fix(infra): resolve NestJS dist build and Prometheus target issues

NestJS: disable incremental in 6 services tsconfig.json to fix dist not emitted when nest-cli deleteOutDir conflicts with tsc tsbuildinfo. classes/iam: import HealthModule in AppModule to fix /healthz 404. classes: rewrite HealthController to Drizzle getDb from TypeORM DI. teacher-bff: add /metrics endpoint for Prometheus scraping. infra: add node/mysql/redis exporters to observability profile. mysql-exporter v0.15.1 uses command-line flags not DATA_SOURCE_NAME. prometheus: enable web.enable-lifecycle for hot reload.
2026-07-09 15:12:15 +08:00
parent b72c8d81d4
commit f1e466a772
16 changed files with 267 additions and 55 deletions
--- a/docs/standards/full-stack-runbook.md
+++ b/docs/standards/full-stack-runbook.md
@@ -195,6 +195,8 @@ Invoke-RestMethod -Uri "http://localhost:3006/readyz"
 ```powershell
 cd e:\Desktop\Edu
 .\scripts\stop-all.ps1
+
+.\scripts\stop-all.ps1 -KillByPort
 ```

 该脚本会关闭所有 `edu-app-*` 标题的终端窗口。
--- a/docs/troubleshooting/known-issues.md
+++ b/docs/troubleshooting/known-issues.md
@@ -187,22 +187,25 @@

 ### 2.2 classes（TS/NestJS，P1 黄金模板）

-| 场景             | 技术/规则                                                                                        |
-| ---------------- | ------------------------------------------------------------------------------------------------ |
-| 黄金模板定位     | P1 完整实现所有横切关注点，后续 8 个服务复制此模板                                               |
-| 黄金模板复制流程 | `cp -r services/classes services/xxx` → 改错误码前缀 → 改 proto → 改业务逻辑 → 改 README → 改 CI |
-| 横切关注点清单   | 错误处理 / 可观测 / 安全 / 契约 / 测试 / 文档 / 配置 / i18n / CI / Dockerfile                    |
-| 错误处理         | `ApplicationError` 基类 + 子类，错误码 `CLASSES_*` 前缀                                          |
-| 可观测           | pino logger + prom-client metrics + OTel tracer                                                  |
-| 安全             | auth.middleware（信任 Gateway 头）+ permission.guard + data-scope.interceptor                    |
-| 配置             | 三层配置 + Zod 校验 env                                                                          |
-| i18n             | `ERROR_CODES` 映射表（错误码 → i18n key）                                                        |
-| 测试四类         | 单元（vitest）+ 集成（Testcontainers）+ 契约（Pact）+ E2E（Playwright）                          |
-| 覆盖率门槛       | 领域逻辑 ≥ 80%，Handler ≥ 60%，整体 ≥ 60%                                                        |
-| Drizzle schema   | `mysqlTable` + `varchar`/`timestamp` + `index`                                                   |
-| ID 生成          | `@paralleldrive/cuid2` 的 `createId()`                                                           |
-| 响应转换         | repository 返回 Date，service 转换为 `createdAt: number`（时间戳）                               |
-| 阶段特有模式回写 | Outbox（P3）/ CDC（P4）/ 长连接（P5）实现后回写黄金模板 README                                   |
+| 场景             | 技术/规则                                                                                          |
+| ---------------- | -------------------------------------------------------------------------------------------------- |
+| 黄金模板定位     | P1 完整实现所有横切关注点，后续 8 个服务复制此模板                                                 |
+| 黄金模板复制流程 | `cp -r services/classes services/xxx` → 改错误码前缀 → 改 proto → 改业务逻辑 → 改 README → 改 CI   |
+| 横切关注点清单   | 错误处理 / 可观测 / 安全 / 契约 / 测试 / 文档 / 配置 / i18n / CI / Dockerfile                      |
+| 错误处理         | `ApplicationError` 基类 + 子类，错误码 `CLASSES_*` 前缀                                            |
+| 可观测           | pino logger + prom-client metrics + OTel tracer                                                    |
+| 安全             | auth.middleware（信任 Gateway 头）+ permission.guard + data-scope.interceptor                      |
+| 配置             | 三层配置 + Zod 校验 env                                                                            |
+| i18n             | `ERROR_CODES` 映射表（错误码 → i18n key）                                                          |
+| 测试四类         | 单元（vitest）+ 集成（Testcontainers）+ 契约（Pact）+ E2E（Playwright）                            |
+| 覆盖率门槛       | 领域逻辑 ≥ 80%，Handler ≥ 60%，整体 ≥ 60%                                                          |
+| Drizzle schema   | `mysqlTable` + `varchar`/`timestamp` + `index`                                                     |
+| ID 生成          | `@paralleldrive/cuid2` 的 `createId()`                                                             |
+| 响应转换         | repository 返回 Date，service 转换为 `createdAt: number`（时间戳）                                 |
+| 阶段特有模式回写 | Outbox（P3）/ CDC（P4）/ 长连接（P5）实现后回写黄金模板 README                                     |
+| 健康检查依赖     | `readyz` 用 Drizzle `getDb().execute(sql\`SELECT 1\`)` 校验，不要依赖 typeorm DataSource DI        |
+| AppModule 注册   | HealthModule 必须在 `app.module.ts` imports 数组显式声明，否则 NestFactory 不扫描 HealthController |
+| 增量编译陷阱     | `tsconfig.json` 显式 `"incremental": false` 覆盖 base，避免 .tsbuildinfo 导致 nest watch 不 emit   |

 ### 2.3 iam（TS/NestJS，P2）

@@ -350,6 +353,7 @@

 | 日期       | 时间 | 模块                            | 做了什么 + 学到什么                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                      |
 | ---------- | ---- | ------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 2026-07-09 | 下午 | classes/全局                    | **一键启动脚本 NestJS dist/ 不生成根因定位 + classes 健康检查修复**：(1) 根因定位：`tsconfig.base.json` 的 `incremental: true` + `nest-cli.json` 的 `deleteOutDir: true` 冲突。`nest start --watch` 启动时先删除 dist/，tsc 读残留 .tsbuildinfo 认为无变化跳过 emit，dist/ 不生成 → `Cannot find module dist/main`。(2) 修复：6 个 NestJS 服务（classes/iam/teacher-bff/core-edu/content/msg）tsconfig.json 显式加 `"incremental": false` 覆盖 base 配置，删除所有残留 .tsbuildinfo 文件。(3) classes AppModule 缺 HealthModule 导入导致 /healthz 404，iam 同样问题，修复 app.module.ts 加 `imports: [..., HealthModule]`。(4) classes HealthController 误用 TypeORM `DataSource` DI（与 iam 不一致），运行时报 `Nest can't resolve dependencies of the HealthController (DataSource)`。修复：改为 Drizzle `getDb()` 函数式调用，与 iam 一致。(5) 一键启动验证：11/11 应用 + 11/11 基础设施 + 5/5 可观测性端点全绿。**学到**：NestJS + TypeScript incremental 编译是陷阱组合——nest-cli deleteOutDir 删 dist 但 tsc 读 tsbuildinfo 认为无变化，必须在服务级 tsconfig 显式 `incremental: false`；HealthModule 必须在 AppModule imports 中显式声明才能被 NestFactory 扫描到；5 个 NestJS 服务的 HealthController 应统一用 Drizzle `getDb()` 函数式调用而非 TypeORM DataSource DI（项目已弃 TypeORM 改 Drizzle）。                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                           |
 | 2026-07-09 | 下午 | 全局                            | **OTel auto-instrumentations 全服务补全**：(1) NestJS 6 服务（iam/classes/core-edu/content/msg/teacher-bff）tracer.ts 补 `getNodeAutoInstrumentations()`，NodeSDK 传 instrumentations 参数自动埋点 HTTP/Express/DB。(2) Python 2 服务（data-ana/ai）main.py 补 `FastAPIInstrumentor.instrument_app(app)`；ai 补缺失的 `opentelemetry-exporter-otlp` 依赖。(3) teacher-bff 从零补完整 OTel：env.ts 加 OTEL_EXPORTER_OTLP_ENDPOINT 字段 + 新建 shared/observability/tracer.ts + main.ts 调用 initTracer/shutdownTracer + package.json 加 sdk-node/exporter/auto-instrumentations 依赖。(4) Go 2 服务（api-gateway/push-gateway）新建 internal/observability/tracer.go（OTLP HTTP exporter + resource + TracerProvider + W3C propagator）+ main.go 调用 InitTracer + otelgin.Middleware 注册 Gin 中间件；push-gateway config.go 补 OTLPEndpoint 字段。(5) 质量校验全通过：TS typecheck 9 服务 + ESLint 6 服务 + ruff 2 服务 + go vet/build 2 服务零错误。**学到**：`getNodeAutoInstrumentations()` 一次注册所有 Node.js 自动埋点（http/express/dns/fs/net/grpc 等），比手动逐个注册 HttpInstrumentation 更简洁；Go OTel 用 `otlptracehttp.WithEndpoint(host)` + `WithInsecure()` 需从 "http://host:port" URL 解析出 host；otelgin.Middleware 必须在 Recovery 之后其他中间件之前注册，确保所有后续 handler 都被 trace；Python FastAPIInstrumentor.instrument_app(app) 在 app 创建后立即调用，lifespan 不受影响。                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                             |
 | 2026-07-09 | 下午 | 全局                            | **P6 硬化：可观测性 + 部署 + CI 硬化**：(1) 可观测性栈完善：5 个 NestJS 服务 main.ts 添加 `/metrics` Prometheus 端点（用 `app.getHttpAdapter().get('/metrics', ...)` 绕过 DI 容器 get 方法）；prometheus.yml 从 2 个目标扩展到 8 个应用服务 + MySQL/Redis + node-exporter + prometheus 自身 + rule_files + alertmanager 关联；monitoring compose 用 Loki + Promtail 替换未配置的 blackbox-exporter；Grafana datasource 新增 Loki；新建 promtail/config.yml 用 docker_sd_configs 仅采集 `edu-*` 容器日志。(2) 部署 compose 扩展：docker-compose.deploy.yml 从 3 服务扩展到 11 服务（+ iam/teacher-bff/core-edu/content/msg/ai/data-ana/push-gateway），每个服务带 healthcheck + depends_on 条件 + edu-net/edu-shared 双网络；deploy.env.example 补全 Neo4j/ES/ClickHouse/LLM/Kafka 可选依赖配置。(3) teacher-bff 补 health.controller.ts（原缺失 /healthz 导致 deploy depends_on service_healthy 失败）。(4) CI 硬化：移除 lint 步骤的 continue-on-error（ESLint 9 flat config 已配置完成），test 保留 continue-on-error（部分服务无 test 脚本）。**学到**：NestJS `app.get('/metrics')` 会被解析为 DI 容器 `get(typeOrToken)`，必须用 `app.getHttpAdapter().get()` 才能注册 Express 路由；Promtail docker_sd_configs 通过 relabel_configs 的 `regex: '/(edu-.*).*'` 过滤容器名前缀；docker-compose.depends_on.condition: service_healthy 要求被依赖服务必须有 healthcheck 配置，否则启动失败。                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                           |
 | 2026-07-09 | 下午 | data-ana/infra                  | **CDC 完整链路实现**：MySQL binlog → Debezium Connect → Kafka → data-ana 消费者 → ClickHouse 宽表。(1) MySQL binlog 配置：log_bin=ON, binlog_format=ROW, binlog_row_image=FULL, server_id=1；用 root 创建 `debezium` 用户授予 REPLICATION SLAVE + REPLICATION CLIENT。(2) Debezium Connect 容器：daocloud 禁用 debezium 镜像改用 `quay.io/debezium/connect:2.7`；MySQL 容器在 edu-minimal_default 网络，需 `docker network connect edu-full_default edu-mysql` 让 Debezium 同时可达；Kafka 必须配置双 listener（INSIDE:kafka:29092 + OUTSIDE:localhost:9092），否则 Debezium 拿到 advertised.listeners 中的 localhost metadata 后切换失败；Debezium 2.x 容器环境变量名用 BOOTSTRAP_SERVERS（不带 KAFKA_ 前缀），通过 envsubst 替换到 connect-distributed.properties。(3) 注册 connector：POST :8083/connectors，配置 topic.prefix=edu-cdc, database.include.list=next_edu_cloud, snapshot.mode=initial，4 张表（core_edu_grades/exams/classes/iam_users）成功产生快照事件。(4) data-ana 消费者实现：新建 cdc_consumer.py 用 aiokafka AIOKafkaConsumer，lifespan 中 asyncio.create_task 后台运行；按 source.table 路由（exams→内存缓存 exam_id→class_id 映射，grades→查缓存填 class_id 后 upsert ClickHouse）；readyz 端点附加 cdc_consumer 状态。(5) ClickHouse 远程访问：默认 default-user.xml 限制 127.0.0.1/::1 无密码，挂载 `clickhouse/users.d/custom-users.xml` 覆盖密码+任意 IP。(6) structlog 24.x API：`make_filtering_bound_logger(level)` 替代废弃的 `make_filtering_logger`。(7) E2E 验证：MySQL INSERT 成绩 → Debezium op=c 事件 → Kafka → 消费者写 ClickHouse 宽表（class_id 通过 exam 缓存正确填充）→ /readyz cdc_consumer=running → /analytics/student/student-002/weakness 返回实时 92 分数据。**学到**：Debezium 2.x 容器 bootstrap.servers 默认值是 0.0.0.0:9092 必须显式覆盖；Kafka 单 listener 配置 localhost 会让容器间通信的客户端拿到 metadata 后切换失败，必须用双 listener；ClickHouse users_xml 存储是 readonly 不能用 ALTER USER 修改密码，必须挂载 users.d 配置文件覆盖；消费者 offset 重置必须先停消费者让 group 处于 Empty 状态才能执行 --reset-offsets。 |
--- a/infra/docker-compose.yml
+++ b/infra/docker-compose.yml
@@ -184,8 +184,14 @@ services:
    container_name: edu-prometheus
    profiles: ["observability"]
    restart: unless-stopped
+    command:
+      - "--config.file=/etc/prometheus/prometheus.yml"
+      - "--storage.tsdb.path=/prometheus"
+      - "--storage.tsdb.retention.time=15d"
+      - "--web.enable-lifecycle"
    volumes:
      - ./prometheus.yml:/etc/prometheus/prometheus.yml:ro
+      - prometheus_data:/prometheus
    ports:
      - "9090:9090"
  grafana:
@@ -199,6 +205,49 @@ services:
      - "3030:3000"
    volumes:
      - grafana_data:/var/lib/grafana
+  # ============================================================
+  # Exporters（observability profile，与 Prometheus 同网络）
+  # ============================================================
+  node-exporter:
+    image: docker.m.daocloud.io/prom/node-exporter:v1.8.2
+    container_name: edu-node-exporter
+    profiles: ["observability"]
+    restart: unless-stopped
+    command:
+      - "--path.rootfs=/host"
+    ports:
+      - "9100:9100"
+    volumes:
+      - /proc:/host/proc:ro
+      - /sys:/host/sys:ro
+      - /:/host:ro
+  mysql-exporter:
+    image: docker.m.daocloud.io/prom/mysqld-exporter:v0.15.1
+    container_name: edu-mysql-exporter
+    profiles: ["observability"]
+    restart: unless-stopped
+    command:
+      - "--mysqld.address=edu-mysql:3306"
+      - "--mysqld.username=edu:changeme"
+    environment:
+      MYSQLD_EXPORTER_PASSWORD: "changeme"
+    ports:
+      - "9104:9104"
+    depends_on:
+      mysql:
+        condition: service_healthy
+  redis-exporter:
+    image: docker.m.daocloud.io/oliver006/redis_exporter:v1.67.0
+    container_name: edu-redis-exporter
+    profiles: ["observability"]
+    restart: unless-stopped
+    environment:
+      REDIS_ADDR: "redis://edu-redis:6379"
+    ports:
+      - "9121:9121"
+    depends_on:
+      redis:
+        condition: service_started
 volumes:
  mysql_data:
  redis_data:
@@ -206,3 +255,4 @@ volumes:
  neo4j_data:
  es_data:
  grafana_data:
+  prometheus_data:
--- a/infra/prometheus.yml
+++ b/infra/prometheus.yml
@@ -89,6 +89,6 @@ scrape_configs:

  - job_name: 'node-exporter'
    static_configs:
-      - targets: ['node-exporter:9100']
+      - targets: ['host.docker.internal:9100']
        labels:
          service: node-exporter
--- a/scripts/start-all.ps1
+++ b/scripts/start-all.ps1
@@ -13,7 +13,8 @@
    .\scripts\start-all.ps1 -SkipInfraCheck
 #>
 param(
-    [switch]$SkipInfraCheck
+    [switch]$SkipInfraCheck,
+    [switch]$Force
 )

 $ErrorActionPreference = "Stop"
@@ -26,7 +27,7 @@ Write-Host ""

 # ===== 1. Infrastructure health check =====
 if (-not $SkipInfraCheck) {
-    Write-Host "[1/4] Checking infrastructure health..." -ForegroundColor Yellow
+    Write-Host "[1/6] Checking infrastructure health..." -ForegroundColor Yellow
    $infraServices = @(
        @{Name="MySQL";       Container="edu-mysql"},
        @{Name="Redis";       Container="edu-redis"},
@@ -64,7 +65,7 @@ if (-not $SkipInfraCheck) {
 }

 # ===== 2. Environment variables =====
-Write-Host "[2/4] Preparing environment variables..." -ForegroundColor Yellow
+Write-Host "[2/6] Preparing environment variables..." -ForegroundColor Yellow

 $env:DEV_MODE = "true"
 $env:OTEL_EXPORTER_OTLP_ENDPOINT = "http://localhost:4318"
@@ -88,8 +89,103 @@ Write-Host "  DEV_MODE=true (dev-token bypass)" -ForegroundColor Green
 Write-Host "  OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:4318" -ForegroundColor Green
 Write-Host ""

-# ===== 3. Start application services =====
-Write-Host "[3/4] Starting application services (11 windows)..." -ForegroundColor Yellow
+# ===== 3. Port conflict check =====
+Write-Host "[3/6] Checking port conflicts..." -ForegroundColor Yellow
+
+$portMap = @{
+    3001="classes"; 3002="iam"; 3003="teacher-bff"; 3004="core-edu"
+    3005="content"; 3006="data-ana"; 3007="msg"; 3008="ai"
+    8080="api-gateway"; 8081="push-gateway"; 3000="teacher-portal"
+}
+
+$conflicts = @()
+foreach ($port in $portMap.Keys | Sort-Object) {
+    $conn = Get-NetTCPConnection -LocalPort $port -State Listen -ErrorAction SilentlyContinue
+    if ($conn) {
+        $svcName = $portMap[$port]
+        $procId = $conn[0].OwningProcess
+        $procName = ""
+        try { $procName = (Get-Process -Id $procId -ErrorAction Stop).ProcessName } catch {}
+        Write-Host "  [WARN] Port $port ($svcName) occupied by PID $procId ($procName)" -ForegroundColor Yellow
+        $conflicts += [PSCustomObject]@{Port=$port; Service=$svcName; PID=$procId; Process=$procName}
+    }
+}
+
+if ($conflicts.Count -gt 0) {
+    Write-Host ""
+    Write-Host "  $($conflicts.Count) port(s) already in use." -ForegroundColor Yellow
+
+    $shouldKill = $false
+    if ($Force) {
+        Write-Host "  -Force specified, killing automatically..." -ForegroundColor White
+        $shouldKill = $true
+    } else {
+        Write-Host "  These services may already be running. Options:" -ForegroundColor White
+        Write-Host "    1. Run .\scripts\stop-all.ps1 -KillByPort first, then re-run start-all" -ForegroundColor White
+        Write-Host "    2. Re-run with -Force to auto-kill and continue" -ForegroundColor White
+        Write-Host ""
+        $answer = Read-Host "  Kill existing processes and continue? (y/N)"
+        if ($answer -eq "y" -or $answer -eq "Y") {
+            $shouldKill = $true
+        }
+    }
+
+    if ($shouldKill) {
+        foreach ($c in $conflicts) {
+            try {
+                Stop-Process -Id $c.PID -Force -ErrorAction Stop
+                Write-Host "  [OK] Killed PID $($c.PID) ($($c.Process)) on port $($c.Port)" -ForegroundColor Green
+                Start-Sleep -Milliseconds 500
+            } catch {
+                Write-Host "  [WARN] Cannot kill PID $($c.PID): $($_.Exception.Message)" -ForegroundColor Yellow
+            }
+        }
+        Start-Sleep -Seconds 2
+    } else {
+        Write-Host "  Aborting. Please stop existing services first." -ForegroundColor Red
+        exit 1
+    }
+} else {
+    Write-Host "  [OK] All app ports are free" -ForegroundColor Green
+}
+Write-Host ""
+
+# ===== 4. Build NestJS services (required for nest start --watch) =====
+Write-Host "[4/6] Building NestJS services (first time required)..." -ForegroundColor Yellow
+
+$nestjsServices = @(
+    "@edu/classes-service",
+    "@edu/iam-service",
+    "@edu/teacher-bff",
+    "@edu/core-edu-service",
+    "@edu/content-service",
+    "@edu/msg-service"
+)
+
+# Clean tsbuildinfo cache (incremental mode leftover causes tsc to skip emit)
+$nestjsDirs = @("classes", "iam", "teacher-bff", "core-edu", "content", "msg")
+foreach ($dir in $nestjsDirs) {
+    $svcPath = Join-Path $ProjectRoot "services\$dir"
+    if (Test-Path $svcPath) {
+        Get-ChildItem -Path $svcPath -Filter "*.tsbuildinfo" -Recurse -ErrorAction SilentlyContinue |
+            Remove-Item -Force -ErrorAction SilentlyContinue
+    }
+}
+
+foreach ($svc in $nestjsServices) {
+    Write-Host "  Building $svc..." -ForegroundColor Gray -NoNewline
+    $buildResult = pnpm --filter $svc build 2>&1
+    if ($LASTEXITCODE -eq 0) {
+        Write-Host " [OK]" -ForegroundColor Green
+    } else {
+        Write-Host " [FAIL]" -ForegroundColor Red
+        Write-Host "  $buildResult" -ForegroundColor DarkGray
+    }
+}
+Write-Host ""
+
+# ===== 5. Start application services =====
+Write-Host "[5/6] Starting application services (11 windows)..." -ForegroundColor Yellow

 $services = @(
    @{Title="edu-app-classes";      Cmd="pnpm"; Args=@("--filter","@edu/classes-service","dev");          Dir="$ProjectRoot"},
@@ -98,8 +194,8 @@ $services = @(
    @{Title="edu-app-core-edu";     Cmd="pnpm"; Args=@("--filter","@edu/core-edu-service","dev");         Dir="$ProjectRoot"},
    @{Title="edu-app-content";      Cmd="pnpm"; Args=@("--filter","@edu/content-service","dev");          Dir="$ProjectRoot"},
    @{Title="edu-app-msg";          Cmd="pnpm"; Args=@("--filter","@edu/msg-service","dev");              Dir="$ProjectRoot"},
-    @{Title="edu-app-data-ana";     Cmd="uv";   Args=@("run","uvicorn","data_ana.main:app","--host","0.0.0.0","--port","3006","--reload"); Dir="$ProjectRoot\services\data-ana"; PyEnv=$true},
-    @{Title="edu-app-ai";           Cmd="uv";   Args=@("run","uvicorn","ai.main:app","--host","0.0.0.0","--port","3008","--reload");       Dir="$ProjectRoot\services\ai";         PyEnv=$true},
+    @{Title="edu-app-data-ana";     Cmd="uv";   Args=@("run","uvicorn","data_ana.main:app","--app-dir","src","--host","0.0.0.0","--port","3006","--reload"); Dir="$ProjectRoot\services\data-ana"; PyEnv=$true},
+    @{Title="edu-app-ai";           Cmd="uv";   Args=@("run","uvicorn","ai.main:app","--app-dir","src","--host","0.0.0.0","--port","3008","--reload");       Dir="$ProjectRoot\services\ai";         PyEnv=$true},
    @{Title="edu-app-api-gateway";  Cmd="go";   Args=@("run",".");                                          Dir="$ProjectRoot\services\api-gateway"; GoEnv=$true},
    @{Title="edu-app-push-gateway"; Cmd="go";   Args=@("run",".");                                          Dir="$ProjectRoot\services\push-gateway";GoEnv=$true},
    @{Title="edu-app-teacher-portal";Cmd="pnpm";Args=@("--filter","teacher-portal","dev");                 Dir="$ProjectRoot"}
@@ -129,11 +225,11 @@ foreach ($svc in $services) {
 }

 Write-Host ""
-Write-Host "  All services started in new windows. Waiting 30s for init..." -ForegroundColor Yellow
-Start-Sleep -Seconds 30
+Write-Host "  All services started in new windows. Waiting 40s for init..." -ForegroundColor Yellow
+Start-Sleep -Seconds 40

-# ===== 4. Health check =====
-Write-Host "[4/4] Health check..." -ForegroundColor Yellow
+# ===== 6. Health check =====
+Write-Host "[6/6] Health check..." -ForegroundColor Yellow
 Write-Host ""

 $healthServices = @(
@@ -152,26 +248,30 @@ $healthServices = @(

 $okCount = 0
 $failCount = 0
+$failedServices = @()
 foreach ($svc in $healthServices) {
    $retries = 0
-    $maxRetries = 5
+    $maxRetries = 3
    $success = $false
+    $lastError = ""
    while ($retries -lt $maxRetries -and -not $success) {
        try {
-            $null = Invoke-RestMethod -Uri $svc.Url -Method Get -TimeoutSec 3 -ErrorAction Stop
+            $null = Invoke-RestMethod -Uri $svc.Url -Method Get -TimeoutSec 5 -ErrorAction Stop
            Write-Host "  [OK] $($svc.Name)" -ForegroundColor Green
            $success = $true
            $okCount++
        } catch {
+            $lastError = $_.Exception.Message
            $retries++
            if ($retries -lt $maxRetries) {
-                Start-Sleep -Seconds 3
+                Start-Sleep -Seconds 5
            }
        }
    }
    if (-not $success) {
-        Write-Host "  [FAIL] $($svc.Name) (failed after $maxRetries retries)" -ForegroundColor Red
+        Write-Host "  [FAIL] $($svc.Name) (after $maxRetries retries: $lastError)" -ForegroundColor Red
        $failCount++
+        $failedServices += $svc.Name
    }
 }

@@ -179,13 +279,23 @@ Write-Host ""
 Write-Host "========================================" -ForegroundColor Cyan
 Write-Host "  Started: [OK] $okCount ready / [FAIL] $failCount failed" -ForegroundColor Cyan
 Write-Host "========================================" -ForegroundColor Cyan
+
+if ($failCount -gt 0) {
+    Write-Host ""
+    Write-Host "Failed services: $($failedServices -join ', ')" -ForegroundColor Red
+    Write-Host ""
+    Write-Host "Troubleshooting:" -ForegroundColor Yellow
+    Write-Host "  1. Check the service window for error output" -ForegroundColor White
+    Write-Host "  2. Verify dependencies: pnpm install / uv sync / go mod tidy" -ForegroundColor White
+    Write-Host "  3. Re-run health check: .\scripts\health-check.ps1" -ForegroundColor White
+    Write-Host "  4. Stop and retry: .\scripts\stop-all.ps1 -KillByPort then .\scripts\start-all.ps1" -ForegroundColor White
+    Write-Host ""
+    exit 1
+}
+
 Write-Host ""
 Write-Host "Next steps:" -ForegroundColor Yellow
 Write-Host "  Health check:  .\scripts\health-check.ps1" -ForegroundColor White
 Write-Host "  CDC test:      .\scripts\test-cdc.ps1" -ForegroundColor White
 Write-Host "  Stop all:      .\scripts\stop-all.ps1" -ForegroundColor White
 Write-Host ""
-
-if ($failCount -gt 0) {
-    exit 1
-}
--- a/services/classes/src/app.module.ts
+++ b/services/classes/src/app.module.ts
@@ -1,7 +1,8 @@
-import { Module } from '@nestjs/common';
-import { ClassesModule } from './classes/classes.module.js';
+import { Module } from "@nestjs/common";
+import { ClassesModule } from "./classes/classes.module.js";
+import { HealthModule } from "./shared/health/health.module.js";

@Module({
-  imports: [ClassesModule],
+  imports: [ClassesModule, HealthModule],
 })
 export class AppModule {}
--- a/services/classes/src/shared/health/health.controller.ts
+++ b/services/classes/src/shared/health/health.controller.ts
@@ -1,7 +1,8 @@
-import { Controller, Get, HttpException, HttpStatus } from '@nestjs/common';
-import { DataSource } from 'typeorm';
+import { Controller, Get, HttpException, HttpStatus } from "@nestjs/common";
+import { sql } from "drizzle-orm";
+import { getDb } from "../../config/database.js";

-const SERVICE_NAME = 'classes';
+const SERVICE_NAME = "classes";

 /**
 * 健康检查端点。
@@ -14,33 +15,37 @@ const SERVICE_NAME = 'classes';
 */
@Controller()
 export class HealthController {
-  constructor(private readonly dataSource: DataSource) {}
-
-  @Get('healthz')
+  @Get("healthz")
  liveness(): { status: string; service: string; timestamp: string } {
    return {
-      status: 'ok',
+      status: "ok",
      service: SERVICE_NAME,
      timestamp: new Date().toISOString(),
    };
  }

-  @Get('readyz')
-  async readiness(): Promise<{ status: string; service: string; timestamp: string }> {
+  @Get("readyz")
+  async readiness(): Promise<{
+    status: string;
+    service: string;
+    timestamp: string;
+  }> {
    try {
-      await this.dataSource.query('SELECT 1');
+      const db = getDb();
+      await db.execute(sql`SELECT 1`);
      return {
-        status: 'ok',
+        status: "ok",
        service: SERVICE_NAME,
        timestamp: new Date().toISOString(),
      };
    } catch (error) {
      throw new HttpException(
        {
-          status: 'error',
+          status: "error",
          service: SERVICE_NAME,
          timestamp: new Date().toISOString(),
-          error: error instanceof Error ? error.message : 'database unreachable',
+          error:
+            error instanceof Error ? error.message : "database unreachable",
        },
        HttpStatus.SERVICE_UNAVAILABLE,
      );
--- a/services/classes/tsconfig.json
+++ b/services/classes/tsconfig.json
@@ -8,6 +8,7 @@
    "emitDecoratorMetadata": true,
    "outDir": "./dist",
    "rootDir": "./src",
+    "incremental": false,
    "types": ["node"]
  },
  "include": ["src/**/*"],
--- a/services/content/tsconfig.json
+++ b/services/content/tsconfig.json
@@ -8,6 +8,7 @@
    "emitDecoratorMetadata": true,
    "outDir": "./dist",
    "rootDir": "./src",
+    "incremental": false,
    "types": ["node"]
  },
  "include": ["src/**/*"],
--- a/services/core-edu/tsconfig.json
+++ b/services/core-edu/tsconfig.json
@@ -8,6 +8,7 @@
    "emitDecoratorMetadata": true,
    "outDir": "./dist",
    "rootDir": "./src",
+    "incremental": false,
    "types": ["node"]
  },
  "include": ["src/**/*"],
--- a/services/iam/src/app.module.ts
+++ b/services/iam/src/app.module.ts
@@ -1,7 +1,8 @@
-import { Module } from '@nestjs/common';
-import { IamModule } from './iam/iam.module.js';
+import { Module } from "@nestjs/common";
+import { IamModule } from "./iam/iam.module.js";
+import { HealthModule } from "./shared/health/health.module.js";

@Module({
-  imports: [IamModule],
+  imports: [IamModule, HealthModule],
 })
 export class AppModule {}
--- a/services/iam/tsconfig.json
+++ b/services/iam/tsconfig.json
@@ -8,6 +8,7 @@
    "emitDecoratorMetadata": true,
    "outDir": "./dist",
    "rootDir": "./src",
+    "incremental": false,
    "types": ["node"]
  },
  "include": ["src/**/*"],
--- a/services/msg/tsconfig.json
+++ b/services/msg/tsconfig.json
@@ -8,6 +8,7 @@
    "emitDecoratorMetadata": true,
    "outDir": "./dist",
    "rootDir": "./src",
+    "incremental": false,
    "types": ["node"]
  },
  "include": ["src/**/*"],
--- a/services/teacher-bff/src/main.ts
+++ b/services/teacher-bff/src/main.ts
@@ -3,6 +3,7 @@ import { NestFactory } from "@nestjs/core";
 import { AppModule } from "./app.module.js";
 import { env } from "./config/env.js";
 import { initTracer, shutdownTracer } from "./shared/observability/tracer.js";
+import { metricsRegistry } from "./shared/observability/metrics.js";

 async function bootstrap(): Promise<void> {
  initTracer();
@@ -12,6 +13,12 @@ async function bootstrap(): Promise<void> {

  app.enableShutdownHooks();

+  // Prometheus 指标端点：不鉴权，供 Prometheus 抓取。
+  app.getHttpAdapter().get("/metrics", async (_req, res) => {
+    res.set("Content-Type", metricsRegistry.contentType);
+    res.end(await metricsRegistry.metrics());
+  });
+
  await app.listen(env.PORT);
  console.log(`Teacher BFF started on port ${env.PORT}`);

--- a/services/teacher-bff/src/shared/observability/metrics.ts
+++ b/services/teacher-bff/src/shared/observability/metrics.ts
@@ -0,0 +1,26 @@
+import promClient from "prom-client";
+
+const registry = new promClient.Registry();
+registry.setDefaultLabels({ service: "teacher-bff" });
+
+registry.registerMetric(
+  new promClient.Counter({
+    name: "teacher_bff_requests_total",
+    help: "Total number of teacher-bff requests",
+    labelNames: ["method", "endpoint", "status"],
+  }),
+);
+
+registry.registerMetric(
+  new promClient.Histogram({
+    name: "teacher_bff_request_duration_seconds",
+    help: "Teacher-bff request duration in seconds",
+    labelNames: ["method", "endpoint"],
+    buckets: [0.01, 0.05, 0.1, 0.3, 0.5, 1, 3, 5],
+  }),
+);
+
+// 自动收集 Node.js 进程级指标（CPU/内存/事件循环/GC等）
+promClient.collectDefaultMetrics({ register: registry });
+
+export { registry as metricsRegistry };
--- a/services/teacher-bff/tsconfig.json
+++ b/services/teacher-bff/tsconfig.json
@@ -8,6 +8,7 @@
    "emitDecoratorMetadata": true,
    "outDir": "./dist",
    "rootDir": "./src",
+    "incremental": false,
    "types": ["node"]
  },
  "include": ["src/**/*"],