fix: preserve container startup diagnostics

back/schedule/health.ts

@@ -1,28 +1,104 @@
 import { ServerUnaryCall, sendUnaryData } from '@grpc/grpc-js';
+import fs from 'fs/promises';
+import path from 'path';
+import { request } from 'undici';
 import { HealthCheckRequest, HealthCheckResponse } from '../protos/health';
 import config from '../config';
-import { promiseExec } from '../config/util';
+
+function formatError(error: unknown): string {
+  if (!(error instanceof Error)) {
+    return String(error);
+  }
+
+  const detailedError = error as Error & {
+    cause?: unknown;
+    code?: string;
+    errors?: unknown[];
+  };
+  const details = [error.message, detailedError.code];
+  if (Array.isArray(detailedError.errors)) {
+    details.push(...detailedError.errors.map(formatError));
+  } else if (detailedError.cause) {
+    details.push(formatError(detailedError.cause));
+  }
+
+  return [...new Set(details.filter(Boolean))].join(': ') || error.name;
+}
+
+async function getRecentSystemLog(lineLimit = 300): Promise<string> {
+  try {
+    const entries = await fs.readdir(config.systemLogPath);
+    const latestLog = entries
+      .filter((entry) => entry.endsWith('.log'))
+      .sort()
+      .at(-1);
+    if (!latestLog) {
+      return `No system log found in ${config.systemLogPath}`;
+    }
+    const content = await fs.readFile(
+      path.join(config.systemLogPath, latestLog),
+      'utf8',
+    );
+    return content.split('\n').slice(-lineLimit).join('\n').trim();
+  } catch (error) {
+    return `Unable to read system log from ${config.systemLogPath}: ${
+      error instanceof Error ? error.message : String(error)
+    }`;
+  }
+}
 
 const check = async (
   call: ServerUnaryCall<HealthCheckRequest, HealthCheckResponse>,
   callback: sendUnaryData<HealthCheckResponse>,
 ) => {
   switch (call.request.service) {
-    case 'cron':
+    case 'cron': {
-      const res = await promiseExec(
+      const healthUrl = `http://localhost:${config.port}${
-        `curl -s --noproxy '*' http://localhost:${config.port}/api/system`,
+        config.baseUrl || ''
-      );
+      }/api/health`;
-
+      let failure = '';
-      if (res.includes('200')) {
+      try {
-        return callback(null, { status: 1 });
+        const response = await request(healthUrl, {
+          method: 'GET',
+          headersTimeout: 5000,
+          bodyTimeout: 5000,
+        });
+        const body = (await response.body.json()) as {
+          code?: number;
+          data?: { status?: string };
+        };
+        if (
+          response.statusCode >= 200 &&
+          response.statusCode < 300 &&
+          body.code === 200 &&
+          body.data?.status === 'ok'
+        ) {
+          return callback(null, { status: 1 });
+        }
+        failure = `HTTP ${response.statusCode}: ${JSON.stringify(body)}`;
+      } catch (error) {
+        failure = formatError(error);
       }
 
-      const qinglongErrLog = await promiseExec(
+      const systemLog = await getRecentSystemLog();
-        `tail -n 300 ~/.pm2/logs/qinglong-error.log`,
+      const containerHint =
-      );
+        process.env.QL_CONTAINER === 'true'
+          ? 'PM2 file logging is disabled in containers. Check `docker logs <container>` for early startup errors.'
+          : 'Check `pm2 logs qinglong --lines 300` for early startup errors.';
       return callback(
-        new Error(`${qinglongErrLog || ''}\n${res}`.trim()),
+        new Error(
+          [
+            `HTTP health check failed: ${healthUrl}`,
+            failure,
+            containerHint,
+            `Recent system log (${config.systemLogPath}):`,
+            systemLog,
+          ]
+            .filter(Boolean)
+            .join('\n'),
+        ),
       );
+    }
 
     default:
       return callback(null, { status: 1 });

ecosystem.config.js

@@ -11,7 +11,9 @@ module.exports = {
       source_map_support: true,
       time: !isContainer,
       out_file: isContainer ? '/dev/null' : undefined,
-      error_file: isContainer ? '/dev/null' : undefined,
+      // Do not persist PM2 logs in containers, but keep early startup errors
+      // visible through `docker logs` before Winston is initialized.
+      error_file: isContainer ? '/proc/1/fd/2' : undefined,
       script: 'static/build/app.js',
       env: {
         http_proxy: '',

shell/check.sh

@@ -23,37 +23,75 @@ copy_dep() {
 }
 
 pm2_log() {
-  t '---> pm2日志'
+  t '---> 服务诊断信息'
-  local panelOut="/root/.pm2/logs/qinglong-out.log"
+  pm2 status || true
-  local panelError="/root/.pm2/logs/qinglong-error.log"
+
-  tail -n 300 "$panelOut"
+  local systemLogDir="$dir_data/syslog"
-  tail -n 300 "$panelError"
+  local latestSystemLog
+  latestSystemLog=$(find "$systemLogDir" -maxdepth 1 -type f -name '*.log' 2>/dev/null | sort | tail -n 1)
+  if [[ -n "$latestSystemLog" ]]; then
+    t '---> 最近的系统日志: %s' "$latestSystemLog"
+    tail -n 300 "$latestSystemLog"
+  else
+    t '---> 未找到系统日志: %s' "$systemLogDir"
+  fi
+
+  if [[ "$QL_CONTAINER" == "true" ]]; then
+    t '---> 容器内 PM2 日志不落盘；早期启动错误请执行 docker logs --tail 300 <容器名>'
+    return
+  fi
+
+  local pm2Home="${PM2_HOME:-$HOME/.pm2}"
+  local panelOut="$pm2Home/logs/qinglong-out.log"
+  local panelError="$pm2Home/logs/qinglong-error.log"
+  [[ -f "$panelOut" ]] && tail -n 300 "$panelOut"
+  [[ -f "$panelError" ]] && tail -n 300 "$panelError"
 }
 
 check_ql() {
-  local api=$(curl -s --noproxy "*" "http://localhost:${ql_port}")
+  local basePath="${ql_base_url%/}"
+  local api
+  local attempt
+  for ((attempt = 1; attempt <= 10; attempt++)); do
+    api=$(curl -s --max-time 2 --noproxy "*" "http://localhost:${ql_port}${basePath}/")
+    [[ $api =~ "<div id=\"root\"></div>" ]] && break
+    sleep 1
+  done
   t '\n=====> 检测面板'
   echo -e "\n\n$api\n"
   if [[ $api =~ "<div id=\"root\"></div>" ]]; then
     t '=====> 面板服务启动正常\n'
+  else
+    t '=====> 面板服务启动异常，请检查上方诊断信息\n'
+    return 1
   fi
 }
 
 check_pm2() {
-  pm2_log
   local currentTimeStamp=$(date +%s)
-  local api=$(
+  local basePath="${ql_base_url%/}"
-    curl -s --noproxy "*" "http://localhost:${ql_port}/api/system?t=$currentTimeStamp" \
+  local api
-      -H 'Accept: */*' \
+  local attempt
-      -H 'User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/90.0.4430.93 Safari/537.36' \
+  for ((attempt = 1; attempt <= 10; attempt++)); do
-      -H "Referer: http://localhost:${ql_port}/crontab" \
+    api=$(
-      -H 'Accept-Language: en-US,en;q=0.9,zh-CN;q=0.8,zh;q=0.7' \
+      curl -s --max-time 2 --noproxy "*" "http://localhost:${ql_port}${basePath}/api/health?t=$currentTimeStamp" \
-      --compressed
+        -H 'Accept: */*' \
-  )
+        -H 'User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/90.0.4430.93 Safari/537.36' \
+        -H "Referer: http://localhost:${ql_port}/crontab" \
+        -H 'Accept-Language: en-US,en;q=0.9,zh-CN;q=0.8,zh;q=0.7' \
+        --compressed
+    )
+    [[ $api == *'"code":200'* && $api == *'"status":"ok"'* ]] && break
+    sleep 1
+  done
   t '\n=====> 检测后台'
   echo -e "\n\n$api\n"
-  if [[ $api =~ "{\"code\"" ]]; then
+  if [[ $api == *'"code":200'* && $api == *'"status":"ok"'* ]]; then
     t '=====> 后台服务启动正常\n'
+  else
+    pm2_log
+    t '=====> 后台服务启动异常，请检查上方诊断信息\n'
+    return 1
   fi
 }
 
@@ -63,10 +101,12 @@ main() {
 
   reset_env
   copy_dep
-  check_ql
-  check_pm2
   reload_pm2
+  local checkStatus=0
+  check_ql || checkStatus=1
+  check_pm2 || checkStatus=1
   t '\n=====> 检测结束\n'
+  return $checkStatus
 }
 
 main

shell/lang/en.sh

@@ -82,10 +82,16 @@ declare -gA LANG_MESSAGES=(
   ['---> 复制一份 %s 为 %s\n']='---> Copying %s to %s\n'
   ['---> 通知文件复制完成\n']='---> Notification files copied\n'
   ['---> pm2日志']='---> pm2 log'
+  ['---> 服务诊断信息']='---> Service diagnostics'
+  ['---> 最近的系统日志: %s']='---> Recent system log: %s'
+  ['---> 未找到系统日志: %s']='---> No system log found: %s'
+  ['---> 容器内 PM2 日志不落盘；早期启动错误请执行 docker logs --tail 300 <容器名>']='---> PM2 logs are not persisted in containers; run docker logs --tail 300 <container> for early startup errors'
   ['\n=====> 检测面板']='\n=====> Checking panel'
   ['=====> 面板服务启动正常\n']='=====> Panel service running normally\n'
+  ['=====> 面板服务启动异常，请检查上方诊断信息\n']='=====> Panel failed to start; check the diagnostics above\n'
   ['\n=====> 检测后台']='\n=====> Checking backend'
   ['=====> 后台服务启动正常\n']='=====> Backend service running normally\n'
+  ['=====> 后台服务启动异常，请检查上方诊断信息\n']='=====> Backend failed to start; check the diagnostics above\n'
   ['=====> 开始检测']='=====> Starting check'
   ['\n=====> 检测结束\n']='\n=====> Check complete\n'
   # rmlog.sh

shell/lang/zh.sh

@@ -82,10 +82,16 @@ declare -gA LANG_MESSAGES=(
   ['---> 复制一份 %s 为 %s\n']='---> 复制一份 %s 为 %s\n'
   ['---> 通知文件复制完成\n']='---> 通知文件复制完成\n'
   ['---> pm2日志']='---> pm2日志'
+  ['---> 服务诊断信息']='---> 服务诊断信息'
+  ['---> 最近的系统日志: %s']='---> 最近的系统日志: %s'
+  ['---> 未找到系统日志: %s']='---> 未找到系统日志: %s'
+  ['---> 容器内 PM2 日志不落盘；早期启动错误请执行 docker logs --tail 300 <容器名>']='---> 容器内 PM2 日志不落盘；早期启动错误请执行 docker logs --tail 300 <容器名>'
   ['\n=====> 检测面板']='\n=====> 检测面板'
   ['=====> 面板服务启动正常\n']='=====> 面板服务启动正常\n'
+  ['=====> 面板服务启动异常，请检查上方诊断信息\n']='=====> 面板服务启动异常，请检查上方诊断信息\n'
   ['\n=====> 检测后台']='\n=====> 检测后台'
   ['=====> 后台服务启动正常\n']='=====> 后台服务启动正常\n'
+  ['=====> 后台服务启动异常，请检查上方诊断信息\n']='=====> 后台服务启动异常，请检查上方诊断信息\n'
   ['=====> 开始检测']='=====> 开始检测'
   ['\n=====> 检测结束\n']='\n=====> 检测结束\n'
   # rmlog.sh

test/back/ecosystem.test.cjs

@@ -13,10 +13,10 @@ function loadConfig(containerValue) {
   return require(configPath).apps[0];
 }
 
-test('container logging relies on pm2-runtime stdout without persistent copies', () => {
+test('container logging discards stdout and forwards startup errors without persistent copies', () => {
   const app = loadConfig('true');
   assert.equal(app.out_file, '/dev/null');
-  assert.equal(app.error_file, '/dev/null');
+  assert.equal(app.error_file, '/proc/1/fd/2');
   assert.equal(app.time, false);
 });
 

test/back/schedule-health.test.cjs

@@ -0,0 +1,86 @@
+const assert = require('node:assert/strict');
+const fs = require('node:fs/promises');
+const http = require('node:http');
+const os = require('node:os');
+const path = require('node:path');
+const test = require('node:test');
+
+const config = require('../../back/config').default;
+const { check } = require('../../back/schedule/health');
+
+function runCheck(service = 'cron') {
+  return new Promise((resolve) => {
+    check({ request: { service } }, (error, response) => {
+      resolve({ error, response });
+    });
+  });
+}
+
+test('schedule health check uses the HTTP health endpoint and system logs', async (t) => {
+  const originalBaseUrl = config.baseUrl;
+  const originalPort = config.port;
+  const originalSystemLogPath = config.systemLogPath;
+  const originalContainer = process.env.QL_CONTAINER;
+
+  t.after(() => {
+    config.baseUrl = originalBaseUrl;
+    config.port = originalPort;
+    config.systemLogPath = originalSystemLogPath;
+    if (originalContainer === undefined) {
+      delete process.env.QL_CONTAINER;
+    } else {
+      process.env.QL_CONTAINER = originalContainer;
+    }
+  });
+
+  await t.test(
+    'returns serving for a healthy prefixed HTTP service',
+    async () => {
+      let requestedUrl = '';
+      const server = http.createServer((request, response) => {
+        requestedUrl = request.url;
+        response.setHeader('Content-Type', 'application/json');
+        response.end(JSON.stringify({ code: 200, data: { status: 'ok' } }));
+      });
+      await new Promise((resolve) => server.listen(0, '127.0.0.1', resolve));
+      t.after(() => new Promise((resolve) => server.close(resolve)));
+
+      config.baseUrl = '/ql';
+      config.port = server.address().port;
+
+      const result = await runCheck();
+      assert.equal(result.error, null);
+      assert.deepEqual(result.response, { status: 1 });
+      assert.equal(requestedUrl, '/ql/api/health');
+    },
+  );
+
+  await t.test(
+    'reports recent system logs when HTTP startup fails',
+    async () => {
+      const logDir = await fs.mkdtemp(path.join(os.tmpdir(), 'ql-health-'));
+      t.after(() => fs.rm(logDir, { recursive: true, force: true }));
+      const lines = Array.from(
+        { length: 305 },
+        (_, index) => `line-${index + 1}`,
+      );
+      await fs.writeFile(path.join(logDir, '2026-08-16.log'), 'older-log');
+      await fs.writeFile(path.join(logDir, '2026-08-17.log'), lines.join('\n'));
+      config.systemLogPath = logDir;
+      config.port = 1;
+      process.env.QL_CONTAINER = 'true';
+
+      const result = await runCheck();
+      assert.equal(result.response, undefined);
+      assert.match(result.error.message, /ECONNREFUSED|connect/);
+      assert.match(
+        result.error.message,
+        /http:\/\/localhost:1\/ql\/api\/health/,
+      );
+      assert.match(result.error.message, /docker logs <container>/);
+      assert.match(result.error.message, /line-305/);
+      assert.doesNotMatch(result.error.message, /line-1\n/);
+      assert.doesNotMatch(result.error.message, /qinglong-error\.log/);
+    },
+  );
+});