    """任务 Worker
+
+    从 Redis 队列获取任务并执行，支持：
+    - 分布式锁防止重复执行
+    - 全局并发控制
+    - 任务重试机制
+    - 锁续租机制
+    - 超时任务回收
+    - 优雅关闭
+    """
+
+    def __init__(self):
+        self._job_manager: Optional[JobManager] = None
+        self._running: bool = False
+        self._current_job_id: Optional[str] = None
+        self._current_lock_token: Optional[str] = None
+        self._lock_renewal_task: Optional[asyncio.Task] = None
+        self._sweeper_task: Optional[asyncio.Task] = None
+
+    async def initialize(self) -> None:
+        """初始化 Worker"""
+        self._job_manager = JobManager()
+        await self._job_manager.initialize()
+        logger.info("Worker 初始化完成")
+
+    async def shutdown(self) -> None:
+        """关闭 Worker"""
+        logger.info("Worker 正在关闭...")
+        self._running = False
+
+        # 取消回收器任务
+        if self._sweeper_task and not self._sweeper_task.done():
+            self._sweeper_task.cancel()
+            try:
+                await self._sweeper_task
+            except asyncio.CancelledError:
+                pass
+
+        # 取消锁续租任务
+        if self._lock_renewal_task and not self._lock_renewal_task.done():
+            self._lock_renewal_task.cancel()
+            try:
+                await self._lock_renewal_task
+            except asyncio.CancelledError:
+                pass
+
+        # 等待当前任务完成
+        if self._current_job_id:
+            logger.info(f"等待当前任务完成: {self._current_job_id}")
+
+        if self._job_manager:
+            await self._job_manager.shutdown()
+
+        logger.info("Worker 已关闭")
+
+    async def run(self) -> None:
+        """运行 Worker 主循环"""
+        self._running = True
+        logger.info(
+            f"Worker 启动，轮询间隔: {settings.worker_poll_interval}s，"
+            f"最大并发: {settings.max_concurrent_jobs}"
+        )
+
+        # 启动超时任务回收器
+        if settings.job_sweeper_enabled:
+            self._sweeper_task = asyncio.create_task(self._sweeper_loop())
+            logger.info(f"超时任务回收器已启动，扫描间隔: {settings.job_sweeper_interval}s")
+
+        while self._running:
+            try:
+                await self._process_next_job()
+            except Exception as e:
+                logger.error(f"Worker 循环异常: {e}", exc_info=True)
+                await asyncio.sleep(settings.worker_poll_interval)
+
+    async def _process_next_job(self) -> None:
+        """处理下一个任务"""
+        if not self._job_manager:
+            logger.error("JobManager 未初始化")
+            await asyncio.sleep(settings.worker_poll_interval)
+            return
+
+        # 从队列获取任务（转移式出队）
+        job_id = await self._job_manager.dequeue_job(timeout=int(settings.worker_poll_interval))
+
+        if not job_id:
+            return
+
+        # 获取任务信息以提取 request_id
+        job_data = await self._job_manager.get_job(job_id)
+        if job_data:
+            request_id = job_data.get("request_id") or job_id
+            set_request_id(request_id)
+        else:
+            set_request_id(job_id)
+
+        logger.info(f"从队列获取任务: {job_id}")
+
+        # 尝试获取分布式锁（返回 token）
+        lock_token = await self._job_manager.acquire_job_lock(job_id)
+        if not lock_token:
+            logger.warning(f"无法获取任务锁，任务可能正在被其他 Worker 执行: {job_id}")
+            # 任务留在 processing 队列，等待回收器处理
+            return
+
+        self._current_lock_token = lock_token
+
+        # 启动锁续租协程
+        self._lock_renewal_task = asyncio.create_task(self._lock_renewal_loop(job_id, lock_token))
+
+        try:
+            # 检查全局并发限制
+            if not await self._job_manager.can_execute():
+                logger.info(f"达到并发限制，任务 NACK 重新入队: {job_id}")
+                await self._job_manager.nack_job(job_id, requeue=True)
+                return
+
+            # 增加并发计数
+            await self._job_manager.increment_concurrency()
+            self._current_job_id = job_id
+
+            try:
+                # 执行任务
+                success = await self._execute_with_retry(job_id)
+                if success:
+                    await self._job_manager.ack_job(job_id)
+                else:
+                    await self._job_manager.increment_job_retry(job_id)
+                    await self._job_manager.nack_job(job_id, requeue=True)
+            finally:
+                # 减少并发计数
+                await self._job_manager.decrement_concurrency()
+                self._current_job_id = None
+
+        finally:
+            # 停止锁续租
+            if self._lock_renewal_task and not self._lock_renewal_task.done():
+                self._lock_renewal_task.cancel()
+                try:
+                    await self._lock_renewal_task
+                except asyncio.CancelledError:
+                    pass
+            self._lock_renewal_task = None
+
+            # 释放分布式锁
+            await self._job_manager.release_job_lock(job_id, lock_token)
+            self._current_lock_token = None
+
+    async def _execute_with_retry(self, job_id: str) -> bool:
+        """执行任务（带重试机制）
+
+        Returns:
+            bool: 任务是否成功执行
+        """
+        if not self._job_manager:
+            return False
+
+        try:
+            # 执行任务
+            await asyncio.wait_for(
+                self._job_manager.execute_job(job_id),
+                timeout=settings.job_execution_timeout,
+            )
+            return True
+        except asyncio.TimeoutError:
+            logger.error(f"任务执行超时: {job_id}")
+            await self._handle_job_failure(job_id, "任务执行超时")
+            return False
+        except Exception as e:
+            logger.error(f"任务执行异常: {job_id}, error={e}", exc_info=True)
+            await self._handle_job_failure(job_id, str(e))
+            return False
+
+    async def _handle_job_failure(self, job_id: str, error: str) -> None:
+        """处理任务失败"""
+        if not self._job_manager:
+            return
+
+        retry_count = await self._job_manager.increment_job_retry(job_id)
+
+        if retry_count < settings.job_max_retries:
+            logger.info(f"任务将重试 ({retry_count}/{settings.job_max_retries}): {job_id}")
+            # 重新入队
+            await self._job_manager.enqueue_job(job_id)
+        else:
+            logger.error(f"任务达到最大重试次数，标记为失败: {job_id}")
+            # 更新任务状态为失败
+            if self._job_manager._redis_client:
+                key = f"job:{job_id}"
+                await self._job_manager._redis_client.hset(
+                    key,
+                    mapping={
+                        "status": "failed",
+                        "error": f"达到最大重试次数 ({settings.job_max_retries}): {error}",
+                    },
+                )
+
+    async def _lock_renewal_loop(self, job_id: str, lock_token: str) -> None:
+        """锁续租协程
+
+        定期续租任务锁，防止长任务执行时锁过期。
+
+        Args:
+            job_id: 任务 ID
+            lock_token: 锁 token
+        """
+        # 续租间隔为锁 TTL 的一半
+        interval = (settings.job_execution_timeout + settings.job_lock_buffer) / 2
+        while True:
+            try:
+                await asyncio.sleep(interval)
+                if not self._job_manager:
+                    break
+                if not await self._job_manager.renew_job_lock(job_id, lock_token):
+                    logger.error(f"锁续租失败，可能已被其他进程获取: {job_id}")
+                    break
+                logger.debug(f"锁续租成功: {job_id}")
+            except asyncio.CancelledError:
+                logger.debug(f"锁续租协程已取消: {job_id}")
+                break
+            except Exception as e:
+                logger.error(f"锁续租异常: {job_id}, error={e}")
+                break
+
+    async def _sweeper_loop(self) -> None:
+        """超时任务回收协程
+
+        定期扫描处理中队列，回收超时任务，并收集队列监控指标。
+        """
+        while self._running:
+            try:
+                await asyncio.sleep(settings.job_sweeper_interval)
+                if not self._job_manager:
+                    continue
+
+                # 回收超时任务
+                recovered = await self._job_manager.recover_stale_jobs()
+                if recovered > 0:
+                    logger.info(f"回收超时任务: {recovered} 个")
+                    # 记录回收指标
+                    from .core.metrics_unified import incr
+
+                    await incr("job_recovered_total", None, recovered)
+
+                # 收集队列监控指标
+                await self._job_manager.collect_queue_metrics()
+
+            except asyncio.CancelledError:
+                logger.debug("超时任务回收协程已取消")
+                break
+            except Exception as e:
+                logger.error(f"超时任务回收异常: {e}")
+
+
+def setup_signal_handlers(
+    worker: JobWorker,
+    health_server: HealthCheckServer,
+    loop: asyncio.AbstractEventLoop,
+) -> None:
+    """设置信号处理器"""
+
+    async def shutdown_all() -> None:
+        """关闭所有服务"""
+        await worker.shutdown()
+        await health_server.stop()
+
+    def signal_handler(sig: signal.Signals) -> None:
+        logger.info(f"收到信号 {sig.name}，准备关闭...")
+        loop.create_task(shutdown_all())
+
+    for sig in (signal.SIGTERM, signal.SIGINT):
+        loop.add_signal_handler(sig, signal_handler, sig)
+
+
+async def main() -> None:
+    """Worker 入口函数"""
+    # 设置日志
+    setup_logging(level=settings.log_level, format_type=settings.log_format)
+
+    # 创建健康检查服务器和 Worker
+    health_server = HealthCheckServer(port=8000)
+    worker = JobWorker()
+
+    # 设置信号处理
+    loop = asyncio.get_running_loop()
+    setup_signal_handlers(worker, health_server, loop)
+
+    try:
+        # 先启动健康检查服务器，确保 FC 健康检查能通过
+        await health_server.start()
+
+        # 初始化并运行 Worker
+        await worker.initialize()
+        await worker.run()
+    except Exception as e:
+        logger.error(f"Worker 异常退出: {e}", exc_info=True)
+        sys.exit(1)
+    finally:
+        await worker.shutdown()
+        await health_server.stop()
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

tests/conftest.py

@@ -2,7 +2,7 @@
 
 import pytest
 from fastapi.testclient import TestClient
-from src.functional_scaffold.main import app
+from functional_scaffold.main import app
 
 
 @pytest.fixture

tests/test_algorithms.py

@@ -1,7 +1,7 @@
 """算法单元测试"""
 
 import pytest
-from src.functional_scaffold.algorithms.prime_checker import PrimeChecker
+from functional_scaffold.algorithms.prime_checker import PrimeChecker
 
 
 class TestPrimeChecker:

tests/test_job_manager.py

@@ -1,17 +1,13 @@
 """异步任务管理器测试"""
 
 import asyncio
-import json
 import pytest
 from unittest.mock import AsyncMock, MagicMock, patch
 from fastapi import status
 
-from src.functional_scaffold.core.job_manager import (
+from functional_scaffold.core.job_manager import (
     JobManager,
-    get_job_manager,
-    shutdown_job_manager,
 )
-from src.functional_scaffold.api.models import JobStatus
 
 
 class TestJobManager:
@@ -186,6 +182,11 @@ class TestJobManagerWithMocks:
         manager._redis_client = mock_redis
         manager._register_algorithms()
 
+        # 初始化 semaphore
+        import asyncio
+
+        manager._semaphore = asyncio.Semaphore(10)
+
         await manager.execute_job("test-job-id")
 
         # 验证状态更新被调用
@@ -199,7 +200,7 @@ class TestJobsAPI:
     def test_create_job_success(self, client):
         """测试成功创建任务"""
         with patch(
-            "src.functional_scaffold.api.routes.get_job_manager", new_callable=AsyncMock
+            "functional_scaffold.api.routes.get_job_manager", new_callable=AsyncMock
         ) as mock_get_manager:
             mock_manager = MagicMock()
             mock_manager.is_available.return_value = True
@@ -213,7 +214,7 @@ class TestJobsAPI:
                     "created_at": "2026-02-02T10:00:00+00:00",
                 }
             )
-            mock_manager.execute_job = AsyncMock()
+            mock_manager.enqueue_job = AsyncMock(return_value=True)
             mock_get_manager.return_value = mock_manager
 
             response = client.post(
@@ -233,7 +234,7 @@ class TestJobsAPI:
     def test_create_job_algorithm_not_found(self, client):
         """测试创建任务时算法不存在"""
         with patch(
-            "src.functional_scaffold.api.routes.get_job_manager", new_callable=AsyncMock
+            "functional_scaffold.api.routes.get_job_manager", new_callable=AsyncMock
         ) as mock_get_manager:
             mock_manager = MagicMock()
             mock_manager.is_available.return_value = True
@@ -255,7 +256,7 @@ class TestJobsAPI:
     def test_create_job_service_unavailable(self, client):
         """测试服务不可用时创建任务"""
         with patch(
-            "src.functional_scaffold.api.routes.get_job_manager", new_callable=AsyncMock
+            "functional_scaffold.api.routes.get_job_manager", new_callable=AsyncMock
         ) as mock_get_manager:
             mock_manager = MagicMock()
             mock_manager.is_available.return_value = False
@@ -274,7 +275,7 @@ class TestJobsAPI:
     def test_get_job_status_success(self, client):
         """测试成功查询任务状态"""
         with patch(
-            "src.functional_scaffold.api.routes.get_job_manager", new_callable=AsyncMock
+            "functional_scaffold.api.routes.get_job_manager", new_callable=AsyncMock
         ) as mock_get_manager:
             mock_manager = MagicMock()
             mock_manager.is_available.return_value = True
@@ -304,7 +305,7 @@ class TestJobsAPI:
     def test_get_job_status_not_found(self, client):
         """测试查询不存在的任务"""
         with patch(
-            "src.functional_scaffold.api.routes.get_job_manager", new_callable=AsyncMock
+            "functional_scaffold.api.routes.get_job_manager", new_callable=AsyncMock
         ) as mock_get_manager:
             mock_manager = MagicMock()
             mock_manager.is_available.return_value = True
@@ -320,7 +321,7 @@ class TestJobsAPI:
     def test_get_job_status_service_unavailable(self, client):
         """测试服务不可用时查询任务"""
         with patch(
-            "src.functional_scaffold.api.routes.get_job_manager", new_callable=AsyncMock
+            "functional_scaffold.api.routes.get_job_manager", new_callable=AsyncMock
         ) as mock_get_manager:
             mock_manager = MagicMock()
             mock_manager.is_available.return_value = False
@@ -391,7 +392,7 @@ class TestWebhook:
         manager._http_client = mock_http
 
         # 使用较短的重试间隔进行测试
-        with patch("src.functional_scaffold.core.job_manager.settings") as mock_settings:
+        with patch("functional_scaffold.core.job_manager.settings") as mock_settings:
             mock_settings.webhook_max_retries = 2
             mock_settings.webhook_timeout = 1
 
@@ -399,3 +400,771 @@ class TestWebhook:
 
         # 验证重试次数
         assert mock_http.post.call_count == 2
+
+
+class TestConcurrencyControl:
+    """测试并发控制功能"""
+
+    @pytest.mark.asyncio
+    async def test_get_concurrency_status(self):
+        """测试获取并发状态"""
+        manager = JobManager()
+
+        # 初始化 semaphore
+        manager._max_concurrent_jobs = 10
+        manager._semaphore = asyncio.Semaphore(10)
+
+        status = manager.get_concurrency_status()
+
+        assert status["max_concurrent"] == 10
+        assert status["available_slots"] == 10
+        assert status["running_jobs"] == 0
+
+    @pytest.mark.asyncio
+    async def test_get_concurrency_status_without_semaphore(self):
+        """测试未初始化 semaphore 时获取并发状态"""
+        manager = JobManager()
+
+        status = manager.get_concurrency_status()
+
+        assert status["max_concurrent"] == 0
+        assert status["available_slots"] == 0
+        assert status["running_jobs"] == 0
+
+    @pytest.mark.asyncio
+    async def test_concurrency_limit(self):
+        """测试并发限制是否生效"""
+        manager = JobManager()
+
+        # 设置较小的并发限制
+        manager._max_concurrent_jobs = 2
+        manager._semaphore = asyncio.Semaphore(2)
+
+        # 模拟 Redis
+        mock_redis = AsyncMock()
+        mock_redis.hgetall = AsyncMock(
+            return_value={
+                "status": "pending",
+                "algorithm": "PrimeChecker",
+                "params": '{"number": 17}',
+                "webhook": "",
+                "request_id": "test-request-id",
+                "created_at": "2026-02-02T10:00:00+00:00",
+            }
+        )
+        mock_redis.hset = AsyncMock()
+        mock_redis.expire = AsyncMock()
+        manager._redis_client = mock_redis
+        manager._register_algorithms()
+
+        # 创建一个慢速任务
+        async def slow_execute():
+            async with manager._semaphore:
+                await asyncio.sleep(0.1)
+
+        # 启动 3 个任务
+        tasks = [asyncio.create_task(slow_execute()) for _ in range(3)]
+
+        # 等待一小段时间，让前两个任务获取 semaphore
+        await asyncio.sleep(0.01)
+
+        # 检查并发状态
+        status = manager.get_concurrency_status()
+        assert status["running_jobs"] == 2  # 只有 2 个任务在运行
+        assert status["available_slots"] == 0  # 没有可用槽位
+
+        # 等待所有任务完成
+        await asyncio.gather(*tasks)
+
+        # 检查最终状态
+        status = manager.get_concurrency_status()
+        assert status["running_jobs"] == 0
+        assert status["available_slots"] == 2
+
+    def test_concurrency_status_api(self, client):
+        """测试并发状态 API 端点"""
+        with patch(
+            "functional_scaffold.api.routes.get_job_manager", new_callable=AsyncMock
+        ) as mock_get_manager:
+            mock_manager = MagicMock()
+            mock_manager.is_available.return_value = True
+            mock_manager.get_concurrency_status.return_value = {
+                "max_concurrent": 10,
+                "available_slots": 8,
+                "running_jobs": 2,
+            }
+            mock_get_manager.return_value = mock_manager
+
+            response = client.get("/jobs/concurrency/status")
+
+            assert response.status_code == status.HTTP_200_OK
+            data = response.json()
+
+            assert "max_concurrent" in data
+            assert "available_slots" in data
+            assert "running_jobs" in data
+            assert isinstance(data["max_concurrent"], int)
+            assert isinstance(data["available_slots"], int)
+            assert isinstance(data["running_jobs"], int)
+
+
+class TestJobQueue:
+    """测试任务队列功能"""
+
+    @pytest.mark.asyncio
+    async def test_enqueue_job(self):
+        """测试任务入队"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.lpush = AsyncMock(return_value=1)
+        manager._redis_client = mock_redis
+
+        result = await manager.enqueue_job("test-job-id")
+
+        assert result is True
+        mock_redis.lpush.assert_called_once()
+
+    @pytest.mark.asyncio
+    async def test_enqueue_job_without_redis(self):
+        """测试 Redis 不可用时入队"""
+        manager = JobManager()
+
+        result = await manager.enqueue_job("test-job-id")
+
+        assert result is False
+
+    @pytest.mark.asyncio
+    async def test_dequeue_job(self):
+        """测试任务出队（使用 BLMOVE）"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.blmove = AsyncMock(return_value="test-job-id")
+        mock_redis.zadd = AsyncMock()
+        manager._redis_client = mock_redis
+
+        result = await manager.dequeue_job(timeout=5)
+
+        assert result == "test-job-id"
+        mock_redis.blmove.assert_called_once()
+        mock_redis.zadd.assert_called_once()
+
+    @pytest.mark.asyncio
+    async def test_dequeue_job_timeout(self):
+        """测试任务出队超时"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.blmove = AsyncMock(return_value=None)
+        manager._redis_client = mock_redis
+
+        result = await manager.dequeue_job(timeout=1)
+
+        assert result is None
+
+    @pytest.mark.asyncio
+    async def test_dequeue_job_without_redis(self):
+        """测试 Redis 不可用时出队"""
+        manager = JobManager()
+
+        result = await manager.dequeue_job(timeout=1)
+
+        assert result is None
+
+
+class TestDistributedLock:
+    """测试分布式锁功能"""
+
+    @pytest.mark.asyncio
+    async def test_acquire_job_lock(self):
+        """测试获取任务锁"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.set = AsyncMock(return_value=True)
+        manager._redis_client = mock_redis
+
+        result = await manager.acquire_job_lock("test-job-id")
+
+        assert result is not None  # 返回 token
+        assert len(result) == 32  # 16 字节的十六进制字符串
+        mock_redis.set.assert_called_once()
+        call_args = mock_redis.set.call_args
+        assert call_args[0][0] == "job:lock:test-job-id"
+        assert call_args[1]["nx"] is True
+        assert "ex" in call_args[1]
+
+    @pytest.mark.asyncio
+    async def test_acquire_job_lock_already_locked(self):
+        """测试获取已被锁定的任务锁"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.set = AsyncMock(return_value=None)  # 锁已存在
+        manager._redis_client = mock_redis
+
+        result = await manager.acquire_job_lock("test-job-id")
+
+        assert result is None
+
+    @pytest.mark.asyncio
+    async def test_release_job_lock(self):
+        """测试释放任务锁"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.eval = AsyncMock(return_value=1)
+        manager._redis_client = mock_redis
+
+        result = await manager.release_job_lock("test-job-id", "valid-token")
+
+        assert result is True
+        mock_redis.eval.assert_called_once()
+
+    @pytest.mark.asyncio
+    async def test_release_job_lock_without_redis(self):
+        """测试 Redis 不可用时释放锁"""
+        manager = JobManager()
+
+        result = await manager.release_job_lock("test-job-id", "token")
+
+        assert result is False
+
+
+class TestGlobalConcurrency:
+    """测试全局并发控制功能"""
+
+    @pytest.mark.asyncio
+    async def test_increment_concurrency(self):
+        """测试增加并发计数"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.incr = AsyncMock(return_value=5)
+        manager._redis_client = mock_redis
+
+        result = await manager.increment_concurrency()
+
+        assert result == 5
+        mock_redis.incr.assert_called_once()
+
+    @pytest.mark.asyncio
+    async def test_decrement_concurrency(self):
+        """测试减少并发计数"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.decr = AsyncMock(return_value=4)
+        manager._redis_client = mock_redis
+
+        result = await manager.decrement_concurrency()
+
+        assert result == 4
+        mock_redis.decr.assert_called_once()
+
+    @pytest.mark.asyncio
+    async def test_decrement_concurrency_prevent_negative(self):
+        """测试防止并发计数变为负数"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.decr = AsyncMock(return_value=-1)
+        mock_redis.set = AsyncMock()
+        manager._redis_client = mock_redis
+
+        result = await manager.decrement_concurrency()
+
+        assert result == 0
+        mock_redis.set.assert_called_once()
+
+    @pytest.mark.asyncio
+    async def test_get_global_concurrency(self):
+        """测试获取全局并发数"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.get = AsyncMock(return_value="7")
+        manager._redis_client = mock_redis
+
+        result = await manager.get_global_concurrency()
+
+        assert result == 7
+
+    @pytest.mark.asyncio
+    async def test_get_global_concurrency_empty(self):
+        """测试获取空的全局并发数"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.get = AsyncMock(return_value=None)
+        manager._redis_client = mock_redis
+
+        result = await manager.get_global_concurrency()
+
+        assert result == 0
+
+    @pytest.mark.asyncio
+    async def test_can_execute(self):
+        """测试检查是否可执行"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.get = AsyncMock(return_value="5")
+        manager._redis_client = mock_redis
+
+        with patch("functional_scaffold.core.job_manager.settings") as mock_settings:
+            mock_settings.max_concurrent_jobs = 10
+
+            result = await manager.can_execute()
+
+            assert result is True
+
+    @pytest.mark.asyncio
+    async def test_can_execute_at_limit(self):
+        """测试达到并发限制时"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.get = AsyncMock(return_value="10")
+        manager._redis_client = mock_redis
+
+        with patch("functional_scaffold.core.job_manager.settings") as mock_settings:
+            mock_settings.max_concurrent_jobs = 10
+
+            result = await manager.can_execute()
+
+            assert result is False
+
+
+class TestJobRetry:
+    """测试任务重试功能"""
+
+    @pytest.mark.asyncio
+    async def test_get_job_retry_count(self):
+        """测试获取任务重试次数"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.hget = AsyncMock(return_value="2")
+        manager._redis_client = mock_redis
+
+        result = await manager.get_job_retry_count("test-job-id")
+
+        assert result == 2
+        mock_redis.hget.assert_called_once_with("job:test-job-id", "retry_count")
+
+    @pytest.mark.asyncio
+    async def test_get_job_retry_count_empty(self):
+        """测试获取空的重试次数"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.hget = AsyncMock(return_value=None)
+        manager._redis_client = mock_redis
+
+        result = await manager.get_job_retry_count("test-job-id")
+
+        assert result == 0
+
+    @pytest.mark.asyncio
+    async def test_increment_job_retry(self):
+        """测试增加任务重试次数"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.hincrby = AsyncMock()
+        mock_redis.hget = AsyncMock(return_value="3")
+        manager._redis_client = mock_redis
+
+        result = await manager.increment_job_retry("test-job-id")
+
+        assert result == 3
+        mock_redis.hincrby.assert_called_once_with("job:test-job-id", "retry_count", 1)
+
+
+class TestTransferDequeue:
+    """测试转移式出队功能"""
+
+    @pytest.mark.asyncio
+    async def test_dequeue_job_with_blmove(self):
+        """测试使用 BLMOVE 转移式出队"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.blmove = AsyncMock(return_value="test-job-id")
+        mock_redis.zadd = AsyncMock()
+        manager._redis_client = mock_redis
+
+        result = await manager.dequeue_job(timeout=5)
+
+        assert result == "test-job-id"
+        mock_redis.blmove.assert_called_once()
+        mock_redis.zadd.assert_called_once()
+
+    @pytest.mark.asyncio
+    async def test_dequeue_job_timeout(self):
+        """测试出队超时"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.blmove = AsyncMock(return_value=None)
+        manager._redis_client = mock_redis
+
+        result = await manager.dequeue_job(timeout=1)
+
+        assert result is None
+        mock_redis.zadd.assert_not_called()
+
+
+class TestTokenBasedLock:
+    """测试带 Token 的安全锁"""
+
+    @pytest.mark.asyncio
+    async def test_acquire_job_lock_returns_token(self):
+        """测试获取锁返回 token"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.set = AsyncMock(return_value=True)
+        manager._redis_client = mock_redis
+
+        result = await manager.acquire_job_lock("test-job-id")
+
+        assert result is not None
+        assert len(result) == 32  # 16 字节的十六进制字符串
+        mock_redis.set.assert_called_once()
+        call_args = mock_redis.set.call_args
+        assert call_args[0][0] == "job:lock:test-job-id"
+        assert call_args[1]["nx"] is True
+
+    @pytest.mark.asyncio
+    async def test_acquire_job_lock_already_locked(self):
+        """测试获取已被锁定的任务锁"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.set = AsyncMock(return_value=None)
+        manager._redis_client = mock_redis
+
+        result = await manager.acquire_job_lock("test-job-id")
+
+        assert result is None
+
+    @pytest.mark.asyncio
+    async def test_release_job_lock_with_token(self):
+        """测试使用 token 释放锁"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.eval = AsyncMock(return_value=1)
+        manager._redis_client = mock_redis
+
+        result = await manager.release_job_lock("test-job-id", "valid-token")
+
+        assert result is True
+        mock_redis.eval.assert_called_once()
+
+    @pytest.mark.asyncio
+    async def test_release_job_lock_invalid_token(self):
+        """测试使用无效 token 释放锁"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.eval = AsyncMock(return_value=0)
+        manager._redis_client = mock_redis
+
+        result = await manager.release_job_lock("test-job-id", "invalid-token")
+
+        assert result is False
+
+    @pytest.mark.asyncio
+    async def test_release_job_lock_without_token(self):
+        """测试不使用 token 释放锁（向后兼容）"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.delete = AsyncMock()
+        manager._redis_client = mock_redis
+
+        result = await manager.release_job_lock("test-job-id")
+
+        assert result is True
+        mock_redis.delete.assert_called_once_with("job:lock:test-job-id")
+
+
+class TestAckNack:
+    """测试 ACK/NACK 机制"""
+
+    @pytest.mark.asyncio
+    async def test_ack_job(self):
+        """测试确认任务完成"""
+        manager = JobManager()
+
+        mock_pipe = AsyncMock()
+        mock_pipe.lrem = MagicMock()
+        mock_pipe.zrem = MagicMock()
+        mock_pipe.execute = AsyncMock()
+        mock_pipe.__aenter__ = AsyncMock(return_value=mock_pipe)
+        mock_pipe.__aexit__ = AsyncMock()
+
+        mock_redis = AsyncMock()
+        mock_redis.pipeline = MagicMock(return_value=mock_pipe)
+        manager._redis_client = mock_redis
+
+        result = await manager.ack_job("test-job-id")
+
+        assert result is True
+        mock_pipe.lrem.assert_called_once()
+        mock_pipe.zrem.assert_called_once()
+
+    @pytest.mark.asyncio
+    async def test_nack_job_requeue(self):
+        """测试拒绝任务并重新入队"""
+        manager = JobManager()
+
+        mock_pipe = AsyncMock()
+        mock_pipe.lrem = MagicMock()
+        mock_pipe.zrem = MagicMock()
+        mock_pipe.lpush = MagicMock()
+        mock_pipe.execute = AsyncMock()
+        mock_pipe.__aenter__ = AsyncMock(return_value=mock_pipe)
+        mock_pipe.__aexit__ = AsyncMock()
+
+        mock_redis = AsyncMock()
+        mock_redis.pipeline = MagicMock(return_value=mock_pipe)
+        mock_redis.hget = AsyncMock(return_value="0")  # retry_count = 0
+        manager._redis_client = mock_redis
+
+        result = await manager.nack_job("test-job-id", requeue=True)
+
+        assert result is True
+        assert mock_pipe.lpush.call_count == 1
+
+    @pytest.mark.asyncio
+    async def test_nack_job_to_dlq(self):
+        """测试拒绝任务进入死信队列"""
+        manager = JobManager()
+
+        mock_pipe = AsyncMock()
+        mock_pipe.lrem = MagicMock()
+        mock_pipe.zrem = MagicMock()
+        mock_pipe.lpush = MagicMock()
+        mock_pipe.execute = AsyncMock()
+        mock_pipe.__aenter__ = AsyncMock(return_value=mock_pipe)
+        mock_pipe.__aexit__ = AsyncMock()
+
+        mock_redis = AsyncMock()
+        mock_redis.pipeline = MagicMock(return_value=mock_pipe)
+        mock_redis.hget = AsyncMock(return_value="5")  # retry_count > max_retries
+        manager._redis_client = mock_redis
+
+        with patch("functional_scaffold.core.job_manager.settings") as mock_settings:
+            mock_settings.job_max_retries = 3
+            mock_settings.job_processing_key = "job:processing"
+            mock_settings.job_processing_ts_key = "job:processing:ts"
+            mock_settings.job_dlq_key = "job:dlq"
+            mock_settings.job_queue_key = "job:queue"
+
+            result = await manager.nack_job("test-job-id", requeue=True)
+
+        assert result is True
+
+
+class TestLockRenewal:
+    """测试锁续租功能"""
+
+    @pytest.mark.asyncio
+    async def test_renew_job_lock_success(self):
+        """测试锁续租成功"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.eval = AsyncMock(return_value=1)
+        manager._redis_client = mock_redis
+
+        result = await manager.renew_job_lock("test-job-id", "valid-token")
+
+        assert result is True
+        mock_redis.eval.assert_called_once()
+
+    @pytest.mark.asyncio
+    async def test_renew_job_lock_invalid_token(self):
+        """测试锁续租失败（token 不匹配）"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.eval = AsyncMock(return_value=0)
+        manager._redis_client = mock_redis
+
+        result = await manager.renew_job_lock("test-job-id", "invalid-token")
+
+        assert result is False
+
+    @pytest.mark.asyncio
+    async def test_renew_job_lock_without_redis(self):
+        """测试 Redis 不可用时续租"""
+        manager = JobManager()
+
+        result = await manager.renew_job_lock("test-job-id", "token")
+
+        assert result is False
+
+
+class TestStaleJobRecovery:
+    """测试超时任务回收功能"""
+
+    @pytest.mark.asyncio
+    async def test_recover_stale_jobs_empty(self):
+        """测试没有超时任务时的回收"""
+        manager = JobManager()
+
+        mock_redis = AsyncMock()
+        mock_redis.zrangebyscore = AsyncMock(return_value=[])
+        manager._redis_client = mock_redis
+
+        result = await manager.recover_stale_jobs()
+
+        assert result == 0
+
+    @pytest.mark.asyncio
+    async def test_recover_stale_jobs_requeue(self):
+        """测试回收超时任务并重新入队"""
+        manager = JobManager()
+
+        mock_pipe = AsyncMock()
+        mock_pipe.lrem = MagicMock()
+        mock_pipe.zrem = MagicMock()
+        mock_pipe.lpush = MagicMock()
+        mock_pipe.execute = AsyncMock()
+        mock_pipe.__aenter__ = AsyncMock(return_value=mock_pipe)
+        mock_pipe.__aexit__ = AsyncMock()
+
+        mock_redis = AsyncMock()
+        mock_redis.zrangebyscore = AsyncMock(return_value=["stale-job-1", "stale-job-2"])
+        mock_redis.hincrby = AsyncMock()
+        mock_redis.hget = AsyncMock(return_value="1")  # retry_count = 1
+        mock_redis.pipeline = MagicMock(return_value=mock_pipe)
+        manager._redis_client = mock_redis
+
+        with patch("functional_scaffold.core.job_manager.settings") as mock_settings:
+            mock_settings.job_execution_timeout = 300
+            mock_settings.job_lock_buffer = 60
+            mock_settings.job_max_retries = 3
+            mock_settings.job_processing_key = "job:processing"
+            mock_settings.job_processing_ts_key = "job:processing:ts"
+            mock_settings.job_dlq_key = "job:dlq"
+            mock_settings.job_queue_key = "job:queue"
+
+            result = await manager.recover_stale_jobs()
+
+        assert result == 2
+
+    @pytest.mark.asyncio
+    async def test_recover_stale_jobs_to_dlq(self):
+        """测试回收超时任务进入死信队列"""
+        manager = JobManager()
+
+        mock_pipe = AsyncMock()
+        mock_pipe.lrem = MagicMock()
+        mock_pipe.zrem = MagicMock()
+        mock_pipe.lpush = MagicMock()
+        mock_pipe.execute = AsyncMock()
+        mock_pipe.__aenter__ = AsyncMock(return_value=mock_pipe)
+        mock_pipe.__aexit__ = AsyncMock()
+
+        mock_redis = AsyncMock()
+        mock_redis.zrangebyscore = AsyncMock(return_value=["stale-job-1"])
+        mock_redis.hincrby = AsyncMock()
+        mock_redis.hget = AsyncMock(return_value="5")  # retry_count > max_retries
+        mock_redis.pipeline = MagicMock(return_value=mock_pipe)
+        manager._redis_client = mock_redis
+
+        with patch("functional_scaffold.core.job_manager.settings") as mock_settings:
+            mock_settings.job_execution_timeout = 300
+            mock_settings.job_lock_buffer = 60
+            mock_settings.job_max_retries = 3
+            mock_settings.job_processing_key = "job:processing"
+            mock_settings.job_processing_ts_key = "job:processing:ts"
+            mock_settings.job_dlq_key = "job:dlq"
+            mock_settings.job_queue_key = "job:queue"
+
+            result = await manager.recover_stale_jobs()
+
+        assert result == 1
+
+    @pytest.mark.asyncio
+    async def test_recover_stale_jobs_without_redis(self):
+        """测试 Redis 不可用时回收"""
+        manager = JobManager()
+
+        result = await manager.recover_stale_jobs()
+
+        assert result == 0
+
+
+class TestQueueMetrics:
+    """测试队列监控指标收集"""
+
+    @pytest.mark.asyncio
+    async def test_collect_queue_metrics(self):
+        """测试收集队列指标"""
+        manager = JobManager()
+
+        mock_pipe = AsyncMock()
+        mock_pipe.llen = MagicMock()
+        mock_pipe.zrange = MagicMock()
+        mock_pipe.execute = AsyncMock(return_value=[5, 2, 1, [("job-1", 1000.0)]])
+        mock_pipe.__aenter__ = AsyncMock(return_value=mock_pipe)
+        mock_pipe.__aexit__ = AsyncMock()
+
+        mock_redis = AsyncMock()
+        mock_redis.pipeline = MagicMock(return_value=mock_pipe)
+        manager._redis_client = mock_redis
+
+        with patch("functional_scaffold.core.job_manager.time") as mock_time:
+            mock_time.time.return_value = 1060.0  # 60 秒后
+
+            with patch("functional_scaffold.core.job_manager.set") as mock_set:
+                result = await manager.collect_queue_metrics()
+
+        assert result["queue_length"] == 5
+        assert result["processing_length"] == 2
+        assert result["dlq_length"] == 1
+        assert result["oldest_waiting_seconds"] == 60.0
+
+    @pytest.mark.asyncio
+    async def test_collect_queue_metrics_empty(self):
+        """测试空队列时收集指标"""
+        manager = JobManager()
+
+        mock_pipe = AsyncMock()
+        mock_pipe.llen = MagicMock()
+        mock_pipe.zrange = MagicMock()
+        mock_pipe.execute = AsyncMock(return_value=[0, 0, 0, []])
+        mock_pipe.__aenter__ = AsyncMock(return_value=mock_pipe)
+        mock_pipe.__aexit__ = AsyncMock()
+
+        mock_redis = AsyncMock()
+        mock_redis.pipeline = MagicMock(return_value=mock_pipe)
+        manager._redis_client = mock_redis
+
+        with patch("functional_scaffold.core.job_manager.set"):
+            result = await manager.collect_queue_metrics()
+
+        assert result["queue_length"] == 0
+        assert result["processing_length"] == 0
+        assert result["dlq_length"] == 0
+        assert result["oldest_waiting_seconds"] == 0
+
+    @pytest.mark.asyncio
+    async def test_collect_queue_metrics_without_redis(self):
+        """测试 Redis 不可用时收集指标"""
+        manager = JobManager()
+
+        result = await manager.collect_queue_metrics()
+
+        assert result["queue_length"] == 0
+        assert result["processing_length"] == 0
+        assert result["dlq_length"] == 0
+        assert result["oldest_waiting_seconds"] == 0

tests/test_metrics_unified.py

@@ -1,158 +1,239 @@
 """metrics_unified 模块单元测试"""
 
 import pytest
-from unittest.mock import MagicMock, patch
+from unittest.mock import AsyncMock, MagicMock, patch
+
+
+@pytest.fixture(autouse=True)
+def reset_manager():
+    """每个测试前后重置管理器"""
+    from functional_scaffold.core.metrics_unified import reset_metrics_manager_sync
+
+    reset_metrics_manager_sync()
+    yield
+    reset_metrics_manager_sync()
 
 
 class TestMetricsManager:
     """MetricsManager 类测试"""
 
-    @pytest.fixture
+    def test_init_loads_default_config(self):
-    def mock_redis(self):
-        """模拟 Redis 客户端"""
-        with patch("redis.Redis") as mock:
-            mock_instance = MagicMock()
-            mock_instance.ping.return_value = True
-            mock_instance.hincrbyfloat.return_value = 1.0
-            mock_instance.hset.return_value = True
-            mock_instance.hgetall.return_value = {}
-            mock_instance.hget.return_value = "0"
-            mock_instance.keys.return_value = []
-            mock_instance.pipeline.return_value = MagicMock()
-            mock.return_value = mock_instance
-            yield mock_instance
-
-    @pytest.fixture
-    def manager(self, mock_redis):
-        """创建测试用的 MetricsManager"""
-        from src.functional_scaffold.core.metrics_unified import (
-            MetricsManager,
-            reset_metrics_manager,
-        )
-
-        reset_metrics_manager()
-        manager = MetricsManager()
-        return manager
-
-    def test_init_loads_default_config(self, manager):
         """测试初始化加载默认配置"""
+        from functional_scaffold.core.metrics_unified import MetricsManager
+
+        manager = MetricsManager()
         assert manager.config is not None
         assert "builtin_metrics" in manager.config or len(manager.metrics_definitions) > 0
 
-    def test_metrics_definitions_registered(self, manager):
+    def test_metrics_definitions_registered(self):
         """测试指标定义已注册"""
+        from functional_scaffold.core.metrics_unified import MetricsManager
+
+        manager = MetricsManager()
         assert "http_requests_total" in manager.metrics_definitions
         assert "http_request_duration_seconds" in manager.metrics_definitions
         assert "algorithm_executions_total" in manager.metrics_definitions
 
-    def test_incr_counter(self, manager, mock_redis):
+    @pytest.mark.asyncio
+    @patch("redis.asyncio.Redis")
+    async def test_incr_counter(self, mock_redis_class):
         """测试计数器增加"""
-        manager.incr("http_requests_total", {"method": "GET", "endpoint": "/", "status": "success"})
+        mock_instance = AsyncMock()
-        mock_redis.hincrbyfloat.assert_called()
+        mock_instance.ping = AsyncMock(return_value=True)
+        mock_instance.hincrbyfloat = AsyncMock(return_value=1.0)
+        mock_instance.close = AsyncMock()
+        mock_redis_class.return_value = mock_instance
 
-    def test_incr_with_invalid_metric_type(self, manager, mock_redis):
+        from functional_scaffold.core.metrics_unified import MetricsManager
+
+        manager = MetricsManager()
+        await manager.initialize()
+
+        await manager.incr(
+            "http_requests_total", {"method": "GET", "endpoint": "/", "status": "success"}
+        )
+        mock_instance.hincrbyfloat.assert_called()
+
+    def test_incr_with_invalid_metric_type(self):
         """测试对非计数器类型调用 incr"""
+        from functional_scaffold.core.metrics_unified import MetricsManager
+
+        manager = MetricsManager()
         # http_request_duration_seconds 是 histogram 类型
-        manager.incr("http_request_duration_seconds", {})
+        # 验证不会抛出异常（因为 Redis 不可用）
-        # 不应该调用 Redis（因为类型不匹配）
-        # 验证没有调用 hincrbyfloat（或者调用次数没有增加）
 
-    def test_set_gauge(self, manager, mock_redis):
+    @pytest.mark.asyncio
+    @patch("redis.asyncio.Redis")
+    async def test_set_gauge(self, mock_redis_class):
         """测试设置仪表盘"""
-        manager.set("http_requests_in_progress", {}, 5)
+        mock_instance = AsyncMock()
-        mock_redis.hset.assert_called()
+        mock_instance.ping = AsyncMock(return_value=True)
+        mock_instance.hset = AsyncMock(return_value=True)
+        mock_instance.close = AsyncMock()
+        mock_redis_class.return_value = mock_instance
 
-    def test_gauge_incr(self, manager, mock_redis):
+        from functional_scaffold.core.metrics_unified import MetricsManager
+
+        manager = MetricsManager()
+        await manager.initialize()
+
+        await manager.set("http_requests_in_progress", {}, 5)
+        mock_instance.hset.assert_called()
+
+    @pytest.mark.asyncio
+    @patch("redis.asyncio.Redis")
+    async def test_gauge_incr(self, mock_redis_class):
         """测试增加仪表盘"""
-        manager.gauge_incr("http_requests_in_progress", {}, 1)
+        mock_instance = AsyncMock()
-        mock_redis.hincrbyfloat.assert_called()
+        mock_instance.ping = AsyncMock(return_value=True)
+        mock_instance.hincrbyfloat = AsyncMock(return_value=1.0)
+        mock_instance.close = AsyncMock()
+        mock_redis_class.return_value = mock_instance
 
-    def test_gauge_decr(self, manager, mock_redis):
+        from functional_scaffold.core.metrics_unified import MetricsManager
+
+        manager = MetricsManager()
+        await manager.initialize()
+
+        await manager.gauge_incr("http_requests_in_progress", {}, 1)
+        mock_instance.hincrbyfloat.assert_called()
+
+    @pytest.mark.asyncio
+    @patch("redis.asyncio.Redis")
+    async def test_gauge_decr(self, mock_redis_class):
         """测试减少仪表盘"""
-        manager.gauge_decr("http_requests_in_progress", {}, 1)
+        mock_instance = AsyncMock()
-        mock_redis.hincrbyfloat.assert_called()
+        mock_instance.ping = AsyncMock(return_value=True)
+        mock_instance.hincrbyfloat = AsyncMock(return_value=1.0)
+        mock_instance.close = AsyncMock()
+        mock_redis_class.return_value = mock_instance
 
-    def test_observe_histogram(self, manager, mock_redis):
+        from functional_scaffold.core.metrics_unified import MetricsManager
+
+        manager = MetricsManager()
+        await manager.initialize()
+
+        await manager.gauge_decr("http_requests_in_progress", {}, 1)
+        mock_instance.hincrbyfloat.assert_called()
+
+    @pytest.mark.asyncio
+    @patch("redis.asyncio.Redis")
+    async def test_observe_histogram(self, mock_redis_class):
         """测试直方图观测"""
-        mock_pipeline = MagicMock()
+        mock_instance = AsyncMock()
-        mock_redis.pipeline.return_value = mock_pipeline
+        mock_instance.ping = AsyncMock(return_value=True)
+        mock_instance.close = AsyncMock()
 
-        manager.observe("http_request_duration_seconds", {"method": "GET", "endpoint": "/"}, 0.05)
+        mock_pipeline = AsyncMock()
+        mock_pipeline.hincrbyfloat = MagicMock()
+        mock_pipeline.execute = AsyncMock(return_value=[])
+        mock_instance.pipeline = MagicMock(return_value=mock_pipeline)
 
-        mock_redis.pipeline.assert_called()
+        mock_redis_class.return_value = mock_instance
-        mock_pipeline.execute.assert_called()
 
-    def test_labels_to_key(self, manager):
+        from functional_scaffold.core.metrics_unified import MetricsManager
+
+        manager = MetricsManager()
+        await manager.initialize()
+
+        await manager.observe(
+            "http_request_duration_seconds", {"method": "GET", "endpoint": "/"}, 0.05
+        )
+        mock_instance.pipeline.assert_called()
+
+    def test_labels_to_key(self):
         """测试标签转换为 key"""
+        from functional_scaffold.core.metrics_unified import MetricsManager
+
+        manager = MetricsManager()
         labels = {"method": "GET", "endpoint": "/api"}
         key = manager._labels_to_key(labels)
         assert "method=GET" in key
         assert "endpoint=/api" in key
 
-    def test_labels_to_key_empty(self, manager):
+    def test_labels_to_key_empty(self):
         """测试空标签转换"""
+        from functional_scaffold.core.metrics_unified import MetricsManager
+
+        manager = MetricsManager()
         key = manager._labels_to_key(None)
         assert key == ""
 
         key = manager._labels_to_key({})
         assert key == ""
 
-    def test_is_available(self, manager):
+    @pytest.mark.asyncio
+    @patch("redis.asyncio.Redis")
+    async def test_is_available(self, mock_redis_class):
         """测试 Redis 可用性检查"""
+        mock_instance = AsyncMock()
+        mock_instance.ping = AsyncMock(return_value=True)
+        mock_instance.close = AsyncMock()
+        mock_redis_class.return_value = mock_instance
+
+        from functional_scaffold.core.metrics_unified import MetricsManager
+
+        manager = MetricsManager()
+        await manager.initialize()
+
         assert manager.is_available() is True
 
 
 class TestConvenienceFunctions:
     """便捷函数测试"""
 
-    @pytest.fixture(autouse=True)
+    @pytest.mark.asyncio
-    def setup(self):
+    @patch("redis.asyncio.Redis")
-        """每个测试前重置管理器"""
+    async def test_incr_function(self, mock_redis_class):
-        from src.functional_scaffold.core.metrics_unified import reset_metrics_manager
-
-        reset_metrics_manager()
-
-    @patch("redis.Redis")
-    def test_incr_function(self, mock_redis_class):
         """测试 incr 便捷函数"""
-        mock_instance = MagicMock()
+        mock_instance = AsyncMock()
-        mock_instance.ping.return_value = True
+        mock_instance.ping = AsyncMock(return_value=True)
+        mock_instance.hincrbyfloat = AsyncMock(return_value=1.0)
+        mock_instance.close = AsyncMock()
         mock_redis_class.return_value = mock_instance
 
-        from src.functional_scaffold.core.metrics_unified import incr, reset_metrics_manager
+        from functional_scaffold.core.metrics_unified import incr
 
-        reset_metrics_manager()
+        await incr(
-        incr("http_requests_total", {"method": "GET", "endpoint": "/", "status": "success"})
+            "http_requests_total", {"method": "GET", "endpoint": "/", "status": "success"}
+        )
 
         mock_instance.hincrbyfloat.assert_called()
 
-    @patch("redis.Redis")
+    @pytest.mark.asyncio
-    def test_set_function(self, mock_redis_class):
+    @patch("redis.asyncio.Redis")
+    async def test_set_function(self, mock_redis_class):
         """测试 set 便捷函数"""
-        mock_instance = MagicMock()
+        mock_instance = AsyncMock()
-        mock_instance.ping.return_value = True
+        mock_instance.ping = AsyncMock(return_value=True)
+        mock_instance.hset = AsyncMock(return_value=True)
+        mock_instance.close = AsyncMock()
         mock_redis_class.return_value = mock_instance
 
-        from src.functional_scaffold.core.metrics_unified import reset_metrics_manager, set
+        from functional_scaffold.core.metrics_unified import set
 
-        reset_metrics_manager()
+        await set("http_requests_in_progress", {}, 10)
-        set("http_requests_in_progress", {}, 10)
 
         mock_instance.hset.assert_called()
 
-    @patch("redis.Redis")
+    @pytest.mark.asyncio
-    def test_observe_function(self, mock_redis_class):
+    @patch("redis.asyncio.Redis")
+    async def test_observe_function(self, mock_redis_class):
         """测试 observe 便捷函数"""
-        mock_instance = MagicMock()
+        mock_instance = AsyncMock()
-        mock_instance.ping.return_value = True
+        mock_instance.ping = AsyncMock(return_value=True)
-        mock_pipeline = MagicMock()
+        mock_instance.close = AsyncMock()
-        mock_instance.pipeline.return_value = mock_pipeline
+
+        mock_pipeline = AsyncMock()
+        mock_pipeline.hincrbyfloat = MagicMock()
+        mock_pipeline.execute = AsyncMock(return_value=[])
+        mock_instance.pipeline = MagicMock(return_value=mock_pipeline)
+
         mock_redis_class.return_value = mock_instance
 
-        from src.functional_scaffold.core.metrics_unified import observe, reset_metrics_manager
+        from functional_scaffold.core.metrics_unified import observe
 
-        reset_metrics_manager()
+        await observe("http_request_duration_seconds", {"method": "GET", "endpoint": "/"}, 0.1)
-        observe("http_request_duration_seconds", {"method": "GET", "endpoint": "/"}, 0.1)
 
         mock_instance.pipeline.assert_called()
 
@@ -160,42 +241,49 @@ class TestConvenienceFunctions:
 class TestExport:
     """导出功能测试"""
 
-    @patch("redis.Redis")
+    @pytest.mark.asyncio
-    def test_export_counter(self, mock_redis_class):
+    @patch("redis.asyncio.Redis")
+    async def test_export_counter(self, mock_redis_class):
         """测试导出计数器"""
-        mock_instance = MagicMock()
+        mock_instance = AsyncMock()
-        mock_instance.ping.return_value = True
+        mock_instance.ping = AsyncMock(return_value=True)
-        mock_instance.hgetall.return_value = {"method=GET,endpoint=/,status=success": "10"}
+        mock_instance.hgetall = AsyncMock(
+            return_value={"method=GET,endpoint=/,status=success": "10"}
+        )
+        mock_instance.hget = AsyncMock(return_value="0")
+        mock_instance.close = AsyncMock()
         mock_redis_class.return_value = mock_instance
 
-        from src.functional_scaffold.core.metrics_unified import export, reset_metrics_manager
+        from functional_scaffold.core.metrics_unified import export
 
-        reset_metrics_manager()
+        output = await export()
-        output = export()
 
         assert "http_requests_total" in output
         assert "HELP" in output
         assert "TYPE" in output
 
-    @patch("redis.Redis")
+    @pytest.mark.asyncio
-    def test_export_histogram(self, mock_redis_class):
+    @patch("redis.asyncio.Redis")
+    async def test_export_histogram(self, mock_redis_class):
         """测试导出直方图"""
-        mock_instance = MagicMock()
+        mock_instance = AsyncMock()
-        mock_instance.ping.return_value = True
+        mock_instance.ping = AsyncMock(return_value=True)
-        mock_instance.hgetall.side_effect = lambda key: (
+
-            {"method=GET,endpoint=/": "5"}
+        async def mock_hgetall(key):
-            if "count" in key
+            if "count" in key:
-            else {"method=GET,endpoint=/": "0.5"}
+                return {"method=GET,endpoint=/": "5"}
-            if "sum" in key
+            elif "sum" in key:
-            else {}
+                return {"method=GET,endpoint=/": "0.5"}
-        )
+            return {}
-        mock_instance.hget.return_value = "3"
+
+        mock_instance.hgetall = mock_hgetall
+        mock_instance.hget = AsyncMock(return_value="3")
+        mock_instance.close = AsyncMock()
         mock_redis_class.return_value = mock_instance
 
-        from src.functional_scaffold.core.metrics_unified import export, reset_metrics_manager
+        from functional_scaffold.core.metrics_unified import export
 
-        reset_metrics_manager()
+        output = await export()
-        output = export()
 
         assert "http_request_duration_seconds" in output
 
@@ -206,7 +294,7 @@ class TestEnvVarSubstitution:
     def test_substitute_env_vars(self):
         """测试环境变量替换"""
         import os
-        from src.functional_scaffold.core.metrics_unified import MetricsManager
+        from functional_scaffold.core.metrics_unified import MetricsManager
 
         # 设置测试环境变量
         os.environ["TEST_VAR"] = "test_value"
@@ -226,21 +314,9 @@ class TestEnvVarSubstitution:
 class TestTrackAlgorithmExecution:
     """track_algorithm_execution 装饰器测试"""
 
-    @patch("redis.Redis")
+    def test_decorator_success(self):
-    def test_decorator_success(self, mock_redis_class):
         """测试装饰器成功执行"""
-        mock_instance = MagicMock()
+        from functional_scaffold.core.metrics_unified import track_algorithm_execution
-        mock_instance.ping.return_value = True
-        mock_pipeline = MagicMock()
-        mock_instance.pipeline.return_value = mock_pipeline
-        mock_redis_class.return_value = mock_instance
-
-        from src.functional_scaffold.core.metrics_unified import (
-            reset_metrics_manager,
-            track_algorithm_execution,
-        )
-
-        reset_metrics_manager()
 
         @track_algorithm_execution("test_algo")
         def test_func():
@@ -249,21 +325,9 @@ class TestTrackAlgorithmExecution:
         result = test_func()
         assert result == "result"
 
-    @patch("redis.Redis")
+    def test_decorator_error(self):
-    def test_decorator_error(self, mock_redis_class):
         """测试装饰器错误处理"""
-        mock_instance = MagicMock()
+        from functional_scaffold.core.metrics_unified import track_algorithm_execution
-        mock_instance.ping.return_value = True
-        mock_pipeline = MagicMock()
-        mock_instance.pipeline.return_value = mock_pipeline
-        mock_redis_class.return_value = mock_instance
-
-        from src.functional_scaffold.core.metrics_unified import (
-            reset_metrics_manager,
-            track_algorithm_execution,
-        )
-
-        reset_metrics_manager()
 
         @track_algorithm_execution("test_algo")
         def test_func():

tests/test_middleware.py

@@ -0,0 +1,97 @@
+"""中间件测试"""
+
+import pytest
+from unittest.mock import patch, MagicMock
+from fastapi.testclient import TestClient
+
+from functional_scaffold.main import app, normalize_path
+
+
+class TestNormalizePath:
+    """测试路径规范化函数"""
+
+    def test_normalize_jobs_path(self):
+        """测试 /jobs/{job_id} 路径规范化"""
+        assert normalize_path("/jobs/a1b2c3d4e5f6") == "/jobs/{job_id}"
+        assert normalize_path("/jobs/123456789012") == "/jobs/{job_id}"
+        assert normalize_path("/jobs/xyz") == "/jobs/{job_id}"
+
+    def test_normalize_other_paths(self):
+        """测试其他路径保持不变"""
+        assert normalize_path("/invoke") == "/invoke"
+        assert normalize_path("/healthz") == "/healthz"
+        assert normalize_path("/readyz") == "/readyz"
+        assert normalize_path("/metrics") == "/metrics"
+        assert normalize_path("/docs") == "/docs"
+
+    def test_normalize_jobs_root(self):
+        """测试 /jobs 根路径"""
+        assert normalize_path("/jobs") == "/jobs"
+
+
+class TestMetricsMiddleware:
+    """测试指标中间件"""
+
+    @patch("functional_scaffold.main.incr")
+    @patch("functional_scaffold.main.observe")
+    @patch("functional_scaffold.main.gauge_incr")
+    @patch("functional_scaffold.main.gauge_decr")
+    def test_skip_health_endpoints(self, mock_gauge_decr, mock_gauge_incr, mock_observe, mock_incr):
+        """测试跳过健康检查端点"""
+        client = TestClient(app)
+
+        # 访问健康检查端点
+        client.get("/healthz")
+        client.get("/readyz")
+        client.get("/metrics")
+
+        # 验证没有记录指标
+        mock_incr.assert_not_called()
+        mock_observe.assert_not_called()
+        mock_gauge_incr.assert_not_called()
+        mock_gauge_decr.assert_not_called()
+
+    @patch("functional_scaffold.main.incr")
+    @patch("functional_scaffold.main.observe")
+    @patch("functional_scaffold.main.gauge_incr")
+    @patch("functional_scaffold.main.gauge_decr")
+    def test_record_normal_endpoints(self, mock_gauge_decr, mock_gauge_incr, mock_observe, mock_incr):
+        """测试记录普通端点"""
+        client = TestClient(app)
+
+        # 访问普通端点
+        client.post("/invoke", json={"number": 17})
+
+        # 验证记录了指标
+        mock_gauge_incr.assert_called_once()
+        mock_gauge_decr.assert_called_once()
+        mock_incr.assert_called_once()
+        mock_observe.assert_called_once()
+
+        # 验证使用了正确的端点路径
+        incr_call_args = mock_incr.call_args
+        assert incr_call_args[0][1]["endpoint"] == "/invoke"
+
+    @patch("functional_scaffold.main.incr")
+    @patch("functional_scaffold.main.observe")
+    @patch("functional_scaffold.main.gauge_incr")
+    @patch("functional_scaffold.main.gauge_decr")
+    @patch("functional_scaffold.core.job_manager.get_job_manager")
+    def test_normalize_job_path(self, mock_get_manager, mock_gauge_decr, mock_gauge_incr, mock_observe, mock_incr):
+        """测试规范化任务路径"""
+        # Mock job manager
+        mock_manager = MagicMock()
+        mock_manager.get_job.return_value = None
+        mock_get_manager.return_value = mock_manager
+
+        client = TestClient(app)
+
+        # 访问任务端点（会返回 404，但中间件应该记录指标）
+        client.get("/jobs/a1b2c3d4e5f6")
+
+        # 验证记录了指标
+        mock_incr.assert_called_once()
+
+        # 验证使用了规范化后的路径
+        incr_call_args = mock_incr.call_args
+        assert incr_call_args[0][1]["endpoint"] == "/jobs/{job_id}"