feat: auto committed
This commit is contained in:
@@ -1,13 +1,15 @@
|
||||
# data-ana 对接契约
|
||||
|
||||
> 负责人:ai11
|
||||
> 关联:[matrix.md](./matrix.md)、[analytics.proto](../../../packages/shared-proto/proto/analytics.proto)、[events.proto](../../../packages/shared-proto/proto/events.proto)
|
||||
> 关联:[matrix.md](../matrix.md)、[coord-cross-review.md](../../coord-cross-review.md)、[analytics.proto](../../../packages/shared-proto/proto/analytics.proto)、[events.proto](../../../packages/shared-proto/proto/events.proto)、[iam.proto](../../../packages/shared-proto/proto/iam.proto)
|
||||
> 对齐文档:[01-understanding.md](../../../services/data-ana/docs/01-understanding.md)、[02-architecture-design.md](../../../services/data-ana/docs/02-architecture-design.md)、[objections/data-ana_issue.md](../objections/data-ana_issue.md)、[worklines/data-ana_workline.md](../worklines/data-ana_workline.md)
|
||||
> 修订:v2(2026-07-10 by ai11)—— 修正 topic 命名 / gRPC 调用声明 / HTTP 端点声明 / 引用断裂,对齐 01/02 v2.1
|
||||
|
||||
---
|
||||
|
||||
## §1 我提供什么(对外接口)
|
||||
|
||||
### 1.1 gRPC 接口(如有)
|
||||
### 1.1 gRPC 接口
|
||||
|
||||
| Service | RPC | 请求 | 响应 | 端口 |
|
||||
| ---------------- | ---------------------- | ----------------------------- | ------------------------- | ----- |
|
||||
@@ -18,31 +20,71 @@
|
||||
| AnalyticsService | GetStudentDashboard | GetStudentDashboardRequest | StudentDashboard | 50055 |
|
||||
| AnalyticsService | GetParentDashboard | GetParentDashboardRequest | ParentDashboard | 50055 |
|
||||
| AnalyticsService | GetAdminDashboard | GetAdminDashboardRequest | AdminDashboard | 50055 |
|
||||
| AnalyticsService | GetWarningList | GetWarningListRequest | WarningListResponse | 50055 |
|
||||
| AnalyticsService | GetWarnings | GetWarningsRequest | WarningList | 50055 |
|
||||
| AnalyticsService | TriggerWarning | TriggerWarningRequest | TriggerWarningResponse | 50055 |
|
||||
| AnalyticsService | GetMasteryDistribution | GetMasteryDistributionRequest | MasteryDistribution | 50055 |
|
||||
| AnalyticsService | GetStudentMastery | GetStudentMasteryRequest | StudentMastery | 50055 |
|
||||
| AnalyticsService | SubscribeMasteryUpdate | SubscribeMasteryUpdateRequest | stream MasteryUpdateEvent | 50055 |
|
||||
|
||||
### 1.2 HTTP 端点(如有)
|
||||
> **proto 状态**:analytics.proto 当前仅 3 RPC(GetClassPerformance / GetStudentWeakness / GetLearningTrend),扩展至 12 RPC 待 coord 补全或 ai11 在 P4 阶段自行补全(见 [ISSUE-003](../objections/data-ana_issue.md))。完整 message 定义见 [02-architecture-design.md §4.2](../../../services/data-ana/docs/02-architecture-design.md)。
|
||||
> **gRPC 启用阶段**:P4 启用(coord-cross-review.md §2.3 裁决),P2-P3 仅 HTTP。
|
||||
> **响应信封**:所有 RPC 返回 ActionState[T](coord-cross-review.md §5.3 P0 整改)。
|
||||
|
||||
无对外 HTTP 端点,仅 gRPC(含 1 个 Server Streaming RPC)。
|
||||
### 1.2 HTTP 端点
|
||||
|
||||
data-ana 保留 HTTP :3006 端点作 Gateway 直连降级(gRPC 不可用时 BFF 可走 HTTP)。共 14 端点(3 基础 + 11 业务):
|
||||
|
||||
| method | path | 权限 | 响应 |
|
||||
| ------ | -------------------------------------------- | ----------------------------- | ----------------------------------- |
|
||||
| GET | `/healthz` | — | `{status, service}` |
|
||||
| GET | `/readyz` | — | `{status, ready, degraded, clickhouse, cdc_consumer, redis, iam_grpc}` |
|
||||
| GET | `/metrics` | — | Prometheus 格式 |
|
||||
| GET | `/analytics/class/{class_id}/performance` | `ANALYTICS_CLASS_READ` | `ActionState<ClassPerformanceData>` |
|
||||
| GET | `/analytics/student/{student_id}/weakness` | `ANALYTICS_STUDENT_READ` | `ActionState<StudentWeaknessData>` |
|
||||
| GET | `/analytics/student/{student_id}/errorbook` | `ANALYTICS_STUDENT_READ` | `ActionState<StudentErrorBookData>` |
|
||||
| GET | `/analytics/student/{student_id}/trend` | `ANALYTICS_STUDENT_READ` | `ActionState<LearningTrendData>` |
|
||||
| GET | `/analytics/student/{student_id}/attendance` | `ANALYTICS_STUDENT_READ` | `ActionState<AttendanceData>` |
|
||||
| GET | `/analytics/dashboard/teacher/{user_id}` | `ANALYTICS_TEACHER_DASHBOARD` | `ActionState<TeacherDashboardData>` |
|
||||
| GET | `/analytics/dashboard/student/{user_id}` | `ANALYTICS_STUDENT_DASHBOARD` | `ActionState<StudentDashboardData>` |
|
||||
| GET | `/analytics/dashboard/parent/{user_id}` | `ANALYTICS_PARENT_DASHBOARD` | `ActionState<ParentDashboardData>` |
|
||||
| GET | `/analytics/dashboard/admin/{user_id}` | `ANALYTICS_ADMIN_DASHBOARD` | `ActionState<AdminDashboardData>` |
|
||||
| GET | `/analytics/warnings` | `ANALYTICS_WARNING_READ` | `ActionState<WarningListData>` |
|
||||
| GET | `/analytics/mastery/distribution` | `ANALYTICS_CLASS_READ` | `ActionState<MasteryDistributionData>` |
|
||||
|
||||
> 完整端点设计见 [02-architecture-design.md §4.1](../../../services/data-ana/docs/02-architecture-design.md)。
|
||||
|
||||
### 1.3 GraphQL schema(如 BFF)
|
||||
|
||||
不适用。
|
||||
不适用。data-ana 是业务服务,不提供 GraphQL。BFF 层(teacher-bff / student-bff / parent-bff)聚合 data-ana gRPC 后对外暴露 GraphQL。
|
||||
|
||||
### 1.4 Kafka 事件发布(如有)
|
||||
### 1.4 Kafka 事件发布
|
||||
|
||||
| Topic | Event | 消费方 |
|
||||
| --------------------------- | --------------------------------------------------------- | -------------- |
|
||||
| edu.data_ana.mastery.events | MasteryEvent(action: mastery.updated/warning.triggered) | core-edu / msg |
|
||||
| Topic | Event | 消费方 | Outbox | 说明 |
|
||||
| -------------------------------- | --------------- | ------------------------------- | ------ | ---- |
|
||||
| `edu.insight.mastery.updated` | MasteryUpdated | core-edu(推荐个性化练习)/ msg | ❌ 豁免 | 掌握度计算完成触发 |
|
||||
| `edu.insight.warning.triggered` | WarningTriggered | msg(推送通知)/ core-edu(标记关注) | ❌ 豁免 | 预警阈值触发 |
|
||||
|
||||
> 注:MasteryEvent 豁免 Outbox 模式(派生数据事件,见 004 §12.2 + §15.3 #6)。
|
||||
> **Outbox 豁免**:派生数据事件(非业务事务写),豁免 Outbox 约束([coord-cross-review.md §3.3](../../coord-cross-review.md) 已仲裁)。直接用 aiokafka AIOKafkaProducer 发布,`idempotent=true` + `transactional_id="data-ana-producer"`。
|
||||
>
|
||||
> **topic 命名说明**:按 004 §7.2 命名规范 `edu.<domain>.<aggregate>.<action>`,data-ana 属于 D6 智能洞察领域(domain=insight),故 `edu.insight.mastery.updated` 符合规范。matrix.md §4 中 `edu.analytics.mastery` 命名不符合规范(缺 action 层级,domain 用了服务名而非领域名),已提请 coord 统一(见 [ISSUE-005](../objections/data-ana_issue.md))。
|
||||
|
||||
### 1.5 错误码前缀
|
||||
|
||||
`DATA_ANA_`(如 DATA_ANA_DASHBOARD_UNAVAILABLE、DATA_ANA_MASTERY_NOT_COMPUTED)
|
||||
`DATA_ANA_`(如 DATA_ANA_DASHBOARD_UNAVAILABLE、DATA_ANA_MASTERY_NOT_COMPUTED、DATA_ANA_CLICKHOUSE_UNAVAILABLE)
|
||||
|
||||
> 来源:[matrix.md §6](../matrix.md) 错误码前缀矩阵。
|
||||
|
||||
### 1.6 ClickHouse 宽表(供 coord 统一管理 DDL)
|
||||
|
||||
| 宽表名 | 用途 | 引擎 |
|
||||
| ------------------------- | -------------------- | ----------------------------- |
|
||||
| student_dashboard_view | 学生学情宽表 | ReplacingMergeTree(last_updated) |
|
||||
| student_errors | 学生错题本 | ReplacingMergeTree(last_error_time) |
|
||||
| mastery_snapshot | 知识点掌握度历史快照 | MergeTree |
|
||||
| attendance_logs | 学生考勤记录 | ReplacingMergeTree(occurred_at) |
|
||||
| ai_usage_log | AI 用量计费记录 | ReplacingMergeTree(occurred_at) |
|
||||
|
||||
> DDL 由 coord 统一管理在 `infra/clickhouse/ddl/`(待 coord 建立),data-ana 提供内容。完整 DDL 见 [02-architecture-design.md §3](../../../services/data-ana/docs/02-architecture-design.md)。
|
||||
|
||||
---
|
||||
|
||||
@@ -50,29 +92,43 @@
|
||||
|
||||
### 2.1 gRPC 调用(同步)
|
||||
|
||||
无主动 gRPC 调用上游。data-ana 通过 CDC + Kafka 事件接收数据,计算后发布 MasteryEvent。
|
||||
| 调用方 | 被调用方 | RPC | 用途 | 端口 | 阶段 | 状态 |
|
||||
| -------- | -------- | ------------------------- | ---------------------------- | ----- | ---- | ---- |
|
||||
| data-ana | iam | GetEffectiveDataScope | DataScope 6 级过滤解析 | 50052 | P4 | ⚠️ iam.proto 当前未实现(ISSUE-001) |
|
||||
|
||||
> **降级兜底**:iam GetEffectiveDataScope 未就绪时,data-ana 按 role 映射默认 DataScope(教师=CLASS,学生=SELF,管理员=SCHOOL),标注 `details.degraded: true`。结果 Redis 缓存 5min(key: `data_ana:datascope:{user_id}`)。
|
||||
>
|
||||
> **裁决依据**:coord-cross-review.md §2.2 #3 已仲裁 iam P4 补全此 RPC。
|
||||
|
||||
### 2.2 Kafka 事件订阅(异步)
|
||||
|
||||
| Topic | Event | 发布方 | mock 策略 |
|
||||
| ---------------------------------- | ------------------- | --------------- | ------------------------------------------------- |
|
||||
| edu.exam.events | ExamEvent | core-edu (ai08) | core-edu 就绪前使用 CDC 模拟数据 + 本地 stub 事件 |
|
||||
| edu.homework.events | HomeworkEvent | core-edu (ai08) | 同上 |
|
||||
| edu.grade.events | GradeEvent | core-edu (ai08) | 同上 |
|
||||
| edu.class.events | ClassEvent | core-edu (ai08) | 同上 |
|
||||
| edu.content.knowledge_point.events | KnowledgePointEvent | content (ai09) | content 就绪前使用内置知识点维度表 |
|
||||
| edu.content.question.events | QuestionEvent | content (ai09) | content 就绪前忽略 |
|
||||
| edu.ai.usage.events | AIUsageEvent | ai (ai12) | ai 就绪前忽略,AI 用量统计为空 |
|
||||
| Topic | Event | 发布方 | mock 策略 |
|
||||
| -------------------------------------- | ------------------- | --------------- | ------------------------------------------------- |
|
||||
| `edu.teaching.exam.published` | ExamEvent | core-edu (ai08) | core-edu 就绪前使用 CDC 模拟数据 + 本地 stub 事件 |
|
||||
| `edu.teaching.homework.assigned` | HomeworkEvent | core-edu (ai08) | 同上 |
|
||||
| `edu.teaching.grade.recorded` | GradeEvent | core-edu (ai08) | 同上 |
|
||||
| `edu.teaching.class.transferred` | ClassEvent | core-edu (ai08) | 同上 |
|
||||
| `edu.content.kp.events` | KnowledgePointEvent | content (ai09) | content 就绪前使用内置知识点维度表 |
|
||||
| `edu.content.question.events` | QuestionEvent | content (ai09) | content 就绪前忽略 |
|
||||
| `edu.insight.ai.usage` | AIUsageEvent | ai (ai12) | ai 就绪前忽略,AI 用量统计为空(P5) |
|
||||
|
||||
> **topic 命名对齐**:core-edu 教学事件 topic 按 coord-cross-review.md §3.1 裁决统一为 `edu.teaching.<aggregate>.<action>` 风格。core-edu 代码实际发布 `edu.exam.events` 等,待 core-edu 整改后切换。
|
||||
>
|
||||
> **AIUsageEvent 缺失**:events.proto 当前缺 AIUsageEvent message(ISSUE-002),P5 前需 coord 补全。
|
||||
|
||||
### 2.3 HTTP 调用(如有)
|
||||
|
||||
无。
|
||||
无。data-ana 不通过 HTTP 调用上游服务。
|
||||
|
||||
### 2.4 CDC 数据源(补充)
|
||||
### 2.4 CDC 数据源
|
||||
|
||||
| 数据源 | 用途 | mock 策略 |
|
||||
| ----------------------------------------------------- | ------------------------------- | --------------------------------------------------------------------------- |
|
||||
| core-edu MySQL(exams/homework/grades/attendance 表) | Debezium CDC → Kafka 同步读模型 | core-edu 就绪前使用 ClickHouse 内置模拟数据集(30 学生 × 5 考试 × 10 作业) |
|
||||
| content MySQL(knowledge_points 表) | Debezium CDC → 知识点元数据同步 | content 就绪前使用内置固定知识点表(数学 50 个知识点) |
|
||||
| iam MySQL(users 表) | Debezium CDC → 用户 dataScope 同步 | iam 就绪前使用硬编码 DataScope 降级 |
|
||||
|
||||
> **CDC topic 命名**:`edu-cdc.next_edu_cloud.<table>`(coord-cross-review.md §3.2 裁决补登 CDC topic 命名规范段,待 coord 在 004 §7.2 落实)。
|
||||
|
||||
---
|
||||
|
||||
@@ -81,17 +137,23 @@
|
||||
### 3.1 我依赖的上游就绪标志
|
||||
|
||||
- [ ] core-edu gRPC 50053 启用(ai08)—— 业务事件 + CDC 数据源
|
||||
- [ ] edu.exam.events / edu.homework.events / edu.grade.events / edu.class.events topic 有事件发布(ai08)
|
||||
- [ ] content gRPC 50054 启用(ai09)—— 知识点维度
|
||||
- [ ] edu.content.knowledge_point.events topic 有事件发布(ai09)
|
||||
- [ ] ai gRPC 50057 启用(ai12)—— AI 用量统计(可选,仪表盘补全)
|
||||
- [ ] core-edu MySQL Debezium CDC 配置(ai08 + SRE)—— CDC 通道前提
|
||||
- [ ] `edu.teaching.exam.published` / `edu.teaching.homework.assigned` / `edu.teaching.grade.recorded` / `edu.teaching.class.transferred` topic 有事件发布(ai08)
|
||||
- [ ] content gRPC 50054 启用(ai09)—— 知识点维度 CDC
|
||||
- [ ] `edu.content.kp.events` topic 有事件发布(ai09)
|
||||
- [ ] iam.proto 补全 GetEffectiveDataScope RPC(ai06/coord,ISSUE-001)—— P4 阻塞项
|
||||
- [ ] analytics.proto 扩展至 12 RPC(coord/ai11,ISSUE-003)—— gRPC stub 生成前提
|
||||
- [ ] events.proto 补全 AIUsageEvent message(coord,ISSUE-002)—— P5 前补全
|
||||
- [ ] ai gRPC 50057 启用(ai12)—— AI 用量统计(P5,可选)
|
||||
|
||||
### 3.2 我的就绪标志(供下游消费)
|
||||
|
||||
- [ ] data-ana gRPC 50055 启用(HealthService.Check 返回 SERVING)
|
||||
- [ ] AnalyticsService 12 RPC 可调用(含 4 端 Dashboard + Server Streaming SubscribeMasteryUpdate)
|
||||
- [ ] GetTeacherDashboard / GetStudentDashboard / GetParentDashboard / GetAdminDashboard 返回结构化数据
|
||||
- [ ] edu.data_ana.mastery.events topic 可发布(mastery.updated / warning.triggered)
|
||||
- [ ] **P4 就绪**:data-ana gRPC 50055 启用(HealthService.Check 返回 SERVING)
|
||||
- [ ] **P4 就绪**:AnalyticsService 12 RPC 可调用(含 4 端 Dashboard + Warning + Mastery + Server Streaming SubscribeMasteryUpdate)
|
||||
- [ ] **P4 就绪**:GetTeacherDashboard / GetStudentDashboard / GetParentDashboard / GetAdminDashboard 返回结构化数据
|
||||
- [ ] **P4 就绪**:`edu.insight.mastery.updated` topic 可发布(mastery.updated / warning.triggered)
|
||||
- [ ] **P5 就绪**:SubscribeMasteryUpdate server-streaming RPC 可订阅
|
||||
- [ ] **P6 就绪**:CDC 多实例水平扩展 + ExamCache Redis 化完成
|
||||
|
||||
---
|
||||
|
||||
@@ -106,16 +168,17 @@
|
||||
- GetStudentDashboard 返回固定仪表盘(avg_score=85.0, class_rank=5, weak_points 3 个)
|
||||
- GetParentDashboard 返回固定仪表盘(child_avg_score=85.0, child_class_rank=5)
|
||||
- GetAdminDashboard 返回固定仪表盘(total_teachers=50, total_students=1200, school_avg_score=80.0)
|
||||
- GetWarningList 返回固定 5 条预警(severity: warning/critical)
|
||||
- GetWarnings 返回固定 5 条预警(severity: warning/critical)
|
||||
- GetMasteryDistribution 返回固定分布(mastered=20, progressing=7, weak=3)
|
||||
- SubscribeMasteryUpdate 返回固定流(每 5 秒推 1 个 MasteryUpdateEvent)
|
||||
- **Kafka mock**:data-ana 就绪前不发布真实 MasteryEvent,msg 使用本地 stub 预警
|
||||
- **Kafka mock**:data-ana 就绪前不发布真实 MasteryUpdated / WarningTriggered,msg 使用本地 stub 预警
|
||||
|
||||
### 4.2 我消费的 mock
|
||||
|
||||
在真实上游就绪前,data-ana 使用以下 mock:
|
||||
|
||||
- 业务数据:ClickHouse 内置模拟数据集(30 学生 × 5 考试 × 10 作业 × 30 天出勤),不依赖 core-edu CDC
|
||||
- 知识点维度:内置固定知识点表(数学 50 个知识点),不依赖 content 事件
|
||||
- AI 用量:AIUsageEvent 为空,仪表盘 AI 用量区块显示"暂无数据"
|
||||
- CDC 通道:core-edu 就绪前 Debezium 不启动,使用 ClickHouse 批量导入模拟数据
|
||||
- **业务数据**:ClickHouse 内置模拟数据集(30 学生 × 5 考试 × 10 作业 × 30 天出勤),不依赖 core-edu CDC
|
||||
- **知识点维度**:内置固定知识点表(数学 50 个知识点),不依赖 content 事件
|
||||
- **AI 用量**:AIUsageEvent 为空,仪表盘 AI 用量区块显示"暂无数据"
|
||||
- **CDC 通道**:core-edu 就绪前 Debezium 不启动,使用 ClickHouse 批量导入模拟数据
|
||||
- **DataScope**:iam GetEffectiveDataScope 未就绪前,按 role 映射默认 DataScope 降级(教师=CLASS,学生=SELF,管理员=SCHOOL),标注 `details.degraded: true`
|
||||
|
||||
@@ -1,24 +1,129 @@
|
||||
# data-ana 问题记录
|
||||
|
||||
> 负责人:ai11
|
||||
> 关联:[coord.md](../coord.md)、[contracts/data-ana_contract.md](../contracts/data-ana_contract.md)
|
||||
> 关联:[coord.md](../coord.md)、[coord-cross-review.md](../../coord-cross-review.md)、[contracts/data-ana_contract.md](../contracts/data-ana_contract.md)
|
||||
> 规则:AI 遇到问题时在此追加条目,coord 仲裁后更新状态
|
||||
|
||||
---
|
||||
|
||||
## 问题列表
|
||||
## §0 已有仲裁核查记录(2026-07-10)
|
||||
|
||||
<!--
|
||||
追加条目格式:
|
||||
> ai11 在编写 objections 前对 coord.md 与 coord-cross-review.md 中涉及 data-ana 的仲裁逐项核查落实状态。
|
||||
|
||||
### ISSUE-[编号]-[AI标识]:[标题]
|
||||
### 0.1 coord.md 仲裁核查
|
||||
|
||||
- **提请方**:aiXX
|
||||
- **日期**:YYYY-MM-DD
|
||||
- **类型**:契约不明确 / 工作量超批 / 前置依赖缺失 / 编号冲突 / 其他
|
||||
- **描述**:[详细描述问题]
|
||||
- **建议方案**:[AI 的建议]
|
||||
- **状态**:待 coord 仲裁 / 已裁决(见 coord.md §X)
|
||||
-->
|
||||
| 编号 | 主题 | 涉及 data-ana | 核查结论 |
|
||||
| ------- | ---------------------------- | ------------- | ------------------------------ |
|
||||
| ARB-001 | teacher-bff GraphQL schema | ❌ 否 | 不涉及,无需 action |
|
||||
| ARB-002 | MF Shell 暴露清单 | ❌ 否 | 不涉及,无需 action |
|
||||
|
||||
(暂无问题)
|
||||
**结论**:coord.md 无 data-ana 直接仲裁。
|
||||
|
||||
### 0.2 coord-cross-review.md 仲裁核查(8 项涉及 data-ana)
|
||||
|
||||
| # | 审查章节 | 裁决内容 | 责任方 | 核查结论 |
|
||||
| -- | ---------------- | ------------------------------------------------------------------ | ------ | ---------------------------------------------------------------------------------------------------- |
|
||||
| 1 | §2.2 #3 | iam 新增 `GetEffectiveDataScope` RPC,P4 补全,data-ana gRPC 调用 | iam | ⚠️ **未落实**:iam.proto 当前仅 4 RPC(Register/Login/RefreshToken/GetUserInfo),无 GetEffectiveDataScope |
|
||||
| 2 | §2.3 P4 行 | content + data-ana P4 启用 gRPC server | ai11 | ⏳ 未到 P4 阶段,待执行 |
|
||||
| 3 | §3.2 | 补登 `edu.insight.ai.usage` topic + events.proto 补 AIUsageEvent | coord | ⚠️ **未落实**:events.proto 当前仅 4 message(Class/Exam/Homework/GradeEvent),缺 AIUsageEvent |
|
||||
| 4 | §3.3 | Python 服务 Outbox 豁免(MasteryUpdated / WarningTriggered) | coord | ✅ **已对齐**:01/02 文档已声明豁免,引用 coord-cross-review.md §3.3 |
|
||||
| 5 | §4.3 | data-ana HTTP=3006 / gRPC=50055 | coord | ✅ **已对齐**:01/02 文档端口声明一致 |
|
||||
| 6 | §5.3 | Python 服务信封改为 ActionState(degraded 放 details 子字段) | ai11 | ✅ **已对齐**:02 §4.3 ActionState 实现已修正,degraded 移至顶层 details |
|
||||
| 7 | §6 #4 | 同 #6,ai06 修正 data-ana/ai 02 文档 + 代码 | ai11 | ✅ **已对齐(文档)**:02 已修正;代码待 P4 实现阶段重构 |
|
||||
| 8 | §6 #7/#8/#9/#10 | coord 在 004 §7.2 补登 topic + §1.2 端口列 + §4.1 gRPC 矩阵 + §12.2 豁免 | coord | ⚠️ **未落实**:004 正文无 §4.2/§7.2 补登段/§11.4/§11.5,01/02 引用断裂已临时改为引 coord-cross-review.md |
|
||||
|
||||
### 0.3 coord-cross-review §8.2 批次 0 产出声明核查
|
||||
|
||||
coord-cross-review.md §8.2 声称批次 0 已完成 proto 补全,ai11 逐文件核查实际状态:
|
||||
|
||||
| 声明产出项 | 声明状态 | 实际文件状态(ai11 核查) | 核查结论 |
|
||||
| -------------------- | ----------------- | --------------------------------------------------------------------------------------------------------------------- | ------------ |
|
||||
| iam.proto 12 RPC | ✅ 12 RPC | **4 RPC**(Register/Login/RefreshToken/GetUserInfo) | ⚠️ 严重不符 |
|
||||
| analytics.proto 扩展 | ✅ 12 RPC(含 Stream) | **3 RPC**(GetClassPerformance/GetStudentWeakness/GetLearningTrend) | ⚠️ 严重不符 |
|
||||
| events.proto 补全 | ✅ 9 message(+AuditEvent) | **4 message**(ClassEvent/ExamEvent/HomeworkEvent/GradeEvent) | ⚠️ 严重不符 |
|
||||
| core_edu.proto 补全 | ✅ 5 service | 未由 ai11 核查(非本模块边界) | — |
|
||||
| buf.gen.yaml 插件 | ✅ go + python | 未由 ai11 核查(coord 维护) | — |
|
||||
|
||||
> **核查说明**:ai11 仅核查与 data-ana 直接相关的 proto(iam/analytics/events)。§8.2 声明与实际文件严重不符,可能原因:(a) 声明为计划态但未执行;(b) 执行后未提交到本 worktree 分支;(c) 在其他分支已执行但未合并。无论哪种原因,data-ana 的 P4 实现依赖这些 proto 补全,当前实际状态构成 P4 阻塞。
|
||||
|
||||
---
|
||||
|
||||
## §1 问题列表
|
||||
|
||||
### ISSUE-001-ai11:iam.proto 缺 GetEffectiveDataScope RPC(P4 阻塞)
|
||||
|
||||
- **提请方**:ai11
|
||||
- **日期**:2026-07-10
|
||||
- **类型**:前置依赖缺失
|
||||
- **描述**:coord-cross-review.md §2.2 #3 已仲裁"iam P4 补全 `GetEffectiveDataScope` RPC,data-ana gRPC 调用",但 iam.proto 当前仅 4 RPC,无此 RPC。data-ana 的 DataScope 6 级过滤(SELF/CLASS/GRADE/SCHOOL/DISTRICT/ALL)依赖此 RPC 解析用户可见数据范围,是 P4 实现的硬阻塞项。
|
||||
- **建议方案**:coord 确认 iam.proto 补全进度。若 iam 侧尚未实现,data-ana P4 阶段将使用硬编码 DataScope 降级(按 role 映射默认 scope),并标注 `details.degraded: true`,待 iam 就绪后切换。
|
||||
- **状态**:待 coord 仲裁(核查已有仲裁 §0.2 #1 未落实)
|
||||
|
||||
---
|
||||
|
||||
### ISSUE-002-ai11:events.proto 缺 AIUsageEvent message(P5 阻塞,P4 预备)
|
||||
|
||||
- **提请方**:ai11
|
||||
- **日期**:2026-07-10
|
||||
- **类型**:前置依赖缺失
|
||||
- **描述**:coord-cross-review.md §3.2 已仲裁"补登 `edu.insight.ai.usage` topic + events.proto 补 `AIUsageEvent` message",但 events.proto 当前仅 4 message(ClassEvent/ExamEvent/HomeworkEvent/GradeEvent),缺 AIUsageEvent。data-ana 需消费此事件落 `ai_usage_log` 宽表,供管理员仪表盘展示 AI 用量统计。
|
||||
- **建议方案**:coord 在 events.proto 补 `AIUsageEvent` message,字段建议:`{event_id, request_id, user_id, provider, model, prompt_tokens, completion_tokens, total_tokens, latency_ms, success, cost_cents, occurred_at}`。P5 前补全即可,P4 仪表盘 AI 用量区块显示"暂无数据"。
|
||||
- **状态**:待 coord 仲裁(核查已有仲裁 §0.2 #3 未落实)
|
||||
|
||||
---
|
||||
|
||||
### ISSUE-003-ai11:analytics.proto 仅 3 RPC,coord-cross-review §8.2 声称已扩展至 12 RPC 但实际未落实(P4 阻塞)
|
||||
|
||||
- **提请方**:ai11
|
||||
- **日期**:2026-07-10
|
||||
- **类型**:前置依赖缺失 + 声明与实际不符
|
||||
- **描述**:coord-cross-review.md §8.2 声称"analytics.proto 扩展 ✅ 12 RPC(含 Stream)",但实际文件仅 3 RPC(GetClassPerformance/GetStudentWeakness/GetLearningTrend)。ai-allocation.md §5 与 matrix.md §2 均要求 data-ana 提供 12 RPC,02-architecture-design.md §4.2 已设计完整 12 RPC 清单(含 4 端 Dashboard + Warning + Mastery + Server Streaming),但 proto 未补全导致无法生成 stub。
|
||||
- **建议方案**:coord 确认 analytics.proto 扩展进度。ai11 可提供 12 RPC 的完整 message 定义提案(见 [02-architecture-design.md §4.2](../../../services/data-ana/docs/02-architecture-design.md)),coord 审议后合并到 analytics.proto。若 coord 未补全,ai11 在 P4 阶段自行补全 proto(本分支内),提请 coord 合并。
|
||||
- **状态**:待 coord 仲裁
|
||||
|
||||
---
|
||||
|
||||
### ISSUE-004-ai11:coord-cross-review §6 整改清单 coord 责任项未落实,导致 004 章节引用断裂
|
||||
|
||||
- **提请方**:ai11
|
||||
- **日期**:2026-07-10
|
||||
- **类型**:契约不明确
|
||||
- **描述**:coord-cross-review.md §6 整改清单中标注"coord"责任的 4 项整改未在 004 正文中落实:
|
||||
- #7:004 §7.2 补登 6 个 topic + CDC 命名规范 → 004 正文无对应段落
|
||||
- #8:004 §1.2 新增 HTTP/gRPC 端口两列 → 004 §1.2 服务清单无端口列
|
||||
- #9:004 §4.1 补充 gRPC 启用阶段矩阵 → 004 正文无 §4.2 子节
|
||||
- #10:004 §12.2 补充派生数据事件 Outbox 豁免条款 → 004 §12.2 未补充
|
||||
|
||||
这导致 01-understanding.md 和 02-architecture-design.md 中引用 004 §4.2/§11.4/§11.5/§15.3 等章节均断裂(004 正文仅到 §14,§15 在 004-p6-addendum.md 但内容不同)。ai11 已在 v2.1 修订中临时改为引用 coord-cross-review.md 对应裁决章节,但这是过渡方案。
|
||||
- **建议方案**:coord 按整改清单 #7/#8/#9/#10 补全 004 对应章节,使 004 成为可信的架构设计意图唯一源。各 AI 文档随后将引用从 coord-cross-review.md 回切到 004 对应章节。
|
||||
- **状态**:待 coord 仲裁
|
||||
|
||||
---
|
||||
|
||||
### ISSUE-005-ai11:data-ana 发布的 MasteryEvent topic 命名三处不一致
|
||||
|
||||
- **提请方**:ai11
|
||||
- **日期**:2026-07-10
|
||||
- **类型**:契约不明确
|
||||
- **描述**:data-ana 发布的掌握度/预警事件 topic 命名在三个文档中不一致:
|
||||
|
||||
| 文档 | topic 命名 |
|
||||
| -------------------------------------- | ------------------------------------- |
|
||||
| 01-understanding.md / 02-architecture-design.md | `edu.insight.mastery.updated` + `edu.insight.warning.triggered` |
|
||||
| matrix.md §4 | `edu.analytics.mastery` |
|
||||
| contracts/data-ana_contract.md(修正前) | `edu.data_ana.mastery.events` |
|
||||
|
||||
按 004 §7.2 命名规范 `edu.<domain>.<aggregate>.<action>`,data-ana 属于 D6 智能洞察领域(domain=insight),故 `edu.insight.mastery.updated` 符合规范。matrix.md 的 `edu.analytics.mastery` 不符合命名规范(缺 action 层级,且 domain 用了服务名而非领域名)。
|
||||
- **建议方案**:coord 裁决统一为 `edu.insight.mastery.updated` + `edu.insight.warning.triggered`,coord 修正 matrix.md §4。ai11 已在 contract.md 中采用此命名。
|
||||
- **状态**:待 coord 仲裁
|
||||
|
||||
---
|
||||
|
||||
### ISSUE-006-ai11:coord-cross-review §8.2 批次 0 产出声明与 proto 实际文件状态严重不符
|
||||
|
||||
- **提请方**:ai11
|
||||
- **日期**:2026-07-10
|
||||
- **类型**:其他(声明与实际不符)
|
||||
- **描述**:coord-cross-review.md §8.2 声称批次 0 已完成 iam.proto(12 RPC)/ analytics.proto(12 RPC)/ events.proto(9 message)补全,但 ai11 逐文件核查发现实际均未补全(详见 §0.3 核查表)。这影响所有依赖这些 proto 的下游 AI 的排期评估——若 AI 信任 §8.2 声明,会在排期中忽略 proto 补全的等待时间,导致排期失真。
|
||||
- **建议方案**:coord 核实 §8.2 声明真实性。若实际已补全但未合并到各 worktree 分支,请协调合并;若实际未补全,请更新 §8.2 状态为"计划中"或"待执行",并明确补全时间点,以便下游 AI 据此排期。
|
||||
- **状态**:待 coord 仲裁
|
||||
|
||||
@@ -1,18 +1,34 @@
|
||||
# data-ana 工作排期
|
||||
|
||||
> 负责人:ai11
|
||||
> 关联:[workline.md](../workline.md)、[coord.md](../coord.md)、[contracts/data-ana_contract.md](../contracts/data-ana_contract.md)
|
||||
> 模式:全并行(各 AI 一口气完成 P2-P6 全部代码,最后统一集成测试)
|
||||
> 关联:[workline.md](../workline.md)、[coord.md](../coord.md)、[contracts/data-ana_contract.md](../contracts/data-ana_contract.md)、[objections/data-ana_issue.md](../objections/data-ana_issue.md)
|
||||
> 模式:全并行(各 AI 一口气完成 P2-P6 全部代码,开发期间用 mock,最后统一集成测试)
|
||||
> 基线日期:批次 0 已完成(2026-07-09),批次 1(P2)2026-07-10 启动
|
||||
|
||||
---
|
||||
|
||||
## §1 总览
|
||||
|
||||
data-ana 是数据分析服务,提供 AnalyticsService(12 个 RPC),基于 ClickHouse 列存查询与 CDC 消费实现实时分析。全阶段目标:P2 ClickHouse 接入+CDC 消费 → P3 AnalyticsService 12 RPC → P4-P6 持续优化。
|
||||
data-ana 是 D6 智能洞察领域的纯读模型服务(Python/FastAPI),基于 ClickHouse ReplacingMergeTree 宽表 + Debezium CDC 消费实现实时学情分析。
|
||||
|
||||
**核心交付物**:
|
||||
- gRPC server :50055 + AnalyticsService 12 RPC(含 1 个 Server Streaming)
|
||||
- HTTP :3006 14 端点(3 基础 + 11 业务,保留作 Gateway 直连降级)
|
||||
- ClickHouse 5 宽表(student_dashboard_view / student_errors / mastery_snapshot / attendance_logs / ai_usage_log)
|
||||
- CDC 消费者(core-edu MySQL binlog → Kafka → ClickHouse 宽表投影)
|
||||
- 掌握度计算(加权滑动平均 + 遗忘曲线)+ 预警评估
|
||||
- 派生数据事件发布(edu.insight.mastery.updated / edu.insight.warning.triggered,豁免 Outbox)
|
||||
|
||||
**阶段里程碑**:
|
||||
- P2 预备:ClickHouse 接入 + CDC 骨架 + ActionState 信封重构 + mock 数据集
|
||||
- P3 预备:CDC 通道接入 + 掌握度算法 v1 + Repository 封装
|
||||
- P4 主战场:gRPC 50055 启用 + 12 RPC + 4 端 Dashboard + Warning + DataScope + 事件发布
|
||||
- P5 扩展:SubscribeMasteryUpdate stream + AI 用量消费 + 手动 commit
|
||||
- P6 硬化:CDC 水平扩展 + 容量规划 + 数据治理 + 监控告警
|
||||
|
||||
---
|
||||
|
||||
## §2 全阶段甘特图(P2-P6,各 AI 自行细化)
|
||||
## §2 全阶段甘特图(P2-P6)
|
||||
|
||||
```mermaid
|
||||
gantt
|
||||
@@ -20,26 +36,317 @@ gantt
|
||||
dateFormat YYYY-MM-DD
|
||||
axisFormat %m-%d
|
||||
|
||||
section P2-P6
|
||||
[阶段任务] :a11a, 2026-07-10, Xd
|
||||
section P2 预备(与批次1并行)
|
||||
2.1 ClickHouse DDL 5宽表建表 :a11a, 2026-07-10, 2d
|
||||
2.2 CDC消费骨架(aiokafka) :a11b, after a11a, 2d
|
||||
2.3 mock数据集(30学生×5考试×10作业) :a11c, after a11a, 1d
|
||||
2.4 ActionState信封重构(P0整改) :crit, a11d, 2026-07-10, 2d
|
||||
2.5 config.py修正(env/Redis/gRPC) :a11e, after a11d, 1d
|
||||
|
||||
section P3 预备(与批次2并行)
|
||||
3.1 gRPC server骨架(3 RPC,不启用) :a11f, after a11b, 2d
|
||||
3.2 core-edu CDC通道接入(grades/exams/homework) :a11g, after a11b, 3d
|
||||
3.3 ExamCache内存LRU :a11h, after a11g, 1d
|
||||
3.4 掌握度算法v1(weighted_moving_avg) :a11i, after a11g, 2d
|
||||
3.5 ClickHouseRepository(FINAL/argMax) :a11j, after a11i, 2d
|
||||
|
||||
section P4 主战场(批次3, 11d)
|
||||
4.1 gRPC 50055正式启用 :crit, a11k, after a11j, 1d
|
||||
4.2 analytics.proto扩展12 RPC :crit, a11l, after a11k, 2d
|
||||
4.3 4端Dashboard RPC实现 :a11m, after a11l, 3d
|
||||
4.4 WarningService+TriggerWarning :a11n, after a11l, 2d
|
||||
4.5 GetMasteryDistribution+GetStudentMastery :a11o, after a11m, 1d
|
||||
4.6 iam.GetEffectiveDataScope集成(降级兜底) :crit, a11p, after a11k, 2d
|
||||
4.7 DataScope 6级WHERE注入 :a11q, after a11p, 1d
|
||||
4.8 attendance+content CDC消费 :a11r, after a11g, 2d
|
||||
4.9 MasteryEvent+WarningTriggered发布 :a11s, after a11n, 1d
|
||||
4.10 HTTP 14端点+readyz硬化 :a11t, after a11m, 2d
|
||||
|
||||
section P5 扩展(批次4并行)
|
||||
5.1 SubscribeMasteryUpdate stream RPC :a11u, after a11t, 3d
|
||||
5.2 AIUsageEvent消费→ai_usage_log :a11v, after a11u, 2d
|
||||
5.3 手动commit替换auto_commit :a11w, after a11v, 1d
|
||||
5.4 Admin Dashboard AI用量区块 :a11x, after a11v, 2d
|
||||
|
||||
section P6 硬化(批次5并行)
|
||||
6.1 CDC多实例水平扩展 :a11y, after a11x, 3d
|
||||
6.2 ExamCache Redis化 :a11z, after a11y, 2d
|
||||
6.3 容量规划+TTL归档策略 :a11aa, after a11z, 2d
|
||||
6.4 监控告警(consumer lag HPA) :a11ab, after a11aa, 2d
|
||||
6.5 readyz深度硬化 :a11ac, after a11ab, 1d
|
||||
```
|
||||
|
||||
> **注意**:以上为 coord 初始规划,ai11 接管后必须自行细化为完整 P2-P6 排期。
|
||||
> **关键路径**(crit):ActionState 信封重构 → gRPC 50055 启用 → analytics.proto 扩展 → iam GetEffectiveDataScope 集成
|
||||
> **总工期**:约 51 天(2026-07-10 ~ 2026-08-30),其中 P4 主战场 11 天为关键交付期
|
||||
|
||||
---
|
||||
|
||||
## §3 详细任务
|
||||
|
||||
### 全阶段任务
|
||||
### 3.1 P2 预备期(2026-07-10 ~ 2026-07-18,8d)
|
||||
|
||||
#### 任务 2.1:ClickHouse DDL 5 宽表建表
|
||||
|
||||
- **负责人**:ai11
|
||||
- **交付物**:⚠️ 由 ai11 自行补充
|
||||
- **依赖**:见 [contracts/data-ana_contract.md](../contracts/data-ana_contract.md)
|
||||
- **验收标准**:⚠️ 由 ai11 自行补充
|
||||
- **依赖**:无(ClickHouse 实例就绪,由 infra 提供)
|
||||
- **交付物**:`infra/clickhouse/ddl/data_ana.sql`(5 宽表 DDL:student_dashboard_view / student_errors / mastery_snapshot / attendance_logs / ai_usage_log)
|
||||
- **验收标准**:5 表在 ClickHouse 中创建成功,ReplacingMergeTree 引擎 + ORDER BY + PARTITION BY 符合 02 §3 DDL 设计
|
||||
|
||||
#### 任务 2.2:CDC 消费骨架
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:Kafka 就绪
|
||||
- **交付物**:`src/data_ana/cdc_consumer.py` 重构(aiokafka AIOKafkaConsumer + EventHandler 路由框架)
|
||||
- **验收标准**:能消费 mock CDC 事件并打印路由日志,consumer group = `data-ana-cdc`
|
||||
|
||||
#### 任务 2.3:mock 数据集
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:任务 2.1
|
||||
- **交付物**:`scripts/seed_clickhouse.py`(批量导入 30 学生 × 5 考试 × 10 作业 × 30 天出勤模拟数据)
|
||||
- **验收标准**:ClickHouse 5 表有数据,可查询返回非空结果
|
||||
|
||||
#### 任务 2.4:ActionState 信封重构(P0 整改,coord-cross-review §5.3)
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:无
|
||||
- **交付物**:`src/data_ana/shared/action_state.py`(ActionState[T] 泛型 + ActionStateError + ok()/fail() 类方法)
|
||||
- **验收标准**:main.py 所有端点返回 `ActionState[T]`,degraded 标记在顶层 `details.degraded`(非 error.details),ruff 零错误
|
||||
|
||||
#### 任务 2.5:config.py 修正
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:无
|
||||
- **交付物**:`src/data_ana/config.py` 修正(env_prefix 补 Redis / gRPC / ClickHouse 配置项,pydantic-settings 校验)
|
||||
- **验收标准**:配置项覆盖 02 §13 配置清单,环境变量缺失时 pydantic-settings 报错
|
||||
|
||||
---
|
||||
|
||||
### 3.2 P3 预备期(2026-07-18 ~ 2026-07-26,8d)
|
||||
|
||||
#### 任务 3.1:gRPC server 骨架(3 RPC,不正式启用)
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:analytics.proto 当前 3 RPC(无需 coord 补全)
|
||||
- **交付物**:`src/data_ana/grpc_server.py`(grpc.aio Server 骨架 + 3 RPC 实现,绑定 :50055 但不启动对外)
|
||||
- **验收标准**:本地可启动 gRPC server,3 RPC 可调用返回 mock 数据
|
||||
|
||||
#### 任务 3.2:core-edu CDC 通道接入
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:core-edu MySQL 就绪 + Debezium CDC 配置(core-edu 就绪前用 mock binlog 事件)
|
||||
- **交付物**:`src/data_ana/cdc_consumer.py` 完善(EventHandler 处理 grades/exams/homework/classes 表 CDC 事件)
|
||||
- **验收标准**:消费 CDC 事件 → 解析 Debezium JSON → 查 ExamCache 填 class_id → upsert ClickHouse 宽表
|
||||
|
||||
#### 任务 3.3:ExamCache 内存 LRU
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:任务 3.2
|
||||
- **交付物**:`src/data_ana/exam_cache.py`(内存 LRU dict,max 10000 条,exam_id → {class_id, subject_id})
|
||||
- **验收标准**:CDC exams 事件触发 ExamCache 更新,grades 事件查 ExamCache 获取 class_id
|
||||
|
||||
#### 任务 3.4:掌握度算法 v1
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:任务 3.2
|
||||
- **交付物**:`src/data_ana/mastery_service.py`(加权滑动平均算法,权重 w_i = 0.6^i,归一化)
|
||||
- **验收标准**:输入学生近期 N 次成绩 → 输出 mastery_level (0.0-1.0) → 写 mastery_snapshot 表
|
||||
|
||||
#### 任务 3.5:ClickHouseRepository 封装
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:任务 2.1
|
||||
- **交付物**:`src/data_ana/clickhouse_client.py` 重构(查询封装 + FINAL/argMax 去重 + DataScope WHERE 注入接口)
|
||||
- **验收标准**:查询 student_dashboard_view 返回去重后最新版本数据
|
||||
|
||||
---
|
||||
|
||||
### 3.3 P4 主战场期(2026-07-26 ~ 2026-08-06,11d,批次 3)
|
||||
|
||||
#### 任务 4.1:gRPC 50055 正式启用
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:任务 3.1
|
||||
- **交付物**:main.py lifespan 启动 gRPC server :50055,HealthService.Check 返回 SERVING
|
||||
- **验收标准**:gRPC server 对外可访问,HealthService.Check = SERVING
|
||||
|
||||
#### 任务 4.2:analytics.proto 扩展 12 RPC
|
||||
|
||||
- **负责人**:ai11(本分支内补全 proto,提请 coord 合并)
|
||||
- **依赖**:ISSUE-003 解决(coord 确认或 ai11 自行补全)
|
||||
- **交付物**:`packages/shared-proto/proto/analytics.proto` 扩展至 12 RPC(3 现有 + 9 新增 message 定义)
|
||||
- **验收标准**:`buf lint` 零错误,`buf generate` 生成 Python stub 成功,12 RPC 全部可调用
|
||||
|
||||
#### 任务 4.3:4 端 Dashboard RPC 实现
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:任务 4.2
|
||||
- **交付物**:GetTeacherDashboard / GetStudentDashboard / GetParentDashboard / GetAdminDashboard 4 RPC 实现
|
||||
- **验收标准**:4 RPC 返回 ActionState[DashboardData],DataScope 过滤生效,降级时返回骨架数据 + degraded: true
|
||||
|
||||
#### 任务 4.4:WarningService + TriggerWarning
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:任务 3.4(掌握度算法)
|
||||
- **交付物**:`src/data_ana/warning_service.py`(预警阈值评估 + TriggerWarning RPC + GetWarnings RPC)
|
||||
- **验收标准**:掌握度 < 0.4 触发 LOW_MASTERY 预警,成绩环比下降 20% 触发 SCORE_DROP,缺勤 ≥ 3 次/周触发 ABSENT_FREQUENT
|
||||
|
||||
#### 任务 4.5:GetMasteryDistribution + GetStudentMastery
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:任务 3.4 + 任务 4.2
|
||||
- **交付物**:2 RPC 实现(班级掌握度分布 + 学生知识点掌握度明细)
|
||||
- **验收标准**:返回 mastered/progressing/weak 三档分布数据
|
||||
|
||||
#### 任务 4.6:iam.GetEffectiveDataScope 集成(降级兜底)
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:ISSUE-001 解决(iam.proto 补全 GetEffectiveDataScope)。若 P4 时 iam 未就绪,使用降级兜底
|
||||
- **交付物**:`src/data_ana/iam_client.py`(gRPC 调 iam.GetEffectiveDataScope + Redis 缓存 5min + 降级兜底)
|
||||
- **验收标准**:iam 可用时调 gRPC 获取 DataScope;iam 不可用时按 role 映射默认 DataScope + degraded: true
|
||||
|
||||
#### 任务 4.7:DataScope 6 级 WHERE 注入
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:任务 4.6 + 任务 3.5
|
||||
- **交付物**:ClickHouseRepository 查询方法注入 DataScope WHERE 子句(SELF/CLASS/GRADE/SCHOOL/DISTRICT/ALL)
|
||||
- **验收标准**:教师只能查自己班级数据,学生只能查自己数据,管理员可查全校数据
|
||||
|
||||
#### 任务 4.8:attendance + content CDC 消费
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:任务 3.2(CDC 框架)
|
||||
- **交付物**:EventHandler 扩展 attendance_logs 表 CDC + content_knowledge_points 表 CDC
|
||||
- **验收标准**:考勤事件落 attendance_logs 表,知识点事件更新 mastery_snapshot 元数据
|
||||
|
||||
#### 任务 4.9:MasteryEvent + WarningTriggered 事件发布
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:任务 3.4 + 任务 4.4
|
||||
- **交付物**:`src/data_ana/kafka_producer.py`(aiokafka AIOKafkaProducer + idempotent + transactional_id)
|
||||
- **验收标准**:掌握度计算完成发布 `edu.insight.mastery.updated`,预警触发发布 `edu.insight.warning.triggered`,豁免 Outbox
|
||||
|
||||
#### 任务 4.10:HTTP 14 端点 + readyz 硬化
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:任务 4.3 + 任务 4.4 + 任务 4.5
|
||||
- **交付物**:main.py 14 个 HTTP 端点全部实现(3 基础 + 11 业务)+ /readyz 检查 4 依赖(clickhouse/cdc_consumer/redis/iam_grpc)
|
||||
- **验收标准**:14 端点返回 ActionState[T],/readyz 依赖检查正确反映服务状态
|
||||
|
||||
---
|
||||
|
||||
### 3.4 P5 扩展期(2026-08-06 ~ 2026-08-19,13d,批次 4 并行)
|
||||
|
||||
#### 任务 5.1:SubscribeMasteryUpdate Server Streaming RPC
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:任务 4.2 + 任务 4.9
|
||||
- **交付物**:SubscribeMasteryUpdate RPC 实现(server-streaming,客户端订阅 student_id/class_id,掌握度更新时推送)
|
||||
- **验收标准**:客户端订阅后,掌握度计算完成时收到 MasteryUpdateEvent 流
|
||||
|
||||
#### 任务 5.2:AIUsageEvent 消费 → ai_usage_log
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:ISSUE-002 解决(events.proto 补 AIUsageEvent)+ ai 服务发布 `edu.insight.ai.usage` topic
|
||||
- **交付物**:EventHandler 扩展 AIUsageEvent 消费 → 落 ai_usage_log 表
|
||||
- **验收标准**:ai 服务发布用量事件后,ai_usage_log 表有数据,Admin Dashboard AI 用量区块可展示
|
||||
|
||||
#### 任务 5.3:手动 commit 替换 auto_commit
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:任务 3.2
|
||||
- **交付物**:cdc_consumer.py 改为 `enable_auto_commit=False` + 手动 commit(at-least-once)
|
||||
- **验收标准**:ClickHouse 写入成功后才 commit offset,重启后无重复消费(依赖 ReplacingMergeTree 去重)
|
||||
|
||||
#### 任务 5.4:Admin Dashboard AI 用量区块
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:任务 5.2
|
||||
- **交付物**:GetAdminDashboard RPC 补全 AI 用量统计区块(按 provider/model/时间窗聚合)
|
||||
- **验收标准**:Admin Dashboard 返回 AI 用量数据,无数据时显示"暂无数据"
|
||||
|
||||
---
|
||||
|
||||
### 3.5 P6 硬化期(2026-08-19 ~ 2026-08-30,11d,批次 5 并行)
|
||||
|
||||
#### 任务 6.1:CDC 多实例水平扩展
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:任务 5.3(手动 commit)
|
||||
- **交付物**:CdcConsumer 支持多实例分摊 partition(consumer group 不变)
|
||||
- **验收标准**:2+ 实例消费同一 topic 无重复无遗漏
|
||||
|
||||
#### 任务 6.2:ExamCache Redis 化
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:任务 6.1
|
||||
- **交付物**:exam_cache.py 改为 Redis 实现(key: `data_ana:exam:{exam_id}` TTL 30 天)
|
||||
- **验收标准**:多实例共享 ExamCache,重启后缓存不丢失
|
||||
|
||||
#### 任务 6.3:容量规划 + TTL 归档策略
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:无
|
||||
- **交付物**:ClickHouse TTL 策略(student_dashboard_view 保留 2 年,ai_usage_log 保留 1 年)+ 冷热数据分离方案
|
||||
- **验收标准**:TTL 配置生效,过期数据自动清理
|
||||
|
||||
#### 任务 6.4:监控告警完善
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:任务 6.1
|
||||
- **交付物**:Prometheus 指标补全(consumer lag histogram + 慢查询 counter + ClickHouse 连接池 gauge)+ Grafana dashboard
|
||||
- **验收标准**:consumer lag 超阈值触发 HPA,慢查询超阈值告警
|
||||
|
||||
#### 任务 6.5:readyz 深度硬化
|
||||
|
||||
- **负责人**:ai11
|
||||
- **依赖**:任务 4.10
|
||||
- **交付物**:/readyz 检查项完善(ClickHouse 查询超时 1s + Redis ping + iam gRPC 超时 2s + CDC consumer lag < 1000)
|
||||
- **验收标准**:任一依赖不健康时 /readyz 返回 503,K8s 摘流量
|
||||
|
||||
---
|
||||
|
||||
## §4 依赖与就绪信号
|
||||
|
||||
- **我依赖**:⚠️ 由 ai11 自行补充(见 contract.md)
|
||||
- **我的就绪信号**:⚠️ 由 ai11 自行补充
|
||||
### 4.1 我依赖的上游就绪标志
|
||||
|
||||
- [ ] **core-edu gRPC 50053 启用**(ai08,批次 2)—— CDC 数据源(grades/exams/homework/attendance 表 binlog)
|
||||
- [ ] **core-edu MySQL Debezium CDC 配置**(ai08 + SRE)—— CDC 通道前提
|
||||
- [ ] **content gRPC 50054 启用**(ai09,批次 3)—— 知识点维度 CDC
|
||||
- [ ] **iam.proto 补全 GetEffectiveDataScope RPC**(ai06/coord,ISSUE-001)—— DataScope 解析
|
||||
- [ ] **analytics.proto 扩展至 12 RPC**(coord/ai11,ISSUE-003)—— gRPC stub 生成前提
|
||||
- [ ] **events.proto 补全 AIUsageEvent message**(coord,ISSUE-002)—— AI 用量消费(P5)
|
||||
- [ ] **ai 服务发布 `edu.insight.ai.usage` topic**(ai12,批次 4)—— AI 用量统计(P5)
|
||||
|
||||
> **降级兜底**:core-edu / content / iam 未就绪时,使用 ClickHouse 内置 mock 数据集 + 硬编码 DataScope 降级,标注 `details.degraded: true`
|
||||
|
||||
### 4.2 我的就绪信号(供下游消费)
|
||||
|
||||
- [ ] **P4 就绪**:data-ana gRPC 50055 启用(HealthService.Check = SERVING)+ AnalyticsService 12 RPC 可调用 + 4 端 Dashboard 返回结构化数据
|
||||
- [ ] **P4 就绪**:`edu.insight.mastery.updated` topic 可发布(mastery.updated / warning.triggered)
|
||||
- [ ] **P5 就绪**:SubscribeMasteryUpdate server-streaming RPC 可订阅
|
||||
- [ ] **P6 就绪**:CDC 多实例水平扩展 + ExamCache Redis 化完成
|
||||
|
||||
### 4.3 下游消费方
|
||||
|
||||
| 下游 | 消费接口 | 就绪依赖阶段 |
|
||||
| -------------------------- | ---------------------------------------------------------------- | ------------ |
|
||||
| teacher-bff(ai03) | gRPC 50055 GetTeacherDashboard / GetClassPerformance 等 | P4 |
|
||||
| student-bff(ai04) | gRPC 50055 GetStudentDashboard / GetStudentWeakness 等 | P4 |
|
||||
| parent-bff(ai05) | gRPC 50055 GetParentDashboard | P4 |
|
||||
| ai 服务(ai12) | gRPC 50055 反向调用查学情(GetStudentMastery / GetLearningTrend) | P5 |
|
||||
| core-edu(ai08) | Kafka `edu.insight.mastery.updated`(推荐个性化练习) | P4 |
|
||||
| msg(ai10) | Kafka `edu.insight.warning.triggered`(推送通知) | P4 |
|
||||
|
||||
---
|
||||
|
||||
## §5 风险与缓解
|
||||
|
||||
| 风险 | 影响 | 缓解措施 |
|
||||
| -------------------------------------------- | ---- | ------------------------------------------------------------------------------------------ |
|
||||
| iam.proto 未补全 GetEffectiveDataScope | P4 | 降级兜底:按 role 映射默认 DataScope + degraded: true(ISSUE-001) |
|
||||
| analytics.proto 未扩展 12 RPC | P4 | ai11 本分支自行补全 proto,提请 coord 合并(ISSUE-003) |
|
||||
| core-edu CDC 通道未就绪 | P3-P4 | mock 数据集降级 + 本地 stub CDC 事件 |
|
||||
| ClickHouse ReplacingMergeTree 去重延迟 | P4 | 查询加 FINAL / argMax 强制去重(02 §3.6 已设计) |
|
||||
| 单实例 CDC 消费者单点故障 | P4 | P6 演进为多实例 + Redis ExamCache;P4 阶段监控 consumer lag 告警 |
|
||||
| 掌握度算法精度不足 | P4 | v1 用加权滑动平均,P5+ 评估引入遗忘曲线 max 叠加(02 §9 已设计 MasteryMethod 枚举预留) |
|
||||
|
||||
Reference in New Issue
Block a user