feat: add docker capacity benchmark mode

2026-07-01 03:31:17 +08:00
parent eb3dca5fb8
commit 3d33c783cd
11 changed files with 242 additions and 6 deletions
--- a/README.md
+++ b/README.md
@@ -557,6 +557,15 @@ docs/refactor/postgresql-4c16g-tuning.md
 docs/refactor/performance-benchmark-runbook.md
 ```

+本地 Docker Desktop 可用时，可以先用受限 API 容器做 4 核 16G shared-host 风格的预演。该入口默认把 API 容器限制为 2 CPU/4G、关闭 legacy `x-user-id`，并用 `Authorization: Bearer <tk_session>` 跑 30 只读、50/100 混合读写矩阵：
+
+```powershell
+$env:DATABASE_URL="postgresql://postgres:postgres@127.0.0.1:54322/postgres"
+npm run perf:api:docker-4c16g
+```
+
+注意：这个入口只限制 API 容器资源，本地 Supabase/PostgreSQL 仍受 Docker Desktop 全局资源影响。正式容量承诺仍要在目标 4 核 16G 云服务器复跑。
+
 PostgreSQL 调参与运行证据采集：

 ```powershell
--- a/apps/api/Dockerfile
+++ b/apps/api/Dockerfile
@@ -12,7 +12,7 @@ RUN npm ci --workspaces --include-workspace-root
 FROM node:20-alpine AS build
 WORKDIR /app
 COPY --from=deps /app/node_modules ./node_modules
-COPY package.json package-lock.json tsconfig.json ./
+COPY package.json package-lock.json ./
 COPY apps/api ./apps/api
 COPY packages ./packages
 RUN npm run build:api
--- a/docker-compose.api.benchmark.yml
+++ b/docker-compose.api.benchmark.yml
@@ -0,0 +1,15 @@
+name: tiku-saas-benchmark
+
+services:
+  api:
+    environment:
+      NODE_ENV: development
+      ALLOW_LEGACY_AUTH_HEADERS: "false"
+      ALLOW_PLATFORM_ADMIN_KEY: "false"
+      AUTH_SESSION_SECRET: ${AUTH_SESSION_SECRET:-development-session-secret-change-me}
+      DB_POOL_MAX: ${DB_POOL_MAX:-10}
+      MAX_JSON_BODY_BYTES: "1048576"
+      MAX_IMPORT_JSON_BODY_BYTES: "10485760"
+    cpus: "${BENCHMARK_API_CPUS:-2.0}"
+    mem_limit: "${BENCHMARK_API_MEMORY:-4g}"
+    memswap_limit: "${BENCHMARK_API_MEMORY:-4g}"
--- a/docker-compose.api.yml
+++ b/docker-compose.api.yml
@@ -11,6 +11,7 @@ services:
      DATABASE_URL: postgresql://postgres:postgres@host.docker.internal:54322/postgres
      DEFAULT_TENANT_SLUG: master
      CORS_ORIGIN: http://127.0.0.1:5173,http://localhost:5173,http://127.0.0.1:5174,http://localhost:5174
+      DB_POOL_MAX: 10
    extra_hosts:
      - host.docker.internal:host-gateway
    ports:
--- a/docs/refactor/README.md
+++ b/docs/refactor/README.md
@@ -16,7 +16,7 @@
 - `apps/api`：新业务 API 服务，内部按 `src/core` 和 `src/features` 分层。
 - `packages/config`、`packages/db`、`packages/domain`：新系统共享基础包。
 - `scripts/import-pocketbase`：PocketBase schema/数据导入工具。
- `docker-compose.api.yml`、`apps/api/Dockerfile`：本地 Docker API 运行入口。
+- `docker-compose.api.yml`、`docker-compose.api.benchmark.yml`、`apps/api/Dockerfile`：本地 Docker API 和受限资源压测入口。
 - `docs/refactor/architecture.md`：新重构目录边界和工程规范。
 - `docs/refactor/ai-development-guardrails.md`：后续 AI/开发者必须遵守的 Supabase-first 架构和安全守则。
 - `docs/refactor/content-import-contract.md`：题目、单词、知识手册导入契约，明确后端校验、旧格式转换和前端职责。
@@ -29,7 +29,7 @@
 - `docs/refactor/taro-frontend-integration.md`：Taro/H5/小程序启动、请求封装、页面/API 映射。
 - `docs/refactor/taro-h5-deployment.md`：Taro H5 三域名部署、运行时配置、Nginx、CSP、缓存和 CORS 边界。
 - `docs/refactor/postgresql-4c16g-tuning.md`：4 核 16G 自托管 PostgreSQL 起步调参、观察 SQL 和回滚方式。
- `docs/refactor/performance-benchmark-runbook.md`：本地/云端 API 压测、4 核 16G 阶梯并发矩阵和报告归档方式。
+- `docs/refactor/performance-benchmark-runbook.md`：本地/云端 API 压测、4 核 16G 阶梯并发矩阵、Docker 受限资源预演和报告归档方式。
 - `docs/refactor/performance-benchmark-summary-20260630.md`：真实迁移数据压测脱敏摘要。
 - `docs/refactor/backend-open-items-and-capacity-20260701.md`：后端剩余功能、已定稿产品口径和最新在线容量估算。
 - `docs/refactor/multitenant-auth-security-contract.md`：多租户隔离、鉴权、权限和资源安全红线。
--- a/docs/refactor/architecture.md
+++ b/docs/refactor/architecture.md
@@ -62,6 +62,14 @@ npm run docker:api:up

 如果 Docker 拉取 `node:20-alpine` 超时，先配置 Docker Desktop 镜像源或代理，再重试 `npm run docker:api:build`。

+本地容量预演可以使用专门的 benchmark override：
+
+```bash
+npm run perf:api:docker-4c16g
+```
+
+该命令会使用 `docker-compose.api.yml` + `docker-compose.api.benchmark.yml` 启动受限 API 容器，默认限制 API 为 2 CPU/4G、`DB_POOL_MAX=10`，关闭 `ALLOW_LEGACY_AUTH_HEADERS`，并用 Bearer `tk_` 会话跑压测。它只用于本地模拟和跑分，不是生产 compose 文件。
+
 验证 API：

 ```bash
@@ -81,6 +89,7 @@ npm run pb:import:validate

 - Docker Desktop 可用。
 - Supabase 本地容器可启动。
+- API Dockerfile 已验证可构建；benchmark override 可启动受限 API 容器并通过短压测 smoke。
 - `supabase db reset` 可完整执行三份 migration 和 seed。
 - `supabase db reset` 可完整执行全部 migration 和 seed。
 - `npm run db:smoke-seed` 可恢复最小业务烟测数据。
--- a/docs/refactor/performance-benchmark-runbook.md
+++ b/docs/refactor/performance-benchmark-runbook.md
@@ -19,6 +19,16 @@ npm run perf:api:local
 - 默认不创建练习 session，不写业务数据。
 - 输出 JSON 和 Markdown 报告到 `docs/refactor/performance-reports/`。该目录已被 `.gitignore` 忽略，不应提交。

+默认本地兼容模式使用 `x-user-id` 作为压测身份，只适合开发机。更接近 Taro/Supabase 接入方式的压测应启用迁移期 Bearer session：
+
+```powershell
+$env:PERF_AUTH_MODE="app_session"
+npm run perf:api:local
+Remove-Item Env:\PERF_AUTH_MODE
+```
+
+`app_session` 模式会在测试库中为压测学生创建 2 小时 `tk_` 会话，然后用 `Authorization: Bearer <tk_...>` 请求业务接口；API 可以关闭 `ALLOW_LEGACY_AUTH_HEADERS`。生产远程压测如果已经有真实 Supabase access token，也可以用 `PERF_AUTH_MODE=bearer` 和 `PERF_BEARER_TOKEN`。
+
 PostgreSQL 调参与运行证据：

 ```bash
@@ -83,6 +93,8 @@ npm run perf:api:local
 | `PERF_INCLUDE_LEADERBOARD` | `false` | 是否加入排行榜接口；排行榜租户默认关闭，仅在租户明确开启并需要专项压测时打开 |
 | `PERF_OUTPUT_DIR` | `docs/refactor/performance-reports` | 报告输出目录 |
 | `PERF_REQUEST_TIMEOUT_MS` | `15000` | 单请求超时 |
+| `PERF_AUTH_MODE` | `legacy` | `legacy` 使用本地 `x-user-id`；`app_session` 创建 `tk_` Bearer session；`bearer` 使用 `PERF_BEARER_TOKEN`；`none` 只适合公开接口 |
+| `PERF_BEARER_TOKEN` | 空 | `PERF_AUTH_MODE=bearer` 时使用 |

 ## 默认工作负载

@@ -138,6 +150,37 @@ npm run perf:api:local

 本地 Docker Desktop 可以先用同一矩阵做跑分，但只能证明代码、索引和本机 Docker 环境的趋势。正式容量承诺必须以目标云服务器、生产 PostgreSQL 参数、生产 API/worker 连接池、对象存储/CDN 和真实网络重新跑。

+## 本地 Docker 4c16g 模拟入口
+
+仓库提供一个 Docker API 受限资源压测入口：
+
+```powershell
+$env:DATABASE_URL="postgresql://postgres:postgres@127.0.0.1:54322/postgres"
+$env:BENCHMARK_API_CPUS="2.0"
+$env:BENCHMARK_API_MEMORY="4g"
+$env:DB_POOL_MAX="10"
+npm run perf:api:docker-4c16g
+Remove-Item Env:\BENCHMARK_API_CPUS
+Remove-Item Env:\BENCHMARK_API_MEMORY
+Remove-Item Env:\DB_POOL_MAX
+```
+
+它会使用 `docker-compose.api.yml` + `docker-compose.api.benchmark.yml` 构建并启动 API 容器，默认限制 API 容器为 2 CPU/4G，关闭 `ALLOW_LEGACY_AUTH_HEADERS`，然后用 `PERF_AUTH_MODE=app_session` 跑：
+
+- 30 worker / 120s / 只读上线门禁。
+- 50 worker / 60s / 10% 刷题闭环写入。
+- 100 worker / 60s / 8% 刷题闭环写入。
+
+这个入口模拟的是“4 核 16G shared-host 中 API 容器的资源约束”，不是完整云服务器复刻。当前本地 Supabase/PostgreSQL 仍跑在 Docker Desktop 的 Supabase stack 中，除非额外手工限制 DB 容器资源，否则数据库容器仍可能使用 Docker Desktop 的全局资源。正式容量承诺必须在目标 4 核 16G 云服务器、生产 PostgreSQL shared-host 参数、真实对象存储/CDN 和真实网络下复跑。
+
+如需保留 API 容器便于排查：
+
+```powershell
+$env:PERF_KEEP_DOCKER_API="true"
+npm run perf:api:docker-4c16g
+Remove-Item Env:\PERF_KEEP_DOCKER_API
+```
+
 PowerShell 示例：

 ```powershell
--- a/package.json
+++ b/package.json
@@ -39,6 +39,7 @@
    "smoke:auth:remote": "node scripts/remote-auth-jwt-smoke.js",
    "test:api": "npm run db:smoke-seed && npm run build:api && node scripts/api-integration-test.js --start-server",
    "perf:api:local": "npm run build:api && node scripts/api-performance-benchmark.js",
+    "perf:api:docker-4c16g": "node scripts/run-docker-4c16g-benchmark.js",
    "perf:postgres:evidence": "node scripts/postgres-tuning-evidence.js",
    "test:worker:crm": "npm run db:smoke-seed && npm run build:worker && node scripts/crm-worker-integration-test.js",
    "test:worker:commerce": "npm run db:smoke-seed && npm run build:worker && node scripts/commerce-worker-integration-test.js",
--- a/packages/db/src/index.ts
+++ b/packages/db/src/index.ts
@@ -9,10 +9,16 @@ export interface DbPoolOptions {
  idleTimeoutMillis?: number;
 }

+function envPoolMax() {
+  const value = Number(process.env.DB_POOL_MAX || 10);
+  if (!Number.isFinite(value) || value <= 0) return 10;
+  return Math.trunc(value);
+}
+
 export function createPool(options: DbPoolOptions = {}) {
  return new Pool({
    connectionString: options.connectionString || process.env.DATABASE_URL || DEFAULT_DATABASE_URL,
-    max: options.max || 10,
+    max: options.max || envPoolMax(),
    idleTimeoutMillis: options.idleTimeoutMillis || 30_000,
  });
 }
--- a/scripts/api-performance-benchmark.js
+++ b/scripts/api-performance-benchmark.js
@@ -1,4 +1,5 @@
 import { spawn } from 'node:child_process';
+import crypto from 'node:crypto';
 import fs from 'node:fs/promises';
 import net from 'node:net';
 import path from 'node:path';
@@ -23,6 +24,9 @@ const TENANT_CODE = process.env.PERF_TENANT_CODE || 'master';
 const QUESTION_LIMIT = envNumber('PERF_QUESTION_LIMIT', 20);
 const MAX_ERRORS_TO_KEEP = envNumber('PERF_MAX_ERRORS', 20);
 const DEFAULT_TIMEOUT_MS = envNumber('PERF_REQUEST_TIMEOUT_MS', 15_000);
+const AUTH_MODE = normalizeAuthMode(process.env.PERF_AUTH_MODE || 'legacy');
+const STATIC_BEARER_TOKEN = process.env.PERF_BEARER_TOKEN || '';
+const AUTH_SESSION_SECRET = process.env.AUTH_SESSION_SECRET || 'development-session-secret-change-me';

 let apiBase = API_BASE_ENV || '';
 let serverProcess = null;
@@ -44,6 +48,14 @@ function envNumber(key, fallback, options = {}) {
  return fallback;
 }

+function normalizeAuthMode(value) {
+  const normalized = String(value || '').toLowerCase().replace(/[-_]/g, '');
+  if (normalized === 'appsession' || normalized === 'session' || normalized === 'tk') return 'app_session';
+  if (normalized === 'bearer' || normalized === 'jwt' || normalized === 'token') return 'bearer';
+  if (normalized === 'none' || normalized === 'anonymous') return 'none';
+  return 'legacy';
+}
+
 function getFreePort() {
  return new Promise((resolve, reject) => {
    const server = net.createServer();
@@ -91,6 +103,7 @@ async function requestJson(endpoint, context, timeoutMs = DEFAULT_TIMEOUT_MS) {
  const started = performance.now();
  const tenantId = context.tenantId || context.tenant?.id;
  const userId = context.userId || context.user?.id;
+  const authToken = context.authToken || '';
  try {
    const response = await fetch(buildUrl(apiBase, endpoint), {
      method: endpoint.method || 'GET',
@@ -98,7 +111,8 @@ async function requestJson(endpoint, context, timeoutMs = DEFAULT_TIMEOUT_MS) {
      headers: {
        'content-type': 'application/json',
        ...(endpoint.tenantHeader === false || !tenantId ? {} : { 'x-tenant-id': tenantId }),
-        ...(endpoint.userHeader === false || !userId ? {} : { 'x-user-id': userId }),
+        ...(AUTH_MODE === 'legacy' && endpoint.userHeader !== false && userId ? { 'x-user-id': userId } : {}),
+        ...(endpoint.authHeader === false || !authToken ? {} : { authorization: `Bearer ${authToken}` }),
        ...(endpoint.headers || {}),
      },
      body: endpoint.body ? JSON.stringify(endpoint.body(context)) : undefined,
@@ -311,6 +325,7 @@ async function startServerIfNeeded() {
      ...process.env,
      PORT: String(port),
      DATABASE_URL,
+      AUTH_SESSION_SECRET,
      MAX_JSON_BODY_BYTES: process.env.MAX_JSON_BODY_BYTES || '1048576',
      MAX_IMPORT_JSON_BODY_BYTES: process.env.MAX_IMPORT_JSON_BODY_BYTES || '10485760',
    },
@@ -342,6 +357,46 @@ async function many(pool, sql, params = []) {
  return result.rows;
 }

+function createSessionToken() {
+  return `tk_${crypto.randomBytes(32).toString('base64url')}`;
+}
+
+function hashSessionToken(token) {
+  return crypto.createHmac('sha256', AUTH_SESSION_SECRET).update(token).digest('hex');
+}
+
+async function createBenchmarkAuthToken(context) {
+  if (AUTH_MODE === 'none' || AUTH_MODE === 'legacy') return '';
+  if (AUTH_MODE === 'bearer') {
+    if (!STATIC_BEARER_TOKEN) throw new Error('PERF_BEARER_TOKEN is required when PERF_AUTH_MODE=bearer.');
+    return STATIC_BEARER_TOKEN;
+  }
+  if (AUTH_MODE !== 'app_session') return '';
+  if (!context.tenant?.id || !context.user?.id) {
+    throw new Error('PERF_AUTH_MODE=app_session requires discovered tenant and user context.');
+  }
+  const token = createSessionToken();
+  const pool = new pg.Pool({ connectionString: DATABASE_URL, max: 2 });
+  try {
+    await pool.query(
+      `
+      insert into app_private.auth_sessions (
+        tenant_id, user_id, token_hash, provider, expires_at, metadata
+      )
+      values (
+        $1, $2, $3, 'performance_benchmark',
+        now() + interval '2 hours',
+        '{"createdBy":"api-performance-benchmark","authMode":"app_session"}'::jsonb
+      )
+      `,
+      [context.tenant.id, context.user.id, hashSessionToken(token)],
+    );
+    return token;
+  } finally {
+    await pool.end();
+  }
+}
+
 async function discoverBenchmarkContext() {
  const pool = new pg.Pool({ connectionString: DATABASE_URL, max: 4 });
  try {
@@ -790,6 +845,7 @@ async function runBenchmark(context) {
      concurrency: CONCURRENCY,
      rampSeconds: RAMP_SECONDS,
      includeWrites: INCLUDE_WRITES,
+      authMode: AUTH_MODE,
      practiceFlowRatio: INCLUDE_WRITES ? PRACTICE_FLOW_RATIO : 0,
      practiceFlowAnswers: INCLUDE_WRITES ? PRACTICE_FLOW_ANSWERS : 0,
      questionLimit: QUESTION_LIMIT,
@@ -906,10 +962,11 @@ async function main() {
      throw new Error('PERF_API_BASE is required when PERF_START_SERVER=false.');
    }
    const context = await discoverBenchmarkContext();
+    context.authToken = await createBenchmarkAuthToken(context);
    await startServerIfNeeded();
    if (!apiBase) throw new Error('API base URL was not resolved.');
    console.log(`[perf] target api=${apiBase} tenant=${context.tenant.slug} user=${context.user.id}`);
-    console.log(`[perf] duration=${DURATION_SECONDS}s concurrency=${CONCURRENCY} includeWrites=${INCLUDE_WRITES}`);
+    console.log(`[perf] duration=${DURATION_SECONDS}s concurrency=${CONCURRENCY} includeWrites=${INCLUDE_WRITES} authMode=${AUTH_MODE}`);
    const report = await runBenchmark(context);
    const files = await writeReport(report);
    console.log(`[perf] requests=${report.summary.requests} ok=${report.summary.ok} errors=${report.summary.errors} rps=${report.summary.throughputRps} p95=${report.summary.latencyOk.p95Ms}ms`);
--- a/scripts/run-docker-4c16g-benchmark.js
+++ b/scripts/run-docker-4c16g-benchmark.js
@@ -0,0 +1,95 @@
+import { spawnSync } from 'node:child_process';
+import fs from 'node:fs';
+import path from 'node:path';
+
+const repoRoot = process.cwd();
+const databaseUrl = process.env.DATABASE_URL || 'postgresql://postgres:postgres@127.0.0.1:54322/postgres';
+const apiBase = process.env.PERF_API_BASE || 'http://127.0.0.1:8787';
+const matrix = [
+  { name: 'read-30', duration: 120, concurrency: 30, ramp: 15, writes: false, ratio: 0 },
+  { name: 'mixed-50', duration: 60, concurrency: 50, ramp: 10, writes: true, ratio: 0.1 },
+  { name: 'mixed-100', duration: 60, concurrency: 100, ramp: 15, writes: true, ratio: 0.08 },
+];
+
+function run(command, args, options = {}) {
+  const result = spawnSync(command, args, {
+    cwd: repoRoot,
+    stdio: 'inherit',
+    shell: process.platform === 'win32',
+    env: {
+      ...process.env,
+      DATABASE_URL: databaseUrl,
+      AUTH_SESSION_SECRET: process.env.AUTH_SESSION_SECRET || 'development-session-secret-change-me',
+      DB_POOL_MAX: process.env.DB_POOL_MAX || '10',
+      BENCHMARK_API_CPUS: process.env.BENCHMARK_API_CPUS || '2.0',
+      BENCHMARK_API_MEMORY: process.env.BENCHMARK_API_MEMORY || '4g',
+      ...options.env,
+    },
+  });
+  if (result.status !== 0) {
+    throw new Error(`${command} ${args.join(' ')} failed with exit code ${result.status}`);
+  }
+}
+
+function latestBenchmarkJson(before) {
+  const dir = path.join(repoRoot, 'docs', 'refactor', 'performance-reports');
+  const files = fs.existsSync(dir)
+    ? fs.readdirSync(dir)
+        .filter(name => /^api-benchmark-\d{8}-\d{6}\.json$/.test(name))
+        .map(name => path.join(dir, name))
+        .filter(file => fs.statSync(file).mtimeMs >= before)
+        .sort((a, b) => fs.statSync(b).mtimeMs - fs.statSync(a).mtimeMs)
+    : [];
+  if (!files[0]) throw new Error('Benchmark report JSON was not generated.');
+  return path.relative(repoRoot, files[0]).replaceAll('\\', '/');
+}
+
+function waitForHealth(timeoutMs = 60_000) {
+  const started = Date.now();
+  while (Date.now() - started < timeoutMs) {
+    const result = spawnSync(process.execPath, ['-e', `fetch('${apiBase}/health').then(r=>process.exit(r.ok?0:1)).catch(()=>process.exit(1))`], {
+      cwd: repoRoot,
+      stdio: 'ignore',
+      shell: process.platform === 'win32',
+    });
+    if (result.status === 0) return;
+    Atomics.wait(new Int32Array(new SharedArrayBuffer(4)), 0, 0, 1000);
+  }
+  throw new Error(`API did not become healthy at ${apiBase}`);
+}
+
+function main() {
+  const generated = [];
+  try {
+    run('docker', ['compose', '-f', 'docker-compose.api.yml', '-f', 'docker-compose.api.benchmark.yml', 'up', '-d', '--build', 'api']);
+    waitForHealth();
+    for (const item of matrix) {
+      const before = Date.now();
+      run('node', ['scripts/api-performance-benchmark.js'], {
+        env: {
+          PERF_START_SERVER: 'false',
+          PERF_API_BASE: apiBase,
+          PERF_AUTH_MODE: 'app_session',
+          PERF_DURATION_SECONDS: String(item.duration),
+          PERF_CONCURRENCY: String(item.concurrency),
+          PERF_RAMP_SECONDS: String(item.ramp),
+          PERF_INCLUDE_WRITES: item.writes ? 'true' : 'false',
+          PERF_PRACTICE_FLOW_RATIO: String(item.ratio),
+          PERF_PRACTICE_FLOW_ANSWERS: '3',
+        },
+      });
+      generated.push({ scenario: item.name, report: latestBenchmarkJson(before) });
+    }
+    console.log(JSON.stringify({ status: 'pass', apiBase, reports: generated }, null, 2));
+  } finally {
+    if (process.env.PERF_KEEP_DOCKER_API !== 'true') {
+      spawnSync('docker', ['compose', '-f', 'docker-compose.api.yml', '-f', 'docker-compose.api.benchmark.yml', 'down'], {
+        cwd: repoRoot,
+        stdio: 'inherit',
+        shell: process.platform === 'win32',
+      });
+    }
+  }
+}
+
+main();