fix(account): address review on OpenAI quota auto-pause

- gate previous_response_id sticky path with quota auto-pause check at both
  the snapshot and DB-recheck stages (previously bypassed, #1)
- skip pausing when the usage window already reset to avoid a stale stuck-pause;
  carry codex_*_reset_at / reset_after_seconds / codex_usage_updated_at through
  the scheduler snapshot whitelist (#2)
- remove the incomplete limit mode; percentage threshold only (#3)
- add global default 5h/7d threshold inputs to the Ops settings dialog with
  validation and en/zh i18n (#4)
- downgrade account_auto_paused_by_quota log from Info to Debug; it fires
  per-candidate on the scheduling hot path (#5)

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

frontend/src/api/admin/ops.ts

@@ -778,9 +778,15 @@ export interface OpsAlertRuntimeSettings {
   thresholds: OpsMetricThresholds // 指标阈值配置
 }
 
+export interface OpsOpenAIAccountQuotaAutoPauseSettings {
+  default_threshold_5h: number // 0~1，0 表示不启用全局默认 5h 阈值
+  default_threshold_7d: number // 0~1，0 表示不启用全局默认 7d 阈值
+}
+
 export interface OpsAdvancedSettings {
   data_retention: OpsDataRetentionSettings
   aggregation: OpsAggregationSettings
+  openai_account_quota_auto_pause: OpsOpenAIAccountQuotaAutoPauseSettings
   ignore_count_tokens_errors: boolean
   ignore_context_canceled: boolean
   ignore_no_available_accounts: boolean

frontend/src/i18n/locales/en.ts

@@ -5193,6 +5193,11 @@ export default {
         aggregation: 'Pre-aggregation Tasks',
         enableAggregation: 'Enable Pre-aggregation',
         aggregationHint: 'Pre-aggregation improves query performance for long time windows',
+        openaiQuotaAutoPause: 'OpenAI Account Quota Auto-pause',
+        openaiQuotaAutoPauseHint: 'When an OpenAI account reaches its 5h / 7d usage threshold, the scheduler skips it automatically and resumes once the window rolls over. Per-account thresholds take precedence over this global default.',
+        openaiQuotaAutoPauseDefault5h: 'Default 5h usage threshold (%)',
+        openaiQuotaAutoPauseDefault7d: 'Default 7d usage threshold (%)',
+        openaiQuotaAutoPauseThresholdHint: 'Value 0-100; leave blank or 0 to disable the global default threshold.',
         errorFiltering: 'Error Filtering',
         ignoreCountTokensErrors: 'Ignore count_tokens errors',
         ignoreCountTokensErrorsHint: 'When enabled, errors from count_tokens requests will not be written to the error log.',
@@ -5223,7 +5228,8 @@ export default {
           slaMinPercentRange: 'SLA minimum percentage must be between 0 and 100',
           ttftP99MaxRange: 'TTFT P99 maximum must be a number ≥ 0',
           requestErrorRateMaxRange: 'Request error rate maximum must be between 0 and 100',
-          upstreamErrorRateMaxRange: 'Upstream error rate maximum must be between 0 and 100'
+          upstreamErrorRateMaxRange: 'Upstream error rate maximum must be between 0 and 100',
+          openaiQuotaAutoPauseRange: 'OpenAI quota auto-pause threshold must be between 0 and 100'
         }
       },
       concurrency: {

frontend/src/i18n/locales/zh.ts

@@ -5352,6 +5352,11 @@ export default {
         aggregation: '预聚合任务',
         enableAggregation: '启用预聚合任务',
         aggregationHint: '预聚合可提升长时间窗口查询性能',
+        openaiQuotaAutoPause: 'OpenAI 账号配额自动暂停',
+        openaiQuotaAutoPauseHint: '当 OpenAI 账号 5h / 7d 用量达到阈值时，调度会自动跳过该账号；窗口滚动后自动恢复。账号级阈值优先于此全局默认值。',
+        openaiQuotaAutoPauseDefault5h: '默认 5h 用量阈值 (%)',
+        openaiQuotaAutoPauseDefault7d: '默认 7d 用量阈值 (%)',
+        openaiQuotaAutoPauseThresholdHint: '取值 0-100，留空或 0 表示不启用全局默认阈值。',
         errorFiltering: '错误过滤',
         ignoreCountTokensErrors: '忽略 count_tokens 错误',
         ignoreCountTokensErrorsHint: '启用后，count_tokens 请求的错误将不会写入错误日志。',
@@ -5383,7 +5388,8 @@ export default {
           slaMinPercentRange: 'SLA最低百分比必须在0-100之间',
           ttftP99MaxRange: 'TTFT P99最大值必须大于等于0',
           requestErrorRateMaxRange: '请求错误率最大值必须在0-100之间',
-          upstreamErrorRateMaxRange: '上游错误率最大值必须在0-100之间'
+          upstreamErrorRateMaxRange: '上游错误率最大值必须在0-100之间',
+          openaiQuotaAutoPauseRange: 'OpenAI 配额自动暂停阈值必须在 0-100 之间'
         }
       },
       concurrency: {

frontend/src/views/admin/ops/components/OpsSettingsDialog.vue

@@ -50,6 +50,10 @@ async function loadAllSettings() {
     runtimeSettings.value = runtime
     emailConfig.value = email
     advancedSettings.value = advanced
+    // 兼容旧 payload：后端未返回该字段时补默认值，保证表单可绑定
+    if (advancedSettings.value && !advancedSettings.value.openai_account_quota_auto_pause) {
+      advancedSettings.value.openai_account_quota_auto_pause = { default_threshold_5h: 0, default_threshold_7d: 0 }
+    }
     // 如果后端返回了阈值，使用后端的值；否则保持默认值
     if (thresholds && Object.keys(thresholds).length > 0) {
         metricThresholds.value = {
@@ -119,6 +123,28 @@ function removeRecipient(target: 'alert' | 'report', email: string) {
   if (idx >= 0) list.splice(idx, 1)
 }
 
+// OpenAI 账号配额自动暂停：后端按 0~1 分数存储，UI 按百分比(0~100)展示
+const quotaAutoPause5hPercent = computed<number | null>({
+  get() {
+    const v = advancedSettings.value?.openai_account_quota_auto_pause?.default_threshold_5h
+    return v && v > 0 ? Math.round(v * 1000) / 10 : null
+  },
+  set(val) {
+    if (!advancedSettings.value?.openai_account_quota_auto_pause) return
+    advancedSettings.value.openai_account_quota_auto_pause.default_threshold_5h = val != null && val > 0 ? val / 100 : 0
+  }
+})
+const quotaAutoPause7dPercent = computed<number | null>({
+  get() {
+    const v = advancedSettings.value?.openai_account_quota_auto_pause?.default_threshold_7d
+    return v && v > 0 ? Math.round(v * 1000) / 10 : null
+  },
+  set(val) {
+    if (!advancedSettings.value?.openai_account_quota_auto_pause) return
+    advancedSettings.value.openai_account_quota_auto_pause.default_threshold_7d = val != null && val > 0 ? val / 100 : 0
+  }
+})
+
 // 验证
 const validation = computed(() => {
   const errors: string[] = []
@@ -145,6 +171,11 @@ const validation = computed(() => {
     if (hourly_metrics_retention_days < 0 || hourly_metrics_retention_days > 365) {
       errors.push(t('admin.ops.settings.validation.retentionDaysRange'))
     }
+
+    const { default_threshold_5h, default_threshold_7d } = advancedSettings.value.openai_account_quota_auto_pause
+    if (default_threshold_5h < 0 || default_threshold_5h > 1 || default_threshold_7d < 0 || default_threshold_7d > 1) {
+      errors.push(t('admin.ops.settings.validation.openaiQuotaAutoPauseRange'))
+    }
   }
 
   // 验证指标阈值
@@ -473,6 +504,40 @@ async function saveAllSettings() {
             </div>
           </div>
 
+          <!-- OpenAI 账号配额自动暂停（全局默认阈值） -->
+          <div class="space-y-3">
+            <h5 class="text-xs font-semibold text-gray-700 dark:text-gray-300">{{ t('admin.ops.settings.openaiQuotaAutoPause') }}</h5>
+            <p class="text-xs text-gray-500">{{ t('admin.ops.settings.openaiQuotaAutoPauseHint') }}</p>
+
+            <div class="grid grid-cols-1 gap-4 md:grid-cols-2">
+              <div>
+                <label class="input-label">{{ t('admin.ops.settings.openaiQuotaAutoPauseDefault5h') }}</label>
+                <input
+                  v-model.number="quotaAutoPause5hPercent"
+                  type="number"
+                  min="0"
+                  max="100"
+                  step="0.1"
+                  class="input"
+                  data-testid="ops-quota-auto-pause-5h"
+                />
+              </div>
+              <div>
+                <label class="input-label">{{ t('admin.ops.settings.openaiQuotaAutoPauseDefault7d') }}</label>
+                <input
+                  v-model.number="quotaAutoPause7dPercent"
+                  type="number"
+                  min="0"
+                  max="100"
+                  step="0.1"
+                  class="input"
+                  data-testid="ops-quota-auto-pause-7d"
+                />
+              </div>
+            </div>
+            <p class="text-xs text-gray-500">{{ t('admin.ops.settings.openaiQuotaAutoPauseThresholdHint') }}</p>
+          </div>
+
           <!-- Error Filtering -->
           <div class="space-y-3">
             <h5 class="text-xs font-semibold text-gray-700 dark:text-gray-300">{{ t('admin.ops.settings.errorFiltering') }}</h5>