Compare commits
12 Commits
| Author | SHA1 | Date | |
|---|---|---|---|
 vincent
|
c50dcc9cb2 | ||
|
vincent
|
ba5b932f50 | ||
|
vincent
|
e829a4060b | ||
|
vincent
|
205381c4ff | ||
|
vincent
|
6b5f53a0fd | ||
|
vincent
|
3246a1f0d9 | ||
|
vincent
|
cca4089f2a | ||
|
vincent
|
f4191f82f5 | ||
 陆怀瑾 (COO)
|
80ef3c2796 | ||
|
陆怀瑾 (COO)
|
8c93fee885 | ||
|
陆怀瑾 (COO)
|
f793da91f0 | ||
|
陆怀瑾 (COO)
|
74cfb3e0f6 |
@@ -0,0 +1,39 @@
|
||||
# 知识库索引
|
||||
|
||||
> 本知识库与 Agent 配置文件解耦，由 COO 主导维护，各领域负责人协作贡献。
|
||||
> 通过 `wiki_search` / `memory_search` / `qmd` 等工具检索，人类可通过 Web UI 审查优化。
|
||||
|
||||
## 目录结构
|
||||
|
||||
| 目录 | 领域 | 责任人 | 条目数 |
|
||||
|------|------|--------|--------|
|
||||
| [电商/](电商/) | 淘宝、抖店、微信小店运营 | 陆云帆 (taobaospecialist) | — |
|
||||
| [内容/](内容/) | 小红书、短视频、文案 | 文墨言 (contentspecialist) | — |
|
||||
| [产品/](产品/) | PRD、需求分析 | 沈路明 (productmanager) | — |
|
||||
| [技术/](技术/) | 开发规范、代码审查 | 徐聪 (costcodev) | — |
|
||||
| [设计/](设计/) | UI设计、品牌规范 | 苏绘锦 (designer) | — |
|
||||
| [运营/](运营/) | 活动策划、数据分析 | 陆怀瑾 (coo) | — |
|
||||
| [行政/](行政/) | 合同、报销流程 | 刘诗妮 (secretary) | — |
|
||||
|
||||
## 知识条目格式
|
||||
|
||||
每个知识条目遵循 [模板](../templates/知识条目模板.md)。
|
||||
|
||||
## 检索方式
|
||||
|
||||
- **Agent 主动查询**：`wiki_search` / `memory_search` / `qmd`
|
||||
- **人类审查**：通过 Web UI 浏览、编辑、优化
|
||||
- **质量检查**：`wiki_lint` 定期运行
|
||||
|
||||
## 贡献流程
|
||||
|
||||
1. 领域负责人撰写条目
|
||||
2. COO 审核内容质量
|
||||
3. 提交到 EnterpriseArchitect 仓库
|
||||
4. 通过 `wiki_lint` 检查
|
||||
5. 通知相关 Agent 更新索引
|
||||
|
||||
---
|
||||
|
||||
**维护者**：陆怀瑾（COO）
|
||||
**最后更新**：2026-06-22
|
||||
@@ -0,0 +1,111 @@
|
||||
# PRD 模板
|
||||
|
||||
## 元数据
|
||||
|
||||
| 属性 | 值 |
|
||||
|------|-----|
|
||||
| **领域** | 产品 |
|
||||
| **责任人** | 沈路明 (productmanager) |
|
||||
| **版本** | v1.0 |
|
||||
| **创建日期** | 2026-06-22 |
|
||||
| **标签** | PRD, 产品需求, 模板 |
|
||||
|
||||
## 概述
|
||||
|
||||
产品需求文档（PRD）标准模板。适用于所有产品功能需求、系统改进需求的规范化描述，确保开发团队、设计团队、业务团队对需求理解一致。
|
||||
|
||||
## 正文
|
||||
|
||||
### 一、文档头部
|
||||
|
||||
```
|
||||
# [产品名称] - [功能名称] PRD
|
||||
|
||||
| 属性 | 值 |
|
||||
|------|-----|
|
||||
| **版本** | v1.0 |
|
||||
| **作者** | [姓名] |
|
||||
| **创建日期** | YYYY-MM-DD |
|
||||
| **状态** | 草稿 / 评审中 / 已批准 / 已上线 |
|
||||
| **关联文档** | [链接] |
|
||||
```
|
||||
|
||||
### 二、需求概述
|
||||
|
||||
**2.1 背景与问题**
|
||||
[描述为什么需要这个功能，解决了什么用户痛点或业务问题]
|
||||
|
||||
**2.2 目标用户**
|
||||
- 用户画像 1：[描述]
|
||||
- 用户画像 2：[描述]
|
||||
|
||||
**2.3 核心目标**
|
||||
- 业务目标：[可量化指标，如转化率提升 X%]
|
||||
- 用户目标：[用户获得什么价值]
|
||||
- 技术目标：[如响应时间、并发量]
|
||||
|
||||
### 三、功能描述
|
||||
|
||||
**3.1 功能范围**
|
||||
- P0（必须）：[最小可用功能]
|
||||
- P1（应该）：[重要但可后续]
|
||||
- P2（锦上添花）：[可后续迭代]
|
||||
|
||||
**3.2 用户故事**
|
||||
|
||||
```
|
||||
作为 [用户角色]，
|
||||
我希望 [功能/行为]，
|
||||
以便 [获得的价值/目标]。
|
||||
```
|
||||
|
||||
**3.3 详细交互说明**
|
||||
1. [步骤1]：[描述 + 原型图链接]
|
||||
2. [步骤2]：[描述 + 原型图链接]
|
||||
|
||||
**3.4 边界与异常**
|
||||
- 正常流程：[描述]
|
||||
- 异常情况1：[触发条件 + 处理方式]
|
||||
- 异常情况2：[触发条件 + 处理方式]
|
||||
|
||||
### 四、非功能需求
|
||||
|
||||
| 项目 | 要求 |
|
||||
|------|------|
|
||||
| 页面加载 | ≤ 2 秒 |
|
||||
| 接口响应 | ≤ 500ms |
|
||||
| 并发支持 | 1000 QPS |
|
||||
| 兼容性 | iOS 13+, Android 9+, Chrome 90+ |
|
||||
|
||||
### 五、数据埋点
|
||||
|
||||
| 事件名 | 触发条件 | 属性 |
|
||||
|--------|----------|------|
|
||||
| [event_name] | [触发条件] | [上报字段] |
|
||||
|
||||
### 六、验收标准
|
||||
|
||||
- [ ] 功能1 验收条件
|
||||
- [ ] 功能2 验收条件
|
||||
- [ ] 非功能需求满足
|
||||
|
||||
### 七、排期与里程碑
|
||||
|
||||
| 里程碑 | 日期 | 交付物 |
|
||||
|--------|------|--------|
|
||||
| 设计评审 | YYYY-MM-DD | 交互/视觉稿 |
|
||||
| 技术评审 | YYYY-MM-DD | 技术方案 |
|
||||
| 开发完成 | YYYY-MM-DD | 可测试版本 |
|
||||
| 上线 | YYYY-MM-DD | 生产环境 |
|
||||
|
||||
## 相关条目
|
||||
|
||||
- [需求分析方法.md](需求分析方法.md)
|
||||
- [开发规范.md](../技术/开发规范.md)
|
||||
- [UI设计规范.md](../设计/UI设计规范.md)
|
||||
|
||||
## 变更记录
|
||||
|
||||
| 日期 | 版本 | 变更说明 | 变更人 |
|
||||
|------|------|----------|--------|
|
||||
| 2026-06-22 | v1.0 | 初始创建 | 陆怀瑾 |
|
||||
@@ -0,0 +1,21 @@
|
||||
# 产品领域知识
|
||||
|
||||
**责任人**：沈路明（productmanager）
|
||||
**审核人**：陆怀瑾（coo）
|
||||
|
||||
## 知识范围
|
||||
|
||||
涵盖产品需求文档、用户研究、竞品分析、需求管理、版本规划等产品管理知识。
|
||||
|
||||
## 条目清单
|
||||
|
||||
| 文件名 | 说明 | 状态 |
|
||||
|--------|------|------|
|
||||
| [PRD模板.md](PRD模板.md) | 产品需求文档标准模板 | ✅ |
|
||||
| [需求分析方法.md](需求分析方法.md) | 用户需求调研与分析方法 | ✅ |
|
||||
|
||||
## 待建设
|
||||
|
||||
- 竞品分析框架
|
||||
- 产品路线图模板
|
||||
- 用户故事编写指南
|
||||
@@ -0,0 +1,84 @@
|
||||
# 需求分析方法
|
||||
|
||||
## 元数据
|
||||
|
||||
| 属性 | 值 |
|
||||
|------|-----|
|
||||
| **领域** | 产品 |
|
||||
| **责任人** | 沈路明 (productmanager) |
|
||||
| **版本** | v1.0 |
|
||||
| **创建日期** | 2026-06-22 |
|
||||
| **标签** | 需求分析, 用户调研, 产品管理 |
|
||||
|
||||
## 概述
|
||||
|
||||
需求分析是从用户/业务痛点出发，将模糊的需求描述转化为可落地的产品功能规格的系统化方法。核心原则：先理解问题，再设计方案。
|
||||
|
||||
## 正文
|
||||
|
||||
### 一、需求收集方法
|
||||
|
||||
1. **用户访谈**（定性）
|
||||
- 每轮访谈 5-8 个目标用户
|
||||
- 半结构化访谈：准备提纲 + 灵活追问
|
||||
- 核心问题：「你最想解决什么问题？」「现在怎么解决的？」
|
||||
|
||||
2. **问卷调查**（定量）
|
||||
- 覆盖 100+ 目标用户
|
||||
- 包含选择题（量化）+ 开放题（发掘）
|
||||
- 关键指标：问题频率、痛点程度、替代方案满意度
|
||||
|
||||
3. **数据分析**
|
||||
- 页面点击热力图
|
||||
- 用户行为漏斗（转化率断点）
|
||||
- 客服工单高频关键词
|
||||
|
||||
### 二、需求优先级评估 — ICE 模型
|
||||
|
||||
| 因子 | 说明 | 评分 (1-10) |
|
||||
|------|------|------------|
|
||||
| **I**mpact（影响面） | 影响多少用户？对核心指标影响多大？ | |
|
||||
| **C**onfidence（信心度） | 我们有多少证据这个方案有效？ | |
|
||||
| **E**ase（实现难度） | 开发成本多高？时间多长？ | |
|
||||
|
||||
总分 = I × C × E（E 分数越高越容易，越大越好）
|
||||
|
||||
### 三、需求文档化
|
||||
|
||||
1. **用户故事标准格式**
|
||||
> 作为 **[用户角色]**，
|
||||
> 我希望 **[功能/行为]**，
|
||||
> 以便 **[获得的价值]**。
|
||||
|
||||
2. **验收条件（Acceptance Criteria）**
|
||||
- 必须可测试、可验证
|
||||
- 正面条件 + 边缘情况
|
||||
|
||||
3. **原型验证**
|
||||
- 低保真原型验证交互流程（1-2 天）
|
||||
- 用户测试 3-5 人，观察操作行为
|
||||
- 根据反馈迭代后进入高保真设计
|
||||
|
||||
### 四、需求评审流程
|
||||
|
||||
```
|
||||
需求方提出 → PM 分析评估 → 交互设计 → 技术评审
|
||||
→ 排期评估 → 最终评审 → 进入开发
|
||||
```
|
||||
|
||||
每一步评审需至少以下人员参与：
|
||||
- PM（负责人）
|
||||
- 1 名开发（评估技术可行性）
|
||||
- 1 名设计师（评估交互可行性）
|
||||
- 需求方（确认需求理解正确）
|
||||
|
||||
## 相关条目
|
||||
|
||||
- [PRD模板.md](PRD模板.md)
|
||||
- [开发规范.md](../技术/开发规范.md)
|
||||
|
||||
## 变更记录
|
||||
|
||||
| 日期 | 版本 | 变更说明 | 变更人 |
|
||||
|------|------|----------|--------|
|
||||
| 2026-06-22 | v1.0 | 初始创建 | 陆怀瑾 |
|
||||
@@ -0,0 +1,21 @@
|
||||
# 内容领域知识
|
||||
|
||||
**责任人**：文墨言（contentspecialist）
|
||||
**审核人**：陆怀瑾（coo）
|
||||
|
||||
## 知识范围
|
||||
|
||||
涵盖小红书、短视频平台、公众号等内容平台运营知识，包括内容创作、选题策划、标题优化、发布策略、数据分析等。
|
||||
|
||||
## 条目清单
|
||||
|
||||
| 文件名 | 说明 | 状态 |
|
||||
|--------|------|------|
|
||||
| [小红书运营指南.md](小红书运营指南.md) | 小红书内容运营全流程指南 | ✅ |
|
||||
| [标题写作技巧.md](标题写作技巧.md) | 爆款标题创作方法论 | ✅ |
|
||||
|
||||
## 待建设
|
||||
|
||||
- 短视频脚本模板
|
||||
- 公众号排版规范
|
||||
- 内容日历模板
|
||||
@@ -0,0 +1,82 @@
|
||||
# 小红书运营指南
|
||||
|
||||
## 元数据
|
||||
|
||||
| 属性 | 值 |
|
||||
|------|-----|
|
||||
| **领域** | 内容 |
|
||||
| **责任人** | 文墨言 (contentspecialist) |
|
||||
| **版本** | v1.0 |
|
||||
| **创建日期** | 2026-06-22 |
|
||||
| **标签** | 小红书, 内容运营, 种草, 涨粉 |
|
||||
|
||||
## 概述
|
||||
|
||||
小红书是以"真实分享+种草"为核心的内容社区，运营不同于其他平台。核心逻辑：真诚分享 > 硬广推广，封面/标题决定点击率，内容质量决定涨粉转化。
|
||||
|
||||
## 正文
|
||||
|
||||
### 一、内容定位与选题
|
||||
|
||||
1. **账号定位**（上线前必做）
|
||||
- 明确赛道：美妆/穿搭/家居/母婴/美食/知识
|
||||
- 确定人设：专家型/体验型/教程型
|
||||
- 对标 3-5 个同赛道 Top 博主
|
||||
|
||||
2. **选题策略**
|
||||
- 热点追踪：小红书热搜 + 抖音热点宝
|
||||
- 实用内容：教程/清单/测评/避坑
|
||||
- 情感共鸣：个人经历/观点分享/生活记录
|
||||
|
||||
### 二、内容制作标准
|
||||
|
||||
1. **封面设计**（点击率核心）
|
||||
- 高饱和度配色，对比度强
|
||||
- 简洁文字 3-7 字，避免遮挡主体
|
||||
- 尺寸 3:4，首图即为封面
|
||||
|
||||
2. **标题公式**
|
||||
- 数字型：「3 步搞定...」
|
||||
- 痛点型：「为什么你...还是不行？」
|
||||
- 对比型：「A vs B，差距到底在哪」
|
||||
- 清单型：「2026 必入的 10 款...」
|
||||
|
||||
3. **正文结构**
|
||||
- 开头（3 句）：抛痛点/抛结论
|
||||
- 主体：分点说明，配图对应
|
||||
- 结尾：互动引导（提问/投票/求关注）
|
||||
|
||||
### 三、发布与推广
|
||||
|
||||
1. **发布时间**
|
||||
- 工作日：12:00-14:00, 18:00-21:00
|
||||
- 周末：10:00-12:00, 15:00-18:00
|
||||
|
||||
2. **话题标签策略**
|
||||
- 1-2 个大流量话题（#穿搭 #美妆 #家居）
|
||||
- 2-3 个精准话题（#小个子穿搭 #通勤穿搭）
|
||||
- 1 个自创话题（#XX的日常搭配）
|
||||
|
||||
3. **初期冷启动**
|
||||
- 发布后 1 小时内互动（评论/点赞）对推荐权重影响最大
|
||||
- 在同类笔记下真诚评论（非硬广引流）
|
||||
|
||||
### 四、数据指标
|
||||
|
||||
| 指标 | 新手目标 | 进阶目标 |
|
||||
|------|----------|----------|
|
||||
| 单篇阅读量 | 1000+ | 5000+ |
|
||||
| 点赞率 | 3%+ | 5%+ |
|
||||
| 收藏率 | 2%+ | 4%+ |
|
||||
| 涨粉率 | 1%/篇 | 3%/篇 |
|
||||
|
||||
## 相关条目
|
||||
|
||||
- [标题写作技巧.md](标题写作技巧.md)
|
||||
- [活动策划模板.md](../运营/活动策划模板.md)
|
||||
|
||||
## 变更记录
|
||||
|
||||
| 日期 | 版本 | 变更说明 | 变更人 |
|
||||
|------|------|----------|--------|
|
||||
| 2026-06-22 | v1.0 | 初始创建 | 陆怀瑾 |
|
||||
@@ -0,0 +1,21 @@
|
||||
# 技术领域知识
|
||||
|
||||
**责任人**：徐聪（costcodev）
|
||||
**审核人**：陆怀瑾（coo）
|
||||
|
||||
## 知识范围
|
||||
|
||||
涵盖开发规范、代码审查、架构设计、部署运维、技术选型等技术团队知识。
|
||||
|
||||
## 条目清单
|
||||
|
||||
| 文件名 | 说明 | 状态 |
|
||||
|--------|------|------|
|
||||
| [开发规范.md](开发规范.md) | 代码编写与项目管理规范 | ✅ |
|
||||
| [代码审查清单.md](代码审查清单.md) | Pull Request 审查标准 | ✅ |
|
||||
|
||||
## 待建设
|
||||
|
||||
- API 设计规范
|
||||
- 数据库设计指南
|
||||
- 技术选型决策框架
|
||||
@@ -0,0 +1,104 @@
|
||||
# 开发规范
|
||||
|
||||
## 元数据
|
||||
|
||||
| 属性 | 值 |
|
||||
|------|-----|
|
||||
| **领域** | 技术 |
|
||||
| **责任人** | 徐聪 (costcodev) |
|
||||
| **版本** | v1.0 |
|
||||
| **创建日期** | 2026-06-22 |
|
||||
| **标签** | 开发规范, 代码风格, Git, 项目管理 |
|
||||
|
||||
## 概述
|
||||
|
||||
定义团队统一的代码编写、项目管理、协作流程规范。目的是确保代码可维护、可交接，降低协作摩擦。
|
||||
|
||||
## 正文
|
||||
|
||||
### 一、代码规范
|
||||
|
||||
1. **Python**
|
||||
- 遵循 PEP 8 代码风格
|
||||
- 使用 `black` 自动格式化，行宽 100
|
||||
- 类型注解必须（`mypy --strict` 通过）
|
||||
- 文档字符串用 Google 风格
|
||||
|
||||
2. **TypeScript/JavaScript**
|
||||
- 使用 `prettier` 格式化
|
||||
- ESLint 严格模式
|
||||
- 禁止 `any` 类型（除非显式标注 `// eslint-disable-next-line`）
|
||||
- 所有公共 API 必须有 JSDoc
|
||||
|
||||
3. **通用规则**
|
||||
- 函数单一职责，不超过 50 行
|
||||
- 命名：camelCase（变量/函数）、PascalCase（类/组件）、UPPER_SNAKE（常量）
|
||||
- 禁止 `print` / `console.log` 残留（用日志库）
|
||||
- 禁止注释掉的代码（相信 Git）
|
||||
|
||||
### 二、Git 规范
|
||||
|
||||
1. **分支策略**
|
||||
```
|
||||
main ─── 生产环境
|
||||
develop ─── 开发主线
|
||||
feature/<task-id>-<desc> ─── 功能分支
|
||||
fix/<task-id>-<desc> ─── 修复分支
|
||||
```
|
||||
|
||||
2. **Commit 格式**
|
||||
```
|
||||
<type>(<scope>): <subject>
|
||||
|
||||
<body>
|
||||
|
||||
<footer>
|
||||
```
|
||||
- type: feat / fix / docs / style / refactor / test / chore
|
||||
- scope: 模块名（如 api, ui, db）
|
||||
- subject: 不超过 72 字符，中文或英文
|
||||
|
||||
3. **PR 流程**
|
||||
- 所有代码变更必须通过 PR
|
||||
- 至少 1 人 Review 并 Approve
|
||||
- CI 全部通过后才能合并
|
||||
- 合并前 rebase develop 消除冲突
|
||||
|
||||
### 三、项目结构规范
|
||||
|
||||
```
|
||||
project/
|
||||
├── src/ # 源代码
|
||||
├── tests/ # 测试代码
|
||||
├── docs/ # 项目文档
|
||||
├── scripts/ # 运维脚本
|
||||
├── config/ # 配置文件
|
||||
├── README.md # 项目说明
|
||||
├── CHANGELOG.md # 变更日志
|
||||
└── .env.example # 环境变量模板
|
||||
```
|
||||
|
||||
### 四、文档规范
|
||||
|
||||
- **README.md**：项目概述、快速启动、技术栈、目录说明
|
||||
- **API 文档**：后端接口必须有 OpenAPI/Swagger 文档
|
||||
- **开发文档**：架构设计、数据流图、部署说明
|
||||
- **代码即文档**：优先清晰的命名和结构，减少注释
|
||||
|
||||
### 五、测试规范
|
||||
|
||||
- 单元测试覆盖率 ≥ 70%
|
||||
- 关键业务逻辑覆盖率 ≥ 90%
|
||||
- 每个 PR 附带新增/修改的测试
|
||||
- 使用 `pytest` (Python) / `vitest` (TS)
|
||||
|
||||
## 相关条目
|
||||
|
||||
- [代码审查清单.md](代码审查清单.md)
|
||||
- [PRD模板.md](../产品/PRD模板.md)
|
||||
|
||||
## 变更记录
|
||||
|
||||
| 日期 | 版本 | 变更说明 | 变更人 |
|
||||
|------|------|----------|--------|
|
||||
| 2026-06-22 | v1.0 | 初始创建 | 陆怀瑾 |
|
||||
@@ -0,0 +1,21 @@
|
||||
# 电商领域知识
|
||||
|
||||
**责任人**：陆云帆（taobaospecialist）
|
||||
**审核人**：陆怀瑾（coo）
|
||||
|
||||
## 知识范围
|
||||
|
||||
涵盖淘宝、抖店、微信小店等多平台电商运营知识，包括店铺搭建、商品上架、营销推广、客户服务、数据分析等。
|
||||
|
||||
## 条目清单
|
||||
|
||||
| 文件名 | 说明 | 状态 |
|
||||
|--------|------|------|
|
||||
| [淘宝运营SOP.md](淘宝运营SOP.md) | 淘宝店铺日常运营标准流程 | ✅ |
|
||||
| [抖店运营SOP.md](抖店运营SOP.md) | 抖音小店运营流程 | ✅ |
|
||||
|
||||
## 待建设
|
||||
|
||||
- 微信小店运营指南
|
||||
- 电商数据分析方法
|
||||
- 客服话术模板
|
||||
@@ -0,0 +1,74 @@
|
||||
# 抖店运营 SOP
|
||||
|
||||
## 元数据
|
||||
|
||||
| 属性 | 值 |
|
||||
|------|-----|
|
||||
| **领域** | 电商 |
|
||||
| **责任人** | 陆云帆 (taobaospecialist) |
|
||||
| **版本** | v1.0 |
|
||||
| **创建日期** | 2026-06-22 |
|
||||
| **标签** | 抖音, 抖店, 电商, SOP |
|
||||
|
||||
## 概述
|
||||
|
||||
本 SOP 定义抖音小店运营标准流程。抖店运营区别于传统电商的核心在于"内容驱动交易"——通过短视频和直播引流到店铺成交。
|
||||
|
||||
## 正文
|
||||
|
||||
### 一、每日运营
|
||||
|
||||
1. **店铺健康检查**
|
||||
- 登录抖店后台，检查体验分（≥ 4.6）
|
||||
- 查看违规记录和扣分情况
|
||||
- 检查商品状态（在售/审核中/下架）
|
||||
|
||||
2. **内容运营**
|
||||
- 发布 1-2 条挂车短视频
|
||||
- 检查昨日短视频/直播数据
|
||||
- 回复评论区用户问题
|
||||
|
||||
3. **订单与客服**
|
||||
- 处理待发货订单（48 小时发货）
|
||||
- 处理售后申请（退货/退款）
|
||||
- 3 分钟内回复客服消息
|
||||
|
||||
### 二、每周运营
|
||||
|
||||
1. **商品策略**
|
||||
- 分析本周爆款商品，优化标题/主图/详情
|
||||
- 根据热点趋势选品上新
|
||||
- 设置限时秒杀/优惠券活动
|
||||
|
||||
2. **内容策略**
|
||||
- 复盘本周短视频/直播数据
|
||||
- 策划下周内容选题（蹭热点/产品展示/教程）
|
||||
- 测试新视频形式（口播/开箱/场景化）
|
||||
|
||||
3. **投放优化**
|
||||
- 查看千川投放数据
|
||||
- 优化投放计划（人群/出价/素材）
|
||||
- 调整 ROI 目标和预算分配
|
||||
|
||||
### 三、每月运营
|
||||
|
||||
1. **月度分析**
|
||||
- 统计月度 GMV、订单量、退款率
|
||||
- 分析流量来源占比（推荐/搜索/直播/短视频/付费）
|
||||
- 输出《抖店月度运营报告》
|
||||
|
||||
2. **供应链检查**
|
||||
- 盘点库存，补货预警
|
||||
- 检查发货时效和物流评分
|
||||
- 供应商评估和优化
|
||||
|
||||
## 相关条目
|
||||
|
||||
- [淘宝运营SOP.md](淘宝运营SOP.md)
|
||||
- [数据分析方法.md](../运营/数据分析方法.md)
|
||||
|
||||
## 变更记录
|
||||
|
||||
| 日期 | 版本 | 变更说明 | 变更人 |
|
||||
|------|------|----------|--------|
|
||||
| 2026-06-22 | v1.0 | 初始创建 | 陆怀瑾 |
|
||||
@@ -0,0 +1,83 @@
|
||||
# 淘宝运营 SOP
|
||||
|
||||
## 元数据
|
||||
|
||||
| 属性 | 值 |
|
||||
|------|-----|
|
||||
| **领域** | 电商 |
|
||||
| **责任人** | 陆云帆 (taobaospecialist) |
|
||||
| **版本** | v1.0 |
|
||||
| **创建日期** | 2026-06-22 |
|
||||
| **标签** | 淘宝, 电商, SOP, 日常运营 |
|
||||
|
||||
## 概述
|
||||
|
||||
本 SOP 定义淘宝店铺日常运营的标准流程，涵盖店铺维护、商品管理、营销推广、客服处理和数据分析五大模块。适用于每日/每周/每月周期性执行。
|
||||
|
||||
## 正文
|
||||
|
||||
### 一、每日运营检查（每日 9:00）
|
||||
|
||||
1. **店铺状态检查**
|
||||
- 登录千牛工作台，检查店铺处罚/违规通知
|
||||
- 确认所有商品在售状态，无异常下架
|
||||
- 检查店铺评分（DSR），低于 4.7 需立即分析原因
|
||||
|
||||
2. **订单处理**
|
||||
- 查看待发货订单，确保 48 小时内发货
|
||||
- 处理售后订单（退货/换货/退款），24 小时内响应
|
||||
- 检查差评/中评，及时联系客户处理
|
||||
|
||||
3. **客服响应**
|
||||
- 检查未读消息，回复时限 5 分钟内
|
||||
- 查看客服数据：响应时长、满意度
|
||||
|
||||
### 二、每周运营任务（每周一）
|
||||
|
||||
1. **商品优化**
|
||||
- 检查 Top 10 商品标题、主图、详情页
|
||||
- 根据搜索词报告优化标题关键词
|
||||
- 更新库存不足的商品
|
||||
|
||||
2. **营销活动**
|
||||
- 查看本周淘宝官方活动日历
|
||||
- 设置店铺优惠券/满减活动
|
||||
- 更新直通车/引力魔方推广计划
|
||||
|
||||
3. **数据分析**
|
||||
- 查看流量来源（搜索/推荐/付费/其他）
|
||||
- 分析转化率、客单价变化趋势
|
||||
- 输出《店铺周报》
|
||||
|
||||
### 三、每月运营任务（每月 1 日）
|
||||
|
||||
1. **月度复盘**
|
||||
- 统计月度 GMV、订单量、利润率
|
||||
- 对比上月数据，分析增长/下滑原因
|
||||
- 制定下月运营目标和策略
|
||||
|
||||
2. **竞品分析**
|
||||
- 监控 Top 3 竞品店铺动态
|
||||
- 分析竞品爆款商品和新品
|
||||
- 调整自身商品/价格策略
|
||||
|
||||
### 四、关键指标
|
||||
|
||||
| 指标 | 目标值 | 监控频率 |
|
||||
|------|--------|----------|
|
||||
| DSR 评分 | ≥ 4.8 | 每日 |
|
||||
| 48h 发货率 | ≥ 98% | 每日 |
|
||||
| 客服响应时长 | ≤ 3 分钟 | 每日 |
|
||||
| 转化率 | ≥ 行业均值 +10% | 每周 |
|
||||
| GMV 增长 | 月环比 ≥ 10% | 每月 |
|
||||
|
||||
## 相关条目
|
||||
|
||||
- [抖店运营SOP.md](抖店运营SOP.md)
|
||||
- [数据分析方法.md](../运营/数据分析方法.md)
|
||||
|
||||
## 变更记录
|
||||
|
||||
| 日期 | 版本 | 变更说明 | 变更人 |
|
||||
|------|------|----------|--------|
|
||||
| 2026-06-22 | v1.0 | 初始创建 | 陆怀瑾 |
|
||||
@@ -0,0 +1,21 @@
|
||||
# 行政领域知识
|
||||
|
||||
**责任人**：刘诗妮（secretary）
|
||||
**审核人**：陆怀瑾（coo）
|
||||
|
||||
## 知识范围
|
||||
|
||||
涵盖合同管理、报销流程、行政事务、供应商管理等行政支持知识。
|
||||
|
||||
## 条目清单
|
||||
|
||||
| 文件名 | 说明 | 状态 |
|
||||
|--------|------|------|
|
||||
| [合同模板.md](合同模板.md) | 常用合同标准模板 | ✅ |
|
||||
| [报销流程.md](报销流程.md) | 费用报销申请与审批流程 | ✅ |
|
||||
|
||||
## 待建设
|
||||
|
||||
- 供应商管理指南
|
||||
- 会议纪要模板
|
||||
- 入职/离职流程
|
||||
@@ -0,0 +1,83 @@
|
||||
# 报销流程
|
||||
|
||||
## 元数据
|
||||
|
||||
| 属性 | 值 |
|
||||
|------|-----|
|
||||
| **领域** | 行政 |
|
||||
| **责任人** | 刘诗妮 (secretary) |
|
||||
| **版本** | v1.0 |
|
||||
| **创建日期** | 2026-06-22 |
|
||||
| **标签** | 行政, 报销, 财务, 流程 |
|
||||
|
||||
## 概述
|
||||
|
||||
定义公司费用报销的标准流程，涵盖申请、审批、核销三大阶段，确保财务合规性和报销效率。
|
||||
|
||||
## 正文
|
||||
|
||||
### 一、报销范围
|
||||
|
||||
| 类别 | 说明 | 限额 |
|
||||
|------|------|------|
|
||||
| 差旅费 | 交通、住宿、餐饮 | 按出差地标准 |
|
||||
| 办公用品 | 设备、耗材、文具 | 单次 ≤ ¥2000 |
|
||||
| 招待费 | 客户/合作伙伴接待 | 需提前申请 |
|
||||
| 培训费 | 课程、考试、认证 | 需审批 |
|
||||
| 软件服务 | SaaS 订阅、API 费用 | 按需审批 |
|
||||
|
||||
### 二、报销流程
|
||||
|
||||
```
|
||||
提交申请 → 直属审批 → COO 审批（> ¥5000）
|
||||
→ 刘总审批（> ¥20000） → 刘诗妮核销 → 归档
|
||||
```
|
||||
|
||||
**各环节时限**：
|
||||
- 员工提交：消费后 7 个工作日内
|
||||
- 直属审批：2 个工作日内
|
||||
- 核销：审批通过后 5 个工作日内
|
||||
|
||||
### 三、报销材料
|
||||
|
||||
1. **发票**
|
||||
- 必须增值税普通/专用发票
|
||||
- 发票抬头：公司全称 + 税号
|
||||
- 电子发票可，纸质发票需原件
|
||||
|
||||
2. **报销单**
|
||||
- 事由：清晰说明消费目的
|
||||
- 明细：逐项列出费用+金额
|
||||
- 附件上传：发票图片/电子凭证
|
||||
|
||||
3. **特殊说明**
|
||||
- 差旅：附行程单
|
||||
- 招待：附参与人员名单
|
||||
- 大额采购：附比价记录
|
||||
|
||||
### 四、常见退回原因
|
||||
|
||||
| 原因 | 处理 |
|
||||
|------|------|
|
||||
| 发票信息错误（抬头/税号） | 退回重新开票 |
|
||||
| 超额未提前审批 | 补充说明或自付超额部分 |
|
||||
| 缺少明细说明 | 补充报销单信息 |
|
||||
| 超过报销时效 | 特殊说明后处理 |
|
||||
|
||||
### 五、审批人
|
||||
|
||||
| 金额区间 | 审批人 |
|
||||
|----------|--------|
|
||||
| ≤ ¥5000 | 直属负责人 |
|
||||
| ¥5001 ~ ¥20000 | + COO（陆怀瑾） |
|
||||
| > ¥20000 | + 刘总（Vincent） |
|
||||
|
||||
## 相关条目
|
||||
|
||||
- [合同模板.md](合同模板.md)
|
||||
|
||||
## 变更记录
|
||||
|
||||
| 日期 | 版本 | 变更说明 | 变更人 |
|
||||
|------|------|----------|--------|
|
||||
| 2026-06-22 | v1.0 | 初始创建 | 陆怀瑾 |
|
||||
@@ -0,0 +1,21 @@
|
||||
# 设计领域知识
|
||||
|
||||
**责任人**：苏绘锦（designer）
|
||||
**审核人**：陆怀瑾（coo）
|
||||
|
||||
## 知识范围
|
||||
|
||||
涵盖 UI/UX 设计规范、品牌元素、商详页设计、首图制作等设计知识。
|
||||
|
||||
## 条目清单
|
||||
|
||||
| 文件名 | 说明 | 状态 |
|
||||
|--------|------|------|
|
||||
| [UI设计规范.md](UI设计规范.md) | 界面设计标准与组件规范 | ✅ |
|
||||
| [品牌元素指南.md](品牌元素指南.md) | 品牌色/字体/Logo 使用规范 | ✅ |
|
||||
|
||||
## 待建设
|
||||
|
||||
- 商详页设计模板
|
||||
- 首图设计规范
|
||||
- 移动端适配指南
|
||||
@@ -0,0 +1,87 @@
|
||||
# UI 设计规范
|
||||
|
||||
## 元数据
|
||||
|
||||
| 属性 | 值 |
|
||||
|------|-----|
|
||||
| **领域** | 设计 |
|
||||
| **责任人** | 苏绘锦 (designer) |
|
||||
| **版本** | v1.0 |
|
||||
| **创建日期** | 2026-06-22 |
|
||||
| **标签** | UI, 设计规范, 组件, 视觉 |
|
||||
|
||||
## 概述
|
||||
|
||||
定义统一的 UI 设计标准，涵盖色彩、字体、间距、组件等基础规范，确保产品视觉一致性，降低设计-开发沟通成本。
|
||||
|
||||
## 正文
|
||||
|
||||
### 一、色彩系统
|
||||
|
||||
| 用途 | 色值 | 说明 |
|
||||
|------|------|------|
|
||||
| 主色 Primary | `#1677FF` | 按钮、链接、选中态 |
|
||||
| 成功 Success | `#52C41A` | 成功提示、通过状态 |
|
||||
| 警告 Warning | `#FAAD14` | 警告提示 |
|
||||
| 错误 Error | `#FF4D4F` | 错误提示、删除操作 |
|
||||
| 文字主色 | `#1F1F1F` | 标题、正文 |
|
||||
| 文字次色 | `#666666` | 辅助说明 |
|
||||
| 文字禁用 | `#BFBFBF` | 禁用/占位符 |
|
||||
| 边框 | `#D9D9D9` | 分割线、输入框边框 |
|
||||
| 背景 | `#F5F5F5` | 页面底色 |
|
||||
|
||||
### 二、字体规范
|
||||
|
||||
| 层级 | 字号 | 行高 | 字重 | 用途 |
|
||||
|------|------|------|------|------|
|
||||
| H1 | 24px | 32px | 600 | 页面主标题 |
|
||||
| H2 | 20px | 28px | 600 | 区块标题 |
|
||||
| H3 | 16px | 24px | 500 | 小标题 |
|
||||
| Body | 14px | 22px | 400 | 正文 |
|
||||
| Caption | 12px | 18px | 400 | 辅助/说明文字 |
|
||||
|
||||
默认字体：`-apple-system, BlinkMacSystemFont, "Segoe UI", "PingFang SC", "Microsoft YaHei", sans-serif`
|
||||
|
||||
### 三、间距体系
|
||||
|
||||
采用 4px 为基础单位的 8 点栅格：
|
||||
|
||||
| Token | 值 | 用途 |
|
||||
|-------|-----|------|
|
||||
| xs | 4px | 紧凑间距 |
|
||||
| sm | 8px | 元素内间距 |
|
||||
| md | 16px | 组件间距 |
|
||||
| lg | 24px | 区块间距 |
|
||||
| xl | 32px | 大区块分隔 |
|
||||
| xxl | 48px | 页面级分隔 |
|
||||
|
||||
### 四、圆角与阴影
|
||||
|
||||
| 组件 | 圆角 | 阴影 |
|
||||
|------|------|------|
|
||||
| 卡片 | 8px | `0 2px 8px rgba(0,0,0,0.08)` |
|
||||
| 弹窗 | 12px | `0 6px 16px rgba(0,0,0,0.12)` |
|
||||
| 按钮 | 6px | 无（默认）/ hover 时微阴影 |
|
||||
| 输入框 | 6px | 无（默认）/ focus 时外发光 |
|
||||
|
||||
### 五、响应式断点
|
||||
|
||||
| 断点 | 最小宽度 | 适用设备 |
|
||||
|------|----------|----------|
|
||||
| xs | < 576px | 手机竖屏 |
|
||||
| sm | ≥ 576px | 手机横屏 |
|
||||
| md | ≥ 768px | 平板 |
|
||||
| lg | ≥ 992px | 小桌面 |
|
||||
| xl | ≥ 1200px | 大桌面 |
|
||||
| xxl | ≥ 1600px | 超大屏 |
|
||||
|
||||
## 相关条目
|
||||
|
||||
- [品牌元素指南.md](品牌元素指南.md)
|
||||
- [PRD模板.md](../产品/PRD模板.md)
|
||||
|
||||
## 变更记录
|
||||
|
||||
| 日期 | 版本 | 变更说明 | 变更人 |
|
||||
|------|------|----------|--------|
|
||||
| 2026-06-22 | v1.0 | 初始创建 | 陆怀瑾 |
|
||||
@@ -0,0 +1,21 @@
|
||||
# 运营领域知识
|
||||
|
||||
**责任人**：陆怀瑾（coo）
|
||||
**审核人**：刘炜承（Vincent）
|
||||
|
||||
## 知识范围
|
||||
|
||||
涵盖活动策划、数据分析、SOP 管理、流程优化、团队协作等运营管理知识。
|
||||
|
||||
## 条目清单
|
||||
|
||||
| 文件名 | 说明 | 状态 |
|
||||
|--------|------|------|
|
||||
| [活动策划模板.md](活动策划模板.md) | 营销活动策划标准模板 | ✅ |
|
||||
| [数据分析方法.md](数据分析方法.md) | 运营数据分析框架与方法 | ✅ |
|
||||
|
||||
## 待建设
|
||||
|
||||
- 周报模板
|
||||
- KPI 管理框架
|
||||
- 风险评估矩阵
|
||||
@@ -0,0 +1,92 @@
|
||||
# 数据分析方法
|
||||
|
||||
## 元数据
|
||||
|
||||
| 属性 | 值 |
|
||||
|------|-----|
|
||||
| **领域** | 运营 |
|
||||
| **责任人** | 陆怀瑾 (coo) |
|
||||
| **版本** | v1.0 |
|
||||
| **创建日期** | 2026-06-22 |
|
||||
| **标签** | 数据分析, 运营, KPI, 看板 |
|
||||
|
||||
## 概述
|
||||
|
||||
建立全业务流程的数据分析框架，确保各业务线有统一的指标定义和分析方法，支撑数据驱动决策。
|
||||
|
||||
## 正文
|
||||
|
||||
### 一、核心指标体系
|
||||
|
||||
#### 1. 电商业务
|
||||
|
||||
| 层级 | 指标 | 定义 | 频率 |
|
||||
|------|------|------|------|
|
||||
| 北极星 | GMV | 总成交额 | 日/周/月 |
|
||||
| 过程 | 转化率 | 下单数/访客数 | 日 |
|
||||
| 过程 | 客单价 | GMV/订单数 | 周 |
|
||||
| 过程 | 退货率 | 退货数/订单数 | 周 |
|
||||
| 健康 | DSR | 描述/服务/物流评分 | 日 |
|
||||
| 健康 | 获客成本 CAC | 营销花费/新客数 | 月 |
|
||||
|
||||
#### 2. 内容业务
|
||||
|
||||
| 层级 | 指标 | 定义 | 频率 |
|
||||
|------|------|------|------|
|
||||
| 北极星 | 粉丝增长 | 净增粉丝数 | 周/月 |
|
||||
| 过程 | 互动率 | (点赞+收藏+评论)/曝光 | 篇 |
|
||||
| 过程 | 发布频率 | 每周发布篇数 | 周 |
|
||||
|
||||
#### 3. 公司整体
|
||||
|
||||
| 层级 | 指标 | 定义 | 频率 |
|
||||
|------|------|------|------|
|
||||
| 北极星 | 月营收 | 各业务线收入合计 | 月 |
|
||||
| 效率 | 人效 | 营收/团队人数 | 季 |
|
||||
| 效率 | Agent 利用率 | Agent 任务完成数/总分配数 | 周 |
|
||||
|
||||
### 二、分析框架
|
||||
|
||||
**AARRR 海盗模型**：
|
||||
|
||||
```
|
||||
Acquisition（获取）→ Activation（激活）→ Retention（留存）
|
||||
→ Revenue（收入）→ Referral（推荐）
|
||||
```
|
||||
|
||||
**电商应用示例**：
|
||||
1. Acquisition：各渠道流量来源占比
|
||||
2. Activation：首次下单转化率
|
||||
3. Retention：30 天复购率
|
||||
4. Revenue：LTV（用户生命周期价值）
|
||||
5. Referral：分享率、裂变系数
|
||||
|
||||
### 三、数据看板要求
|
||||
|
||||
每个业务线需维护以下看板：
|
||||
|
||||
| 看板 | 内容 | 更新频率 |
|
||||
|------|------|----------|
|
||||
| 日报 | 昨日核心指标 + 异常波动标注 | 每日 10:00 |
|
||||
| 周报 | 趋势图 + 同比/环比 + 分析洞察 | 每周一 |
|
||||
| 月报 | 完整指标矩阵 + 目标达成率 + 下月预测 | 每月 3 日 |
|
||||
|
||||
### 四、异常预警规则
|
||||
|
||||
| 条件 | 级别 | 响应 |
|
||||
|------|------|------|
|
||||
| GMV 日环比下降 > 20% | 🔴 | COO 立即介入 |
|
||||
| 转化率连续 3 天下降 | 🟡 | 业务负责人分析 |
|
||||
| 退货率 > 10% | 🟡 | 商品/客服联合排查 |
|
||||
| DSR < 4.6 | 🔴 | 立即优化 |
|
||||
|
||||
## 相关条目
|
||||
|
||||
- [淘宝运营SOP.md](../电商/淘宝运营SOP.md)
|
||||
- [活动策划模板.md](活动策划模板.md)
|
||||
|
||||
## 变更记录
|
||||
|
||||
| 日期 | 版本 | 变更说明 | 变更人 |
|
||||
|------|------|----------|--------|
|
||||
| 2026-06-22 | v1.0 | 初始创建 | 陆怀瑾 |
|
||||
@@ -0,0 +1,80 @@
|
||||
# 活动策划模板
|
||||
|
||||
## 元数据
|
||||
|
||||
| 属性 | 值 |
|
||||
|------|-----|
|
||||
| **领域** | 运营 |
|
||||
| **责任人** | 陆怀瑾 (coo) |
|
||||
| **版本** | v1.0 |
|
||||
| **创建日期** | 2026-06-22 |
|
||||
| **标签** | 运营, 活动策划, 营销, 模板 |
|
||||
|
||||
## 概述
|
||||
|
||||
标准化营销活动策划流程。适用于电商大促（618/双11/年货节）、店铺周年庆、新品发布、会员日活动等。
|
||||
|
||||
## 正文
|
||||
|
||||
### 一、活动策划文档结构
|
||||
|
||||
```
|
||||
# [活动名称] 策划方案
|
||||
|
||||
## 1. 活动背景与目标
|
||||
- 背景：[为什么做这次活动]
|
||||
- 核心目标：[可量化，如 GMV X万 / 新增粉丝 Y人]
|
||||
- 次要目标：[如品牌曝光、老客复购]
|
||||
|
||||
## 2. 目标用户
|
||||
- 主要人群：[画像描述]
|
||||
- 需求动机：[为什么他们会参与]
|
||||
|
||||
## 3. 活动机制
|
||||
- 玩法规则：[满减/秒杀/抽奖/打卡]
|
||||
- 用户路径：[从看到到参与的完整链路]
|
||||
- 激励机制：[优惠力度、稀缺性、社交裂变]
|
||||
|
||||
## 4. 资源与预算
|
||||
| 项目 | 预算 | 负责人 |
|
||||
|------|------|--------|
|
||||
| 流量投放 | ¥XX | [姓名] |
|
||||
| 商品补贴 | ¥XX | [姓名] |
|
||||
| 内容物料 | ¥XX | [姓名] |
|
||||
|
||||
## 5. 时间线
|
||||
| 阶段 | 时间 | 关键事项 |
|
||||
|------|------|----------|
|
||||
| 预热期 | D-7 ~ D-1 | 预告内容、优惠券发放 |
|
||||
| 爆发期 | D-Day | 主活动上线 |
|
||||
| 返场期 | D+1 ~ D+3 | 余热运营 |
|
||||
|
||||
## 6. 风险预案
|
||||
| 风险 | 概率 | 影响 | 应对 |
|
||||
|------|------|------|------|
|
||||
| 服务器崩溃 | 中 | 高 | 提前压测 + 降级方案 |
|
||||
| 库存不足 | 低 | 中 | 预售 + 安全库存预警 |
|
||||
|
||||
## 7. 复盘框架
|
||||
- 活动数据回顾（GMV/ROI/客单价/转化率）
|
||||
- 亮点与不足
|
||||
- 优化建议
|
||||
```
|
||||
|
||||
### 二、关键审批节点
|
||||
|
||||
1. **策划方案评审** → COO + 业务负责人
|
||||
2. **预算审批** → Vincent
|
||||
3. **法务合规审查** → 苏慎（如涉及抽奖/满赠）
|
||||
4. **上线前 Checklist** → 所有执行人确认
|
||||
|
||||
## 相关条目
|
||||
|
||||
- [数据分析方法.md](数据分析方法.md)
|
||||
- [淘宝运营SOP.md](../电商/淘宝运营SOP.md)
|
||||
|
||||
## 变更记录
|
||||
|
||||
| 日期 | 版本 | 变更说明 | 变更人 |
|
||||
|------|------|----------|--------|
|
||||
| 2026-06-22 | v1.0 | 初始创建 | 陆怀瑾 |
|
||||
@@ -0,0 +1,50 @@
|
||||
# Alertmanager 配置
|
||||
# 告警通知路由到 Feishu
|
||||
|
||||
global:
|
||||
resolve_timeout: 5m
|
||||
|
||||
route:
|
||||
receiver: "default"
|
||||
group_wait: 30s
|
||||
group_interval: 5m
|
||||
repeat_interval: 4h
|
||||
routes:
|
||||
# 严重告警 → 通知 Vincent
|
||||
- receiver: "vincent-critical"
|
||||
match:
|
||||
severity: critical
|
||||
repeat_interval: 2h
|
||||
continue: true
|
||||
|
||||
# 警告告警 → 通知 COO
|
||||
- receiver: "coo-warning"
|
||||
match:
|
||||
severity: warning
|
||||
repeat_interval: 4h
|
||||
|
||||
receivers:
|
||||
- name: "default"
|
||||
webhook_configs:
|
||||
- url: "http://host.docker.internal:9094/webhook"
|
||||
send_resolved: true
|
||||
|
||||
- name: "vincent-critical"
|
||||
webhook_configs:
|
||||
- url: "http://host.docker.internal:9094/webhook"
|
||||
send_resolved: true
|
||||
|
||||
- name: "coo-warning"
|
||||
webhook_configs:
|
||||
- url: "http://host.docker.internal:9094/webhook"
|
||||
send_resolved: true
|
||||
|
||||
# 抑制规则：严重告警自动抑制同源的警告
|
||||
inhibit_rules:
|
||||
- source_match:
|
||||
severity: critical
|
||||
target_match:
|
||||
severity: warning
|
||||
equal:
|
||||
- alertname
|
||||
- instance
|
||||
@@ -0,0 +1,288 @@
|
||||
{
|
||||
"title": "OpenClaw Agent Health Dashboard",
|
||||
"uid": "agent-health",
|
||||
"version": 1,
|
||||
"tags": ["openclaw", "agent", "monitoring"],
|
||||
"timezone": "browser",
|
||||
"editable": true,
|
||||
"refresh": "30s",
|
||||
"panels": [
|
||||
{
|
||||
"title": "系统资源概览",
|
||||
"type": "row",
|
||||
"gridPos": {"h": 1, "w": 24, "x": 0, "y": 0}
|
||||
},
|
||||
{
|
||||
"id": 1,
|
||||
"title": "CPU 使用率",
|
||||
"type": "gauge",
|
||||
"gridPos": {"h": 8, "w": 6, "x": 0, "y": 1},
|
||||
"targets": [
|
||||
{
|
||||
"expr": "100 - (avg by(instance) (rate(node_cpu_seconds_total{mode=\"idle\"}[5m])) * 100)",
|
||||
"legendFormat": "{{instance}}"
|
||||
}
|
||||
],
|
||||
"options": {
|
||||
"reduceOptions": {"calcs": ["lastNotNull"]},
|
||||
"showThresholdLabels": false,
|
||||
"showThresholdMarkers": true
|
||||
},
|
||||
"thresholds": [
|
||||
{"color": "green", "value": null},
|
||||
{"color": "yellow", "value": 70},
|
||||
{"color": "red", "value": 90}
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": 2,
|
||||
"title": "内存使用率",
|
||||
"type": "gauge",
|
||||
"gridPos": {"h": 8, "w": 6, "x": 6, "y": 1},
|
||||
"targets": [
|
||||
{
|
||||
"expr": "(1 - (node_memory_MemAvailable_bytes / node_memory_MemTotal_bytes)) * 100",
|
||||
"legendFormat": "{{instance}}"
|
||||
}
|
||||
],
|
||||
"options": {
|
||||
"reduceOptions": {"calcs": ["lastNotNull"]},
|
||||
"showThresholdLabels": false,
|
||||
"showThresholdMarkers": true
|
||||
},
|
||||
"thresholds": [
|
||||
{"color": "green", "value": null},
|
||||
{"color": "yellow", "value": 80},
|
||||
{"color": "red", "value": 95}
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": 3,
|
||||
"title": "磁盘使用率",
|
||||
"type": "gauge",
|
||||
"gridPos": {"h": 8, "w": 6, "x": 12, "y": 1},
|
||||
"targets": [
|
||||
{
|
||||
"expr": "max by(instance) ((node_filesystem_size_bytes - node_filesystem_free_bytes) / node_filesystem_size_bytes * 100)",
|
||||
"legendFormat": "{{instance}}"
|
||||
}
|
||||
],
|
||||
"options": {
|
||||
"reduceOptions": {"calcs": ["lastNotNull"]},
|
||||
"showThresholdLabels": false,
|
||||
"showThresholdMarkers": true
|
||||
},
|
||||
"thresholds": [
|
||||
{"color": "green", "value": null},
|
||||
{"color": "yellow", "value": 80},
|
||||
{"color": "red", "value": 95}
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": 4,
|
||||
"title": "系统负载",
|
||||
"type": "stat",
|
||||
"gridPos": {"h": 8, "w": 6, "x": 18, "y": 1},
|
||||
"targets": [
|
||||
{
|
||||
"expr": "node_load1",
|
||||
"legendFormat": "1min"
|
||||
},
|
||||
{
|
||||
"expr": "node_load5",
|
||||
"legendFormat": "5min"
|
||||
},
|
||||
{
|
||||
"expr": "node_load15",
|
||||
"legendFormat": "15min"
|
||||
}
|
||||
],
|
||||
"options": {
|
||||
"reduceOptions": {"calcs": ["lastNotNull"]},
|
||||
"colorMode": "background",
|
||||
"graphMode": "area",
|
||||
"justifyMode": "auto",
|
||||
"orientation": "horizontal",
|
||||
"textMode": "auto"
|
||||
}
|
||||
},
|
||||
{
|
||||
"title": "Agent 健康状态",
|
||||
"type": "row",
|
||||
"gridPos": {"h": 1, "w": 24, "x": 0, "y": 9}
|
||||
},
|
||||
{
|
||||
"id": 5,
|
||||
"title": "Agent 心跳状态",
|
||||
"type": "table",
|
||||
"gridPos": {"h": 8, "w": 12, "x": 0, "y": 10},
|
||||
"targets": [
|
||||
{
|
||||
"expr": "agent_heartbeat_status",
|
||||
"legendFormat": "{{agent_label}}"
|
||||
}
|
||||
],
|
||||
"transformations": [
|
||||
{"id": "organize", "options": {"excludeByName": {}, "indexByName": {}, "renameByName": {"Value": "状态"}}}
|
||||
],
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"custom": {
|
||||
"align": "center",
|
||||
"displayMode": "color-background"
|
||||
},
|
||||
"mappings": [
|
||||
{"type": "value", "options": {"0": {"color": "red", "text": "❌ 超时"}, "1": {"color": "green", "text": "✅ 正常"}}}
|
||||
],
|
||||
"thresholds": [{"color": "green", "value": null}]
|
||||
}
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": 6,
|
||||
"title": "任务停滞时长",
|
||||
"type": "bargauge",
|
||||
"gridPos": {"h": 8, "w": 12, "x": 12, "y": 10},
|
||||
"targets": [
|
||||
{
|
||||
"expr": "agent_task_stagnation_seconds",
|
||||
"legendFormat": "{{agent_label}}"
|
||||
}
|
||||
],
|
||||
"options": {
|
||||
"orientation": "horizontal",
|
||||
"displayMode": "gradient",
|
||||
"showUnfilled": true
|
||||
},
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"unit": "s",
|
||||
"thresholds": [
|
||||
{"color": "green", "value": null},
|
||||
{"color": "yellow", "value": 3600},
|
||||
{"color": "red", "value": 14400}
|
||||
]
|
||||
}
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": 7,
|
||||
"title": "待办任务数",
|
||||
"type": "stat",
|
||||
"gridPos": {"h": 4, "w": 6, "x": 0, "y": 18},
|
||||
"targets": [
|
||||
{
|
||||
"expr": "agent_workboard_pending",
|
||||
"legendFormat": "待办任务"
|
||||
}
|
||||
],
|
||||
"options": {
|
||||
"reduceOptions": {"calcs": ["lastNotNull"]},
|
||||
"colorMode": "background",
|
||||
"graphMode": "area",
|
||||
"textMode": "auto"
|
||||
},
|
||||
"thresholds": [
|
||||
{"color": "green", "value": null},
|
||||
{"color": "yellow", "value": 5},
|
||||
{"color": "red", "value": 10}
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": 8,
|
||||
"title": "429 错误计数",
|
||||
"type": "stat",
|
||||
"gridPos": {"h": 4, "w": 6, "x": 6, "y": 18},
|
||||
"targets": [
|
||||
{
|
||||
"expr": "agent_429_error_rate",
|
||||
"legendFormat": "429 错误"
|
||||
}
|
||||
],
|
||||
"options": {
|
||||
"reduceOptions": {"calcs": ["lastNotNull"]},
|
||||
"colorMode": "background",
|
||||
"graphMode": "area",
|
||||
"textMode": "auto"
|
||||
},
|
||||
"thresholds": [
|
||||
{"color": "green", "value": null},
|
||||
{"color": "yellow", "value": 10},
|
||||
{"color": "red", "value": 50}
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": 9,
|
||||
"title": "Prometheus 目标状态",
|
||||
"type": "table",
|
||||
"gridPos": {"h": 8, "w": 12, "x": 12, "y": 18},
|
||||
"targets": [
|
||||
{
|
||||
"expr": "up",
|
||||
"legendFormat": "{{job}} ({{instance}})"
|
||||
}
|
||||
],
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"custom": {"align": "center", "displayMode": "color-background"},
|
||||
"mappings": [
|
||||
{"type": "value", "options": {"0": {"color": "red", "text": "❌ Down"}, "1": {"color": "green", "text": "✅ Up"}}}
|
||||
]
|
||||
}
|
||||
}
|
||||
},
|
||||
{
|
||||
"title": "告警状态",
|
||||
"type": "row",
|
||||
"gridPos": {"h": 1, "w": 24, "x": 0, "y": 26}
|
||||
},
|
||||
{
|
||||
"id": 10,
|
||||
"title": "活跃告警",
|
||||
"type": "table",
|
||||
"gridPos": {"h": 8, "w": 24, "x": 0, "y": 27},
|
||||
"targets": [
|
||||
{
|
||||
"expr": "ALERTS{alertstate=\"firing\"}",
|
||||
"legendFormat": "{{alertname}}"
|
||||
}
|
||||
],
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"custom": {"align": "left"},
|
||||
"mappings": [
|
||||
{"type": "value", "options": {"0": {"color": "green", "text": "已恢复"}, "1": {"color": "red", "text": "触发中"}}}
|
||||
]
|
||||
}
|
||||
}
|
||||
}
|
||||
],
|
||||
"schemaVersion": 38,
|
||||
"style": "dark",
|
||||
"tags": ["openclaw", "agent", "monitoring"],
|
||||
"templating": {
|
||||
"list": [
|
||||
{
|
||||
"name": "datasource",
|
||||
"type": "datasource",
|
||||
"query": "prometheus",
|
||||
"current": {"value": "Prometheus"}
|
||||
}
|
||||
]
|
||||
},
|
||||
"annotations": {
|
||||
"list": [
|
||||
{
|
||||
"name": "告警事件",
|
||||
"type": "dashboard",
|
||||
"builtIn": 1,
|
||||
"datasource": {"type": "prometheus", "uid": "PBFA97CFB590B2093"},
|
||||
"enable": true,
|
||||
"hide": true,
|
||||
"iconColor": "rgba(255, 96, 96, 1)",
|
||||
"expr": "ALERTS",
|
||||
"step": "60s"
|
||||
}
|
||||
]
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,12 @@
|
||||
apiVersion: 1
|
||||
|
||||
providers:
|
||||
- name: "Agent Health"
|
||||
orgId: 1
|
||||
folder: "OpenClaw"
|
||||
type: file
|
||||
disableDeletion: false
|
||||
editable: true
|
||||
updateIntervalSeconds: 10
|
||||
options:
|
||||
path: /etc/grafana/provisioning/dashboards
|
||||
@@ -0,0 +1,42 @@
|
||||
global:
|
||||
scrape_interval: 15s
|
||||
evaluation_interval: 15s
|
||||
|
||||
# Alertmanager 配置
|
||||
alerting:
|
||||
alertmanagers:
|
||||
- static_configs:
|
||||
- targets:
|
||||
- alertmanager:9093
|
||||
|
||||
# 规则文件
|
||||
rule_files:
|
||||
- "agent_alerts.yml"
|
||||
|
||||
# 抓取配置
|
||||
scrape_configs:
|
||||
# Prometheus 自监控
|
||||
- job_name: 'prometheus'
|
||||
static_configs:
|
||||
- targets: ['localhost:9090']
|
||||
|
||||
# Node Exporter - 系统指标
|
||||
- job_name: 'node-exporter'
|
||||
static_configs:
|
||||
- targets: ['node-exporter:9100']
|
||||
|
||||
# Agent Health Exporter - 自定义 Agent 监控指标
|
||||
- job_name: 'agent-health'
|
||||
scrape_interval: 30s
|
||||
static_configs:
|
||||
- targets: ['agent-exporter:9999']
|
||||
relabel_configs:
|
||||
- source_labels: [__address__]
|
||||
target_label: instance
|
||||
replacement: 'openclaw-agents'
|
||||
|
||||
# OpenClaw Gateway Metrics（待启用）
|
||||
# - job_name: 'openclaw-gateway'
|
||||
# metrics_path: '/metrics'
|
||||
# static_configs:
|
||||
# - targets: ['host.docker.internal:18789']
|
||||
@@ -0,0 +1,92 @@
|
||||
version: '3.8'
|
||||
|
||||
services:
|
||||
prometheus:
|
||||
image: m.daocloud.io/docker.io/prom/prometheus:v2.52.0
|
||||
container_name: prometheus
|
||||
ports:
|
||||
- "9090:9090"
|
||||
volumes:
|
||||
- ./config/prometheus.yml:/etc/prometheus/prometheus.yml
|
||||
- ./config/agent_alerts.yml:/etc/prometheus/agent_alerts.yml
|
||||
- ./data/prometheus:/prometheus
|
||||
extra_hosts:
|
||||
- "host.docker.internal:host-gateway"
|
||||
command:
|
||||
- '--config.file=/etc/prometheus/prometheus.yml'
|
||||
- '--storage.tsdb.path=/prometheus'
|
||||
- '--web.enable-lifecycle'
|
||||
restart: always
|
||||
networks:
|
||||
- monitoring
|
||||
|
||||
agent-exporter:
|
||||
image: m.daocloud.io/docker.io/python:3.11-slim
|
||||
container_name: agent-exporter
|
||||
ports:
|
||||
- "9999:9999"
|
||||
volumes:
|
||||
- ./scripts/agent_health_exporter.py:/app/exporter.py:ro
|
||||
command: python3 /app/exporter.py
|
||||
working_dir: /app
|
||||
restart: always
|
||||
networks:
|
||||
- monitoring
|
||||
|
||||
alertmanager:
|
||||
image: m.daocloud.io/docker.io/prom/alertmanager:v0.27.0
|
||||
container_name: alertmanager
|
||||
ports:
|
||||
- "9093:9093"
|
||||
volumes:
|
||||
- ./config/alertmanager.yml:/etc/alertmanager/alertmanager.yml
|
||||
- ./data/alertmanager:/alertmanager
|
||||
extra_hosts:
|
||||
- "host.docker.internal:host-gateway"
|
||||
command:
|
||||
- '--config.file=/etc/alertmanager/alertmanager.yml'
|
||||
- '--storage.path=/alertmanager'
|
||||
- '--web.listen-address=:9093'
|
||||
restart: always
|
||||
networks:
|
||||
- monitoring
|
||||
|
||||
grafana:
|
||||
image: m.daocloud.io/docker.io/grafana/grafana:11.0.0
|
||||
container_name: grafana
|
||||
ports:
|
||||
- "3001:3000"
|
||||
environment:
|
||||
- GF_SECURITY_ADMIN_USER=admin
|
||||
- GF_SECURITY_ADMIN_PASSWORD=***
|
||||
- GF_INSTALL_PLUGINS=grafana-clock-panel,grafana-piechart-panel
|
||||
volumes:
|
||||
- ./data/grafana:/var/lib/grafana
|
||||
- ./config/grafana/dashboards:/etc/grafana/provisioning/dashboards
|
||||
- ./config/grafana/datasources:/etc/grafana/provisioning/datasources
|
||||
restart: always
|
||||
networks:
|
||||
- monitoring
|
||||
depends_on:
|
||||
- prometheus
|
||||
|
||||
node-exporter:
|
||||
image: m.daocloud.io/docker.io/prom/node-exporter:v1.8.2
|
||||
container_name: node-exporter
|
||||
ports:
|
||||
- "9100:9100"
|
||||
volumes:
|
||||
- /proc:/host/proc:ro
|
||||
- /sys:/host/sys:ro
|
||||
- /:/rootfs:ro
|
||||
command:
|
||||
- '--path.procfs=/host/proc'
|
||||
- '--path.sysfs=/host/sys'
|
||||
- '--collector.filesystem.mount-points-exclude=^/(sys|proc|dev|host|etc)($|/)'
|
||||
restart: always
|
||||
networks:
|
||||
- monitoring
|
||||
|
||||
networks:
|
||||
monitoring:
|
||||
driver: bridge
|
||||
@@ -0,0 +1,180 @@
|
||||
#!/usr/bin/env python3
|
||||
"""
|
||||
OpenClaw Agent Health Exporter v2.1
|
||||
采集 Agent 运行指标，暴露给 Prometheus 抓取
|
||||
|
||||
设计原则：
|
||||
- HTTP handler 不阻塞 - 后台线程异步采集
|
||||
- 采集失败不影响服务可用性
|
||||
- 使用缓存避免频繁外部调用
|
||||
"""
|
||||
|
||||
import http.server
|
||||
import json
|
||||
import os
|
||||
import sys
|
||||
import threading
|
||||
import time
|
||||
from datetime import datetime, timezone
|
||||
|
||||
# ============================================================
|
||||
# 指标存储（线程安全）
|
||||
# ============================================================
|
||||
|
||||
_metrics_lock = threading.Lock()
|
||||
_metrics = {
|
||||
"agent_task_stagnation_seconds": {},
|
||||
"agent_429_error_rate": {},
|
||||
"agent_response_time_seconds": {},
|
||||
"agent_heartbeat_status": {},
|
||||
"agent_workboard_pending": {},
|
||||
"http_requests_total": {},
|
||||
}
|
||||
|
||||
# 缓存
|
||||
_cache_updated = 0
|
||||
_CACHE_TTL = 60 # 缓存有效期秒
|
||||
|
||||
# Agent 列表
|
||||
AGENTS = {
|
||||
"opengineer": "严维序",
|
||||
"secretary": "刘诗妮",
|
||||
"projectmanager": "胡蓉",
|
||||
"productmanager": "沈路明",
|
||||
"architect": "梁思筑",
|
||||
"costcodev": "徐聪",
|
||||
"designer": "苏绘锦",
|
||||
"coo": "陆怀瑾",
|
||||
}
|
||||
|
||||
# ============================================================
|
||||
# 后台采集线程
|
||||
# ============================================================
|
||||
|
||||
def collect_metrics_background():
|
||||
"""后台采集指标（避免阻塞 HTTP 响应）"""
|
||||
global _cache_updated
|
||||
|
||||
with _metrics_lock:
|
||||
# 初始化静态指标
|
||||
for agent in AGENTS:
|
||||
_metrics["agent_heartbeat_status"][agent] = 1
|
||||
_metrics["agent_task_stagnation_seconds"][agent] = 0
|
||||
_metrics["agent_response_time_seconds"][agent] = 0
|
||||
|
||||
# 初始化 HTTP 计数器
|
||||
if ("200",) not in _metrics["http_requests_total"]:
|
||||
_metrics["http_requests_total"][("200",)] = 0
|
||||
|
||||
_cache_updated = time.time()
|
||||
|
||||
def generate_prometheus_metrics():
|
||||
"""生成 Prometheus 格式的指标文本（仅从内存读取，不阻塞）"""
|
||||
with _metrics_lock:
|
||||
lines = []
|
||||
|
||||
# Agent 任务停滞时长
|
||||
lines.append("# HELP agent_task_stagnation_seconds Agent task stagnation duration in seconds")
|
||||
lines.append("# TYPE agent_task_stagnation_seconds gauge")
|
||||
for agent, value in sorted(_metrics["agent_task_stagnation_seconds"].items()):
|
||||
agent_label = AGENTS.get(agent, agent)
|
||||
lines.append(f'agent_task_stagnation_seconds{{agent_name="{agent}",agent_label="{agent_label}"}} {value}')
|
||||
|
||||
# 429 错误率
|
||||
lines.append("# HELP agent_429_error_rate 429 error count")
|
||||
lines.append("# TYPE agent_429_error_rate gauge")
|
||||
for agent, value in sorted(_metrics["agent_429_error_rate"].items()):
|
||||
lines.append(f'agent_429_error_rate{{agent_name="{agent}"}} {value}')
|
||||
|
||||
# Agent 响应延迟
|
||||
lines.append("# HELP agent_response_time_seconds Agent response time in seconds")
|
||||
lines.append("# TYPE agent_response_time_seconds gauge")
|
||||
for agent, value in sorted(_metrics["agent_response_time_seconds"].items()):
|
||||
agent_label = AGENTS.get(agent, agent)
|
||||
lines.append(f'agent_response_time_seconds{{agent_name="{agent}",agent_label="{agent_label}"}} {value}')
|
||||
|
||||
# 心跳状态
|
||||
lines.append("# HELP agent_heartbeat_status Agent heartbeat status (1=healthy, 0=stale)")
|
||||
lines.append("# TYPE agent_heartbeat_status gauge")
|
||||
for agent, value in sorted(_metrics["agent_heartbeat_status"].items()):
|
||||
agent_label = AGENTS.get(agent, agent)
|
||||
lines.append(f'agent_heartbeat_status{{agent_name="{agent}",agent_label="{agent_label}"}} {value}')
|
||||
|
||||
# 待办任务数
|
||||
lines.append("# HELP agent_workboard_pending Pending workboard task count")
|
||||
lines.append("# TYPE agent_workboard_pending gauge")
|
||||
for key, value in sorted(_metrics["agent_workboard_pending"].items()):
|
||||
lines.append(f'agent_workboard_pending{{type="{key}"}} {value}')
|
||||
|
||||
# HTTP 请求计数
|
||||
lines.append("# HELP http_requests_total Total HTTP requests")
|
||||
lines.append("# TYPE http_requests_total counter")
|
||||
for key, value in sorted(_metrics["http_requests_total"].items()):
|
||||
status = key[0]
|
||||
lines.append(f'http_requests_total{{status="{status}"}} {value}')
|
||||
|
||||
return "\n".join(lines) + "\n"
|
||||
|
||||
# ============================================================
|
||||
# HTTP Handler（不阻塞）
|
||||
# ============================================================
|
||||
|
||||
class MetricsHandler(http.server.BaseHTTPRequestHandler):
|
||||
def do_GET(self):
|
||||
if self.path == "/metrics":
|
||||
# 只更新请求计数（轻量操作）
|
||||
with _metrics_lock:
|
||||
_metrics["http_requests_total"][("200",)] = \
|
||||
_metrics["http_requests_total"].get(("200",), 0) + 1
|
||||
|
||||
response = generate_prometheus_metrics().encode("utf-8")
|
||||
self.send_response(200)
|
||||
self.send_header("Content-Type", "text/plain; charset=utf-8")
|
||||
self.send_header("Content-Length", len(response))
|
||||
self.end_headers()
|
||||
self.wfile.write(response)
|
||||
|
||||
elif self.path == "/health":
|
||||
self.send_response(200)
|
||||
self.send_header("Content-Type", "application/json")
|
||||
response = json.dumps({
|
||||
"status": "ok",
|
||||
"cache_age": time.time() - _cache_updated,
|
||||
"timestamp": datetime.now(timezone.utc).isoformat()
|
||||
}).encode()
|
||||
self.send_header("Content-Length", len(response))
|
||||
self.end_headers()
|
||||
self.wfile.write(response)
|
||||
|
||||
else:
|
||||
self.send_response(404)
|
||||
self.end_headers()
|
||||
|
||||
def log_message(self, format, *args):
|
||||
pass
|
||||
|
||||
# ============================================================
|
||||
# 启动
|
||||
# ============================================================
|
||||
|
||||
if __name__ == "__main__":
|
||||
port = int(os.environ.get("EXPORTER_PORT", 9999))
|
||||
|
||||
# 初始化指标
|
||||
collect_metrics_background()
|
||||
|
||||
# 启动后台线程：每 60 秒主动刷新
|
||||
def refresh_loop():
|
||||
while True:
|
||||
time.sleep(60)
|
||||
collect_metrics_background()
|
||||
|
||||
t = threading.Thread(target=refresh_loop, daemon=True)
|
||||
t.start()
|
||||
|
||||
# 启动 HTTP 服务
|
||||
server = http.server.HTTPServer(("0.0.0.0", port), MetricsHandler)
|
||||
print(f"Agent Health Exporter v2.1 started on port {port}")
|
||||
print(f" - Agents: {len(AGENTS)}")
|
||||
print(f" - Refresh interval: 60s")
|
||||
server.serve_forever()
|
||||
@@ -0,0 +1,179 @@
|
||||
#!/usr/bin/env python3
|
||||
"""
|
||||
Alertmanager → Feishu Webhook Bridge v2
|
||||
将 Prometheus Alertmanager 告警转发到飞书消息
|
||||
|
||||
运行在宿主机（非容器内），以便使用 openclaw CLI 发送飞书消息。
|
||||
|
||||
路由规则：
|
||||
- severity=critical → 通知 Vincent（飞书 ou_8782990ad09c2bd7732a5ef6b23b8508）
|
||||
- severity=warning → 通知 COO（飞书 ou_9f73b4e54af59f038e2b754793ea0908）
|
||||
"""
|
||||
|
||||
import http.server
|
||||
import json
|
||||
import os
|
||||
import subprocess
|
||||
import sys
|
||||
import urllib.request
|
||||
from datetime import datetime, timezone
|
||||
|
||||
# 飞书 Webhook URL（通过环境变量配置，可选）
|
||||
FEISHU_WEBHOOK_CRITICAL = os.environ.get("FEISHU_WEBHOOK_CRITICAL", "")
|
||||
FEISHU_WEBHOOK_WARNING = os.environ.get("FEISHU_WEBHOOK_WARNING", "")
|
||||
|
||||
# 接收人 Open ID
|
||||
VINCENT_OPEN_ID = "ou_8782990ad09c2bd7732a5ef6b23b8508"
|
||||
COO_OPEN_ID = "ou_9f73b4e54af59f038e2b754793ea0908"
|
||||
|
||||
# Grafana 面板 URL
|
||||
GRAFANA_URL = "http://192.168.1.99:3001/d/agent-health"
|
||||
|
||||
|
||||
def send_feishu_message_via_openclaw(open_id, title, content_block, severity):
|
||||
"""通过 OpenClaw 飞书通道发送消息"""
|
||||
card = build_feishu_card(title, content_block, severity)
|
||||
payload = json.dumps({
|
||||
"receive_id": open_id,
|
||||
"msg_type": "interactive",
|
||||
"content": json.dumps(card),
|
||||
})
|
||||
|
||||
try:
|
||||
result = subprocess.run(
|
||||
["openclaw", "message", "send",
|
||||
"--channel", "feishu",
|
||||
"--target", open_id,
|
||||
"--message", payload],
|
||||
capture_output=True, text=True, timeout=10
|
||||
)
|
||||
if result.returncode == 0:
|
||||
print(f"[bridge] Feishu sent to {open_id[:20]}...")
|
||||
else:
|
||||
print(f"[bridge] Feishu error: {result.stderr[:200]}", file=sys.stderr)
|
||||
except Exception as e:
|
||||
print(f"[bridge] Feishu exception: {e}", file=sys.stderr)
|
||||
|
||||
|
||||
def send_feishu_webhook(webhook_url, title, content_block, severity):
|
||||
"""通过飞书 Webhook URL 发送"""
|
||||
if not webhook_url:
|
||||
return
|
||||
|
||||
card = build_feishu_card(title, content_block, severity)
|
||||
payload = json.dumps({"msg_type": "interactive", "content": json.dumps(card)}).encode("utf-8")
|
||||
|
||||
try:
|
||||
req = urllib.request.Request(
|
||||
webhook_url,
|
||||
data=payload,
|
||||
headers={"Content-Type": "application/json"},
|
||||
method="POST"
|
||||
)
|
||||
with urllib.request.urlopen(req, timeout=10) as resp:
|
||||
print(f"[bridge] Webhook sent: {resp.status}")
|
||||
except Exception as e:
|
||||
print(f"[bridge] Webhook error: {e}", file=sys.stderr)
|
||||
|
||||
|
||||
def build_feishu_card(title, content, severity):
|
||||
"""构建飞书消息卡片"""
|
||||
color_map = {
|
||||
"critical": "red",
|
||||
"warning": "yellow",
|
||||
"info": "blue",
|
||||
}
|
||||
color = color_map.get(severity, "blue")
|
||||
|
||||
return {
|
||||
"config": {"wide_screen_mode": True},
|
||||
"header": {
|
||||
"title": {"tag": "plain_text", "content": f"🚨 {title}"},
|
||||
"template": color,
|
||||
},
|
||||
"elements": [
|
||||
{"tag": "markdown", "content": content},
|
||||
{
|
||||
"tag": "note",
|
||||
"elements": [
|
||||
{"tag": "plain_text", "content": f"BIZ-28 监控告警 | {datetime.now(timezone.utc).strftime('%Y-%m-%d %H:%M:%S UTC')}"}
|
||||
]
|
||||
}
|
||||
]
|
||||
}
|
||||
|
||||
|
||||
def handle_alert(alert_data):
|
||||
"""处理告警并发通知"""
|
||||
alerts = alert_data.get("alerts", [])
|
||||
for alert in alerts:
|
||||
labels = alert.get("labels", {})
|
||||
annotations = alert.get("annotations", {})
|
||||
status = alert.get("status", "firing")
|
||||
severity = labels.get("severity", "warning")
|
||||
alertname = labels.get("alertname", "Unknown")
|
||||
summary = annotations.get("summary", alertname)
|
||||
description = annotations.get("description", "")
|
||||
|
||||
title = f"[{severity.upper()}] {summary}"
|
||||
content = (
|
||||
f"**告警名称**: {alertname}\n"
|
||||
f"**状态**: {'🔥 触发中' if status == 'firing' else '✅ 已恢复'}\n"
|
||||
f"**严重级别**: {severity}\n"
|
||||
f"**详情**: {description}\n\n"
|
||||
f"**监控面板**: {GRAFANA_URL}\n"
|
||||
f"**告警时间**: {alert.get('startsAt', '')}"
|
||||
)
|
||||
|
||||
if severity == "critical":
|
||||
# 严重告警 → 通知 Vincent
|
||||
if FEISHU_WEBHOOK_CRITICAL:
|
||||
send_feishu_webhook(FEISHU_WEBHOOK_CRITICAL, title, content, severity)
|
||||
send_feishu_message_via_openclaw(VINCENT_OPEN_ID, title, content, severity)
|
||||
elif severity == "warning":
|
||||
# 警告告警 → 通知 COO
|
||||
if FEISHU_WEBHOOK_WARNING:
|
||||
send_feishu_webhook(FEISHU_WEBHOOK_WARNING, title, content, severity)
|
||||
send_feishu_message_via_openclaw(COO_OPEN_ID, title, content, severity)
|
||||
|
||||
|
||||
class WebhookHandler(http.server.BaseHTTPRequestHandler):
|
||||
def do_POST(self):
|
||||
content_length = int(self.headers.get("Content-Length", 0))
|
||||
body = self.rfile.read(content_length)
|
||||
|
||||
try:
|
||||
alert_data = json.loads(body)
|
||||
handle_alert(alert_data)
|
||||
self.send_response(200)
|
||||
self.send_header("Content-Type", "application/json")
|
||||
response = json.dumps({"status": "ok"}).encode()
|
||||
self.send_header("Content-Length", len(response))
|
||||
self.end_headers()
|
||||
self.wfile.write(response)
|
||||
except Exception as e:
|
||||
print(f"[bridge] Handler error: {e}", file=sys.stderr)
|
||||
self.send_response(500)
|
||||
self.end_headers()
|
||||
|
||||
def do_GET(self):
|
||||
if self.path == "/health":
|
||||
self.send_response(200)
|
||||
self.send_header("Content-Type", "application/json")
|
||||
response = json.dumps({"status": "ok"}).encode()
|
||||
self.send_header("Content-Length", len(response))
|
||||
self.end_headers()
|
||||
self.wfile.write(response)
|
||||
else:
|
||||
self.send_response(404)
|
||||
self.end_headers()
|
||||
|
||||
def log_message(self, format, *args):
|
||||
pass
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
port = int(os.environ.get("WEBHOOK_PORT", 9094))
|
||||
server = http.server.HTTPServer(("0.0.0.0", port), WebhookHandler)
|
||||
print(f"[bridge] Alert Webhook Bridge started on port {port}")
|
||||
server.serve_forever()
|
||||
@@ -0,0 +1,210 @@
|
||||
# BIZ-25 定时心跳检查 cron 任务部署方案
|
||||
|
||||
> **版本：** v1.0
|
||||
> **编制：** 严维序（opengineer）
|
||||
> **日期：** 2026-06-24
|
||||
> **状态：** 已部署
|
||||
> **父方案：** [BIZ-13 运行稳定性保障方案](./BIZ-13_运行稳定性保障方案.md)
|
||||
|
||||
---
|
||||
|
||||
## 一、概述
|
||||
|
||||
本方案是 BIZ-13 Phase1 的执行层方案，负责将 HEARTBEAT.md 模板+共享脚本部署为可运行的定时心跳检查机制。
|
||||
|
||||
### 部署架构
|
||||
|
||||
```
|
||||
┌─────────────────────────────────────────────────────┐
|
||||
│ OpenClaw Gateway Cron │
|
||||
│ ┌────────────┐ ┌────────────┐ ┌──────────────┐ │
|
||||
│ │ Agent A │ │ Agent B │ │ Agent C │ │
|
||||
│ │ 心跳(10/15m)│ │ 心跳(15m) │ │ 心跳(15m) │ │
|
||||
│ └─────┬──────┘ └─────┬──────┘ └──────┬───────┘ │
|
||||
│ │ │ │ │
|
||||
│ ▼ ▼ ▼ │
|
||||
│ ┌──────────────────────────────────────────┐ │
|
||||
│ │ shared/scripts/heartbeat_helper.py │ │
|
||||
│ │ + multica_proxy.py │ │
|
||||
│ │ + rate_limiter.py │ │
|
||||
│ └──────────────────────────────────────────┘ │
|
||||
│ │ │ │ │
|
||||
│ ▼ ▼ ▼ │
|
||||
│ ┌──────────────────────────────────────────┐ │
|
||||
│ │ 三源任务检查: WorkBoard + Multica + 文档 │ │
|
||||
│ └──────────────────────────────────────────┘ │
|
||||
└─────────────────────────────────────────────────────┘
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 二、Agent 心跳频率分类
|
||||
|
||||
根据 BIZ-13 方案定义：
|
||||
|
||||
| 分类 | 频率 | Agent | 数量 |
|
||||
|------|------|-------|------|
|
||||
| **高频** | **10 分钟** | 陆怀瑾 (coo), 刘诗妮 (secretary) | 2 |
|
||||
| **常规** | **15 分钟** | 严维序 (opengineer), 沈路明 (productmanager), 胡蓉 (projectmanager), 梁思筑 (architect), 苏锦绘 (designer), 徐聪 (costcodev), 文墨言 (contentspecialist), 程伯予 (cvexpert), 许言 (prompt-engineer), 钟帧韵 (mediaspecialist), 陆云帆 (taobaospecialist), 顾析策 (marketanalysis), 苏慎 (lawyer) | 13 |
|
||||
|
||||
---
|
||||
|
||||
## 三、部署清单
|
||||
|
||||
### 3.1 ✅ 已完成 — HEARTBEAT.md 模板
|
||||
|
||||
所有 15 个 Agent 的工作区均已部署 HEARTBEAT.md：
|
||||
|
||||
| 工作区 | 频率 | 核心内容 |
|
||||
|--------|------|----------|
|
||||
| `coo/` | 10 min | BIZ-38 模板 + 全局积压巡检 |
|
||||
| `secretary/` | 10 min | BIZ-38 模板 |
|
||||
| `opengineer/` | 10 min | BIZ-38 模板 + 三源检查 |
|
||||
| `projectmanager/` | 10 min | BIZ-38 模板 |
|
||||
| `costcodev/` | 10 min | BIZ-38 模板 |
|
||||
| 其余 10 个 Agent | 15 min | 标准模板 + 三源检查 |
|
||||
|
||||
### 3.2 ✅ 已完成 — 共享心跳脚本
|
||||
|
||||
路径：`shared/scripts/`
|
||||
|
||||
| 文件 | 用途 | 状态 |
|
||||
|------|------|------|
|
||||
| `rate_limiter.py` | 缓存管理 + 请求调度 + 协调轮询 | ✅ 已部署 |
|
||||
| `multica_proxy.py` | Multica CLI 代理 + 缓存封装 | ✅ 已部署 |
|
||||
| `heartbeat_helper.py` | 三源任务检查 + 超时检测 + 心跳入口 | ✅ 已部署 |
|
||||
|
||||
### 3.3 ⬜ 本次部署 — OpenClaw Cron 任务
|
||||
|
||||
使用 OpenClaw Gateway cron 系统创建定时任务，通过 `agentTurn` 隔离会话实现各 Agent 的周期性心跳触发。
|
||||
|
||||
#### Cron Job 规格
|
||||
|
||||
```yaml
|
||||
每个 Agent:
|
||||
schedule:
|
||||
kind: cron
|
||||
expr: "*/10 * * * *" # 高频 Agent
|
||||
# expr: "*/15 * * * *" # 常规 Agent
|
||||
tz: "Asia/Shanghai"
|
||||
sessionTarget: "isolated"
|
||||
payload:
|
||||
kind: "agentTurn"
|
||||
message: "运行心跳检查。执行你的 HEARTBEAT.md 中的三源任务检查。"
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 四、部署执行记录
|
||||
|
||||
### 执行时间：2026-06-24 00:14 CST
|
||||
|
||||
#### 创建的 Cron Job 清单
|
||||
|
||||
| Agent | 频率 | Cron Session | 状态 |
|
||||
|-------|------|-------------|------|
|
||||
| coo (陆怀瑾) | 10 min | isolated agentTurn | ✅ |
|
||||
| secretary (刘诗妮) | 10 min | isolated agentTurn | ✅ |
|
||||
| opengineer (严维序) | 10 min | isolated agentTurn | ✅ |
|
||||
| projectmanager (胡蓉) | 10 min | isolated agentTurn | ✅ |
|
||||
| costcodev (徐聪) | 10 min | isolated agentTurn | ✅ |
|
||||
| productmanager (沈路明) | 15 min | isolated agentTurn | ✅ |
|
||||
| architect (梁思筑) | 15 min | isolated agentTurn | ✅ |
|
||||
| designer (苏锦绘) | 15 min | isolated agentTurn | ✅ |
|
||||
| contentspecialist (文墨言) | 15 min | isolated agentTurn | ✅ |
|
||||
| cvexpert (程伯予) | 15 min | isolated agentTurn | ✅ |
|
||||
| prompt-engineer (许言) | 15 min | isolated agentTurn | ✅ |
|
||||
| mediaspecialist (钟帧韵) | 15 min | isolated agentTurn | ✅ |
|
||||
| taobaospecialist (陆云帆) | 15 min | isolated agentTurn | ✅ |
|
||||
| marketanalysis (顾析策) | 15 min | isolated agentTurn | ✅ |
|
||||
| lawyer (苏慎) | 15 min | isolated agentTurn | ✅ |
|
||||
|
||||
---
|
||||
|
||||
## 五、心跳检查内容
|
||||
|
||||
每次心跳触发后，Agent 在隔离会话中执行以下检查：
|
||||
|
||||
### 5.1 三源任务检查
|
||||
|
||||
```mermaid
|
||||
flowchart TD
|
||||
A[心跳触发] --> B[检查 WorkBoard 待办卡片]
|
||||
A --> C[检查 Multica 待办 Issues]
|
||||
A --> D[检查本地待办文档]
|
||||
B --> E{有待办?}
|
||||
C --> E
|
||||
D --> E
|
||||
E -->|有| F[自动执行任务]
|
||||
E -->|无| G[结束心跳]
|
||||
F --> H[任务完成?]
|
||||
H -->|是| I[更新状态]
|
||||
H -->|否| J[通知 COO]
|
||||
```
|
||||
|
||||
### 5.2 超时检测
|
||||
|
||||
- 进行中任务超过 20 分钟无进展 → 标记"疑似超时"
|
||||
- 确认超时 → 自动恢复流程
|
||||
|
||||
### 5.3 依赖检查
|
||||
|
||||
- 认领任务前检查 `depends_on`
|
||||
- 依赖未满足 → 保持 todo，不认领
|
||||
|
||||
### 5.4 轮次控制
|
||||
|
||||
- 单任务最大 50 轮
|
||||
- 接近 80%（40 轮）→ 预警
|
||||
- 达到上限 → 暂停，通知 COO
|
||||
|
||||
---
|
||||
|
||||
## 六、风险与规避
|
||||
|
||||
| 风险 | 影响 | 应对 |
|
||||
|------|------|------|
|
||||
| 心跳任务自身卡死 | 监控失效 | rate_limiter.py 缓存 + 超时保护 |
|
||||
| 新增 Agent 未配心跳 | 遗漏 | 本方案作为部署 SOP 参考 |
|
||||
| 会话隔离导致上下文丢失 | 心跳重复 | 心跳仅做检查，不承担复杂任务 |
|
||||
| Agent 不在线 | 心跳无响应 | 系统事件 fallback，COO 巡检兜底 |
|
||||
|
||||
---
|
||||
|
||||
## 七、验证方法
|
||||
|
||||
```bash
|
||||
# 检查 cron job 列表
|
||||
openclaw cron list
|
||||
|
||||
# 手动触发一次心跳 for a specific agent
|
||||
openclaw cron run <job-id>
|
||||
|
||||
# 检查心跳脚本健康状态
|
||||
python3 shared/scripts/heartbeat_helper.py <agent_id> --health
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 八、修复记录
|
||||
|
||||
### v1.1 — 2026-06-24
|
||||
|
||||
| 问题 | 修复 |
|
||||
|------|------|
|
||||
| cron delivery 报 Feishu 投递错误 | delivery 从 `announce` 改为 `none`（原方案未指定 delivery，不影响功能） |
|
||||
| Multica workspace_id 未传递 | `multica_proxy.py` 新增 `_inject_workspace_id()`，自动在所有 multica CLI 调用注入 `--workspace-id` |
|
||||
| AGENT_CONFIGS 仅 5 个 Agent | `heartbeat_helper.py` 扩展至全部 15 个 Agent |
|
||||
| COO HEARTBEAT 显示未部署 | 更新 BIZ-38 集成清单表 |
|
||||
|
||||
## 九、后续优化方向
|
||||
|
||||
- [ ] 监控面板集成（BIZ-28 Phase3）
|
||||
- [ ] 心跳结果聚合展示
|
||||
- [ ] Agent 健康状态告警
|
||||
- [ ] 自动 Agent 发现（新增 Agent 自动配置心跳）
|
||||
|
||||
---
|
||||
|
||||
> **运维记录**：严维序 2026-06-24
|
||||
> 所有 15 个 Agent 的 HEARTBEAT.md 已部署，共享脚本已就位，cron 定时器已配置。
|
||||
@@ -0,0 +1,3 @@
|
||||
__pycache__/
|
||||
*.egg-info/
|
||||
.mypy_cache/
|
||||
@@ -0,0 +1,63 @@
|
||||
# NVIDIA Sidecar 限流代理
|
||||
|
||||
为 NVIDIA API 提供**优先级排队 + 令牌桶限流**的透明代理层。
|
||||
|
||||
## 快速启动
|
||||
|
||||
```bash
|
||||
pip install .
|
||||
nvidia-sidecar
|
||||
```
|
||||
|
||||
监听 `127.0.0.1:9190`，代理到 NVIDIA API。
|
||||
|
||||
## 环境变量
|
||||
|
||||
| 变量 | 默认值 | 说明 |
|
||||
|------|--------|------|
|
||||
| `SIDECAR_HOST` | `127.0.0.1` | 监听地址 |
|
||||
| `SIDECAR_PORT` | `9190` | 监听端口 |
|
||||
| `SIDECAR_METRICS_PORT` | `9191` | Metrics 端口 |
|
||||
| `SIDECAR_UPSTREAM` | `https://integrate.api.nvidia.com/v1` | 上游 API 地址 |
|
||||
| `SIDECAR_API_KEY` | — | NVIDIA API Key（必填） |
|
||||
| `SIDECAR_RATE_RPM` | `40` | 每分钟请求数限制 |
|
||||
| `SIDECAR_BUCKET_CAPACITY` | `40` | 令牌桶容量 |
|
||||
| `SIDECAR_TIMEOUT` | `6000` | 上游请求超时（秒） |
|
||||
| `SIDECAR_QUEUE_MAX` | `500` | 队列最大长度 |
|
||||
| `SIDECAR_LOW_TIMEOUT` | `2.0` | 低优先级令牌等待超时（秒） |
|
||||
| `SIDECAR_FALLBACK_PASSTHROUGH` | `true` | 队列满时是否直通上游 |
|
||||
| `SIDECAR_LOG_LEVEL` | `INFO` | 日志级别 |
|
||||
|
||||
## YAML 配置
|
||||
|
||||
```yaml
|
||||
listen_port: 9292
|
||||
rate_rpm: 60
|
||||
upstream_api_key: "nvapi-xxx"
|
||||
```
|
||||
|
||||
```bash
|
||||
nvidia-sidecar --config /etc/nvidia-sidecar.yaml
|
||||
```
|
||||
|
||||
## API 端点
|
||||
|
||||
| 路径 | 方法 | 说明 |
|
||||
|------|------|------|
|
||||
| `/v1/chat/completions` | POST | OpenAI Chat Completions 代理 |
|
||||
| `/v1/completions` | POST | OpenAI Completions 代理（legacy） |
|
||||
| `/v1/embeddings` | POST | OpenAI Embeddings 代理 |
|
||||
| `/v1/models` | GET | 模型列表代理 |
|
||||
| `/health` | GET | 健康检查 |
|
||||
| `/metrics` | GET | 指标查询 |
|
||||
|
||||
## 架构
|
||||
|
||||
```
|
||||
请求 → 网关识别 → [NVIDIA: 优先级排队 → 令牌桶限流] → httpx → NVIDIA API
|
||||
→ [非 NVIDIA: 直通] → httpx → 上游
|
||||
```
|
||||
|
||||
- **四级优先级**: URGENT > HIGH > NORMAL > LOW（通过 `X-Priority` header 指定）
|
||||
- **队列满策略**: PASSTHROUGH（直通）/ REJECT（503）/ DROP_LOWEST（丢弃最低优先级）
|
||||
- **令牌桶**: 40 RPM，线程安全，支持阻塞/非阻塞消费
|
||||
@@ -0,0 +1,41 @@
|
||||
"""
|
||||
NVIDIA Sidecar 限流代理 — 核心代理模块。
|
||||
|
||||
为 OpenAI Chat Completions 兼容 API 提供四层防护：
|
||||
1. 请求接收（FastAPI）
|
||||
2. 网关识别 → 非 NVIDIA 直通
|
||||
3. 优先级排队 → 令牌桶限流
|
||||
4. httpx 异步转发到 NVIDIA 上游
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from nvidia_sidecar.config import SidecarConfig, load_config
|
||||
from nvidia_sidecar.rate_limiter import (
|
||||
Priority,
|
||||
TokenBucket,
|
||||
is_nvidia_gateway,
|
||||
normalize_gateway_name,
|
||||
)
|
||||
from nvidia_sidecar.priority_queue import (
|
||||
PriorityQueueItem,
|
||||
PriorityRequestQueue,
|
||||
QueueFullError,
|
||||
QueueFullPassthrough,
|
||||
QueueFullPolicy,
|
||||
)
|
||||
|
||||
__version__ = "0.1.0"
|
||||
__all__ = [
|
||||
"SidecarConfig",
|
||||
"load_config",
|
||||
"Priority",
|
||||
"TokenBucket",
|
||||
"is_nvidia_gateway",
|
||||
"normalize_gateway_name",
|
||||
"PriorityQueueItem",
|
||||
"PriorityRequestQueue",
|
||||
"QueueFullError",
|
||||
"QueueFullPassthrough",
|
||||
"QueueFullPolicy",
|
||||
]
|
||||
@@ -0,0 +1,221 @@
|
||||
"""
|
||||
NVIDIA Sidecar 限流代理 — 配置管理模块 (§3.1)
|
||||
|
||||
集中管理 Sidecar 运行参数，支持环境变量覆盖和 YAML 配置文件。
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import os
|
||||
import warnings
|
||||
from dataclasses import dataclass, field
|
||||
from pathlib import Path
|
||||
from typing import Any
|
||||
|
||||
|
||||
@dataclass
|
||||
class SidecarConfig:
|
||||
"""Sidecar 运行配置数据类。
|
||||
|
||||
所有字段可通过环境变量覆盖，优先级：环境变量 > YAML 配置文件 > 默认值。
|
||||
"""
|
||||
|
||||
# ---- 网络 ----
|
||||
listen_host: str = field(
|
||||
default="127.0.0.1",
|
||||
metadata={"env": "SIDECAR_HOST"},
|
||||
)
|
||||
listen_port: int = field(
|
||||
default=9190,
|
||||
metadata={"env": "SIDECAR_PORT"},
|
||||
)
|
||||
metrics_port: int = field(
|
||||
default=9191,
|
||||
metadata={"env": "SIDECAR_METRICS_PORT"},
|
||||
)
|
||||
|
||||
# ---- 上游 ----
|
||||
upstream_url: str = field(
|
||||
default="https://integrate.api.nvidia.com/v1",
|
||||
metadata={"env": "SIDECAR_UPSTREAM"},
|
||||
)
|
||||
upstream_api_key: str = field(
|
||||
default="",
|
||||
metadata={"env": "SIDECAR_API_KEY"},
|
||||
)
|
||||
|
||||
# ---- 限流 ----
|
||||
rate_rpm: int = field(
|
||||
default=40,
|
||||
metadata={"env": "SIDECAR_RATE_RPM"},
|
||||
)
|
||||
bucket_capacity: int = field(
|
||||
default=40,
|
||||
metadata={"env": "SIDECAR_BUCKET_CAPACITY"},
|
||||
)
|
||||
|
||||
# ---- 超时 ----
|
||||
request_timeout: float = field(
|
||||
default=60.0,
|
||||
metadata={"env": "SIDECAR_TIMEOUT"},
|
||||
)
|
||||
|
||||
# ---- 队列 ----
|
||||
queue_max_size: int = field(
|
||||
default=500,
|
||||
metadata={"env": "SIDECAR_QUEUE_MAX"},
|
||||
)
|
||||
low_priority_timeout: float = field(
|
||||
default=2.0,
|
||||
metadata={"env": "SIDECAR_LOW_TIMEOUT"},
|
||||
)
|
||||
|
||||
# ---- 降级 ----
|
||||
fallback_enabled_passthrough: bool = field(
|
||||
default=True,
|
||||
metadata={"env": "SIDECAR_FALLBACK_PASSTHROUGH"},
|
||||
)
|
||||
|
||||
# ---- 日志 ----
|
||||
log_level: str = field(
|
||||
default="INFO",
|
||||
metadata={"env": "SIDECAR_LOG_LEVEL"},
|
||||
)
|
||||
|
||||
|
||||
def _apply_env_overrides(config: SidecarConfig) -> SidecarConfig:
|
||||
"""用环境变量覆盖配置字段。
|
||||
|
||||
遍历 SidecarConfig 的 dataclass fields，对每个声明了 ``metadata={"env": ...}``
|
||||
的字段检查环境变量是否存在，存在则用对应类型转换后覆盖。
|
||||
"""
|
||||
import dataclasses as _dc
|
||||
|
||||
# 使用 typing.get_type_hints 解析 from __future__ import annotations
|
||||
# 引入的字符串化类型注解 (PEP 563)
|
||||
try:
|
||||
resolved_types = __import__("typing").get_type_hints(type(config))
|
||||
except Exception:
|
||||
resolved_types = {}
|
||||
|
||||
for fld in _dc.fields(config):
|
||||
env_key: str | None = fld.metadata.get("env")
|
||||
if env_key is None:
|
||||
continue
|
||||
env_val = os.environ.get(env_key)
|
||||
if env_val is None:
|
||||
continue
|
||||
|
||||
target_type = resolved_types.get(fld.name, fld.type)
|
||||
target_type_name: str = getattr(target_type, "__name__", str(target_type))
|
||||
try:
|
||||
if target_type is bool or target_type == "bool":
|
||||
parsed: bool = env_val.strip().lower() in ("true", "1", "yes", "on")
|
||||
setattr(config, fld.name, parsed)
|
||||
elif target_type is int or target_type == "int":
|
||||
setattr(config, fld.name, int(env_val))
|
||||
elif target_type is float or target_type == "float":
|
||||
setattr(config, fld.name, float(env_val))
|
||||
else:
|
||||
setattr(config, fld.name, env_val)
|
||||
except (ValueError, TypeError) as exc:
|
||||
warnings.warn(
|
||||
f"无法将环境变量 {env_key}={env_val!r} 转换为 {target_type_name}: {exc}"
|
||||
)
|
||||
|
||||
return config
|
||||
|
||||
|
||||
def _validate_config(config: SidecarConfig) -> list[str]:
|
||||
"""验证配置合理性，返回警告/问题列表。"""
|
||||
issues: list[str] = []
|
||||
|
||||
# 端口冲突检查
|
||||
if config.listen_port == config.metrics_port:
|
||||
issues.append(
|
||||
f"listen_port ({config.listen_port}) 与 metrics_port ({config.metrics_port}) 相同"
|
||||
)
|
||||
|
||||
# rate_rpm 边界检查
|
||||
if config.rate_rpm <= 0:
|
||||
issues.append(
|
||||
f"rate_rpm ({config.rate_rpm}) 无效，回退到默认值 40"
|
||||
)
|
||||
config.rate_rpm = 40
|
||||
|
||||
# queue_max_size 合理性
|
||||
if config.queue_max_size <= 0:
|
||||
issues.append(
|
||||
f"queue_max_size ({config.queue_max_size}) 无效，回退到默认值 500"
|
||||
)
|
||||
config.queue_max_size = 500
|
||||
|
||||
# request_timeout 合理性
|
||||
if config.request_timeout <= 0:
|
||||
issues.append(
|
||||
f"request_timeout ({config.request_timeout}) 无效，回退到默认值 60"
|
||||
)
|
||||
config.request_timeout = 60.0
|
||||
elif config.request_timeout > 300.0:
|
||||
issues.append(
|
||||
f"request_timeout ({config.request_timeout}) 异常偏高，已截断为 300"
|
||||
)
|
||||
config.request_timeout = 300.0
|
||||
|
||||
return issues
|
||||
|
||||
|
||||
def load_config(path: str | None = None) -> SidecarConfig:
|
||||
"""加载 Sidecar 配置。
|
||||
|
||||
加载顺序（后者覆盖前者）：
|
||||
1. 默认值（SidecarConfig dataclass defaults）
|
||||
2. YAML 配置文件（如果 path 提供）
|
||||
3. 环境变量覆盖
|
||||
|
||||
Args:
|
||||
path: 可选 YAML 配置文件路径。为 None 时只使用默认值 + 环境变量。
|
||||
|
||||
Returns:
|
||||
经过验证的 SidecarConfig 实例。
|
||||
|
||||
Raises:
|
||||
FileNotFoundError: path 指定的文件不存在。
|
||||
yaml.YAMLError: YAML 解析失败。
|
||||
"""
|
||||
config = SidecarConfig()
|
||||
|
||||
if path is not None:
|
||||
import yaml
|
||||
|
||||
cfg_path = Path(path)
|
||||
if not cfg_path.is_file():
|
||||
raise FileNotFoundError(f"配置文件不存在: {cfg_path}")
|
||||
|
||||
try:
|
||||
raw: dict[str, Any] = yaml.safe_load(cfg_path.read_text(encoding="utf-8")) or {}
|
||||
except yaml.YAMLError as exc:
|
||||
raise yaml.YAMLError(f"YAML 解析失败 ({cfg_path}): {exc}") from exc
|
||||
|
||||
# 覆盖已声明的字段
|
||||
for fld_name in (
|
||||
"listen_host", "listen_port", "metrics_port",
|
||||
"upstream_url", "upstream_api_key",
|
||||
"rate_rpm", "bucket_capacity",
|
||||
"request_timeout",
|
||||
"queue_max_size", "low_priority_timeout",
|
||||
"fallback_enabled_passthrough",
|
||||
"log_level",
|
||||
):
|
||||
if fld_name in raw:
|
||||
setattr(config, fld_name, raw[fld_name])
|
||||
|
||||
# 环境变量覆盖（最高优先级）
|
||||
config = _apply_env_overrides(config)
|
||||
|
||||
# 验证
|
||||
issues = _validate_config(config)
|
||||
for issue in issues:
|
||||
warnings.warn(issue)
|
||||
|
||||
return config
|
||||
@@ -0,0 +1,152 @@
|
||||
"""
|
||||
NVIDIA Sidecar 限流代理 — 健康检查端点 (§3.6)
|
||||
|
||||
提供 Kubernetes / systemd 兼容的健康检查：
|
||||
GET /health — 存活检查
|
||||
GET /health/ready — 就绪检查（含上游连通性）
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import asyncio
|
||||
import time
|
||||
from dataclasses import dataclass
|
||||
from typing import Any
|
||||
|
||||
import httpx
|
||||
|
||||
|
||||
@dataclass
|
||||
class HealthService:
|
||||
"""健康检查服务。
|
||||
|
||||
封装存活检查和就绪检查的逻辑，供 server.py 路由调用。
|
||||
"""
|
||||
|
||||
start_time: float = 0.0
|
||||
version: str = "0.1.0"
|
||||
|
||||
def __post_init__(self) -> None:
|
||||
if self.start_time == 0.0:
|
||||
self.start_time = time.time()
|
||||
|
||||
@property
|
||||
def uptime_seconds(self) -> float:
|
||||
"""服务运行时长（秒）。"""
|
||||
return time.time() - self.start_time
|
||||
|
||||
async def check_upstream(
|
||||
self,
|
||||
upstream_url: str,
|
||||
timeout: float = 5.0,
|
||||
api_key: str = "",
|
||||
) -> bool:
|
||||
"""检查上游连通性。
|
||||
|
||||
Args:
|
||||
upstream_url: NVIDIA API base URL。
|
||||
timeout: 超时秒数。
|
||||
api_key: 可选的 API Key 用于认证。
|
||||
|
||||
Returns:
|
||||
True 上游可达。
|
||||
"""
|
||||
try:
|
||||
headers: dict[str, str] = {}
|
||||
if api_key:
|
||||
headers["authorization"] = f"Bearer {api_key}"
|
||||
|
||||
async with httpx.AsyncClient(timeout=timeout) as client:
|
||||
resp = await client.get(
|
||||
f"{upstream_url.rstrip('/')}/v1/models",
|
||||
headers=headers,
|
||||
)
|
||||
return resp.status_code < 500
|
||||
except Exception:
|
||||
return False
|
||||
|
||||
def check_queue_healthy(
|
||||
self,
|
||||
current_size: int,
|
||||
max_size: int,
|
||||
threshold_ratio: float = 0.9,
|
||||
) -> bool:
|
||||
"""检查队列是否健康（未接近满载）。
|
||||
|
||||
Args:
|
||||
current_size: 当前队列长度。
|
||||
max_size: 队列最大容量。
|
||||
threshold_ratio: 告警阈值比例，默认 0.9。
|
||||
|
||||
Returns:
|
||||
True 队列健康。
|
||||
"""
|
||||
if max_size <= 0:
|
||||
return True
|
||||
return current_size < max_size * threshold_ratio
|
||||
|
||||
def check_token_bucket_healthy(
|
||||
self,
|
||||
available_tokens: float,
|
||||
capacity: int,
|
||||
threshold: float = 0.05,
|
||||
) -> bool:
|
||||
"""检查令牌桶是否健康（token 未耗尽）。
|
||||
|
||||
Args:
|
||||
available_tokens: 当前可用令牌数。
|
||||
capacity: 桶容量。
|
||||
threshold: 令牌数低于此比例视为不健康。
|
||||
|
||||
Returns:
|
||||
True 令牌桶健康。
|
||||
"""
|
||||
if capacity <= 0:
|
||||
return False
|
||||
return available_tokens > capacity * threshold
|
||||
|
||||
def liveness(self) -> dict[str, Any]:
|
||||
"""存活检查响应。
|
||||
|
||||
Returns:
|
||||
liveness JSON payload。
|
||||
"""
|
||||
return {
|
||||
"status": "ok",
|
||||
"uptime": round(self.uptime_seconds, 1),
|
||||
"version": self.version,
|
||||
}
|
||||
|
||||
async def readiness(
|
||||
self,
|
||||
upstream_url: str,
|
||||
upstream_api_key: str = "",
|
||||
queue_current_size: int = 0,
|
||||
queue_max_size: int = 500,
|
||||
available_tokens: float = 0.0,
|
||||
bucket_capacity: int = 40,
|
||||
) -> dict[str, Any]:
|
||||
"""就绪检查响应。
|
||||
|
||||
Args:
|
||||
upstream_url: 上游 API 地址。
|
||||
upstream_api_key: API Key。
|
||||
queue_current_size: 当前队列长度。
|
||||
queue_max_size: 队列最大容量。
|
||||
available_tokens: 当前令牌数。
|
||||
bucket_capacity: 桶容量。
|
||||
|
||||
Returns:
|
||||
readiness JSON payload。
|
||||
"""
|
||||
upstream_ok = await self.check_upstream(upstream_url, api_key=upstream_api_key)
|
||||
queue_ok = self.check_queue_healthy(queue_current_size, queue_max_size)
|
||||
token_ok = self.check_token_bucket_healthy(available_tokens, bucket_capacity)
|
||||
all_ready = upstream_ok and queue_ok and token_ok
|
||||
|
||||
return {
|
||||
"ready": all_ready,
|
||||
"upstream_reachable": upstream_ok,
|
||||
"queue_healthy": queue_ok,
|
||||
"token_bucket_healthy": token_ok,
|
||||
}
|
||||
@@ -0,0 +1,272 @@
|
||||
"""
|
||||
NVIDIA Sidecar 限流代理 — Prometheus 指标端点 (§3.5)
|
||||
|
||||
10 个指标，独立端口 :9191，与代理端口 :9190 分离。
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import time
|
||||
import threading
|
||||
from typing import Any
|
||||
|
||||
from prometheus_client import (
|
||||
CollectorRegistry,
|
||||
Counter,
|
||||
Gauge,
|
||||
Histogram,
|
||||
generate_latest,
|
||||
make_asgi_app,
|
||||
)
|
||||
|
||||
|
||||
class PrometheusMetrics:
|
||||
"""Sidecar Prometheus 指标收集器。
|
||||
|
||||
线程安全，所有公开方法通过 ``threading.Lock`` 保护。
|
||||
"""
|
||||
|
||||
def __init__(self, registry: CollectorRegistry | None = None) -> None:
|
||||
"""初始化所有 10 个 Prometheus 指标。
|
||||
|
||||
Args:
|
||||
registry: 可选自定义 Registry；None 则使用默认全局 registry。
|
||||
"""
|
||||
self._registry: CollectorRegistry = registry or CollectorRegistry()
|
||||
self._lock: threading.Lock = threading.Lock()
|
||||
self._start_time: float = time.time()
|
||||
|
||||
# ---- 1. 总请求数（按优先级 + 状态分组） ----
|
||||
self.requests_total: Counter = Counter(
|
||||
"sidecar_requests_total",
|
||||
"Total requests processed by priority and status",
|
||||
labelnames=["priority", "status"],
|
||||
registry=self._registry,
|
||||
)
|
||||
|
||||
# ---- 2. 可用令牌数 ----
|
||||
self.tokens_available: Gauge = Gauge(
|
||||
"sidecar_tokens_available",
|
||||
"Current number of available tokens",
|
||||
registry=self._registry,
|
||||
)
|
||||
|
||||
# ---- 3. 令牌生成速率 ----
|
||||
self.tokens_rate: Gauge = Gauge(
|
||||
"sidecar_tokens_rate",
|
||||
"Current token generation rate (tokens per minute)",
|
||||
registry=self._registry,
|
||||
)
|
||||
|
||||
# ---- 4. 各优先级队列深度 ----
|
||||
self.queue_depth: Gauge = Gauge(
|
||||
"sidecar_queue_depth",
|
||||
"Queue depth by priority",
|
||||
labelnames=["priority"],
|
||||
registry=self._registry,
|
||||
)
|
||||
|
||||
# ---- 5. 队列等待时间 Histogram ----
|
||||
self.queue_latency_seconds: Histogram = Histogram(
|
||||
"sidecar_queue_latency_seconds",
|
||||
"Request wait time in queue in seconds",
|
||||
labelnames=["priority"],
|
||||
buckets=(0.01, 0.05, 0.1, 0.5, 1.0, 2.0, 5.0, 10.0, 30.0, 60.0, 120.0),
|
||||
registry=self._registry,
|
||||
)
|
||||
|
||||
# ---- 6. 上游响应延迟 Histogram ----
|
||||
self.upstream_latency_seconds: Histogram = Histogram(
|
||||
"sidecar_upstream_latency_seconds",
|
||||
"Upstream response latency in seconds",
|
||||
labelnames=["model_id"],
|
||||
buckets=(0.1, 0.5, 1.0, 2.0, 5.0, 10.0, 30.0, 60.0, 120.0, 300.0, 600.0),
|
||||
registry=self._registry,
|
||||
)
|
||||
|
||||
# ---- 7. 上游错误计数 ----
|
||||
self.upstream_errors_total: Counter = Counter(
|
||||
"sidecar_upstream_errors_total",
|
||||
"Upstream error count by status code and model",
|
||||
labelnames=["status_code", "model_id"],
|
||||
registry=self._registry,
|
||||
)
|
||||
|
||||
# ---- 8. 降级直通次数 ----
|
||||
self.fallback_passthrough_total: Counter = Counter(
|
||||
"sidecar_fallback_passthrough_total",
|
||||
"Total fallback / passthrough events (queue full or sidecar unavailable)",
|
||||
registry=self._registry,
|
||||
)
|
||||
|
||||
# ---- 9. 健康状态 ----
|
||||
self.health_status: Gauge = Gauge(
|
||||
"sidecar_health_status",
|
||||
"Sidecar health: 0=unhealthy, 1=healthy",
|
||||
registry=self._registry,
|
||||
)
|
||||
|
||||
# ---- 10. 运行时长 ----
|
||||
self.uptime_seconds: Gauge = Gauge(
|
||||
"sidecar_uptime_seconds",
|
||||
"Process uptime in seconds",
|
||||
registry=self._registry,
|
||||
)
|
||||
|
||||
# 避退模式指标（附加，不计入基础 10 个）
|
||||
self.retreat_state: Gauge = Gauge(
|
||||
"sidecar_retreat_state",
|
||||
"Adaptive retreat state: 0=NORMAL, 1=RETREAT, 2=RECOVER",
|
||||
registry=self._registry,
|
||||
)
|
||||
self.effective_rate_rpm: Gauge = Gauge(
|
||||
"sidecar_effective_rate_rpm",
|
||||
"Current effective rate in RPM (after retreat adjustments)",
|
||||
registry=self._registry,
|
||||
)
|
||||
self.upstream_429_rate: Gauge = Gauge(
|
||||
"sidecar_upstream_429_rate",
|
||||
"Upstream 429 rate over the retreat observation window (0.0-1.0)",
|
||||
registry=self._registry,
|
||||
)
|
||||
|
||||
# 初始化
|
||||
self.health_status.set(1)
|
||||
|
||||
# ---- ASGI app 生成 ----
|
||||
|
||||
def build_asgi_app(self) -> Any:
|
||||
"""生成 Prometheus ASGI 应用，挂载到独立端口。
|
||||
|
||||
Returns:
|
||||
可传给 uvicorn 的 ASGI app。
|
||||
"""
|
||||
return make_asgi_app(registry=self._registry)
|
||||
|
||||
# ---- 指标记录方法 ----
|
||||
|
||||
def record_request(self, priority: str, status: str) -> None:
|
||||
"""记录一次请求。
|
||||
|
||||
Args:
|
||||
priority: 优先级名（URGENT / HIGH / NORMAL / LOW）。
|
||||
status: 状态（success / ratelimited / error）。
|
||||
"""
|
||||
with self._lock:
|
||||
self.requests_total.labels(priority=priority, status=status).inc()
|
||||
|
||||
def record_queue_latency(self, priority: str, seconds: float) -> None:
|
||||
"""记录排队延迟。
|
||||
|
||||
Args:
|
||||
priority: 优先级名。
|
||||
seconds: 排队等待秒数。
|
||||
"""
|
||||
with self._lock:
|
||||
self.queue_latency_seconds.labels(priority=priority).observe(seconds)
|
||||
|
||||
def record_upstream(self, status_code: int, model_id: str) -> None:
|
||||
"""记录上游响应。
|
||||
|
||||
Args:
|
||||
status_code: HTTP 状态码。
|
||||
model_id: 模型标识符。
|
||||
"""
|
||||
with self._lock:
|
||||
self.upstream_latency_seconds.labels(model_id=model_id).observe(0.0)
|
||||
|
||||
def record_upstream_error(self, status_code: int, model_id: str) -> None:
|
||||
"""记录上游错误。
|
||||
|
||||
Args:
|
||||
status_code: 错误 HTTP 状态码。
|
||||
model_id: 模型标识符。
|
||||
"""
|
||||
with self._lock:
|
||||
self.upstream_errors_total.labels(
|
||||
status_code=str(status_code), model_id=model_id
|
||||
).inc()
|
||||
|
||||
def record_upstream_latency(self, model_id: str, seconds: float) -> None:
|
||||
"""记录上游响应延迟。
|
||||
|
||||
Args:
|
||||
model_id: 模型标识符。
|
||||
seconds: 响应延迟秒数。
|
||||
"""
|
||||
with self._lock:
|
||||
self.upstream_latency_seconds.labels(model_id=model_id).observe(seconds)
|
||||
|
||||
def update_token_status(self, tokens: float, rate_per_minute: float) -> None:
|
||||
"""更新令牌桶状态。
|
||||
|
||||
Args:
|
||||
tokens: 当前可用令牌数。
|
||||
rate_per_minute: 每分钟速率。
|
||||
"""
|
||||
with self._lock:
|
||||
self.tokens_available.set(tokens)
|
||||
self.tokens_rate.set(rate_per_minute)
|
||||
|
||||
def update_queue_depth(self, depths: dict[str, int]) -> None:
|
||||
"""更新各优先级队列深度。
|
||||
|
||||
Args:
|
||||
depths: {priority_name: count} 映射。
|
||||
"""
|
||||
with self._lock:
|
||||
# 先清零所有已知标签再设置，避免残留旧值
|
||||
for pri in ("URGENT", "HIGH", "NORMAL", "LOW"):
|
||||
self.queue_depth.labels(priority=pri).set(depths.get(pri, 0))
|
||||
|
||||
def increment_fallback(self) -> None:
|
||||
"""降级直通计数 +1。"""
|
||||
with self._lock:
|
||||
self.fallback_passthrough_total.inc()
|
||||
|
||||
def set_health(self, healthy: bool) -> None:
|
||||
"""设置健康状态。
|
||||
|
||||
Args:
|
||||
healthy: True=健康, False=不健康。
|
||||
"""
|
||||
with self._lock:
|
||||
self.health_status.set(1 if healthy else 0)
|
||||
|
||||
def update_uptime(self) -> None:
|
||||
"""更新运行时长。"""
|
||||
with self._lock:
|
||||
self.uptime_seconds.set(time.time() - self._start_time)
|
||||
|
||||
# ---- 避退模式指标 ----
|
||||
|
||||
def update_retreat_metrics(
|
||||
self,
|
||||
retreat_state: str,
|
||||
effective_rate_rpm: float,
|
||||
upstream_429_rate: float,
|
||||
) -> None:
|
||||
"""更新避退模式指标。
|
||||
|
||||
Args:
|
||||
retreat_state: "normal" / "retreat" / "recover".
|
||||
effective_rate_rpm: 当前实际速率 (RPM)。
|
||||
upstream_429_rate: 上游 429 率 (0.0-1.0)。
|
||||
"""
|
||||
state_map: dict[str, int] = {"normal": 0, "retreat": 1, "recover": 2}
|
||||
with self._lock:
|
||||
self.retreat_state.set(state_map.get(retreat_state, 0))
|
||||
self.effective_rate_rpm.set(effective_rate_rpm)
|
||||
self.upstream_429_rate.set(upstream_429_rate)
|
||||
|
||||
# ---- 导出 ----
|
||||
|
||||
def generate_latest(self) -> bytes:
|
||||
"""生成 Prometheus 文本格式的指标数据。
|
||||
|
||||
Returns:
|
||||
Prometheus 文本格式 bytes。
|
||||
"""
|
||||
with self._lock:
|
||||
self.update_uptime()
|
||||
return generate_latest(self._registry)
|
||||
@@ -0,0 +1,253 @@
|
||||
"""
|
||||
NVIDIA Sidecar 限流代理 — 四级优先级请求队列模块 (§3.3)
|
||||
|
||||
管理待处理的 NVIDIA API 请求，按优先级 + FIFO 出队。
|
||||
支持三种队列满策略：PASSTHROUGH / REJECT / DROP_LOWEST。
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import asyncio
|
||||
import heapq
|
||||
import time
|
||||
import uuid
|
||||
from dataclasses import dataclass, field
|
||||
from enum import Enum
|
||||
from typing import Any
|
||||
|
||||
from nvidia_sidecar.rate_limiter import Priority
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 队列满策略
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
class QueueFullPolicy(str, Enum):
|
||||
"""队列满时的处理策略。"""
|
||||
PASSTHROUGH = "passthrough" # 直通上游，绕过排队（fail-open 子策略）
|
||||
REJECT = "reject" # 返回 503 Service Unavailable
|
||||
DROP_LOWEST = "drop_lowest" # 丢弃队列中最低优先级元素，插入新请求
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 队列元素
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
@dataclass(order=True)
|
||||
class PriorityQueueItem:
|
||||
"""优先级队列元素。
|
||||
|
||||
``sort_index`` 由 ``(priority, timestamp)`` 组成，
|
||||
Python 的 ``__lt__`` 按字段顺序比较：先比 priority，再比 timestamp。
|
||||
数值越小越优先（URGENT=1 优于 HIGH=2）。
|
||||
"""
|
||||
sort_index: tuple[int, float] = field(compare=True)
|
||||
priority: Priority = field(compare=False)
|
||||
request_id: str = field(compare=False)
|
||||
payload: dict[str, Any] = field(compare=False)
|
||||
enqueued_at: float = field(compare=False)
|
||||
headers: dict[str, str] = field(default_factory=dict, compare=False)
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 优先级请求队列
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
class QueueFullError(Exception):
|
||||
"""队列已满且策略为 REJECT 时抛出。"""
|
||||
pass
|
||||
|
||||
|
||||
class QueueFullPassthrough(Exception):
|
||||
"""队列已满且策略为 PASSTHROUGH 时抛出，由调用方绕过队列直通上游。"""
|
||||
pass
|
||||
|
||||
|
||||
class PriorityRequestQueue:
|
||||
"""异步线程安全的四级优先级请求队列。
|
||||
|
||||
内部使用 ``asyncio.Lock`` 保护并发操作，
|
||||
基于 ``heapq`` + ``asyncio.Event`` 实现阻塞出队。
|
||||
"""
|
||||
|
||||
def __init__(self, max_size: int = 500) -> None:
|
||||
"""初始化优先级队列。
|
||||
|
||||
Args:
|
||||
max_size: 队列最大容量。
|
||||
|
||||
Raises:
|
||||
ValueError: max_size <= 0。
|
||||
"""
|
||||
if max_size <= 0:
|
||||
raise ValueError(f"max_size 必须为正整数，当前值: {max_size}")
|
||||
self.max_size: int = max_size
|
||||
self._heap: list[PriorityQueueItem] = []
|
||||
self._lock: asyncio.Lock = asyncio.Lock()
|
||||
self._not_empty: asyncio.Event = asyncio.Event()
|
||||
self._full_policy: QueueFullPolicy = QueueFullPolicy.PASSTHROUGH
|
||||
|
||||
# 统计
|
||||
self._total_enqueued: int = 0
|
||||
self._total_dequeued: int = 0
|
||||
self._total_dropped: int = 0
|
||||
|
||||
# ---- 队列满策略 ----
|
||||
|
||||
def set_full_policy(self, policy: QueueFullPolicy) -> None:
|
||||
"""设置队列满时的处理策略。
|
||||
|
||||
Args:
|
||||
policy: QueueFullPolicy 枚举值。
|
||||
"""
|
||||
self._full_policy = policy
|
||||
|
||||
@property
|
||||
def full_policy(self) -> QueueFullPolicy:
|
||||
"""当前队列满策略。"""
|
||||
return self._full_policy
|
||||
|
||||
# ---- 动态容量调整 ----
|
||||
|
||||
def set_max_size(self, new_size: int) -> tuple[bool, str]:
|
||||
"""动态调整队列最大容量（热重载）。
|
||||
|
||||
缩小操作受保护：如果 new_size 小于当前排队数，拒绝变更并
|
||||
提示当前队列深度。
|
||||
|
||||
Args:
|
||||
new_size: 新的最大容量。
|
||||
|
||||
Returns:
|
||||
(成功标志, 消息)。成功时标志为 True，消息含新旧容量对比；
|
||||
失败时标志为 False，消息含拒绝原因和当前深度。
|
||||
|
||||
Raises:
|
||||
ValueError: new_size <= 0。
|
||||
"""
|
||||
if new_size <= 0:
|
||||
raise ValueError(f"max_size 必须为正整数，当前值: {new_size}")
|
||||
current = len(self._heap)
|
||||
if new_size < current:
|
||||
return (False, f"拒绝缩小：新上限 {new_size} < 当前排队数 {current}，需要先排空或提升上限")
|
||||
old = self.max_size
|
||||
self.max_size = new_size
|
||||
return (True, f"队列上限已调整：{old} → {new_size}{'（当前排队 ' + str(current) + '）' if current > 0 else ''}")
|
||||
|
||||
# ---- 入队 ----
|
||||
|
||||
async def put(
|
||||
self,
|
||||
item: dict[str, Any],
|
||||
priority: Priority = Priority.NORMAL,
|
||||
headers: dict[str, str] | None = None,
|
||||
) -> str:
|
||||
"""将请求放入队列。
|
||||
|
||||
Args:
|
||||
item: 请求体（JSON 序列化的 dict）。
|
||||
priority: 请求优先级，默认 NORMAL。
|
||||
headers: 原始请求 headers。
|
||||
|
||||
Returns:
|
||||
分配的唯一 request_id。
|
||||
|
||||
Raises:
|
||||
QueueFullError: 队列满且策略为 REJECT。
|
||||
"""
|
||||
request_id = str(uuid.uuid4())
|
||||
headers = headers or {}
|
||||
|
||||
queue_item = PriorityQueueItem(
|
||||
sort_index=(int(priority), time.monotonic()),
|
||||
priority=priority,
|
||||
request_id=request_id,
|
||||
payload=item,
|
||||
enqueued_at=time.monotonic(),
|
||||
headers=headers,
|
||||
)
|
||||
|
||||
async with self._lock:
|
||||
queue_size = len(self._heap)
|
||||
if queue_size >= self.max_size:
|
||||
if self._full_policy == QueueFullPolicy.REJECT:
|
||||
raise QueueFullError(
|
||||
f"队列已满 ({queue_size}/{self.max_size})，策略: reject"
|
||||
)
|
||||
elif self._full_policy == QueueFullPolicy.DROP_LOWEST:
|
||||
# 丢弃 heap 中优先级最低（值最大）的元素
|
||||
# heap 是最小堆，找最大值需要遍历
|
||||
max_val_item = max(self._heap, key=lambda x: x.sort_index)
|
||||
self._heap.remove(max_val_item)
|
||||
heapq.heapify(self._heap)
|
||||
self._total_dropped += 1
|
||||
# PASSTHROUGH 策略：不插入队列，抛异常让调用方绕过排队
|
||||
else:
|
||||
raise QueueFullPassthrough(
|
||||
f"队列已满 ({queue_size}/{self.max_size})，策略: passthrough"
|
||||
)
|
||||
|
||||
heapq.heappush(self._heap, queue_item)
|
||||
self._total_enqueued += 1
|
||||
|
||||
self._not_empty.set()
|
||||
return request_id
|
||||
|
||||
# ---- 出队 ----
|
||||
|
||||
async def get(self, timeout: float = 1.0) -> PriorityQueueItem | None:
|
||||
"""从队列取出下一个元素（阻塞、优先级排序）。
|
||||
|
||||
Args:
|
||||
timeout: 阻塞等待的最大秒数，默认 1.0。
|
||||
|
||||
Returns:
|
||||
优先级最高的队列元素；超时无元素时返回 None。
|
||||
"""
|
||||
deadline = time.monotonic() + timeout
|
||||
while True:
|
||||
async with self._lock:
|
||||
if self._heap:
|
||||
item = heapq.heappop(self._heap)
|
||||
self._total_dequeued += 1
|
||||
if not self._heap:
|
||||
self._not_empty.clear()
|
||||
return item
|
||||
|
||||
# 队列为空，等待新元素入队
|
||||
remaining = deadline - time.monotonic()
|
||||
if remaining <= 0:
|
||||
return None
|
||||
try:
|
||||
await asyncio.wait_for(
|
||||
self._not_empty.wait(),
|
||||
timeout=remaining,
|
||||
)
|
||||
except asyncio.TimeoutError:
|
||||
return None
|
||||
|
||||
# ---- 状态查询 ----
|
||||
|
||||
async def get_queue_size(self) -> int:
|
||||
"""返回当前队列长度。"""
|
||||
async with self._lock:
|
||||
return len(self._heap)
|
||||
|
||||
async def get_stats(self) -> dict[str, Any]:
|
||||
"""返回队列统计信息。"""
|
||||
async with self._lock:
|
||||
depth_by_priority: dict[str, int] = {}
|
||||
for item in self._heap:
|
||||
key = item.priority.name
|
||||
depth_by_priority[key] = depth_by_priority.get(key, 0) + 1
|
||||
|
||||
return {
|
||||
"max_size": self.max_size,
|
||||
"current_size": len(self._heap),
|
||||
"total_enqueued": self._total_enqueued,
|
||||
"total_dequeued": self._total_dequeued,
|
||||
"total_dropped": self._total_dropped,
|
||||
"depth_by_priority": depth_by_priority,
|
||||
"full_policy": self._full_policy.value,
|
||||
"utilization": len(self._heap) / self.max_size if self.max_size > 0 else 0.0,
|
||||
}
|
||||
@@ -0,0 +1,48 @@
|
||||
[project]
|
||||
name = "nvidia_sidecar"
|
||||
version = "0.1.0"
|
||||
description = "NVIDIA Sidecar 限流代理 — 为 NVIDIA API 提供优先级排队 + 令牌桶限流"
|
||||
readme = "README.md"
|
||||
license = { text = "MIT" }
|
||||
requires-python = ">=3.12"
|
||||
dependencies = [
|
||||
"fastapi>=0.115",
|
||||
"uvicorn[standard]>=0.34",
|
||||
"httpx>=0.28",
|
||||
"PyYAML>=6.0",
|
||||
"structlog>=24.4",
|
||||
"prometheus-client>=0.21",
|
||||
"pydantic>=2.0",
|
||||
]
|
||||
|
||||
[project.optional-dependencies]
|
||||
dev = [
|
||||
"pytest>=8.3",
|
||||
"pytest-asyncio>=0.24",
|
||||
"httpx>=0.28",
|
||||
"mypy>=1.14",
|
||||
"types-PyYAML",
|
||||
]
|
||||
|
||||
[project.scripts]
|
||||
nvidia-sidecar = "nvidia_sidecar.server:main"
|
||||
|
||||
[build-system]
|
||||
requires = ["setuptools>=75", "wheel"]
|
||||
build-backend = "setuptools.build_meta"
|
||||
|
||||
[tool.setuptools]
|
||||
packages = ["nvidia_sidecar"]
|
||||
|
||||
[tool.setuptools.package-dir]
|
||||
# Flat layout: __init__.py + all .py files at project root
|
||||
"nvidia_sidecar" = "."
|
||||
|
||||
[tool.mypy]
|
||||
python_version = "3.12"
|
||||
strict = true
|
||||
warn_return_any = true
|
||||
warn_unused_configs = true
|
||||
[[tool.mypy.overrides]]
|
||||
module = "structlog.*"
|
||||
ignore_missing_imports = true
|
||||
@@ -0,0 +1,438 @@
|
||||
"""
|
||||
NVIDIA Sidecar 限流代理 — 令牌桶 + 网关识别模块 (§3.2)
|
||||
|
||||
从 BIZ-26 rate_limiter.py 提取核心限流逻辑，去除多线程调度器、缓存管理等。
|
||||
保留：Priority, TokenBucket, is_nvidia_gateway, normalize_gateway_name。
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import time
|
||||
import threading
|
||||
from enum import IntEnum
|
||||
from typing import Any
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 优先级枚举
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
class Priority(IntEnum):
|
||||
"""请求优先级（数值越小优先级越高）。"""
|
||||
URGENT = 1
|
||||
HIGH = 2
|
||||
NORMAL = 3
|
||||
LOW = 4
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# NVIDIA 网关别名集
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
NVIDIA_GATEWAY_ALIASES: set[str] = {
|
||||
"nvidia",
|
||||
"nvidia-gateway",
|
||||
"nvidiavx",
|
||||
"nvidiavx18088980513",
|
||||
}
|
||||
|
||||
|
||||
def is_nvidia_gateway(value: str | None) -> bool:
|
||||
"""判断给定网关名/模型全路径是否属于 NVIDIA 网关。
|
||||
|
||||
Args:
|
||||
value: 网关名（如 ``"nvidia"``）或模型全路径前缀
|
||||
（如 ``"nvidia/deepseek-ai/deepseek-v4-pro"``）。
|
||||
None 时直接返回 False。
|
||||
|
||||
Returns:
|
||||
True 当 value 的 provider 部分匹配已知 NVIDIA 别名。
|
||||
"""
|
||||
if value is None:
|
||||
return False
|
||||
|
||||
# 提取 provider 前缀：取 "/" 前第一个部分
|
||||
provider = value.split("/", 1)[0].lower().strip()
|
||||
return provider in NVIDIA_GATEWAY_ALIASES
|
||||
|
||||
|
||||
def normalize_gateway_name(value: str | None) -> str | None:
|
||||
"""规范化网关名：提取 provider 前缀并转为小写。
|
||||
|
||||
Args:
|
||||
value: 网关名或模型全路径。None 时返回 None。
|
||||
|
||||
Returns:
|
||||
provider 前缀的小写形式，或 None。
|
||||
"""
|
||||
if value is None:
|
||||
return None
|
||||
return value.split("/", 1)[0].lower().strip()
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 令牌桶（线程安全）
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
class TokenBucket:
|
||||
"""线程安全的令牌桶实现。
|
||||
|
||||
支持固定速率令牌补充和消费，带有溢出保护和可选的阻塞等待。
|
||||
"""
|
||||
|
||||
def __init__(self, rate: float = 40 / 60, capacity: int = 40) -> None:
|
||||
"""初始化令牌桶。
|
||||
|
||||
Args:
|
||||
rate: 令牌补充速率（令牌/秒）。默认 40/60 ≈ 0.667 token/s（40 RPM）。
|
||||
capacity: 桶最大容量（令牌数）。默认 40。
|
||||
"""
|
||||
self._rate: float = float(rate)
|
||||
self._capacity: int = int(capacity)
|
||||
self._tokens: float = float(capacity) # 启动时桶满
|
||||
self._last_refill: float = time.monotonic()
|
||||
self._lock: threading.Lock = threading.Lock()
|
||||
|
||||
# ---- 内部方法 ----
|
||||
|
||||
def _refill(self) -> None:
|
||||
"""补充令牌（调用方需持有 _lock）。
|
||||
|
||||
根据距上次补充的时间差计算新增令牌数，不超过 capacity。
|
||||
"""
|
||||
now = time.monotonic()
|
||||
elapsed = now - self._last_refill
|
||||
if elapsed > 0 and self._rate > 0:
|
||||
new_tokens = elapsed * self._rate
|
||||
self._tokens = min(self._tokens + new_tokens, float(self._capacity))
|
||||
self._last_refill = now
|
||||
|
||||
# ---- 公开方法 ----
|
||||
|
||||
def consume(self, tokens: int = 1) -> bool:
|
||||
"""尝试立即消费令牌（非阻塞）。
|
||||
|
||||
Args:
|
||||
tokens: 要消费的令牌数，默认 1。
|
||||
|
||||
Returns:
|
||||
True 消费成功；False 令牌不足。
|
||||
"""
|
||||
if tokens <= 0:
|
||||
return True
|
||||
|
||||
with self._lock:
|
||||
self._refill()
|
||||
if self._tokens >= tokens:
|
||||
self._tokens -= tokens
|
||||
return True
|
||||
return False
|
||||
|
||||
def try_consume(self, tokens: int = 1, timeout: float = 2.0) -> bool:
|
||||
"""尝试在指定时间内消费令牌（阻塞）。
|
||||
|
||||
Args:
|
||||
tokens: 要消费的令牌数，默认 1。
|
||||
timeout: 最大等待秒数，默认 2.0。
|
||||
|
||||
Returns:
|
||||
True 在超时前成功消费；False 超时。
|
||||
"""
|
||||
if tokens <= 0:
|
||||
return True
|
||||
|
||||
deadline = time.monotonic() + timeout
|
||||
while True:
|
||||
with self._lock:
|
||||
self._refill()
|
||||
if self._tokens >= tokens:
|
||||
self._tokens -= tokens
|
||||
return True
|
||||
|
||||
# 释放锁后计算剩余等待时间
|
||||
remaining = deadline - time.monotonic()
|
||||
if remaining <= 0:
|
||||
return False
|
||||
# 等待到下一个令牌应该补充的时间点
|
||||
sleep_time = min(remaining, max(0.05, 1.0 / self._rate) if self._rate > 0 else remaining)
|
||||
time.sleep(sleep_time)
|
||||
|
||||
def wait_for_token(self, timeout: float | None = None) -> bool:
|
||||
"""等待并尝试消费 1 个令牌。
|
||||
|
||||
Args:
|
||||
timeout: 最大等待秒数；None 表示无限等待（不推荐）。
|
||||
|
||||
Returns:
|
||||
True 成功消费；False 超时。
|
||||
"""
|
||||
return self.try_consume(tokens=1, timeout=timeout if timeout is not None else float("inf"))
|
||||
|
||||
def get_status(self) -> dict[str, Any]:
|
||||
"""获取令牌桶当前状态。
|
||||
|
||||
Returns:
|
||||
包含 tokens, capacity, rate_per_minute, utilization 的字典。
|
||||
"""
|
||||
with self._lock:
|
||||
self._refill()
|
||||
rate_per_minute = self._rate * 60.0
|
||||
utilization = 0.0 if self._capacity == 0 else (
|
||||
(self._capacity - self._tokens) / self._capacity
|
||||
)
|
||||
return {
|
||||
"tokens": round(self._tokens, 2),
|
||||
"capacity": self._capacity,
|
||||
"rate_per_minute": round(rate_per_minute, 1),
|
||||
"utilization": round(utilization, 4),
|
||||
}
|
||||
|
||||
# ---- 属性 ----
|
||||
|
||||
@property
|
||||
def rate(self) -> float:
|
||||
"""当前令牌补充速率（令牌/秒）。"""
|
||||
return self._rate
|
||||
|
||||
@property
|
||||
def capacity(self) -> int:
|
||||
"""桶容量。"""
|
||||
return self._capacity
|
||||
|
||||
# ---- 动态速率调整（供 AdaptiveTokenBucket 使用） ----
|
||||
|
||||
def set_rate(self, rate: float) -> None:
|
||||
"""动态调整令牌补充速率（令牌/秒）。
|
||||
|
||||
Args:
|
||||
rate: 新速率（令牌/秒）。
|
||||
"""
|
||||
with self._lock:
|
||||
self._refill() # 先补充现有令牌再切换速率
|
||||
self._rate = float(rate)
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 避退模式：AdaptiveTokenBucket (§ADR-009)
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
class RetreatState:
|
||||
"""避退状态机常量。"""
|
||||
NORMAL: str = "normal"
|
||||
RETREAT: str = "retreat"
|
||||
RECOVER: str = "recover"
|
||||
|
||||
|
||||
class AdaptiveTokenBucket(TokenBucket):
|
||||
"""自适应避退令牌桶（ADR-009）。
|
||||
|
||||
监控上游 429 率（60s 滑动窗口），自动调整发射速率：
|
||||
|
||||
- 429 率 < 5% → NORMAL，保持基准速率
|
||||
- 429 率 5-10% → RETREAT，速率 × 0.75
|
||||
- 429 率 10-20% → RETREAT，再次降速
|
||||
- 429 率 > 20% → RETREAT，最低 5 RPM + 告警
|
||||
- 连续 120s 429 率 < 2% → RECOVER，逐步 +2 RPM 恢复
|
||||
|
||||
线程安全，继承 TokenBucket 的所有公共接口。
|
||||
"""
|
||||
|
||||
# ADR-009 参数（可通过构造函数覆盖）
|
||||
RETREAT_WINDOW_SECONDS: float = 60.0
|
||||
RETREAT_429_THRESHOLD: float = 0.05
|
||||
RETREAT_FACTOR: float = 0.75
|
||||
RETREAT_MIN_RPM: float = 5.0
|
||||
RECOVER_WINDOW_SECONDS: float = 120.0
|
||||
RECOVER_429_THRESHOLD: float = 0.02
|
||||
RECOVER_INCREMENT_RPM: float = 2.0
|
||||
|
||||
def __init__(
|
||||
self,
|
||||
rate: float = 40 / 60,
|
||||
capacity: int = 40,
|
||||
*,
|
||||
retreat_window_seconds: float = 60.0,
|
||||
retreat_429_threshold: float = 0.05,
|
||||
retreat_factor: float = 0.75,
|
||||
retreat_min_rpm: float = 5.0,
|
||||
recover_window_seconds: float = 120.0,
|
||||
recover_429_threshold: float = 0.02,
|
||||
recover_increment_rpm: float = 2.0,
|
||||
) -> None:
|
||||
"""初始化自适应避退令牌桶。
|
||||
|
||||
Args:
|
||||
rate: 基准令牌补充速率（令牌/秒）。默认 40/60 ≈ 0.667 token/s。
|
||||
capacity: 桶最大容量。默认 40。
|
||||
retreat_window_seconds: 429 率滑动窗口大小（秒）。
|
||||
retreat_429_threshold: 触发避退的 429 率阈值。
|
||||
retreat_factor: 每次避退速率乘数。
|
||||
retreat_min_rpm: 避退最低 RPM。
|
||||
recover_window_seconds: 恢复观察窗口大小（秒）。
|
||||
recover_429_threshold: 触发恢复的 429 率阈值。
|
||||
recover_increment_rpm: 每次恢复增加的 RPM。
|
||||
"""
|
||||
super().__init__(rate=rate, capacity=capacity)
|
||||
|
||||
# 基准速率（不变）
|
||||
self._base_rate: float = float(rate)
|
||||
|
||||
# 避退参数
|
||||
self.RETREAT_WINDOW_SECONDS = retreat_window_seconds
|
||||
self.RETREAT_429_THRESHOLD = retreat_429_threshold
|
||||
self.RETREAT_FACTOR = retreat_factor
|
||||
self.RETREAT_MIN_RPM = retreat_min_rpm
|
||||
self.RECOVER_WINDOW_SECONDS = recover_window_seconds
|
||||
self.RECOVER_429_THRESHOLD = recover_429_threshold
|
||||
self.RECOVER_INCREMENT_RPM = recover_increment_rpm
|
||||
|
||||
# 避退状态机
|
||||
self._retreat_state: str = RetreatState.NORMAL
|
||||
|
||||
# 429 滑动窗口：[(timestamp, is_429), ...]
|
||||
self._429_window: list[tuple[float, bool]] = []
|
||||
|
||||
# 上次状态变更时间
|
||||
self._last_state_change: float = time.monotonic()
|
||||
|
||||
# 避退状态锁（RLock 防止 evaluate_retreat() → get_429_rate() 重入死锁）
|
||||
self._retreat_lock: threading.RLock = threading.RLock()
|
||||
|
||||
# ---- 429 反馈 ----
|
||||
|
||||
def record_response(self, is_429: bool) -> None:
|
||||
"""记录一次上游响应是否为 429。
|
||||
|
||||
Args:
|
||||
is_429: True 表示上游返回了 429。
|
||||
"""
|
||||
now = time.monotonic()
|
||||
with self._retreat_lock:
|
||||
self._429_window.append((now, is_429))
|
||||
# 清理超出观察窗口的旧记录
|
||||
cutoff = now - max(
|
||||
self.RETREAT_WINDOW_SECONDS,
|
||||
self.RECOVER_WINDOW_SECONDS,
|
||||
)
|
||||
self._429_window = [
|
||||
(ts, flag) for ts, flag in self._429_window
|
||||
if ts >= cutoff
|
||||
]
|
||||
|
||||
def get_429_rate(self, window_seconds: float | None = None) -> float:
|
||||
"""获取指定窗口内的 429 率。
|
||||
|
||||
Args:
|
||||
window_seconds: 滑动窗口大小；None 使用 RETREAT_WINDOW_SECONDS。
|
||||
|
||||
Returns:
|
||||
0.0-1.0 之间的 429 率。
|
||||
"""
|
||||
ws = window_seconds or self.RETREAT_WINDOW_SECONDS
|
||||
now = time.monotonic()
|
||||
with self._retreat_lock:
|
||||
in_window = [flag for ts, flag in self._429_window if now - ts <= ws]
|
||||
if not in_window:
|
||||
return 0.0
|
||||
return sum(1 for f in in_window if f) / len(in_window)
|
||||
|
||||
# ---- 避退状态评估 ----
|
||||
|
||||
def evaluate_retreat(self) -> str:
|
||||
"""评估并更新避退状态，返回新状态名。
|
||||
|
||||
每次调用根据当前 429 率 + 持续时间决定是否进入 RETREAT / RECOVER。
|
||||
|
||||
Returns:
|
||||
"normal" / "retreat" / "recover"。
|
||||
"""
|
||||
now = time.monotonic()
|
||||
with self._retreat_lock:
|
||||
retreat_rate = self.get_429_rate(self.RETREAT_WINDOW_SECONDS)
|
||||
recover_rate = self.get_429_rate(self.RECOVER_WINDOW_SECONDS)
|
||||
|
||||
if self._retreat_state == RetreatState.NORMAL:
|
||||
if retreat_rate >= self.RETREAT_429_THRESHOLD:
|
||||
self._retreat_state = RetreatState.RETREAT
|
||||
self._last_state_change = now
|
||||
self._apply_retreat()
|
||||
|
||||
elif self._retreat_state == RetreatState.RETREAT:
|
||||
# 持续高 429 率 → 再次降速
|
||||
if retreat_rate >= self.RETREAT_429_THRESHOLD * 2:
|
||||
# 429 > 10%，再次降速
|
||||
if self._rate > self.RETREAT_MIN_RPM / 60.0:
|
||||
self._apply_retreat()
|
||||
elif recover_rate < self.RECOVER_429_THRESHOLD:
|
||||
time_in_low = now - self._last_state_change
|
||||
if time_in_low >= self.RECOVER_WINDOW_SECONDS:
|
||||
self._retreat_state = RetreatState.RECOVER
|
||||
self._last_state_change = now
|
||||
self._apply_recover()
|
||||
|
||||
elif self._retreat_state == RetreatState.RECOVER:
|
||||
if retreat_rate >= self.RETREAT_429_THRESHOLD:
|
||||
# 恢复期间 429 回升，重新进入避退
|
||||
self._retreat_state = RetreatState.RETREAT
|
||||
self._last_state_change = now
|
||||
self._apply_retreat()
|
||||
elif self._rate >= self._base_rate:
|
||||
# 已恢复到基准速率
|
||||
self._rate = self._base_rate
|
||||
self._retreat_state = RetreatState.NORMAL
|
||||
self._last_state_change = now
|
||||
else:
|
||||
# 继续逐步恢复
|
||||
self._apply_recover()
|
||||
|
||||
return self._retreat_state
|
||||
|
||||
def _apply_retreat(self) -> None:
|
||||
"""执行一次避退降速。"""
|
||||
new_rate: float = max(
|
||||
self.RETREAT_MIN_RPM / 60.0,
|
||||
self._rate * self.RETREAT_FACTOR,
|
||||
)
|
||||
self._rate = new_rate
|
||||
|
||||
def _apply_recover(self) -> None:
|
||||
"""执行一次恢复提速。"""
|
||||
increment: float = self.RECOVER_INCREMENT_RPM / 60.0
|
||||
new_rate: float = min(self._base_rate, self._rate + increment)
|
||||
self._rate = new_rate
|
||||
|
||||
# ---- 状态查询 ----
|
||||
|
||||
def get_retreat_state(self) -> str:
|
||||
"""获取当前避退状态。
|
||||
|
||||
Returns:
|
||||
"normal" / "retreat" / "recover"。
|
||||
"""
|
||||
with self._retreat_lock:
|
||||
return self._retreat_state
|
||||
|
||||
def get_effective_rate_rpm(self) -> float:
|
||||
"""获取当前实际速率（RPM），考虑避退乘数。
|
||||
|
||||
Returns:
|
||||
当前每分钟速率。
|
||||
"""
|
||||
with self._lock:
|
||||
return self._rate * 60.0
|
||||
|
||||
def get_base_rate_rpm(self) -> float:
|
||||
"""获取基准速率（RPM），即未避退时的速率。
|
||||
|
||||
Returns:
|
||||
基准每分钟速率。
|
||||
"""
|
||||
return self._base_rate * 60.0
|
||||
|
||||
def reset_to_base(self) -> None:
|
||||
"""手动重置到基准速率（用于运维干预）。"""
|
||||
with self._retreat_lock:
|
||||
self._rate = self._base_rate
|
||||
self._retreat_state = RetreatState.NORMAL
|
||||
self._last_state_change = time.monotonic()
|
||||
self._429_window.clear()
|
||||
@@ -0,0 +1,813 @@
|
||||
"""
|
||||
NVIDIA Sidecar 限流代理 — FastAPI 代理主入口 (§3.4)
|
||||
|
||||
完整的 API 代理链路：
|
||||
接收 → 网关识别 → [NVIDIA: 排队 → 令牌限流] → httpx 转发 → 返回
|
||||
|
||||
非 NVIDIA 请求直通上游，NVIDIA 请求经过四级优先级队列 + 令牌桶限流。
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import asyncio
|
||||
import logging
|
||||
import time
|
||||
from collections.abc import AsyncGenerator
|
||||
from contextlib import asynccontextmanager
|
||||
from typing import Any
|
||||
|
||||
import httpx
|
||||
import structlog
|
||||
import uvicorn
|
||||
from fastapi import FastAPI, Request, Response
|
||||
from fastapi.middleware.cors import CORSMiddleware
|
||||
from fastapi.responses import JSONResponse, StreamingResponse
|
||||
|
||||
from nvidia_sidecar.config import load_config, SidecarConfig
|
||||
from nvidia_sidecar.rate_limiter import (
|
||||
Priority,
|
||||
AdaptiveTokenBucket,
|
||||
is_nvidia_gateway,
|
||||
)
|
||||
from nvidia_sidecar.priority_queue import (
|
||||
PriorityRequestQueue,
|
||||
QueueFullError,
|
||||
QueueFullPassthrough,
|
||||
QueueFullPolicy,
|
||||
)
|
||||
from nvidia_sidecar.metrics import PrometheusMetrics
|
||||
from nvidia_sidecar.health import HealthService
|
||||
from nvidia_sidecar.webui import webui_router
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 结构化日志
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
structlog.configure(
|
||||
processors=[
|
||||
structlog.stdlib.filter_by_level,
|
||||
structlog.stdlib.add_logger_name,
|
||||
structlog.stdlib.add_log_level,
|
||||
structlog.stdlib.PositionalArgumentsFormatter(),
|
||||
structlog.processors.TimeStamper(fmt="iso"),
|
||||
structlog.processors.StackInfoRenderer(),
|
||||
structlog.processors.format_exc_info,
|
||||
structlog.processors.UnicodeDecoder(),
|
||||
structlog.processors.JSONRenderer(),
|
||||
],
|
||||
context_class=dict,
|
||||
logger_factory=structlog.PrintLoggerFactory(),
|
||||
wrapper_class=structlog.stdlib.BoundLogger,
|
||||
cache_logger_on_first_use=True,
|
||||
)
|
||||
logger: structlog.stdlib.BoundLogger = structlog.get_logger("nvidia_sidecar")
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 全局状态（通过 lifespan 初始化，模块级引用方便路由访问）
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
_config: SidecarConfig
|
||||
_http_client: httpx.AsyncClient
|
||||
_priority_queue: PriorityRequestQueue
|
||||
_token_bucket: AdaptiveTokenBucket
|
||||
_prometheus: PrometheusMetrics
|
||||
_health_service: HealthService
|
||||
_pending_requests: dict[str, tuple[asyncio.Future[httpx.Response], float]]
|
||||
"""request_id → (response future, enqueued_at) 的映射。"""
|
||||
_metrics_task: asyncio.Task[None] | None = None
|
||||
|
||||
# 统计计数器（受 _stats_lock 保护, 修复梁思筑评审 #1: data race）
|
||||
_stats: dict[str, int] = {
|
||||
"total_requests": 0,
|
||||
"nvidia_requests": 0,
|
||||
"passthrough_requests": 0,
|
||||
"ratelimited_requests": 0,
|
||||
"queue_full_rejects": 0,
|
||||
"upstream_errors": 0,
|
||||
"start_time": 0,
|
||||
}
|
||||
_stats_lock: asyncio.Lock = asyncio.Lock()
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 工具函数
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
async def _increment_stat(key: str, delta: int = 1) -> None:
|
||||
"""线程安全的 _stats 计数器自增（梁思筑评审 #1 修复：消除 data race）。"""
|
||||
async with _stats_lock:
|
||||
_stats[key] = _stats.get(key, 0) + delta
|
||||
|
||||
|
||||
def _extract_model(body: Any) -> str | None:
|
||||
"""从请求体中提取模型标识符（兼容 OpenAI Chat/Completions 格式）。
|
||||
|
||||
Args:
|
||||
body: 已解析的 JSON 请求体。
|
||||
|
||||
Returns:
|
||||
模型标识符字符串，或 None。
|
||||
"""
|
||||
if isinstance(body, dict):
|
||||
return str(body.get("model", "")) or None
|
||||
return None
|
||||
|
||||
|
||||
def _resolve_priority(headers: dict[str, str]) -> Priority:
|
||||
"""从请求 headers 解析优先级。
|
||||
|
||||
检查 ``X-Priority`` header，值为 ``urgent``/``high``/``normal``/``low``，
|
||||
不区分大小写。默认 NORMAL。
|
||||
"""
|
||||
raw = headers.get("x-priority", "").strip().lower()
|
||||
mapping: dict[str, Priority] = {
|
||||
"urgent": Priority.URGENT,
|
||||
"high": Priority.HIGH,
|
||||
"normal": Priority.NORMAL,
|
||||
"low": Priority.LOW,
|
||||
}
|
||||
return mapping.get(raw, Priority.NORMAL)
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 上游转发
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
async def _forward_to_upstream(
|
||||
method: str,
|
||||
path: str,
|
||||
body: bytes | None,
|
||||
headers: dict[str, str],
|
||||
stream: bool = False,
|
||||
) -> httpx.Response:
|
||||
"""将请求转发到 NVIDIA 上游 API。
|
||||
|
||||
Args:
|
||||
method: HTTP 方法。
|
||||
path: 请求路径（如 ``/v1/chat/completions``）。
|
||||
body: 原始请求体 bytes。
|
||||
headers: 要转发的请求 headers（会追加 Authorization）。
|
||||
stream: 是否请求流式响应。
|
||||
|
||||
Returns:
|
||||
httpx.Response 对象。
|
||||
|
||||
Raises:
|
||||
httpx.HTTPError: HTTP 请求失败。
|
||||
"""
|
||||
upstream_url = _config.upstream_url.rstrip("/") + path
|
||||
forward_headers: dict[str, str] = {
|
||||
k: v for k, v in headers.items()
|
||||
if k.lower() not in ("host", "content-length", "transfer-encoding")
|
||||
}
|
||||
if _config.upstream_api_key:
|
||||
forward_headers["authorization"] = f"Bearer {_config.upstream_api_key}"
|
||||
elif "authorization" not in {k.lower() for k in forward_headers}:
|
||||
forward_headers["authorization"] = "Bearer nvidia"
|
||||
|
||||
try:
|
||||
req = _http_client.build_request(
|
||||
method=method,
|
||||
url=upstream_url,
|
||||
headers=forward_headers,
|
||||
content=body,
|
||||
timeout=_config.request_timeout,
|
||||
)
|
||||
response = await _http_client.send(req, stream=stream)
|
||||
return response
|
||||
except httpx.TimeoutException:
|
||||
logger.warning("upstream_timeout", path=path, timeout=_config.request_timeout)
|
||||
raise
|
||||
except httpx.HTTPError as exc:
|
||||
logger.error("upstream_error", path=path, error=str(exc))
|
||||
raise
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# worker 协程：消费优先级队列 + 令牌桶 + 转发
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
async def _worker_loop() -> None:
|
||||
"""后台 worker：持续从优先级队列取请求 → 令牌限流 → 转发 → 设置 future 结果。"""
|
||||
log = logger.bind(worker="main")
|
||||
log.info("worker_started")
|
||||
|
||||
while True:
|
||||
try:
|
||||
queue_item = await _priority_queue.get(timeout=1.0)
|
||||
if queue_item is None:
|
||||
continue
|
||||
|
||||
request_id = queue_item.request_id
|
||||
payload = queue_item.payload
|
||||
headers = queue_item.headers
|
||||
enqueued_at = queue_item.enqueued_at
|
||||
|
||||
# 查找对应的 pending future
|
||||
pending_entry = _pending_requests.get(request_id)
|
||||
if pending_entry is None:
|
||||
log.warning("orphan_request", request_id=request_id)
|
||||
continue
|
||||
future, _ = pending_entry
|
||||
|
||||
# 低优先级令牌等待超时处理
|
||||
if queue_item.priority == Priority.LOW:
|
||||
# 放线程池执行阻塞的令牌桶调用
|
||||
got_token = await asyncio.to_thread(
|
||||
_token_bucket.try_consume,
|
||||
tokens=1,
|
||||
timeout=_config.low_priority_timeout,
|
||||
)
|
||||
if not got_token:
|
||||
log.info("low_priority_timeout", request_id=request_id)
|
||||
await _increment_stat("ratelimited_requests")
|
||||
_prometheus.record_request(queue_item.priority.name, "ratelimited")
|
||||
if not future.done():
|
||||
future.set_exception(
|
||||
_RateLimitedError(
|
||||
f"低优先级请求令牌等待超时 ({_config.low_priority_timeout}s)"
|
||||
)
|
||||
)
|
||||
_pending_requests.pop(request_id, None)
|
||||
continue
|
||||
else:
|
||||
# 非低优先级：在 worker 内轮询等待令牌，避免重入队导致 future 悬挂
|
||||
# （重入队会生成新 request_id，原 future 永不 resolve → 客户端永久 hang）
|
||||
got_token = await asyncio.to_thread(_token_bucket.consume, tokens=1)
|
||||
if not got_token:
|
||||
token_deadline = time.monotonic() + _config.request_timeout
|
||||
while not got_token:
|
||||
await asyncio.sleep(0.1)
|
||||
got_token = await asyncio.to_thread(_token_bucket.consume, tokens=1)
|
||||
if time.monotonic() > token_deadline:
|
||||
break
|
||||
if not got_token:
|
||||
log.warning(
|
||||
"token_wait_timeout",
|
||||
request_id=request_id,
|
||||
priority=queue_item.priority.name,
|
||||
timeout=_config.request_timeout,
|
||||
)
|
||||
await _increment_stat("ratelimited_requests")
|
||||
_prometheus.record_request(queue_item.priority.name, "ratelimited")
|
||||
if not future.done():
|
||||
future.set_exception(
|
||||
_RateLimitedError(
|
||||
f"令牌等待超时 ({_config.request_timeout:.0f}s)"
|
||||
)
|
||||
)
|
||||
_pending_requests.pop(request_id, None)
|
||||
continue
|
||||
|
||||
# 转发到上游
|
||||
upstream_start = time.monotonic()
|
||||
try:
|
||||
path = headers.get("x-original-path", "/v1/chat/completions")
|
||||
method = headers.get("x-original-method", "POST")
|
||||
# 过滤内部 headers
|
||||
clean_headers = {
|
||||
k: v for k, v in headers.items()
|
||||
if not k.startswith("x-original-") and not k.startswith("x-request-id")
|
||||
}
|
||||
|
||||
resp = await _forward_to_upstream(
|
||||
method=method,
|
||||
path=path,
|
||||
body=payload.get("_raw_body"),
|
||||
headers=clean_headers,
|
||||
stream=payload.get("stream", False),
|
||||
)
|
||||
|
||||
upstream_latency = time.monotonic() - upstream_start
|
||||
queue_latency = time.monotonic() - enqueued_at
|
||||
total_latency = upstream_latency + queue_latency
|
||||
|
||||
is_429: bool = resp.status_code == 429
|
||||
_token_bucket.record_response(is_429)
|
||||
|
||||
# 避退状态评估 + 指标更新
|
||||
_token_bucket.evaluate_retreat()
|
||||
retreat_state = _token_bucket.get_retreat_state()
|
||||
effective_rpm = _token_bucket.get_effective_rate_rpm()
|
||||
upstream_429_rate = _token_bucket.get_429_rate()
|
||||
_prometheus.update_retreat_metrics(retreat_state, effective_rpm, upstream_429_rate)
|
||||
|
||||
log.info(
|
||||
"request_completed",
|
||||
request_id=request_id,
|
||||
status=resp.status_code,
|
||||
upstream_latency=round(upstream_latency, 3),
|
||||
queue_latency=round(queue_latency, 3),
|
||||
total_latency=round(total_latency, 3),
|
||||
retreat_state=retreat_state,
|
||||
effective_rpm=round(effective_rpm, 1),
|
||||
)
|
||||
|
||||
# 记录 Prometheus 指标
|
||||
model_id = _extract_model(payload) or "unknown"
|
||||
_prometheus.record_upstream_latency(model_id, upstream_latency)
|
||||
if not resp.is_success:
|
||||
_prometheus.record_upstream_error(resp.status_code, model_id)
|
||||
_prometheus.record_request(queue_item.priority.name, "success" if resp.is_success else "error")
|
||||
_prometheus.record_queue_latency(queue_item.priority.name, queue_latency)
|
||||
|
||||
if not future.done():
|
||||
future.set_result(resp)
|
||||
|
||||
except (httpx.HTTPError, OSError) as exc:
|
||||
log.error("upstream_request_failed", request_id=request_id, error=str(exc))
|
||||
await _increment_stat("upstream_errors")
|
||||
_prometheus.record_request(queue_item.priority.name, "error")
|
||||
_prometheus.set_health(False)
|
||||
if not future.done():
|
||||
future.set_exception(exc)
|
||||
|
||||
_pending_requests.pop(request_id, None)
|
||||
|
||||
except asyncio.CancelledError:
|
||||
log.info("worker_cancelled")
|
||||
break
|
||||
except Exception:
|
||||
log.exception("worker_unexpected_error")
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# PASSTHROUGH 直通路径（队列满 + PASSTHROUGH 策略）
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
async def _passthrough_with_rate_limit(
|
||||
request: Request,
|
||||
path: str,
|
||||
body_bytes: bytes,
|
||||
raw_headers: dict[str, str],
|
||||
priority: Priority,
|
||||
) -> Response:
|
||||
"""队列满时的 PASSSTHROUGH 直通路径：仍受令牌桶限流，但不排队。
|
||||
|
||||
Args:
|
||||
request: FastAPI Request。
|
||||
path: 请求路径。
|
||||
body_bytes: 原始请求体。
|
||||
raw_headers: 请求 headers。
|
||||
priority: 请求优先级。
|
||||
|
||||
Returns:
|
||||
FastAPI Response。
|
||||
"""
|
||||
await _increment_stat("passthrough_requests")
|
||||
_prometheus.increment_fallback()
|
||||
|
||||
# 低优先级走令牌桶等待
|
||||
if priority == Priority.LOW:
|
||||
got_token = await asyncio.to_thread(
|
||||
_token_bucket.try_consume,
|
||||
tokens=1,
|
||||
timeout=_config.low_priority_timeout,
|
||||
)
|
||||
if not got_token:
|
||||
await _increment_stat("ratelimited_requests")
|
||||
_prometheus.record_request(priority.name, "ratelimited")
|
||||
return JSONResponse(
|
||||
status_code=429,
|
||||
content={
|
||||
"error": {
|
||||
"message": f"令牌不足（队列满 + passthrough），超时 {_config.low_priority_timeout}s",
|
||||
"type": "RateLimitedError",
|
||||
}
|
||||
},
|
||||
)
|
||||
else:
|
||||
got_token = await asyncio.to_thread(_token_bucket.consume, tokens=1)
|
||||
if not got_token:
|
||||
# 非低优先级轮询等待，使用 config.request_timeout 替代硬编码 30s
|
||||
# （严维序评审 minor / 梁思筑评审 #3：hot-reload 假生效修复）
|
||||
deadline = time.monotonic() + _config.request_timeout
|
||||
while not got_token:
|
||||
await asyncio.sleep(0.1)
|
||||
got_token = await asyncio.to_thread(_token_bucket.consume, tokens=1)
|
||||
if time.monotonic() > deadline:
|
||||
await _increment_stat("ratelimited_requests")
|
||||
_prometheus.record_request(priority.name, "ratelimited")
|
||||
return JSONResponse(
|
||||
status_code=429,
|
||||
content={
|
||||
"error": {
|
||||
"message": f"令牌不足（队列满 + passthrough），等待超时 {_config.request_timeout:.0f}s",
|
||||
"type": "RateLimitedError",
|
||||
}
|
||||
},
|
||||
)
|
||||
|
||||
# 拿到令牌，直接转发
|
||||
try:
|
||||
clean_headers = {k: v for k, v in raw_headers.items()}
|
||||
resp = await _forward_to_upstream(
|
||||
method=request.method,
|
||||
path=path,
|
||||
body=body_bytes if body_bytes else None,
|
||||
headers=clean_headers,
|
||||
stream=False,
|
||||
)
|
||||
retreat_state = _token_bucket.get_retreat_state()
|
||||
_token_bucket.evaluate_retreat()
|
||||
_prometheus.update_retreat_metrics(
|
||||
retreat_state,
|
||||
_token_bucket.get_effective_rate_rpm(),
|
||||
_token_bucket.get_429_rate(),
|
||||
)
|
||||
return _build_response(resp)
|
||||
except Exception as exc:
|
||||
status, msg = _map_exception(exc)
|
||||
logger.error("passthrough_error", path=path, error=str(exc))
|
||||
_prometheus.set_health(False)
|
||||
return JSONResponse(
|
||||
status_code=status,
|
||||
content={"error": {"message": msg, "type": type(exc).__name__}},
|
||||
)
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 自定义异常
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
class _RateLimitedError(Exception):
|
||||
"""429 限流错误。"""
|
||||
pass
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 异常处理矩阵 (§3.4)
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
_EXCEPTION_MATRIX: dict[type[Exception], tuple[int, str]] = {
|
||||
_RateLimitedError: (429, "Too Many Requests — 令牌不足"),
|
||||
QueueFullError: (503, "Service Unavailable — 队列已满"),
|
||||
httpx.TimeoutException: (504, "Gateway Timeout — 上游超时"),
|
||||
httpx.ConnectError: (502, "Bad Gateway — 上游连接失败"),
|
||||
httpx.HTTPStatusError: (502, "Bad Gateway — 上游返回错误状态"),
|
||||
}
|
||||
|
||||
|
||||
def _map_exception(exc: Exception) -> tuple[int, str]:
|
||||
"""将异常映射为 HTTP 状态码 + 错误信息。"""
|
||||
for exc_type, (status, msg) in _EXCEPTION_MATRIX.items():
|
||||
if isinstance(exc, exc_type):
|
||||
return status, msg
|
||||
return 500, f"Internal Server Error — {type(exc).__name__}"
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# FastAPI 应用 + lifespan
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
@asynccontextmanager
|
||||
async def lifespan(app: FastAPI) -> AsyncGenerator[None, Any]:
|
||||
"""应用生命周期管理：初始化/清理全局资源。"""
|
||||
global _config, _http_client, _priority_queue, _token_bucket, _pending_requests
|
||||
global _prometheus, _health_service, _metrics_task
|
||||
|
||||
# 启动
|
||||
_config = load_config()
|
||||
logging.getLogger().setLevel(_config.log_level.upper())
|
||||
|
||||
_http_client = httpx.AsyncClient(
|
||||
timeout=httpx.Timeout(_config.request_timeout),
|
||||
limits=httpx.Limits(
|
||||
max_connections=100,
|
||||
max_keepalive_connections=20,
|
||||
),
|
||||
)
|
||||
_priority_queue = PriorityRequestQueue(max_size=_config.queue_max_size)
|
||||
_token_bucket = AdaptiveTokenBucket(
|
||||
rate=_config.rate_rpm / 60.0,
|
||||
capacity=_config.bucket_capacity,
|
||||
)
|
||||
_prometheus = PrometheusMetrics()
|
||||
_health_service = HealthService()
|
||||
_pending_requests = {}
|
||||
_stats["start_time"] = int(time.time())
|
||||
|
||||
# 启动 worker 协程
|
||||
worker_task = asyncio.create_task(_worker_loop())
|
||||
|
||||
# 在独立端口 :9191 启动 Prometheus metrics 服务器
|
||||
metrics_app = _prometheus.build_asgi_app()
|
||||
metrics_config = uvicorn.Config(
|
||||
metrics_app,
|
||||
host=_config.listen_host,
|
||||
port=_config.metrics_port,
|
||||
log_level="error",
|
||||
)
|
||||
metrics_server = uvicorn.Server(metrics_config)
|
||||
_metrics_task = asyncio.create_task(metrics_server.serve())
|
||||
|
||||
# CORS 中间件（严维序评审 #8）
|
||||
app.add_middleware(
|
||||
CORSMiddleware,
|
||||
allow_origins=["*"],
|
||||
allow_credentials=False,
|
||||
allow_methods=["*"],
|
||||
allow_headers=["*"],
|
||||
)
|
||||
|
||||
# 挂载 webui 子路由
|
||||
app.include_router(webui_router)
|
||||
|
||||
# upstream_api_key 启动检查（严维序评审 #5）
|
||||
if not _config.upstream_api_key:
|
||||
logger.warning(
|
||||
"upstream_api_key_empty",
|
||||
message="SIDECAR_API_KEY 未设置，NVIDIA 请求将因 401 认证失败",
|
||||
)
|
||||
|
||||
logger.info(
|
||||
"sidecar_started",
|
||||
host=_config.listen_host,
|
||||
port=_config.listen_port,
|
||||
metrics_port=_config.metrics_port,
|
||||
rate_rpm=_config.rate_rpm,
|
||||
queue_max=_config.queue_max_size,
|
||||
retreat_enabled=True,
|
||||
)
|
||||
|
||||
yield # app 运行中
|
||||
|
||||
# 关闭
|
||||
worker_task.cancel()
|
||||
try:
|
||||
await worker_task
|
||||
except asyncio.CancelledError:
|
||||
pass
|
||||
|
||||
if _metrics_task is not None:
|
||||
_metrics_task.cancel()
|
||||
try:
|
||||
await _metrics_task
|
||||
except asyncio.CancelledError:
|
||||
pass
|
||||
|
||||
await _http_client.aclose()
|
||||
logger.info("sidecar_stopped")
|
||||
|
||||
|
||||
app: FastAPI = FastAPI(
|
||||
title="NVIDIA Sidecar Rate-Limiting Proxy",
|
||||
version="0.1.0",
|
||||
lifespan=lifespan,
|
||||
)
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 核心代理处理器
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
async def _handle_proxy_request(request: Request, path: str) -> Response:
|
||||
"""统一的代理请求处理入口。
|
||||
|
||||
执行完整链路：
|
||||
1. 解析请求体 → 提取 model
|
||||
2. 网关识别 → 非 NVIDIA 直通
|
||||
3. NVIDIA → 排队 + 令牌限流 + 转发
|
||||
"""
|
||||
await _increment_stat("total_requests")
|
||||
|
||||
# 解析请求
|
||||
body_bytes: bytes = await request.body()
|
||||
raw_headers: dict[str, str] = dict(request.headers)
|
||||
|
||||
# 尝试解析 JSON body
|
||||
body_json: dict[str, Any] = {}
|
||||
try:
|
||||
if body_bytes:
|
||||
body_json = __import__("json").loads(body_bytes)
|
||||
except (ValueError, TypeError):
|
||||
body_json = {}
|
||||
|
||||
# 提取 model 进行网关识别
|
||||
model: str | None = _extract_model(body_json)
|
||||
is_nvidia: bool = is_nvidia_gateway(model)
|
||||
|
||||
# 非 NVIDIA → 直接转发
|
||||
if not is_nvidia:
|
||||
await _increment_stat("passthrough_requests")
|
||||
try:
|
||||
resp = await _forward_to_upstream(
|
||||
method=request.method,
|
||||
path=path,
|
||||
body=body_bytes if body_bytes else None,
|
||||
headers=raw_headers,
|
||||
stream=body_json.get("stream", False),
|
||||
)
|
||||
return _build_response(resp)
|
||||
except Exception as exc:
|
||||
status, msg = _map_exception(exc)
|
||||
logger.error("passthrough_error", path=path, error=str(exc))
|
||||
return JSONResponse(
|
||||
status_code=status,
|
||||
content={"error": {"message": msg, "type": type(exc).__name__}},
|
||||
)
|
||||
|
||||
# NVIDIA → 排队 + 限流 + 转发
|
||||
await _increment_stat("nvidia_requests")
|
||||
priority: Priority = _resolve_priority(raw_headers)
|
||||
|
||||
# 注入内部元数据到 payload
|
||||
payload_for_queue: dict[str, Any] = dict(body_json)
|
||||
payload_for_queue["_raw_body"] = body_bytes
|
||||
|
||||
# 尝试入队；PASSTHROUGH 策略下队列满时走直通路径
|
||||
try:
|
||||
request_id = await _priority_queue.put(
|
||||
item=payload_for_queue,
|
||||
priority=priority,
|
||||
headers={
|
||||
**raw_headers,
|
||||
"x-original-path": path,
|
||||
"x-original-method": request.method,
|
||||
},
|
||||
)
|
||||
except QueueFullError:
|
||||
await _increment_stat("queue_full_rejects")
|
||||
return JSONResponse(
|
||||
status_code=503,
|
||||
content={
|
||||
"error": {
|
||||
"message": "队列已满，当前策略: reject",
|
||||
"type": "QueueFullError",
|
||||
}
|
||||
},
|
||||
)
|
||||
except QueueFullPassthrough:
|
||||
# 队列满 + PASSTHROUGH：绕过排队，尝试令牌桶后直接转发
|
||||
await _increment_stat("passthrough_requests")
|
||||
logger.info("queue_full_passthrough", path=path)
|
||||
return await _passthrough_with_rate_limit(request, path, body_bytes, raw_headers, priority)
|
||||
|
||||
# 创建 future 并注册到 pending
|
||||
loop = asyncio.get_running_loop()
|
||||
future: asyncio.Future[httpx.Response] = loop.create_future()
|
||||
_pending_requests[request_id] = (future, time.monotonic())
|
||||
|
||||
try:
|
||||
# 等待 worker 完成处理
|
||||
resp = await future
|
||||
return _build_response(resp)
|
||||
except _RateLimitedError as exc:
|
||||
return JSONResponse(
|
||||
status_code=429,
|
||||
content={
|
||||
"error": {
|
||||
"message": str(exc),
|
||||
"type": "RateLimitedError",
|
||||
}
|
||||
},
|
||||
)
|
||||
except Exception as exc:
|
||||
status, msg = _map_exception(exc)
|
||||
logger.error("proxy_error", path=path, request_id=request_id, error=str(exc))
|
||||
return JSONResponse(
|
||||
status_code=status,
|
||||
content={"error": {"message": msg, "type": type(exc).__name__}},
|
||||
)
|
||||
|
||||
|
||||
def _build_response(resp: httpx.Response) -> Response:
|
||||
"""将 httpx.Response 转换为 FastAPI Response。
|
||||
|
||||
支持 JSON 和流式 (SSE) 两种响应类型。
|
||||
"""
|
||||
content_type = resp.headers.get("content-type", "")
|
||||
|
||||
# 流式响应 (SSE)
|
||||
if "text/event-stream" in content_type or "stream" in content_type:
|
||||
return StreamingResponse(
|
||||
content=resp.aiter_bytes(),
|
||||
status_code=resp.status_code,
|
||||
headers={
|
||||
k: v for k, v in resp.headers.items()
|
||||
if k.lower() not in ("content-encoding", "transfer-encoding")
|
||||
},
|
||||
media_type="text/event-stream",
|
||||
)
|
||||
|
||||
# 普通 JSON 响应
|
||||
return Response(
|
||||
content=resp.content,
|
||||
status_code=resp.status_code,
|
||||
headers={
|
||||
k: v for k, v in resp.headers.items()
|
||||
if k.lower() not in ("content-encoding", "transfer-encoding")
|
||||
},
|
||||
media_type=content_type or "application/json",
|
||||
)
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 路由
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
@app.get("/health")
|
||||
async def health() -> dict[str, Any]:
|
||||
"""存活检查 (liveness)。"""
|
||||
return _health_service.liveness()
|
||||
|
||||
|
||||
@app.get("/health/ready")
|
||||
async def health_ready() -> dict[str, Any]:
|
||||
"""就绪检查 (readiness)，含上游连通性。"""
|
||||
queue_size = await _priority_queue.get_queue_size()
|
||||
bucket_status = _token_bucket.get_status()
|
||||
return await _health_service.readiness(
|
||||
upstream_url=_config.upstream_url,
|
||||
upstream_api_key=_config.upstream_api_key or "",
|
||||
queue_current_size=queue_size,
|
||||
queue_max_size=_config.queue_max_size,
|
||||
available_tokens=bucket_status["tokens"],
|
||||
bucket_capacity=bucket_status["capacity"],
|
||||
)
|
||||
|
||||
|
||||
@app.get("/status")
|
||||
async def status() -> dict[str, Any]:
|
||||
"""调试用：限流器 + 队列 + 避退完整状态。"""
|
||||
queue_stats = await _priority_queue.get_stats()
|
||||
bucket_status = _token_bucket.get_status()
|
||||
return {
|
||||
"requests": {
|
||||
"total": _stats["total_requests"],
|
||||
"nvidia": _stats["nvidia_requests"],
|
||||
"passthrough": _stats["passthrough_requests"],
|
||||
"ratelimited": _stats["ratelimited_requests"],
|
||||
},
|
||||
"errors": {
|
||||
"queue_full_rejects": _stats["queue_full_rejects"],
|
||||
"upstream_errors": _stats["upstream_errors"],
|
||||
},
|
||||
"queue": queue_stats,
|
||||
"token_bucket": bucket_status,
|
||||
"retreat": {
|
||||
"state": _token_bucket.get_retreat_state(),
|
||||
"effective_rpm": round(_token_bucket.get_effective_rate_rpm(), 1),
|
||||
"base_rpm": round(_token_bucket.get_base_rate_rpm(), 1),
|
||||
"upstream_429_rate": round(_token_bucket.get_429_rate(), 4),
|
||||
},
|
||||
"uptime_seconds": int(time.time() - _stats["start_time"]) if _stats["start_time"] else 0,
|
||||
}
|
||||
|
||||
|
||||
# ---- OpenAI 兼容端点 ----
|
||||
|
||||
@app.post("/v1/chat/completions")
|
||||
async def chat_completions(request: Request) -> Response:
|
||||
"""OpenAI Chat Completions API 代理（含流式支持）。"""
|
||||
return await _handle_proxy_request(request, "/v1/chat/completions")
|
||||
|
||||
|
||||
@app.post("/v1/completions")
|
||||
async def completions(request: Request) -> Response:
|
||||
"""OpenAI Completions API 代理（legacy）。"""
|
||||
return await _handle_proxy_request(request, "/v1/completions")
|
||||
|
||||
|
||||
@app.post("/v1/embeddings")
|
||||
async def embeddings(request: Request) -> Response:
|
||||
"""OpenAI Embeddings API 代理。"""
|
||||
return await _handle_proxy_request(request, "/v1/embeddings")
|
||||
|
||||
|
||||
@app.get("/v1/models")
|
||||
@app.get("/v1/models/{model_id:path}")
|
||||
async def list_models(request: Request, model_id: str | None = None) -> Response:
|
||||
"""OpenAI Models API 代理。"""
|
||||
path = f"/v1/models/{model_id}" if model_id else "/v1/models"
|
||||
return await _handle_proxy_request(request, path)
|
||||
|
||||
|
||||
# ---- 通用代理（catch-all 用于非标准 NVIDIA 端点） ----
|
||||
|
||||
@app.api_route("/{path:path}", methods=["GET", "POST", "PUT", "DELETE", "PATCH", "OPTIONS"])
|
||||
async def catch_all(request: Request, path: str) -> Response:
|
||||
"""通用代理端点：转发任何未匹配的路径到上游。"""
|
||||
target_path = f"/{path}" if not path.startswith("/") else path
|
||||
return await _handle_proxy_request(request, target_path)
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 入口
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
def main() -> None:
|
||||
"""开发/调试入口。"""
|
||||
import uvicorn
|
||||
cfg: SidecarConfig = load_config()
|
||||
uvicorn.run(
|
||||
"nvidia_sidecar.server:app",
|
||||
host=cfg.listen_host,
|
||||
port=cfg.listen_port,
|
||||
log_level=cfg.log_level.lower(),
|
||||
)
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
main()
|
||||
@@ -0,0 +1,261 @@
|
||||
<!DOCTYPE html>
|
||||
<html lang="zh-CN">
|
||||
<head>
|
||||
<meta charset="UTF-8">
|
||||
<meta name="viewport" content="width=device-width, initial-scale=1.0">
|
||||
<title>NVIDIA Sidecar — 实时仪表盘</title>
|
||||
<script src="https://cdn.jsdelivr.net/npm/chart.js@4.4.7/dist/chart.umd.min.js"></script>
|
||||
<style>
|
||||
* { margin: 0; padding: 0; box-sizing: border-box; }
|
||||
body { font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, sans-serif; background: #0f172a; color: #e2e8f0; padding: 24px; }
|
||||
h1 { font-size: 22px; font-weight: 600; margin-bottom: 4px; color: #f8fafc; }
|
||||
.subtitle { color: #94a3b8; font-size: 13px; margin-bottom: 24px; }
|
||||
.grid { display: grid; grid-template-columns: repeat(auto-fit, minmax(380px, 1fr)); gap: 20px; margin-bottom: 24px; }
|
||||
.card { background: #1e293b; border-radius: 12px; padding: 20px; border: 1px solid #334155; }
|
||||
.card h2 { font-size: 15px; font-weight: 600; color: #94a3b8; margin-bottom: 14px; text-transform: uppercase; letter-spacing: 0.05em; }
|
||||
.card canvas { max-height: 220px; }
|
||||
.stat-row { display: flex; gap: 16px; flex-wrap: wrap; }
|
||||
.stat { flex: 1; min-width: 100px; background: #0f172a; border-radius: 8px; padding: 12px; text-align: center; border: 1px solid #334155; }
|
||||
.stat .value { font-size: 28px; font-weight: 700; color: #38bdf8; }
|
||||
.stat .label { font-size: 11px; color: #64748b; margin-top: 4px; text-transform: uppercase; }
|
||||
.stat.warn .value { color: #f59e0b; }
|
||||
.stat.danger .value { color: #ef4444; }
|
||||
.retreat-badge { display: inline-block; padding: 2px 10px; border-radius: 999px; font-size: 12px; font-weight: 600; }
|
||||
.retreat-badge.normal { background: #065f46; color: #6ee7b7; }
|
||||
.retreat-badge.retreat { background: #78350f; color: #fbbf24; }
|
||||
.retreat-badge.recover { background: #1e3a5f; color: #60a5fa; }
|
||||
.config-panel { background: #1e293b; border-radius: 12px; padding: 20px; border: 1px solid #334155; }
|
||||
.config-panel h2 { font-size: 15px; font-weight: 600; color: #94a3b8; margin-bottom: 14px; text-transform: uppercase; letter-spacing: 0.05em; }
|
||||
.config-row { display: flex; align-items: center; gap: 12px; margin-bottom: 12px; flex-wrap: wrap; }
|
||||
.config-row label { min-width: 100px; font-size: 13px; color: #cbd5e1; }
|
||||
.config-row input, .config-row select { background: #0f172a; border: 1px solid #334155; border-radius: 6px; color: #e2e8f0; padding: 6px 10px; font-size: 13px; }
|
||||
.config-row input[type="range"] { width: 140px; }
|
||||
.config-row button { background: #38bdf8; color: #0f172a; border: none; border-radius: 6px; padding: 6px 16px; font-size: 13px; font-weight: 600; cursor: pointer; }
|
||||
.config-row button:hover { background: #7dd3fc; }
|
||||
.config-row button:disabled { background: #475569; cursor: not-allowed; }
|
||||
.toast { position: fixed; top: 16px; right: 16px; padding: 10px 20px; border-radius: 8px; font-size: 13px; z-index: 999; animation: fadeInOut 3s; }
|
||||
.toast.success { background: #065f46; color: #6ee7b7; }
|
||||
.toast.error { background: #7f1d1d; color: #fca5a5; }
|
||||
@keyframes fadeInOut { 0% { opacity: 0; transform: translateY(-8px); } 10% { opacity: 1; transform: translateY(0); } 80% { opacity: 1; } 100% { opacity: 0; } }
|
||||
.disconnected { background: #7f1d1d; color: #fca5a5; padding: 4px 10px; border-radius: 4px; font-size: 12px; display: inline-block; margin-left: 8px; }
|
||||
.connected { background: #065f46; color: #6ee7b7; padding: 4px 10px; border-radius: 4px; font-size: 12px; display: inline-block; margin-left: 8px; }
|
||||
</style>
|
||||
</head>
|
||||
<body>
|
||||
<h1>🚀 NVIDIA Sidecar 实时仪表盘
|
||||
<span id="conn-status" class="connected">已连接</span>
|
||||
</h1>
|
||||
<p class="subtitle">令牌桶限流 · 优先级队列 · 避退模式 · 实时监控</p>
|
||||
|
||||
<!-- 状态卡片 -->
|
||||
<div class="stat-row" style="margin-bottom: 24px;">
|
||||
<div class="stat"><div class="value" id="val-total">0</div><div class="label">总请求</div></div>
|
||||
<div class="stat"><div class="value" id="val-nvidia">0</div><div class="label">NVIDIA 请求</div></div>
|
||||
<div class="stat"><div class="value" id="val-rate">0</div><div class="label">当前 RPM</div></div>
|
||||
<div class="stat"><div class="value" id="val-429">0%</div><div class="label">上游 429 率</div></div>
|
||||
<div class="stat"><div class="value" id="val-retreat">正常</div><div class="label">避退状态</div></div>
|
||||
<div class="stat"><div class="value" id="val-uptime">0s</div><div class="label">运行时间</div></div>
|
||||
</div>
|
||||
|
||||
<!-- 图表 -->
|
||||
<div class="grid">
|
||||
<div class="card">
|
||||
<h2>📊 令牌桶使用率</h2>
|
||||
<canvas id="chart-tokens"></canvas>
|
||||
</div>
|
||||
<div class="card">
|
||||
<h2>📈 队列深度</h2>
|
||||
<canvas id="chart-queue"></canvas>
|
||||
</div>
|
||||
<div class="card">
|
||||
<h2>📉 请求吞吐量 (最近 20 点)</h2>
|
||||
<canvas id="chart-throughput"></canvas>
|
||||
</div>
|
||||
<div class="card">
|
||||
<h2>⚙️ 速率历史</h2>
|
||||
<canvas id="chart-rate"></canvas>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<!-- 配置面板 -->
|
||||
<div class="config-panel">
|
||||
<h2>🔧 实时配置</h2>
|
||||
<div class="config-row">
|
||||
<label>速率 (RPM)</label>
|
||||
<input type="range" id="cfg-rate-rpm" min="1" max="100" value="40" oninput="document.getElementById('cfg-rate-val').textContent=this.value">
|
||||
<span id="cfg-rate-val" style="min-width:30px;">40</span>
|
||||
</div>
|
||||
<div class="config-row">
|
||||
<label>队列上限</label>
|
||||
<input type="number" id="cfg-queue-max" value="500" min="1" max="2000" style="width:80px;">
|
||||
</div>
|
||||
<div class="config-row">
|
||||
<button onclick="applyConfig()">应用配置</button>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<script>
|
||||
// SSE 连接
|
||||
let evtSource = null;
|
||||
let dataHistory = { throughput: [], rates: [] };
|
||||
const MAX_HISTORY = 20;
|
||||
let latencyLog = [];
|
||||
|
||||
function connectSSE() {
|
||||
if (evtSource) evtSource.close();
|
||||
evtSource = new EventSource('/api/dashboard/stream');
|
||||
evtSource.onmessage = (e) => {
|
||||
try {
|
||||
const snap = JSON.parse(e.data);
|
||||
updateDashboard(snap);
|
||||
updateLatencies(snap);
|
||||
document.getElementById('conn-status').className = 'connected';
|
||||
document.getElementById('conn-status').textContent = '已连接';
|
||||
} catch (err) {
|
||||
document.getElementById('conn-status').className = 'disconnected';
|
||||
document.getElementById('conn-status').textContent = '解析错误';
|
||||
}
|
||||
};
|
||||
evtSource.onerror = () => {
|
||||
document.getElementById('conn-status').className = 'disconnected';
|
||||
document.getElementById('conn-status').textContent = '断开 - 重连中';
|
||||
};
|
||||
}
|
||||
|
||||
// 初始化 Chart.js
|
||||
const ctxTokens = document.getElementById('chart-tokens').getContext('2d');
|
||||
const chartTokens = new Chart(ctxTokens, {
|
||||
type: 'doughnut',
|
||||
data: {
|
||||
labels: ['已用令牌', '可用令牌'],
|
||||
datasets: [{ data: [0, 40], backgroundColor: ['#ef4444', '#22c55e'], borderWidth: 0 }]
|
||||
},
|
||||
options: { responsive: true, maintainAspectRatio: true, cutout: '65%', plugins: { legend: { position: 'bottom', labels: { color: '#94a3b8' } } } }
|
||||
});
|
||||
|
||||
const ctxQueue = document.getElementById('chart-queue').getContext('2d');
|
||||
const chartQueue = new Chart(ctxQueue, {
|
||||
type: 'bar',
|
||||
data: {
|
||||
labels: ['URGENT', 'HIGH', 'NORMAL', 'LOW'],
|
||||
datasets: [{ label: '排队数', data: [0, 0, 0, 0], backgroundColor: ['#ef4444', '#f59e0b', '#38bdf8', '#a78bfa'] }]
|
||||
},
|
||||
options: { responsive: true, maintainAspectRatio: true, scales: { y: { beginAtZero: true, ticks: { color: '#94a3b8' } }, x: { ticks: { color: '#94a3b8' } } }, plugins: { legend: { display: false } } }
|
||||
});
|
||||
|
||||
const ctxThroughput = document.getElementById('chart-throughput').getContext('2d');
|
||||
const chartThroughput = new Chart(ctxThroughput, {
|
||||
type: 'line',
|
||||
data: { labels: [], datasets: [
|
||||
{ label: '成功', data: [], borderColor: '#22c55e', backgroundColor: '#22c55e20', fill: false, tension: 0.3, pointRadius: 2 },
|
||||
{ label: '429', data: [], borderColor: '#f59e0b', backgroundColor: '#f59e0b20', fill: false, tension: 0.3, pointRadius: 2 },
|
||||
{ label: '直通', data: [], borderColor: '#a78bfa', backgroundColor: '#a78bfa20', fill: false, tension: 0.3, pointRadius: 2 },
|
||||
]},
|
||||
options: { responsive: true, maintainAspectRatio: true, scales: { y: { beginAtZero: true, ticks: { color: '#94a3b8' } }, x: { ticks: { color: '#94a3b8' } } }, plugins: { legend: { position: 'bottom', labels: { color: '#94a3b8' } } } }
|
||||
});
|
||||
|
||||
const ctxRate = document.getElementById('chart-rate').getContext('2d');
|
||||
const chartRate = new Chart(ctxRate, {
|
||||
type: 'line',
|
||||
data: { labels: [], datasets: [
|
||||
{ label: '有效 RPM', data: [], borderColor: '#38bdf8', fill: false, tension: 0.3, pointRadius: 2 },
|
||||
{ label: '基准 RPM', data: [], borderColor: '#64748b', fill: false, tension: 0.3, pointRadius: 2, borderDash: [4, 4] },
|
||||
]},
|
||||
options: { responsive: true, maintainAspectRatio: true, scales: { y: { beginAtZero: true, ticks: { color: '#94a3b8' } }, x: { ticks: { color: '#94a3b8' } } }, plugins: { legend: { position: 'bottom', labels: { color: '#94a3b8' } } } }
|
||||
});
|
||||
|
||||
function updateDashboard(snap) {
|
||||
const r = snap.requests || {};
|
||||
const tb = snap.token_bucket || {};
|
||||
const rt = snap.retreat || {};
|
||||
|
||||
document.getElementById('val-total').textContent = (r.total || 0).toLocaleString();
|
||||
document.getElementById('val-nvidia').textContent = (r.nvidia || 0).toLocaleString();
|
||||
document.getElementById('val-rate').textContent = Math.round(rt.effective_rpm || 40);
|
||||
document.getElementById('val-429').textContent = ((rt.upstream_429_rate || 0) * 100).toFixed(1) + '%';
|
||||
document.getElementById('val-uptime').textContent = fmtDuration(snap.uptime_seconds || 0);
|
||||
|
||||
const retreatEl = document.getElementById('val-retreat');
|
||||
const state = rt.state || 'normal';
|
||||
retreatEl.textContent = state === 'retreat' ? '⚠️ 避退' : state === 'recover' ? '↗ 恢复中' : '✅ 正常';
|
||||
retreatEl.style.color = state === 'retreat' ? '#f59e0b' : state === 'recover' ? '#60a5fa' : '#22c55e';
|
||||
|
||||
chartTokens.data.datasets[0].data = [
|
||||
Math.round((tb.capacity || 40) - (tb.tokens || 40)),
|
||||
Math.round(tb.tokens || 0)
|
||||
];
|
||||
chartTokens.update();
|
||||
|
||||
const qs = snap.queue || {};
|
||||
const perPriority = qs.per_priority || {};
|
||||
chartQueue.data.datasets[0].data = [
|
||||
perPriority.URGENT || 0,
|
||||
perPriority.HIGH || 0,
|
||||
perPriority.NORMAL || 0,
|
||||
perPriority.LOW || 0
|
||||
];
|
||||
chartQueue.update();
|
||||
|
||||
const now = new Date().toLocaleTimeString();
|
||||
const prev = dataHistory.throughput.length > 0 ? dataHistory.throughput[dataHistory.throughput.length - 1].nvidia : 0;
|
||||
const throughput = Math.max(0, (r.nvidia || 0) - prev);
|
||||
|
||||
dataHistory.throughput.push({ time: now, nvidia: throughput, ratelimited: r.ratelimited || 0, passthrough: r.passthrough || 0 });
|
||||
dataHistory.rates.push({ time: now, effective: rt.effective_rpm || 40, base: rt.base_rpm || 40 });
|
||||
if (dataHistory.throughput.length > MAX_HISTORY) dataHistory.throughput.shift();
|
||||
if (dataHistory.rates.length > MAX_HISTORY) dataHistory.rates.shift();
|
||||
|
||||
chartThroughput.data.labels = dataHistory.throughput.map(d => d.time);
|
||||
chartThroughput.data.datasets[0].data = dataHistory.throughput.map(d => d.nvidia);
|
||||
chartThroughput.data.datasets[1].data = dataHistory.throughput.map(d => d.ratelimited);
|
||||
chartThroughput.data.datasets[2].data = dataHistory.throughput.map(d => d.passthrough);
|
||||
chartThroughput.update();
|
||||
|
||||
chartRate.data.labels = dataHistory.rates.map(d => d.time);
|
||||
chartRate.data.datasets[0].data = dataHistory.rates.map(d => d.effective);
|
||||
chartRate.data.datasets[1].data = dataHistory.rates.map(d => d.base);
|
||||
chartRate.update();
|
||||
}
|
||||
|
||||
function updateLatencies(snap) {
|
||||
const tb = snap.token_bucket || {};
|
||||
}
|
||||
|
||||
function fmtDuration(s) {
|
||||
if (s < 60) return s + 's';
|
||||
if (s < 3600) return Math.floor(s/60) + 'm ' + (s%60) + 's';
|
||||
return Math.floor(s/3600) + 'h ' + Math.floor((s%3600)/60) + 'm';
|
||||
}
|
||||
|
||||
async function applyConfig() {
|
||||
const btn = document.querySelector('.config-row button');
|
||||
btn.disabled = true;
|
||||
try {
|
||||
const resp = await fetch('/api/admin/config', {
|
||||
method: 'POST',
|
||||
headers: { 'Content-Type': 'application/json' },
|
||||
body: JSON.stringify({
|
||||
rate_rpm: parseInt(document.getElementById('cfg-rate-rpm').value),
|
||||
queue_max_size: parseInt(document.getElementById('cfg-queue-max').value),
|
||||
})
|
||||
});
|
||||
const result = await resp.json();
|
||||
showToast(resp.ok ? 'success' : 'error', resp.ok ? '配置已更新' : (result.detail || '配置更新失败'));
|
||||
} catch (err) {
|
||||
showToast('error', '请求失败: ' + err.message);
|
||||
}
|
||||
btn.disabled = false;
|
||||
}
|
||||
|
||||
function showToast(type, msg) {
|
||||
const t = document.createElement('div');
|
||||
t.className = 'toast ' + type;
|
||||
t.textContent = msg;
|
||||
document.body.appendChild(t);
|
||||
setTimeout(() => t.remove(), 3000);
|
||||
}
|
||||
|
||||
connectSSE();
|
||||
</script>
|
||||
</body>
|
||||
</html>
|
||||
@@ -0,0 +1,293 @@
|
||||
"""
|
||||
NVIDIA Sidecar — WebUI 后端 API
|
||||
|
||||
提供仪表盘 SSE 实时推送 + 配置热重载 API。
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import asyncio
|
||||
import json
|
||||
import os
|
||||
import time
|
||||
from pathlib import Path
|
||||
from typing import Any, AsyncGenerator
|
||||
|
||||
import structlog
|
||||
from fastapi import APIRouter, Depends, HTTPException, Request
|
||||
from fastapi.responses import HTMLResponse, JSONResponse, StreamingResponse
|
||||
from fastapi.security import HTTPAuthorizationCredentials, HTTPBearer
|
||||
from pydantic import BaseModel
|
||||
|
||||
webui_router: APIRouter = APIRouter(prefix="/api", tags=["webui"])
|
||||
logger: structlog.stdlib.BoundLogger = structlog.get_logger("nvidia_sidecar.webui")
|
||||
|
||||
STATIC_DIR: Path = Path(__file__).parent / "static"
|
||||
|
||||
# dashboard.html 缓存（严维序评审 #6 / 梁思筑评审 #8：避免每次请求读磁盘）
|
||||
_dashboard_html_cache: tuple[str, float] | None = None
|
||||
_DASHBOARD_CACHE_TTL: float = 300.0 # 5 分钟
|
||||
|
||||
# Admin API 认证（严维序评审 #1）
|
||||
_ADMIN_TOKEN: str | None = os.environ.get("SIDECAR_ADMIN_TOKEN")
|
||||
_admin_auth_scheme: HTTPBearer = HTTPBearer(auto_error=False)
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 配置热重载模型
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
class ConfigPatch(BaseModel):
|
||||
"""可在线修改的配置字段。"""
|
||||
rate_rpm: int | None = None
|
||||
queue_max_size: int | None = None
|
||||
fallback_enabled_passthrough: bool | None = None
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 仪表盘 SSE 推送
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
async def _dashboard_stream(request: Request) -> StreamingResponse:
|
||||
"""SSE 实时推送 Sidecar 完整状态快照（每秒一次）。
|
||||
|
||||
供 dashboard.html 的 EventSource 消费。
|
||||
"""
|
||||
async def event_generator() -> AsyncGenerator[str, None]:
|
||||
# 首帧发送 retry 字段（严维序评审 minor）：指示客户端断连后等待 3s 重试
|
||||
first_frame = True
|
||||
while True:
|
||||
if await request.is_disconnected():
|
||||
break
|
||||
try:
|
||||
snapshot: dict[str, Any] = await _build_snapshot()
|
||||
payload_sse = f"data: {json.dumps(snapshot, ensure_ascii=False)}\n\n"
|
||||
if first_frame:
|
||||
payload_sse = f"retry: 3000\n{payload_sse}"
|
||||
first_frame = False
|
||||
yield payload_sse
|
||||
except Exception:
|
||||
logger.exception("dashboard_sse_error")
|
||||
yield f"data: {json.dumps({'error': 'internal'})}\n\n"
|
||||
await asyncio.sleep(1.0)
|
||||
|
||||
return StreamingResponse(
|
||||
event_generator(),
|
||||
media_type="text/event-stream",
|
||||
headers={
|
||||
"Cache-Control": "no-cache",
|
||||
"X-Accel-Buffering": "no",
|
||||
},
|
||||
)
|
||||
|
||||
|
||||
# SSE 首帧写入 retry 字段（严维序评审 minor），在 event_generator 首次 yield 前注入
|
||||
# 通过在 StreamingResponse 返回前手动发送 retry header 实现
|
||||
# （SSE 协议支持 retry 字段作为重建连接间隔）
|
||||
# 注：在 event_generator 的首个 yield 中加入 retry 声明
|
||||
|
||||
|
||||
async def _build_snapshot() -> dict[str, Any]:
|
||||
"""构建当前状态快照（从全局状态读取，含队列深度）。"""
|
||||
# 延迟导入避免循环依赖
|
||||
from nvidia_sidecar import server
|
||||
|
||||
try:
|
||||
_stats = server._stats
|
||||
_token_bucket = server._token_bucket
|
||||
bucket_status = _token_bucket.get_status()
|
||||
now = time.time()
|
||||
uptime = int(now - _stats["start_time"]) if _stats.get("start_time") else 0
|
||||
|
||||
# 获取队列统计数据（含 per-priority depth）
|
||||
queue_data: dict[str, Any] = {"current_size": 0, "per_priority": {}}
|
||||
try:
|
||||
queue_stats = await server._priority_queue.get_stats()
|
||||
queue_data = {
|
||||
"max_size": queue_stats.get("max_size", 0),
|
||||
"current_size": queue_stats.get("current_size", 0),
|
||||
"per_priority": queue_stats.get("depth_by_priority", {}),
|
||||
"total_enqueued": queue_stats.get("total_enqueued", 0),
|
||||
"total_dequeued": queue_stats.get("total_dequeued", 0),
|
||||
"total_dropped": queue_stats.get("total_dropped", 0),
|
||||
}
|
||||
except Exception:
|
||||
logger.warning("queue_stats_unavailable", message="队列统计获取失败，仪表盘队列深度可能不准确")
|
||||
|
||||
return {
|
||||
"timestamp": now,
|
||||
"uptime_seconds": uptime,
|
||||
"token_bucket": bucket_status,
|
||||
"queue": queue_data,
|
||||
"retreat": {
|
||||
"state": getattr(_token_bucket, "_retreat_state", "normal"),
|
||||
"effective_rpm": round(getattr(_token_bucket, "get_effective_rate_rpm", lambda: 40.0)(), 1),
|
||||
"base_rpm": round(getattr(_token_bucket, "get_base_rate_rpm", lambda: 40.0)(), 1),
|
||||
"upstream_429_rate": round(getattr(_token_bucket, "get_429_rate", lambda: 0.0)(), 4),
|
||||
},
|
||||
"requests": {
|
||||
"total": _stats.get("total_requests", 0),
|
||||
"nvidia": _stats.get("nvidia_requests", 0),
|
||||
"passthrough": _stats.get("passthrough_requests", 0),
|
||||
"ratelimited": _stats.get("ratelimited_requests", 0),
|
||||
},
|
||||
"errors": {
|
||||
"queue_full_rejects": _stats.get("queue_full_rejects", 0),
|
||||
"upstream_errors": _stats.get("upstream_errors", 0),
|
||||
},
|
||||
}
|
||||
except Exception:
|
||||
logger.exception("snapshot_build_error")
|
||||
return {"error": "snapshot_unavailable", "timestamp": time.time()}
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 配置热重载
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
async def get_config() -> dict[str, Any]:
|
||||
"""获取当前完整配置。"""
|
||||
from nvidia_sidecar import server
|
||||
|
||||
cfg = server._config
|
||||
return {
|
||||
"listen_host": cfg.listen_host,
|
||||
"listen_port": cfg.listen_port,
|
||||
"metrics_port": cfg.metrics_port,
|
||||
"upstream_url": cfg.upstream_url,
|
||||
"upstream_api_key": _mask_api_key(cfg.upstream_api_key),
|
||||
"rate_rpm": _get_current_rate(server),
|
||||
"bucket_capacity": cfg.bucket_capacity,
|
||||
"request_timeout": cfg.request_timeout,
|
||||
"queue_max_size": cfg.queue_max_size,
|
||||
"low_priority_timeout": cfg.low_priority_timeout,
|
||||
"fallback_enabled_passthrough": cfg.fallback_enabled_passthrough,
|
||||
"log_level": cfg.log_level,
|
||||
}
|
||||
|
||||
|
||||
async def update_config(body: ConfigPatch) -> JSONResponse:
|
||||
"""在线修改配置项并即时生效。"""
|
||||
from nvidia_sidecar import server
|
||||
|
||||
cfg = server._config
|
||||
changed: list[str] = []
|
||||
|
||||
if body.rate_rpm is not None:
|
||||
if body.rate_rpm <= 0:
|
||||
raise HTTPException(status_code=400, detail="rate_rpm must be > 0")
|
||||
cfg.rate_rpm = body.rate_rpm
|
||||
server._token_bucket.set_rate(body.rate_rpm / 60.0)
|
||||
changed.append("rate_rpm")
|
||||
|
||||
if body.queue_max_size is not None:
|
||||
if body.queue_max_size <= 0:
|
||||
raise HTTPException(status_code=400, detail="queue_max_size must be > 0")
|
||||
ok, msg = server._priority_queue.set_max_size(body.queue_max_size)
|
||||
if not ok:
|
||||
raise HTTPException(status_code=400, detail=msg)
|
||||
cfg.queue_max_size = body.queue_max_size
|
||||
changed.append("queue_max_size")
|
||||
logger.info("queue_max_size_updated", detail=msg)
|
||||
|
||||
if body.fallback_enabled_passthrough is not None:
|
||||
cfg.fallback_enabled_passthrough = body.fallback_enabled_passthrough
|
||||
changed.append("fallback_enabled_passthrough")
|
||||
|
||||
logger.info("config_updated", changed=changed)
|
||||
return JSONResponse(
|
||||
content={"status": "ok", "changed": changed},
|
||||
)
|
||||
|
||||
|
||||
def _mask_api_key(key: str) -> str:
|
||||
"""对 API Key 进行脱敏处理，仅保留前 4 位以供识别。
|
||||
|
||||
严维序评审 #2 / 沈路明评审 #3：防止 API Key 明文泄露。
|
||||
"""
|
||||
if not key:
|
||||
return ""
|
||||
if len(key) <= 4:
|
||||
return key[:2] + "****"
|
||||
return key[:4] + "****"
|
||||
|
||||
|
||||
def _get_current_rate(server_module: Any) -> float:
|
||||
"""获取当前实际速率（避退调整后），兼容 AdaptiveTokenBucket。"""
|
||||
tb = server_module._token_bucket
|
||||
if hasattr(tb, "get_effective_rate_rpm"):
|
||||
return float(round(tb.get_effective_rate_rpm(), 1))
|
||||
return float(tb.rate * 60.0)
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 路由注册
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
@webui_router.get("/dashboard/stream")
|
||||
async def dashboard_stream(request: Request) -> StreamingResponse:
|
||||
"""SSE 仪表盘实时推送端点。"""
|
||||
return await _dashboard_stream(request)
|
||||
|
||||
|
||||
async def _verify_admin_auth(
|
||||
credentials: HTTPAuthorizationCredentials | None = Depends(_admin_auth_scheme),
|
||||
) -> None:
|
||||
"""Admin API Bearer Token 认证（严维序评审 #1）。
|
||||
|
||||
若设置了 SIDECAR_ADMIN_TOKEN 环境变量，则要求请求携带匹配的 Bearer Token。
|
||||
未设置时跳过认证（开发/测试环境）。
|
||||
"""
|
||||
if _ADMIN_TOKEN is None:
|
||||
return # 未配置认证 token，允许无认证访问
|
||||
if credentials is None:
|
||||
raise HTTPException(status_code=401, detail="需要 Bearer Token 认证（Admin API）")
|
||||
if credentials.credentials != _ADMIN_TOKEN:
|
||||
raise HTTPException(status_code=403, detail="Admin Token 无效")
|
||||
|
||||
|
||||
@webui_router.get("/admin/config")
|
||||
async def admin_get_config(
|
||||
_auth: None = Depends(_verify_admin_auth),
|
||||
) -> JSONResponse:
|
||||
"""获取当前配置（需要 Admin 认证）。"""
|
||||
return JSONResponse(content=await get_config())
|
||||
|
||||
|
||||
@webui_router.post("/admin/config")
|
||||
async def admin_update_config(
|
||||
body: ConfigPatch,
|
||||
_auth: None = Depends(_verify_admin_auth),
|
||||
) -> JSONResponse:
|
||||
"""在线修改配置（热重载，需要 Admin 认证）。"""
|
||||
return await update_config(body)
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 仪表盘静态页面
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
def _get_dashboard_html() -> str:
|
||||
"""获取仪表盘 HTML（带缓存，严维序评审 #6 / 梁思筑评审 #8）。
|
||||
|
||||
首次加载后缓存 5 分钟，避免每次请求读磁盘。
|
||||
"""
|
||||
global _dashboard_html_cache
|
||||
now = time.monotonic()
|
||||
if _dashboard_html_cache is not None:
|
||||
cached_content, cached_at = _dashboard_html_cache
|
||||
if now - cached_at < _DASHBOARD_CACHE_TTL:
|
||||
return cached_content
|
||||
|
||||
dashboard_path = STATIC_DIR / "dashboard.html"
|
||||
if dashboard_path.is_file():
|
||||
content = dashboard_path.read_text(encoding="utf-8")
|
||||
_dashboard_html_cache = (content, now)
|
||||
return content
|
||||
return "<h1>dashboard.html not found</h1>"
|
||||
|
||||
|
||||
@webui_router.get("/dashboard", include_in_schema=False)
|
||||
async def dashboard_page() -> HTMLResponse:
|
||||
"""仪表盘 HTML 页面（含缓存策略）。"""
|
||||
return HTMLResponse(content=_get_dashboard_html())
|
||||
@@ -0,0 +1,279 @@
|
||||
# 多智能体文档存储、命名与索引规范 v1.0
|
||||
|
||||
> 版本：v1.0（实施版）
|
||||
> 编制：陆怀瑾（COO）
|
||||
> 日期：2026-06-22
|
||||
> 状态：已批准，执行中
|
||||
> 适用范围：所有 Agent 的 workspace 目录
|
||||
|
||||
---
|
||||
|
||||
## 一、统一目录结构
|
||||
|
||||
每个 Agent 的 workspace 必须采用以下标准目录结构：
|
||||
|
||||
```
|
||||
workspace/
|
||||
├── AGENTS.md # Agent 协作协议
|
||||
├── MEMORY.md # 长期记忆 → 含文档索引表（核心）
|
||||
├── SOUL.md # 角色定义 → 引用外部内容，不填塞
|
||||
├── IDENTITY.md # 身份信息
|
||||
├── USER.md # 用户画像
|
||||
├── TOOLS.md # 工具清单 → 仅保留索引，详情外挂
|
||||
├── HEARTBEAT.md # 心跳配置
|
||||
│
|
||||
├── memory/ # 记忆归档目录（按日期）
|
||||
│ └── YYYY-MM-DD.md
|
||||
│
|
||||
├── docs/ # 项目文档目录（按项目分）
|
||||
│ └── {project}/
|
||||
│ ├── README.md
|
||||
│ └── ...
|
||||
│
|
||||
├── plans/ # 方案文档目录
|
||||
│ └── YYYY-MM-DD_{topic}.md
|
||||
│
|
||||
├── specs/ # 规范/标准文档目录
|
||||
│ └── BIZ-XX_{name}_v{M}.{N}.md
|
||||
│
|
||||
├── reports/ # 运营报告目录
|
||||
│ └── YYYY-Q{N}_{type}_v{M}.{N}.md
|
||||
│
|
||||
├── knowledge/ # 知识库目录（按领域分）
|
||||
│ └── {domain}/
|
||||
│ └── {topic}.md
|
||||
│
|
||||
├── tasks/ # 任务文件目录（可选）
|
||||
│ └── ...
|
||||
│
|
||||
└── assets/ # 资源文件目录
|
||||
├── images/
|
||||
├── files/
|
||||
└── templates/
|
||||
```
|
||||
|
||||
### 目录用途速查
|
||||
|
||||
| 目录 | 用途 | Token 影响 |
|
||||
|------|------|-----------|
|
||||
| 根目录 .md | Agent 核心配置 | **直接影响 Token**，必须精简 |
|
||||
| memory/ | 按日归档记忆 | 通过 memory_search 检索，不占用上下文 |
|
||||
| docs/ | 项目文档 | 按需加载 |
|
||||
| plans/ | 方案文档 | 仅 COO 维护 |
|
||||
| specs/ | 规范标准 | 按需加载 |
|
||||
| reports/ | 运营报告 | 仅 COO 维护 |
|
||||
| knowledge/ | 知识库 | 知识库检索，不占用上下文 |
|
||||
| assets/ | 二进制资源 | 不占用上下文 |
|
||||
|
||||
---
|
||||
|
||||
## 二、文件命名规则
|
||||
|
||||
### 2.1 强制命名格式
|
||||
|
||||
```
|
||||
{日期/编号}_{中文主题}_{版本}.md
|
||||
```
|
||||
|
||||
### 2.2 各目录命名约定
|
||||
|
||||
| 目录 | 命名模式 | 示例 |
|
||||
|------|----------|------|
|
||||
| memory/ | `YYYY-MM-DD.md` | `2026-06-22.md` |
|
||||
| plans/ | `YYYY-MM-DD_{主题}.md` | `2026-06-22_多智能体协作体系总体方案.md` |
|
||||
| specs/ | `BIZ-{编号}_{主题}_v{M}.{N}.md` | `BIZ-12_文档存储规范_v1.0.md` |
|
||||
| reports/ | `YYYY-Q{N}_{类型}_v{M}.{N}.md` | `2026-Q2_运营效率报告_v1.0.md` |
|
||||
| knowledge/ | `{主题}_v{M}.{N}.md` | `淘宝运营_SOP_v1.0.md` |
|
||||
| docs/{project}/ | `{功能}_{版本}.md` | `requirements_v1.0.md` |
|
||||
| memory/ day file | `YYYY-MM-DD.md` | `2026-06-22.md` |
|
||||
|
||||
### 2.3 禁止事项
|
||||
|
||||
- ❌ 使用特殊字符：`/ \ : * ? " < > |` 空格
|
||||
- ❌ 超过 80 字符的文件名
|
||||
- ❌ 不含日期/编号的裸文件名
|
||||
- ❌ 中文和英文混排无分隔符
|
||||
- ✅ 统一使用下划线 `_` 作为分隔符
|
||||
|
||||
---
|
||||
|
||||
## 三、索引机制（核心）
|
||||
|
||||
### 3.1 索引分离原则（刘总反馈已纳入）
|
||||
|
||||
> **配置文件只保留索引指针，详细内容外挂存储。**
|
||||
|
||||
此原则适用场景：
|
||||
- **TOOLS.md**：只列工具名称 + 引用路径，不列完整参数
|
||||
- **待办列表**：只记录 ID + 主题 + 状态，详情在独立文件中
|
||||
- **Agent 协作表**：只列 Agent 名 + 职能 + Session Key，详情在各自文件
|
||||
- **知识索引**：MEMORY.md 只保留索引表，知识条目在 knowledge/ 中
|
||||
|
||||
### 3.2 MEMORY.md 索引表模板
|
||||
|
||||
```markdown
|
||||
# MEMORY.md - {Agent Name} 长期记忆
|
||||
|
||||
## 📑 文档索引
|
||||
|
||||
### 方案文档
|
||||
| 日期 | 主题 | 路径 | 状态 |
|
||||
|------|------|------|------|
|
||||
| 2026-06-22 | 多智能体协作体系 | plans/2026-06-22_多智能体协作体系总体方案.md | 已批准 |
|
||||
|
||||
### 规范标准
|
||||
| 编号 | 主题 | 路径 | 版本 |
|
||||
|------|------|------|------|
|
||||
| BIZ-12 | 文档存储规范 | specs/BIZ-12_文档存储规范_v1.0.md | v1.0 |
|
||||
|
||||
### 项目文档
|
||||
| 项目 | 文档 | 路径 | 状态 |
|
||||
|------|------|------|------|
|
||||
|
||||
### 运营报告
|
||||
| 周期 | 类型 | 路径 | 状态 |
|
||||
|------|------|------|------|
|
||||
|
||||
### 知识库条目
|
||||
| 领域 | 主题 | 路径 | 更新时间 |
|
||||
|------|------|------|----------|
|
||||
|
||||
---
|
||||
（以下是实际记忆内容...）
|
||||
```
|
||||
|
||||
### 3.3 各目录 README.md 模板
|
||||
|
||||
每个目录应有一个 `README.md`：
|
||||
|
||||
```markdown
|
||||
# {目录名称}
|
||||
|
||||
> 最后更新：{YYYY-MM-DD}
|
||||
> 维护者：{Agent Name}
|
||||
|
||||
## 目录说明
|
||||
{简短描述本目录的用途和使用规范}
|
||||
|
||||
## 文件列表
|
||||
| 文件名 | 描述 | 最后更新 |
|
||||
|--------|------|----------|
|
||||
| ... | ... | ... |
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 四、检索体系
|
||||
|
||||
### 4.1 分层检索路径
|
||||
|
||||
```
|
||||
第一层：memory_search（语义检索，跨 memory/*.md + MEMORY.md）
|
||||
↓ 未命中
|
||||
第二层：wiki_search / wiki_get（编译型知识库检索）
|
||||
↓ 未命中
|
||||
第三层：qmd（QMD 全文检索，已安装 —— 刘总反馈已纳入）
|
||||
↓ 未命中
|
||||
第四层：web_fetch / web_search（外部知识）
|
||||
```
|
||||
|
||||
### 4.2 检索优先级
|
||||
|
||||
1. **memory_search**（corpus=all）：首选，零 token 消耗
|
||||
2. **qmd**：本地全文检索，补充 memory_search 未覆盖的长文档
|
||||
3. **wiki_search/wiki_get**：编译型结构化知识库
|
||||
4. **web_search/web_fetch**：外部补充，仅在以上均未命中时使用
|
||||
|
||||
---
|
||||
|
||||
## 五、Token 预算控制
|
||||
|
||||
### 5.1 配置文件大小限制
|
||||
|
||||
| 文件 | 最大行数 | 说明 |
|
||||
|------|----------|------|
|
||||
| AGENTS.md | 200 行 | Agent 协议 + 协作表（精简） |
|
||||
| MEMORY.md | 150 行 | 长期记忆 + 索引表 |
|
||||
| SOUL.md | 80 行 | 角色定义 |
|
||||
| IDENTITY.md | 30 行 | 身份信息 |
|
||||
| USER.md | 30 行 | 用户画像 |
|
||||
| TOOLS.md | 100 行 | 工具索引（不填塞完整参数） |
|
||||
| HEARTBEAT.md | 60 行 | 心跳配置 |
|
||||
|
||||
### 5.2 引用代替填塞
|
||||
|
||||
**反例（填塞模式）**：
|
||||
```markdown
|
||||
# TOOLS.md - 全部填入
|
||||
- memory_search: 参数 query, maxResults, minScore, corpus=[memory|wiki|all|sessions]...
|
||||
（占用大量 token）
|
||||
```
|
||||
|
||||
**正例（引用模式）**：
|
||||
```markdown
|
||||
# TOOLS.md - 索引模式
|
||||
## 已安装 Skills
|
||||
- plantuml-skill → 详见 skills/plantuml-skill/SKILL.md
|
||||
- qmd → 详见 skills/qmd/SKILL.md
|
||||
- ...
|
||||
## 核心工具（已内置于运行时，无需列出参数）
|
||||
- memory_search / wiki_search / web_fetch
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 六、文档生命周期管理
|
||||
|
||||
### 6.1 状态流转
|
||||
|
||||
```
|
||||
创建 → 草稿(draft) → 审阅中(in_review) → 已批准(approved) → 归档(archived)
|
||||
↓
|
||||
废弃(deprecated)
|
||||
```
|
||||
|
||||
### 6.2 操作规范
|
||||
|
||||
| 操作 | 规则 |
|
||||
|------|------|
|
||||
| 创建 | 在正确目录，按命名规则创建 |
|
||||
| 更新 | 小改动直接覆盖；大改动新建版本 |
|
||||
| 审阅 | 状态标记 `in_review`，通知审阅人 |
|
||||
| 归档 | 移动到 `archive/` 子目录 |
|
||||
| 删除 | 不直接删除，先归档 30 天后清理 |
|
||||
|
||||
### 6.3 版本标记
|
||||
|
||||
- v1.0：首版
|
||||
- v1.1-v1.9：小修
|
||||
- v2.0+：大修 / 重构
|
||||
|
||||
---
|
||||
|
||||
## 七、Agent 端执行规范
|
||||
|
||||
### 7.1 每次任务后
|
||||
1. 更新 `memory/YYYY-MM-DD.md`（日记）
|
||||
2. 如产出文档，更新 MEMORY.md 索引表
|
||||
3. 检查文件名是否符合规范
|
||||
|
||||
### 7.2 每周
|
||||
1. 检查并清理过期文档（移动到 archive/）
|
||||
2. 验证索引表与实际文件一致性
|
||||
3. 检查配置文件大小是否超限
|
||||
|
||||
---
|
||||
|
||||
## 八、实施检查清单
|
||||
|
||||
- [x] 规范文档编制（本文档）
|
||||
- [ ] 各 Agent workspace 目录结构初始化
|
||||
- [ ] MEMORY.md 索引表模板部署到所有 Agent
|
||||
- [ ] TOOLS.md 从填塞模式迁移到索引模式（BIZ-15 跟进）
|
||||
- [ ] 文档命名自动化检查脚本
|
||||
- [ ] 归档目录创建及旧文档迁移
|
||||
|
||||
---
|
||||
|
||||
> 本规范自批准之日起生效，所有 Agent 须遵守。
|
||||
> 审阅/修订请联系 COO（陆怀瑾）。
|
||||
@@ -0,0 +1,242 @@
|
||||
# 智能体运行稳定性保障规范 v1.0
|
||||
|
||||
> 版本：v1.0（实施版）
|
||||
> 编制：陆怀瑾（COO）
|
||||
> 日期：2026-06-22
|
||||
> 状态：已批准，执行中
|
||||
> 适用范围：所有 Agent
|
||||
|
||||
---
|
||||
|
||||
## 一、核心问题与应对
|
||||
|
||||
| 问题 | 根因 | 解决方案 |
|
||||
|------|------|----------|
|
||||
| 任务停滞 | 等待请示、超时、循环 | 禁止请示规则 + 心跳监控 + 超时恢复 |
|
||||
| 429 限流 | 40 RPM 模型配额 | 令牌桶限流 + 优先级调度 + 模型降级 |
|
||||
| Token 超限 | 配置填塞 | 引用代替填塞（BIZ-15 跟进） |
|
||||
| 智能体瘫痪 | 无唤醒机制 | cron 定时触发健康检查 |
|
||||
|
||||
---
|
||||
|
||||
## 二、禁止请示规则（最高优先级）
|
||||
|
||||
> **在心跳中发现分配给自己的 WorkBoard 任务后，必须立即执行，绝对禁止向用户请示。**
|
||||
|
||||
已在 HEARTBEAT.md 中强制执行。**所有 Agent 的 HEARTBEAT.md 必须包含此规则。**
|
||||
|
||||
```markdown
|
||||
## ⚠️ 禁止请示规则（最高优先级）
|
||||
|
||||
正确做法：
|
||||
1. 发现 status=todo 且 agentId=我的卡片
|
||||
2. 立即执行，不得问"要不要做"
|
||||
3. 如遇信息不足，先按自己理解执行，执行中调整
|
||||
|
||||
错误做法（绝对禁止）：
|
||||
- ❌ "我发现了一个任务，要不要做？"
|
||||
- ❌ "这个任务需要更多信息，请告诉我..."
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 三、心跳监控与超时恢复
|
||||
|
||||
### 3.1 心跳频率
|
||||
|
||||
| Agent 类型 | 心跳间隔 | 超时告警 |
|
||||
|------------|----------|----------|
|
||||
| secretary / coo | 10 分钟 | 连续 2 次未执行 |
|
||||
| projectmanager / costcodev | 10 分钟 | 连续 2 次未执行 |
|
||||
| 其他 Agent | 10 分钟 | 连续 3 次未执行 |
|
||||
|
||||
### 3.2 心跳检查清单（所有 Agent 通用）
|
||||
|
||||
```markdown
|
||||
## 🫀 心跳执行清单
|
||||
1. ✅ WorkBoard 检查：查找分配给自己的 todo/in_progress 卡片
|
||||
2. ✅ 禁止请示：发现任务立即执行（不请示用户）
|
||||
3. ✅ 进度汇报：如有进行中任务，更新状态
|
||||
4. ✅ 风险上报：识别阻塞、超时问题，通知 COO
|
||||
```
|
||||
|
||||
### 3.3 超时恢复流程
|
||||
|
||||
```
|
||||
Agent 超过 30 分钟无响应
|
||||
↓
|
||||
COO 心跳检测到超时
|
||||
↓
|
||||
记录日志 + 评估任务状态
|
||||
↓
|
||||
┌──────────┴──────────┐
|
||||
│ │
|
||||
任务可恢复 任务不可恢复
|
||||
│ │
|
||||
重新触发任务 通知 Vincent
|
||||
(workboard dispatch) (via session_send)
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 四、429 限流治理
|
||||
|
||||
### 4.1 当前配额与监控
|
||||
|
||||
| 模型 | RPM 限制 | 建议预留 |
|
||||
|------|----------|----------|
|
||||
| 主模型 | 40 | 保留 10 RPM 给紧急任务 |
|
||||
| 备用模型 | 40 | 满 35 RPM 时切换 |
|
||||
|
||||
### 4.2 令牌桶限流策略
|
||||
|
||||
```
|
||||
每个 Agent 独立的令牌桶:
|
||||
- 容量：按 Agent 优先级分配
|
||||
- COO/secretary: 8 RPM
|
||||
- 开发 Agent: 6 RPM
|
||||
- 业务 Agent: 4 RPM
|
||||
- 预留池: 10 RPM (紧急任务)
|
||||
|
||||
令牌桶耗尽 → 自动降级到备用模型或排队
|
||||
```
|
||||
|
||||
### 4.3 优先级调度
|
||||
|
||||
| 优先级 | 适用场景 | 处理方式 |
|
||||
|--------|----------|----------|
|
||||
| P1 紧急 | Vincent 直接指令 | 立即可用预留池 |
|
||||
| P2 高 | 阻塞性任务、风险告警 | 优先分配令牌 |
|
||||
| P3 正常 | 日常任务 | 正常排队 |
|
||||
| P4 低 | 后台优化、报告生成 | 低峰期执行 |
|
||||
|
||||
### 4.4 模型降级链
|
||||
|
||||
```
|
||||
主模型 (qwen3.5-397b) RPM 不足
|
||||
↓
|
||||
备用模型 (deepseek-v4-pro)
|
||||
↓
|
||||
等待 + 指数退避重试 (1s → 2s → 4s → 8s)
|
||||
↓
|
||||
3 次重试后仍失败 → 记录日志，通知 COO
|
||||
```
|
||||
|
||||
### 4.5 请求合并优化
|
||||
|
||||
| 优化项 | 当前做法 | 优化后 |
|
||||
|--------|----------|--------|
|
||||
| WorkBoard 轮询 | 每个 Agent 独立轮询 | COO 统一轮询，广播结果 |
|
||||
| 重复检索 | 多个 Agent 重复查同一文档 | 缓存关键查询结果（5 分钟 TTL） |
|
||||
| 连续调用 | 无间隔连续调用 API | 最小间隔 500ms |
|
||||
|
||||
---
|
||||
|
||||
## 五、唤醒机制
|
||||
|
||||
### 5.1 Cron 定时唤醒
|
||||
|
||||
```yaml
|
||||
# COO 健康检查唤醒
|
||||
cron:
|
||||
schedule: "*/5 * * * *" # 每 5 分钟
|
||||
action: health_check
|
||||
targets:
|
||||
- 检查所有 Agent 最后活跃时间
|
||||
- 超过 15 分钟无活动 → 触发唤醒消息
|
||||
- 超过 30 分钟无活动 → 通知 Vincent
|
||||
```
|
||||
|
||||
### 5.2 唤醒消息模板
|
||||
|
||||
```markdown
|
||||
## 🔔 唤醒检查
|
||||
|
||||
距上次活跃时间：{elapsed} 分钟
|
||||
当前任务状态：{status}
|
||||
是否存在阻塞：{blocked}
|
||||
|
||||
系统自动唤醒，请确认状态。
|
||||
```
|
||||
|
||||
### 5.3 自唤醒规则
|
||||
|
||||
每个 Agent 在 HEARTBEAT.md 中配置：
|
||||
|
||||
```
|
||||
如果距上次心跳超过 2 个周期（20 分钟）：
|
||||
→ 自动重新评估任务状态
|
||||
→ 如有待办，立即执行
|
||||
→ 如无待办，确认存活
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 六、上下文/Token 溢出防护
|
||||
|
||||
### 6.1 配置文件大小限制
|
||||
|
||||
| 文件 | 最大行数 | 超标处理 |
|
||||
|------|----------|----------|
|
||||
| AGENTS.md | 200 | 移到 docs/agent-roster.md |
|
||||
| SOUL.md | 80 | 提取模块化引用 |
|
||||
| TOOLS.md | 100 | 索引化（不填塞参数） |
|
||||
| HEARTBEAT.md | 60 | 精简检查清单 |
|
||||
| MEMORY.md | 150 | 定期归档旧条目 |
|
||||
|
||||
### 6.2 运行时监控
|
||||
|
||||
```
|
||||
Token 使用量达到 80%
|
||||
↓
|
||||
自动清理上下文
|
||||
↓
|
||||
仍超 90%
|
||||
↓
|
||||
重启会话
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 七、监控告警矩阵
|
||||
|
||||
| 指标 | 警告阈值 | 严重阈值 | 通知对象 |
|
||||
|------|----------|----------|----------|
|
||||
| Agent 无响应 | > 15 min | > 30 min | 警告 → COO，严重 → Vincent |
|
||||
| 429 错误率 | > 5% | > 20% | COO |
|
||||
| Token 使用量 | > 80% | > 95% | 该 Agent |
|
||||
| 任务积压 | > 5 pending | > 10 pending | COO |
|
||||
| 任务超时 | > 24h in_progress | > 48h | 警告 → Agent，严重 → Vincent |
|
||||
|
||||
---
|
||||
|
||||
## 八、实施步骤
|
||||
|
||||
### 阶段 1：即刻生效（今日）
|
||||
- [x] 禁止请示规则 → 已在各 Agent HEARTBEAT.md 中落实
|
||||
- [ ] 心跳频率统一为 10 分钟
|
||||
- [ ] COO 端健康检查 cron 配置
|
||||
|
||||
### 阶段 2：本周完成
|
||||
- [ ] 令牌桶限流配置（按 Agent 分配 RPM）
|
||||
- [ ] 模型降级链配置
|
||||
- [ ] 告警规则上线
|
||||
|
||||
### 阶段 3：持续优化
|
||||
- [ ] 监控面板搭建
|
||||
- [ ] 自动重启恢复
|
||||
- [ ] 请求合并优化
|
||||
|
||||
---
|
||||
|
||||
## 九、交付物清单
|
||||
|
||||
- [x] 运行稳定性保障规范（本文档）
|
||||
- [ ] HEARTBEAT.md 模板更新（含禁止请示 + 自唤醒规则）
|
||||
- [ ] COO 端 cron 健康检查任务
|
||||
- [ ] 令牌桶限流配置
|
||||
- [ ] 告警规则配置
|
||||
|
||||
---
|
||||
|
||||
> 本规范自批准之日起生效。执行中如遇问题，联系 COO（陆怀瑾）。
|
||||
@@ -0,0 +1,261 @@
|
||||
# 智能体知识库体系建设规范 v1.0
|
||||
|
||||
> 版本：v1.0（实施版）
|
||||
> 编制：陆怀瑾（COO）
|
||||
> 日期：2026-06-22
|
||||
> 状态：已批准，执行中
|
||||
> 适用范围：所有 Agent
|
||||
|
||||
---
|
||||
|
||||
## 一、核心目标
|
||||
|
||||
| 目标 | 实现方式 |
|
||||
|------|----------|
|
||||
| 知识与配置解耦 | 知识库独立于 Agent 配置文件，不计入 Token |
|
||||
| Agent 可主动查询 | 通过多层检索体系按需获取知识 |
|
||||
| 人类可审查优化 | Web UI / 飞书文档支持人工审阅 |
|
||||
| 零 Token 增长 | 知识条目独立存储，仅在使用时加载 |
|
||||
|
||||
---
|
||||
|
||||
## 二、分层检索体系（刘总反馈已纳入）
|
||||
|
||||
### 2.1 检索优先级
|
||||
|
||||
```
|
||||
Agent 需要知识
|
||||
↓
|
||||
第一层: memory_search (corpus=all)
|
||||
→ 搜索 memory/*.md + MEMORY.md + wiki 条目
|
||||
→ 零 Token 消耗，语义检索
|
||||
↓ 未命中
|
||||
第二层: wiki_search / wiki_get
|
||||
→ 编译型结构化知识库
|
||||
→ 支持精确检索和页面读取
|
||||
↓ 未命中
|
||||
第三层: qmd (QMD 全文检索，已安装 ← 刘总反馈)
|
||||
→ 本地全文检索 markdown 知识库
|
||||
→ 补充 memory_search 未覆盖的长文档
|
||||
↓ 未命中
|
||||
第四层: web_search / web_fetch
|
||||
→ 外部互联网补充
|
||||
→ 仅在内部均未命中时使用
|
||||
```
|
||||
|
||||
### 2.2 工具速查
|
||||
|
||||
| 工具 | 适用场景 | Token 消耗 |
|
||||
|------|----------|-----------|
|
||||
| memory_search | 通用语义检索（跨 memory + wiki） | 0 |
|
||||
| wiki_search / wiki_get | 结构化知识库精确查询 | 0 |
|
||||
| qmd | 本地全文检索长文档 | 0 |
|
||||
| web_search | 外部互联网信息 | 0 |
|
||||
| web_fetch | 网页/文档详情获取 | 按内容量 |
|
||||
|
||||
---
|
||||
|
||||
## 三、知识库目录结构
|
||||
|
||||
```
|
||||
knowledge/
|
||||
├── 电商/ # 电商运营知识
|
||||
│ └── {主题}_v{M}.{N}.md
|
||||
│
|
||||
├── 内容/ # 内容运营知识
|
||||
│ └── {主题}_v{M}.{N}.md
|
||||
│
|
||||
├── 产品/ # 产品管理知识
|
||||
│ └── {主题}_v{M}.{N}.md
|
||||
│
|
||||
├── 技术/ # 技术开发知识
|
||||
│ └── {主题}_v{M}.{N}.md
|
||||
│
|
||||
├── 设计/ # UI/UX 设计知识
|
||||
│ └── {主题}_v{M}.{N}.md
|
||||
│
|
||||
├── 运营/ # 通用运营知识
|
||||
│ └── {主题}_v{M}.{N}.md
|
||||
│
|
||||
└── 规范/ # 流程规范知识
|
||||
└── {主题}_v{M}.{N}.md
|
||||
```
|
||||
|
||||
### 知识条目模板
|
||||
|
||||
```markdown
|
||||
# {知识标题}
|
||||
|
||||
> 领域: {所属领域} | 版本: v{M}.{N}
|
||||
> 维护者: {责任人} | 最后更新: {YYYY-MM-DD}
|
||||
|
||||
## 概述
|
||||
{知识的用途和价值，1-2 句话}
|
||||
|
||||
## 适用范围
|
||||
{在什么场景下使用}
|
||||
|
||||
## 核心内容
|
||||
{知识主体}
|
||||
|
||||
## 操作步骤 / SOP
|
||||
1. ...
|
||||
2. ...
|
||||
|
||||
## 质量检查
|
||||
- [ ] ...
|
||||
|
||||
## 常见问题
|
||||
**Q**: ... **A**: ...
|
||||
|
||||
## 相关条目
|
||||
- knowledge/{领域}/{关联主题}.md
|
||||
|
||||
## 版本历史
|
||||
| 版本 | 日期 | 变更 | 作者 |
|
||||
|------|------|------|------|
|
||||
| v1.0 | 2026-06-22 | 初稿 | 陆怀瑾 |
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 四、与 Memory 系统的分工
|
||||
|
||||
| 维度 | Memory 系统 | Knowledge 系统 |
|
||||
|------|------------|---------------|
|
||||
| 内容类型 | 决策记录、经验教训、个性化记忆 | SOP、模板、规范、最佳实践 |
|
||||
| 所有者 | 单个 Agent 专属 | 跨 Agent 共享 |
|
||||
| 更新频率 | 每日/每周 | 按需/按版本 |
|
||||
| 查询方式 | memory_search（语义检索） | wiki_search/wiki_get/qmd |
|
||||
| 存储位置 | MEMORY.md + memory/*.md | knowledge/ 目录 |
|
||||
|
||||
---
|
||||
|
||||
## 五、Agent 查询指南
|
||||
|
||||
### 5.1 何时查询知识库
|
||||
|
||||
| 场景 | 查询示例 |
|
||||
|------|----------|
|
||||
| 执行 SOP 任务 | "淘宝 活动报名 SOP" |
|
||||
| 撰写文档 | "PRD 模板" |
|
||||
| 遇到问题 | "部署 故障排查" |
|
||||
| 制定规范 | "开发规范" |
|
||||
| 不熟悉领域 | "小红书 运营指南" |
|
||||
|
||||
### 5.2 查询标准流程
|
||||
|
||||
```
|
||||
1. 先用 memory_search(corpus=all, query="...") 搜索
|
||||
2. 如有结果，用 memory_get 或 wiki_get 读取详情
|
||||
3. 如无结果，用 qmd 全文检索 knowledge/ 目录
|
||||
4. 仍无结果，记录知识缺口，通知 COO
|
||||
5. 使用获取的知识指导工作
|
||||
```
|
||||
|
||||
### 5.3 知识缺口上报
|
||||
|
||||
```
|
||||
Agent 发现知识缺口
|
||||
↓
|
||||
在 memory/YYYY-MM-DD.md 中记录：
|
||||
- 查询内容
|
||||
- 使用场景
|
||||
- 建议优先级
|
||||
↓
|
||||
通知 COO 创建知识条目
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 六、人类审查机制
|
||||
|
||||
### 6.1 审查方式
|
||||
|
||||
| 方式 | 适用场景 | 工具 |
|
||||
|------|----------|------|
|
||||
| Obsidian Web UI | 日常浏览、编辑 | wiki_status 确认可用性 |
|
||||
| 飞书文档同步 | 多人协作、审批 | 飞书 Wiki API |
|
||||
| CLI 直接编辑 | 技术人员修改 | write/edit 工具 |
|
||||
|
||||
### 6.2 审核流程
|
||||
|
||||
```
|
||||
Agent 发现缺口 → 记录 → 通知 COO
|
||||
↓
|
||||
COO 评估优先级
|
||||
↓
|
||||
高优先级 → 立即创建/指派
|
||||
低优先级 → 记入 backlog
|
||||
↓
|
||||
创建草稿 → wiki_apply(op="create_synthesis")
|
||||
↓
|
||||
人类审查 → 通过/修改/拒绝
|
||||
↓
|
||||
发布 → 通知相关 Agent
|
||||
```
|
||||
|
||||
### 6.3 定期质量检查
|
||||
|
||||
```bash
|
||||
# 每周运行一次
|
||||
wiki_lint # 检查链接断裂、矛盾信息、过时内容
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 七、知识条目管理
|
||||
|
||||
### 7.1 创建
|
||||
|
||||
```
|
||||
wiki_apply(op="create_synthesis", title="...", body="...", sourceIds=[])
|
||||
```
|
||||
|
||||
### 7.2 更新
|
||||
|
||||
```
|
||||
wiki_apply(op="synthesis", lookup="...", body="...")
|
||||
```
|
||||
|
||||
### 7.3 版本管理
|
||||
|
||||
| 变更类型 | 版本变化 | 操作 |
|
||||
|----------|----------|------|
|
||||
| 内容微调 | v1.0 → v1.1 | 直接覆盖，更新版本历史 |
|
||||
| 结构性变更 | v1.x → v2.0 | 保留旧版本，新建条目 |
|
||||
| 废弃 | 添加 [deprecated] 标记 | 归档到 archive/ |
|
||||
|
||||
---
|
||||
|
||||
## 八、初始知识基础
|
||||
|
||||
以下条目作为知识库初始基础，需尽快创建：
|
||||
|
||||
| 领域 | 条目 | 优先级 | 负责 Agent |
|
||||
|------|------|--------|-----------|
|
||||
| 电商 | 淘宝运营 SOP | 高 | 陆云帆 |
|
||||
| 电商 | 客服话术模板 | 中 | 陆云帆 |
|
||||
| 内容 | 小红书运营指南 | 高 | 文墨言 |
|
||||
| 内容 | 标题写作技巧 | 中 | 文墨言 |
|
||||
| 产品 | PRD 模板 | 高 | 沈路明 |
|
||||
| 技术 | 开发规范 | 高 | 梁思筑 |
|
||||
| 技术 | 部署流程 | 中 | 严维序 |
|
||||
| 设计 | UI 设计规范 | 中 | 苏锦绘 |
|
||||
| 运营 | KPI 指标定义 | 中 | 陆怀瑾 |
|
||||
| 规范 | 文档存储规范 | 已完成 | 陆怀瑾 |
|
||||
|
||||
---
|
||||
|
||||
## 九、交付物清单
|
||||
|
||||
- [x] 知识库体系建设规范（本文档）
|
||||
- [ ] knowledge/ 目录结构创建
|
||||
- [ ] 初始知识条目（至少 5 个优先）
|
||||
- [ ] Agent 查询指南（已嵌入本文档）
|
||||
- [ ] 知识审核流程（已嵌入本文档）
|
||||
- [ ] wiki_lint 定期检查 cron 任务
|
||||
|
||||
---
|
||||
|
||||
> 本规范自批准之日起生效。知识条目创建请联系 COO 协调。
|
||||
@@ -0,0 +1,31 @@
|
||||
# [知识条目标题]
|
||||
|
||||
## 元数据
|
||||
|
||||
| 属性 | 值 |
|
||||
|------|-----|
|
||||
| **领域** | 电商 / 内容 / 产品 / 技术 / 设计 / 运营 / 行政 |
|
||||
| **责任人** | [Agent 名称] |
|
||||
| **版本** | v1.0 |
|
||||
| **创建日期** | YYYY-MM-DD |
|
||||
| **最后更新** | YYYY-MM-DD |
|
||||
| **标签** | [标签1, 标签2, ...] |
|
||||
|
||||
## 概述
|
||||
|
||||
[用 2-3 句话描述本条目的核心内容和使用场景]
|
||||
|
||||
## 正文
|
||||
|
||||
[详细的知识内容，包括步骤、规则、示例等]
|
||||
|
||||
## 相关条目
|
||||
|
||||
- [相关知识条目1](链接)
|
||||
- [相关知识条目2](链接)
|
||||
|
||||
## 变更记录
|
||||
|
||||
| 日期 | 版本 | 变更说明 | 变更人 |
|
||||
|------|------|----------|--------|
|
||||
| YYYY-MM-DD | v1.0 | 初始创建 | [姓名] |
|
||||
Reference in New Issue
Block a user