Merge pull request #251 from IanShaw027/fix/ops-bugs

feat(ops): 运维看板功能增强 - 实时流量监控与指标阈值配置

backend/internal/handler/admin/ops_realtime_handler.go

@@ -118,3 +118,96 @@ func (h *OpsHandler) GetAccountAvailability(c *gin.Context) {
 	}
 	response.Success(c, payload)
 }
+
+func parseOpsRealtimeWindow(v string) (time.Duration, string, bool) {
+	switch strings.ToLower(strings.TrimSpace(v)) {
+	case "", "1min", "1m":
+		return 1 * time.Minute, "1min", true
+	case "5min", "5m":
+		return 5 * time.Minute, "5min", true
+	case "30min", "30m":
+		return 30 * time.Minute, "30min", true
+	case "1h", "60m", "60min":
+		return 1 * time.Hour, "1h", true
+	default:
+		return 0, "", false
+	}
+}
+
+// GetRealtimeTrafficSummary returns QPS/TPS current/peak/avg for the selected window.
+// GET /api/v1/admin/ops/realtime-traffic
+//
+// Query params:
+// - window: 1min|5min|30min|1h (default: 1min)
+// - platform: optional
+// - group_id: optional
+func (h *OpsHandler) GetRealtimeTrafficSummary(c *gin.Context) {
+	if h.opsService == nil {
+		response.Error(c, http.StatusServiceUnavailable, "Ops service not available")
+		return
+	}
+	if err := h.opsService.RequireMonitoringEnabled(c.Request.Context()); err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+
+	windowDur, windowLabel, ok := parseOpsRealtimeWindow(c.Query("window"))
+	if !ok {
+		response.BadRequest(c, "Invalid window")
+		return
+	}
+
+	platform := strings.TrimSpace(c.Query("platform"))
+	var groupID *int64
+	if v := strings.TrimSpace(c.Query("group_id")); v != "" {
+		id, err := strconv.ParseInt(v, 10, 64)
+		if err != nil || id <= 0 {
+			response.BadRequest(c, "Invalid group_id")
+			return
+		}
+		groupID = &id
+	}
+
+	endTime := time.Now().UTC()
+	startTime := endTime.Add(-windowDur)
+
+	if !h.opsService.IsRealtimeMonitoringEnabled(c.Request.Context()) {
+		disabledSummary := &service.OpsRealtimeTrafficSummary{
+			Window:    windowLabel,
+			StartTime: startTime,
+			EndTime:   endTime,
+			Platform:  platform,
+			GroupID:   groupID,
+			QPS:       service.OpsRateSummary{},
+			TPS:       service.OpsRateSummary{},
+		}
+		response.Success(c, gin.H{
+			"enabled":   false,
+			"summary":   disabledSummary,
+			"timestamp": endTime,
+		})
+		return
+	}
+
+	filter := &service.OpsDashboardFilter{
+		StartTime: startTime,
+		EndTime:   endTime,
+		Platform:  platform,
+		GroupID:   groupID,
+		QueryMode: service.OpsQueryModeRaw,
+	}
+
+	summary, err := h.opsService.GetRealtimeTrafficSummary(c.Request.Context(), filter)
+	if err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+	if summary != nil {
+		summary.Window = windowLabel
+	}
+	response.Success(c, gin.H{
+		"enabled":   true,
+		"summary":   summary,
+		"timestamp": endTime,
+	})
+}

backend/internal/handler/admin/ops_settings_handler.go

@@ -146,3 +146,49 @@ func (h *OpsHandler) UpdateAdvancedSettings(c *gin.Context) {
 	}
 	response.Success(c, updated)
 }
+
+// GetMetricThresholds returns Ops metric thresholds (DB-backed).
+// GET /api/v1/admin/ops/settings/metric-thresholds
+func (h *OpsHandler) GetMetricThresholds(c *gin.Context) {
+	if h.opsService == nil {
+		response.Error(c, http.StatusServiceUnavailable, "Ops service not available")
+		return
+	}
+	if err := h.opsService.RequireMonitoringEnabled(c.Request.Context()); err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+
+	cfg, err := h.opsService.GetMetricThresholds(c.Request.Context())
+	if err != nil {
+		response.Error(c, http.StatusInternalServerError, "Failed to get metric thresholds")
+		return
+	}
+	response.Success(c, cfg)
+}
+
+// UpdateMetricThresholds updates Ops metric thresholds (DB-backed).
+// PUT /api/v1/admin/ops/settings/metric-thresholds
+func (h *OpsHandler) UpdateMetricThresholds(c *gin.Context) {
+	if h.opsService == nil {
+		response.Error(c, http.StatusServiceUnavailable, "Ops service not available")
+		return
+	}
+	if err := h.opsService.RequireMonitoringEnabled(c.Request.Context()); err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+
+	var req service.OpsMetricThresholds
+	if err := c.ShouldBindJSON(&req); err != nil {
+		response.BadRequest(c, "Invalid request body")
+		return
+	}
+
+	updated, err := h.opsService.UpdateMetricThresholds(c.Request.Context(), &req)
+	if err != nil {
+		response.Error(c, http.StatusBadRequest, err.Error())
+		return
+	}
+	response.Success(c, updated)
+}

backend/internal/repository/ops_repo_realtime_traffic.go

@@ -0,0 +1,129 @@
+package repository
+
+import (
+	"context"
+	"fmt"
+	"strings"
+	"time"
+
+	"github.com/Wei-Shaw/sub2api/internal/service"
+)
+
+func (r *opsRepository) GetRealtimeTrafficSummary(ctx context.Context, filter *service.OpsDashboardFilter) (*service.OpsRealtimeTrafficSummary, error) {
+	if r == nil || r.db == nil {
+		return nil, fmt.Errorf("nil ops repository")
+	}
+	if filter == nil {
+		return nil, fmt.Errorf("nil filter")
+	}
+	if filter.StartTime.IsZero() || filter.EndTime.IsZero() {
+		return nil, fmt.Errorf("start_time/end_time required")
+	}
+
+	start := filter.StartTime.UTC()
+	end := filter.EndTime.UTC()
+	if start.After(end) {
+		return nil, fmt.Errorf("start_time must be <= end_time")
+	}
+
+	window := end.Sub(start)
+	if window <= 0 {
+		return nil, fmt.Errorf("invalid time window")
+	}
+	if window > time.Hour {
+		return nil, fmt.Errorf("window too large")
+	}
+
+	usageJoin, usageWhere, usageArgs, next := buildUsageWhere(filter, start, end, 1)
+	errorWhere, errorArgs, _ := buildErrorWhere(filter, start, end, next)
+
+	q := `
+WITH usage_buckets AS (
+  SELECT
+    date_trunc('minute', ul.created_at) AS bucket,
+    COALESCE(COUNT(*), 0) AS success_count,
+    COALESCE(SUM(input_tokens + output_tokens + cache_creation_tokens + cache_read_tokens), 0) AS token_sum
+  FROM usage_logs ul
+  ` + usageJoin + `
+  ` + usageWhere + `
+  GROUP BY 1
+),
+error_buckets AS (
+  SELECT
+    date_trunc('minute', created_at) AS bucket,
+    COALESCE(COUNT(*), 0) AS error_count
+  FROM ops_error_logs
+  ` + errorWhere + `
+    AND COALESCE(status_code, 0) >= 400
+  GROUP BY 1
+),
+combined AS (
+  SELECT
+    COALESCE(u.bucket, e.bucket) AS bucket,
+    COALESCE(u.success_count, 0) AS success_count,
+    COALESCE(u.token_sum, 0) AS token_sum,
+    COALESCE(e.error_count, 0) AS error_count,
+    COALESCE(u.success_count, 0) + COALESCE(e.error_count, 0) AS request_total
+  FROM usage_buckets u
+  FULL OUTER JOIN error_buckets e ON u.bucket = e.bucket
+)
+SELECT
+  COALESCE(SUM(success_count), 0) AS success_total,
+  COALESCE(SUM(error_count), 0) AS error_total,
+  COALESCE(SUM(token_sum), 0) AS token_total,
+  COALESCE(MAX(request_total), 0) AS peak_requests_per_min,
+  COALESCE(MAX(token_sum), 0) AS peak_tokens_per_min
+FROM combined`
+
+	args := append(usageArgs, errorArgs...)
+	var successCount int64
+	var errorTotal int64
+	var tokenConsumed int64
+	var peakRequestsPerMin int64
+	var peakTokensPerMin int64
+	if err := r.db.QueryRowContext(ctx, q, args...).Scan(
+		&successCount,
+		&errorTotal,
+		&tokenConsumed,
+		&peakRequestsPerMin,
+		&peakTokensPerMin,
+	); err != nil {
+		return nil, err
+	}
+
+	windowSeconds := window.Seconds()
+	if windowSeconds <= 0 {
+		windowSeconds = 1
+	}
+
+	requestCountTotal := successCount + errorTotal
+	qpsAvg := roundTo1DP(float64(requestCountTotal) / windowSeconds)
+	tpsAvg := roundTo1DP(float64(tokenConsumed) / windowSeconds)
+
+	// Keep "current" consistent with the dashboard overview semantics: last 1 minute.
+	// This remains "within the selected window" since end=start+window.
+	qpsCurrent, tpsCurrent, err := r.queryCurrentRates(ctx, filter, end)
+	if err != nil {
+		return nil, err
+	}
+
+	qpsPeak := roundTo1DP(float64(peakRequestsPerMin) / 60.0)
+	tpsPeak := roundTo1DP(float64(peakTokensPerMin) / 60.0)
+
+	return &service.OpsRealtimeTrafficSummary{
+		StartTime: start,
+		EndTime:   end,
+		Platform:  strings.TrimSpace(filter.Platform),
+		GroupID:   filter.GroupID,
+		QPS: service.OpsRateSummary{
+			Current: qpsCurrent,
+			Peak:    qpsPeak,
+			Avg:     qpsAvg,
+		},
+		TPS: service.OpsRateSummary{
+			Current: tpsCurrent,
+			Peak:    tpsPeak,
+			Avg:     tpsAvg,
+		},
+	}, nil
+}

backend/internal/server/routes/admin.go

@@ -73,6 +73,7 @@ func registerOpsRoutes(admin *gin.RouterGroup, h *handler.Handlers) {
 		// Realtime ops signals
 		ops.GET("/concurrency", h.Admin.Ops.GetConcurrencyStats)
 		ops.GET("/account-availability", h.Admin.Ops.GetAccountAvailability)
+		ops.GET("/realtime-traffic", h.Admin.Ops.GetRealtimeTrafficSummary)
 
 		// Alerts (rules + events)
 		ops.GET("/alert-rules", h.Admin.Ops.ListAlertRules)
@@ -96,6 +97,13 @@ func registerOpsRoutes(admin *gin.RouterGroup, h *handler.Handlers) {
 		ops.GET("/advanced-settings", h.Admin.Ops.GetAdvancedSettings)
 		ops.PUT("/advanced-settings", h.Admin.Ops.UpdateAdvancedSettings)
 
+		// Settings group (DB-backed)
+		settings := ops.Group("/settings")
+		{
+			settings.GET("/metric-thresholds", h.Admin.Ops.GetMetricThresholds)
+			settings.PUT("/metric-thresholds", h.Admin.Ops.UpdateMetricThresholds)
+		}
+
 		// WebSocket realtime (QPS/TPS)
 		ws := ops.Group("/ws")
 		{

backend/internal/service/ops_port.go

@@ -17,6 +17,8 @@ type OpsRepository interface {
 
 	// Lightweight window stats (for realtime WS / quick sampling).
 	GetWindowStats(ctx context.Context, filter *OpsDashboardFilter) (*OpsWindowStats, error)
+	// Lightweight realtime traffic summary (for the Ops dashboard header card).
+	GetRealtimeTrafficSummary(ctx context.Context, filter *OpsDashboardFilter) (*OpsRealtimeTrafficSummary, error)
 
 	GetDashboardOverview(ctx context.Context, filter *OpsDashboardFilter) (*OpsDashboardOverview, error)
 	GetThroughputTrend(ctx context.Context, filter *OpsDashboardFilter, bucketSeconds int) (*OpsThroughputTrendResponse, error)

backend/internal/service/ops_realtime_traffic.go

@@ -0,0 +1,36 @@
+package service
+
+import (
+	"context"
+	"time"
+
+	infraerrors "github.com/Wei-Shaw/sub2api/internal/pkg/errors"
+)
+
+// GetRealtimeTrafficSummary returns QPS/TPS current/peak/avg for the provided window.
+// This is used by the Ops dashboard "Realtime Traffic" card and is intentionally lightweight.
+func (s *OpsService) GetRealtimeTrafficSummary(ctx context.Context, filter *OpsDashboardFilter) (*OpsRealtimeTrafficSummary, error) {
+	if err := s.RequireMonitoringEnabled(ctx); err != nil {
+		return nil, err
+	}
+	if s.opsRepo == nil {
+		return nil, infraerrors.ServiceUnavailable("OPS_REPO_UNAVAILABLE", "Ops repository not available")
+	}
+	if filter == nil {
+		return nil, infraerrors.BadRequest("OPS_FILTER_REQUIRED", "filter is required")
+	}
+	if filter.StartTime.IsZero() || filter.EndTime.IsZero() {
+		return nil, infraerrors.BadRequest("OPS_TIME_RANGE_REQUIRED", "start_time/end_time are required")
+	}
+	if filter.StartTime.After(filter.EndTime) {
+		return nil, infraerrors.BadRequest("OPS_TIME_RANGE_INVALID", "start_time must be <= end_time")
+	}
+	if filter.EndTime.Sub(filter.StartTime) > time.Hour {
+		return nil, infraerrors.BadRequest("OPS_TIME_RANGE_TOO_LARGE", "invalid time range: max window is 1 hour")
+	}
+
+	// Realtime traffic summary always uses raw logs (minute granularity peaks).
+	filter.QueryMode = OpsQueryModeRaw
+
+	return s.opsRepo.GetRealtimeTrafficSummary(ctx, filter)
+}

backend/internal/service/ops_realtime_traffic_models.go

@@ -0,0 +1,19 @@
+package service
+
+import "time"
+
+// OpsRealtimeTrafficSummary is a lightweight summary used by the Ops dashboard "Realtime Traffic" card.
+// It reports QPS/TPS current/peak/avg for the requested time window.
+type OpsRealtimeTrafficSummary struct {
+	// Window is a normalized label (e.g. "1min", "5min", "30min", "1h").
+	Window string `json:"window"`
+
+	StartTime time.Time `json:"start_time"`
+	EndTime   time.Time `json:"end_time"`
+
+	Platform string `json:"platform"`
+	GroupID  *int64 `json:"group_id"`
+
+	QPS OpsRateSummary `json:"qps"`
+	TPS OpsRateSummary `json:"tps"`
+}

backend/internal/service/ops_settings.go

@@ -463,3 +463,93 @@ func (s *OpsService) UpdateOpsAdvancedSettings(ctx context.Context, cfg *OpsAdva
 	_ = json.Unmarshal(raw, updated)
 	return updated, nil
 }
+
+// =========================
+// Metric thresholds
+// =========================
+
+const SettingKeyOpsMetricThresholds = "ops_metric_thresholds"
+
+func defaultOpsMetricThresholds() *OpsMetricThresholds {
+	slaMin := 99.5
+	latencyMax := 2000.0
+	ttftMax := 500.0
+	reqErrMax := 5.0
+	upstreamErrMax := 5.0
+	return &OpsMetricThresholds{
+		SLAPercentMin:               &slaMin,
+		LatencyP99MsMax:             &latencyMax,
+		TTFTp99MsMax:                &ttftMax,
+		RequestErrorRatePercentMax:  &reqErrMax,
+		UpstreamErrorRatePercentMax: &upstreamErrMax,
+	}
+}
+
+func (s *OpsService) GetMetricThresholds(ctx context.Context) (*OpsMetricThresholds, error) {
+	defaultCfg := defaultOpsMetricThresholds()
+	if s == nil || s.settingRepo == nil {
+		return defaultCfg, nil
+	}
+	if ctx == nil {
+		ctx = context.Background()
+	}
+
+	raw, err := s.settingRepo.GetValue(ctx, SettingKeyOpsMetricThresholds)
+	if err != nil {
+		if errors.Is(err, ErrSettingNotFound) {
+			if b, mErr := json.Marshal(defaultCfg); mErr == nil {
+				_ = s.settingRepo.Set(ctx, SettingKeyOpsMetricThresholds, string(b))
+			}
+			return defaultCfg, nil
+		}
+		return nil, err
+	}
+
+	cfg := &OpsMetricThresholds{}
+	if err := json.Unmarshal([]byte(raw), cfg); err != nil {
+		return defaultCfg, nil
+	}
+
+	return cfg, nil
+}
+
+func (s *OpsService) UpdateMetricThresholds(ctx context.Context, cfg *OpsMetricThresholds) (*OpsMetricThresholds, error) {
+	if s == nil || s.settingRepo == nil {
+		return nil, errors.New("setting repository not initialized")
+	}
+	if ctx == nil {
+		ctx = context.Background()
+	}
+	if cfg == nil {
+		return nil, errors.New("invalid config")
+	}
+
+	// Validate thresholds
+	if cfg.SLAPercentMin != nil && (*cfg.SLAPercentMin < 0 || *cfg.SLAPercentMin > 100) {
+		return nil, errors.New("sla_percent_min must be between 0 and 100")
+	}
+	if cfg.LatencyP99MsMax != nil && *cfg.LatencyP99MsMax < 0 {
+		return nil, errors.New("latency_p99_ms_max must be >= 0")
+	}
+	if cfg.TTFTp99MsMax != nil && *cfg.TTFTp99MsMax < 0 {
+		return nil, errors.New("ttft_p99_ms_max must be >= 0")
+	}
+	if cfg.RequestErrorRatePercentMax != nil && (*cfg.RequestErrorRatePercentMax < 0 || *cfg.RequestErrorRatePercentMax > 100) {
+		return nil, errors.New("request_error_rate_percent_max must be between 0 and 100")
+	}
+	if cfg.UpstreamErrorRatePercentMax != nil && (*cfg.UpstreamErrorRatePercentMax < 0 || *cfg.UpstreamErrorRatePercentMax > 100) {
+		return nil, errors.New("upstream_error_rate_percent_max must be between 0 and 100")
+	}
+
+	raw, err := json.Marshal(cfg)
+	if err != nil {
+		return nil, err
+	}
+	if err := s.settingRepo.Set(ctx, SettingKeyOpsMetricThresholds, string(raw)); err != nil {
+		return nil, err
+	}
+
+	updated := &OpsMetricThresholds{}
+	_ = json.Unmarshal(raw, updated)
+	return updated, nil
+}

backend/internal/service/ops_settings_models.go

@@ -61,11 +61,20 @@ type OpsAlertSilencingSettings struct {
 	Entries []OpsAlertSilenceEntry `json:"entries,omitempty"`
 }
 
+type OpsMetricThresholds struct {
+	SLAPercentMin               *float64 `json:"sla_percent_min,omitempty"`                 // SLA低于此值变红
+	LatencyP99MsMax             *float64 `json:"latency_p99_ms_max,omitempty"`              // 延迟P99高于此值变红
+	TTFTp99MsMax                *float64 `json:"ttft_p99_ms_max,omitempty"`                 // TTFT P99高于此值变红
+	RequestErrorRatePercentMax  *float64 `json:"request_error_rate_percent_max,omitempty"`  // 请求错误率高于此值变红
+	UpstreamErrorRatePercentMax *float64 `json:"upstream_error_rate_percent_max,omitempty"` // 上游错误率高于此值变红
+}
+
 type OpsAlertRuntimeSettings struct {
 	EvaluationIntervalSeconds int `json:"evaluation_interval_seconds"`
 
 	DistributedLock OpsDistributedLockSettings `json:"distributed_lock"`
 	Silencing       OpsAlertSilencingSettings  `json:"silencing"`
+	Thresholds      OpsMetricThresholds        `json:"thresholds"` // 指标阈值配置
 }
 
 // OpsAdvancedSettings stores advanced ops configuration (data retention, aggregation).

frontend/.eslintignore

@@ -0,0 +1,14 @@
+# 忽略编译后的文件
+vite.config.js
+vite.config.d.ts
+
+# 忽略依赖
+node_modules/
+
+# 忽略构建输出
+dist/
+../backend/internal/web/dist/
+
+# 忽略缓存
+.cache/
+.vite/

frontend/src/api/admin/ops.ts

@@ -362,6 +362,45 @@ export async function getAccountAvailabilityStats(platform?: string, groupId?: n
   return data
 }
 
+export interface OpsRateSummary {
+  current: number
+  peak: number
+  avg: number
+}
+
+export interface OpsRealtimeTrafficSummary {
+  window: string
+  start_time: string
+  end_time: string
+  platform: string
+  group_id?: number | null
+  qps: OpsRateSummary
+  tps: OpsRateSummary
+}
+
+export interface OpsRealtimeTrafficSummaryResponse {
+  enabled: boolean
+  summary: OpsRealtimeTrafficSummary | null
+  timestamp?: string
+}
+
+export async function getRealtimeTrafficSummary(
+  window: string,
+  platform?: string,
+  groupId?: number | null
+): Promise<OpsRealtimeTrafficSummaryResponse> {
+  const params: Record<string, any> = { window }
+  if (platform) {
+    params.platform = platform
+  }
+  if (typeof groupId === 'number' && groupId > 0) {
+    params.group_id = groupId
+  }
+
+  const { data } = await apiClient.get<OpsRealtimeTrafficSummaryResponse>('/admin/ops/realtime-traffic', { params })
+  return data
+}
+
 /**
  * Subscribe to realtime QPS updates via WebSocket.
  *
@@ -661,6 +700,14 @@ export interface EmailNotificationConfig {
   }
 }
 
+export interface OpsMetricThresholds {
+  sla_percent_min?: number | null                // SLA低于此值变红
+  latency_p99_ms_max?: number | null             // 延迟P99高于此值变红
+  ttft_p99_ms_max?: number | null                // TTFT P99高于此值变红
+  request_error_rate_percent_max?: number | null // 请求错误率高于此值变红
+  upstream_error_rate_percent_max?: number | null // 上游错误率高于此值变红
+}
+
 export interface OpsDistributedLockSettings {
   enabled: boolean
   key: string
@@ -681,6 +728,7 @@ export interface OpsAlertRuntimeSettings {
       reason: string
     }>
   }
+  thresholds: OpsMetricThresholds // 指标阈值配置
 }
 
 export interface OpsAdvancedSettings {
@@ -929,6 +977,17 @@ export async function updateAdvancedSettings(config: OpsAdvancedSettings): Promi
   return data
 }
 
+// ==================== Metric Thresholds ====================
+
+async function getMetricThresholds(): Promise<OpsMetricThresholds> {
+  const { data } = await apiClient.get<OpsMetricThresholds>('/admin/ops/settings/metric-thresholds')
+  return data
+}
+
+async function updateMetricThresholds(thresholds: OpsMetricThresholds): Promise<void> {
+  await apiClient.put('/admin/ops/settings/metric-thresholds', thresholds)
+}
+
 export const opsAPI = {
   getDashboardOverview,
   getThroughputTrend,
@@ -937,6 +996,7 @@ export const opsAPI = {
   getErrorDistribution,
   getConcurrencyStats,
   getAccountAvailabilityStats,
+  getRealtimeTrafficSummary,
   subscribeQPS,
   listErrorLogs,
   getErrorLogDetail,
@@ -952,7 +1012,9 @@ export const opsAPI = {
   getAlertRuntimeSettings,
   updateAlertRuntimeSettings,
   getAdvancedSettings,
-  updateAdvancedSettings
+  updateAdvancedSettings,
+  getMetricThresholds,
+  updateMetricThresholds
 }
 
 export default opsAPI

frontend/src/components/icons/Icon.vue

@@ -124,7 +124,8 @@ const icons = {
   chatBubble: 'M8 10h.01M12 10h.01M16 10h.01M9 16H5a2 2 0 01-2-2V6a2 2 0 012-2h14a2 2 0 012 2v8a2 2 0 01-2 2h-5l-5 5v-5z',
   calculator: 'M9 7h6m0 10v-3m-3 3h.01M9 17h.01M9 14h.01M12 14h.01M15 11h.01M12 11h.01M9 11h.01M7 21h10a2 2 0 002-2V5a2 2 0 00-2-2H7a2 2 0 00-2 2v14a2 2 0 002 2z',
   fire: 'M17.657 18.657A8 8 0 016.343 7.343S7 9 9 10c0-2 .5-5 2.986-7C14 5 16.09 5.777 17.656 7.343A7.975 7.975 0 0120 13a7.975 7.975 0 01-2.343 5.657z',
-  badge: 'M9 12.75L11.25 15 15 9.75M21 12c0 1.268-.63 2.39-1.593 3.068a3.745 3.745 0 01-1.043 3.296 3.745 3.745 0 01-3.296 1.043A3.745 3.745 0 0112 21c-1.268 0-2.39-.63-3.068-1.593a3.746 3.746 0 01-3.296-1.043 3.745 3.745 0 01-1.043-3.296A3.745 3.745 0 013 12c0-1.268.63-2.39 1.593-3.068a3.745 3.745 0 011.043-3.296 3.746 3.746 0 013.296-1.043A3.746 3.746 0 0112 3c1.268 0 2.39.63 3.068 1.593a3.746 3.746 0 013.296 1.043 3.746 3.746 0 011.043 3.296A3.745 3.745 0 0121 12z'
+  badge: 'M9 12.75L11.25 15 15 9.75M21 12c0 1.268-.63 2.39-1.593 3.068a3.745 3.745 0 01-1.043 3.296 3.745 3.745 0 01-3.296 1.043A3.745 3.745 0 0112 21c-1.268 0-2.39-.63-3.068-1.593a3.746 3.746 0 01-3.296-1.043 3.745 3.745 0 01-1.043-3.296A3.745 3.745 0 013 12c0-1.268.63-2.39 1.593-3.068a3.745 3.745 0 011.043-3.296 3.746 3.746 0 013.296-1.043A3.746 3.746 0 0112 3c1.268 0 2.39.63 3.068 1.593a3.746 3.746 0 013.296 1.043 3.746 3.746 0 011.043 3.296A3.745 3.745 0 0121 12z',
+  brain: 'M9.75 3.104v5.714a2.25 2.25 0 01-.659 1.591L5 14.5M9.75 3.104c-.251.023-.501.05-.75.082m.75-.082a24.301 24.301 0 014.5 0m0 0v5.714c0 .597.237 1.17.659 1.591L19.8 15.3M14.25 3.104c.251.023.501.05.75.082M19.8 15.3l-1.57.393A9.065 9.065 0 0112 15a9.065 9.065 0 00-6.23.693L5 14.5m0 0l-2.69 2.689c-1.232 1.232-.65 3.318 1.067 3.611A48.309 48.309 0 0012 21c2.773 0 5.491-.235 8.135-.687 1.718-.293 2.3-2.379 1.067-3.61L19.8 15.3M12 8.25a1.5 1.5 0 100-3 1.5 1.5 0 000 3zm0 0v3m-3-1.5a1.5 1.5 0 100-3 1.5 1.5 0 000 3zm0 0h6m-3 4.5a1.5 1.5 0 100-3 1.5 1.5 0 000 3z'
 } as const
 
 const iconPath = computed(() => icons[props.name])

frontend/src/i18n/locales/en.ts

@@ -156,6 +156,7 @@ export default {
         unknownError: 'Unknown error occurred',
         saving: 'Saving...', 
         selectedCount: '({count} selected)',    refresh: 'Refresh',
+    settings: 'Settings',
     notAvailable: 'N/A',
     now: 'Now',
     unknown: 'Unknown',
@@ -1906,6 +1907,7 @@ export default {
       max: 'max:',
       qps: 'QPS',
       requests: 'Requests',
+      requestsTitle: 'Requests',
       upstream: 'Upstream',
       client: 'Client',
       system: 'System',
@@ -2118,7 +2120,10 @@ export default {
         empty: 'No alert rules',
         loadFailed: 'Failed to load alert rules',
         saveFailed: 'Failed to save alert rule',
+        saveSuccess: 'Alert rule saved successfully',
         deleteFailed: 'Failed to delete alert rule',
+        deleteSuccess: 'Alert rule deleted successfully',
+        manage: 'Manage Alert Rules',
         create: 'Create Rule',
         createTitle: 'Create Alert Rule',
         editTitle: 'Edit Alert Rule',
@@ -2301,6 +2306,54 @@ export default {
           accountHealthThresholdRange: 'Account health threshold must be between 0 and 100'
         }
       },
+      settings: {
+        title: 'Ops Monitoring Settings',
+        loadFailed: 'Failed to load settings',
+        saveSuccess: 'Ops monitoring settings saved successfully',
+        saveFailed: 'Failed to save settings',
+        dataCollection: 'Data Collection',
+        evaluationInterval: 'Evaluation Interval (seconds)',
+        evaluationIntervalHint: 'Frequency of detection tasks, recommended to keep default',
+        alertConfig: 'Alert Configuration',
+        enableAlert: 'Enable Alerts',
+        alertRecipients: 'Alert Recipient Emails',
+        emailPlaceholder: 'Enter email address',
+        recipientsHint: 'If empty, the system will use the first admin email as default recipient',
+        minSeverity: 'Minimum Severity',
+        reportConfig: 'Report Configuration',
+        enableReport: 'Enable Reports',
+        reportRecipients: 'Report Recipient Emails',
+        dailySummary: 'Daily Summary',
+        weeklySummary: 'Weekly Summary',
+        metricThresholds: 'Metric Thresholds',
+        metricThresholdsHint: 'Configure alert thresholds for metrics, values exceeding thresholds will be displayed in red',
+        slaMinPercent: 'SLA Minimum Percentage',
+        slaMinPercentHint: 'SLA below this value will be displayed in red (default: 99.5%)',
+        latencyP99MaxMs: 'Latency P99 Maximum (ms)',
+        latencyP99MaxMsHint: 'Latency P99 above this value will be displayed in red (default: 2000ms)',
+        ttftP99MaxMs: 'TTFT P99 Maximum (ms)',
+        ttftP99MaxMsHint: 'TTFT P99 above this value will be displayed in red (default: 500ms)',
+        requestErrorRateMaxPercent: 'Request Error Rate Maximum (%)',
+        requestErrorRateMaxPercentHint: 'Request error rate above this value will be displayed in red (default: 5%)',
+        upstreamErrorRateMaxPercent: 'Upstream Error Rate Maximum (%)',
+        upstreamErrorRateMaxPercentHint: 'Upstream error rate above this value will be displayed in red (default: 5%)',
+        advancedSettings: 'Advanced Settings',
+        dataRetention: 'Data Retention Policy',
+        enableCleanup: 'Enable Data Cleanup',
+        cleanupSchedule: 'Cleanup Schedule (Cron)',
+        cleanupScheduleHint: 'Example: 0 2 * * * means 2 AM daily',
+        errorLogRetentionDays: 'Error Log Retention Days',
+        minuteMetricsRetentionDays: 'Minute Metrics Retention Days',
+        hourlyMetricsRetentionDays: 'Hourly Metrics Retention Days',
+        retentionDaysHint: 'Recommended 7-90 days, longer periods will consume more storage',
+        aggregation: 'Pre-aggregation Tasks',
+        enableAggregation: 'Enable Pre-aggregation',
+        aggregationHint: 'Pre-aggregation improves query performance for long time windows',
+        validation: {
+          title: 'Please fix the following issues',
+          retentionDaysRange: 'Retention days must be between 1-365 days'
+        }
+      },
       concurrency: {
         title: 'Concurrency / Queue',
         byPlatform: 'By Platform',
@@ -2334,12 +2387,13 @@ export default {
         accountError: 'Error'
       },
       tooltips: {
+        totalRequests: 'Total number of requests (including both successful and failed requests) in the selected time window.',
         throughputTrend: 'Requests/QPS + Tokens/TPS in the selected window.',
         latencyHistogram: 'Latency distribution (duration_ms) for successful requests.',
         errorTrend: 'Error counts over time (SLA scope excludes business limits; upstream excludes 429/529).',
         errorDistribution: 'Error distribution by status code.',
         goroutines:
-          'Number of Go runtime goroutines (lightweight threads). There is no absolute “safe” number—use your historical baseline. Heuristic: <2k is common; 2k–8k watch; >8k plus rising queue/latency often suggests blocking/leaks.',
+          'Number of Go runtime goroutines (lightweight threads). There is no absolute "safe" number—use your historical baseline. Heuristic: <2k is common; 2k–8k watch; >8k plus rising queue/latency often suggests blocking/leaks.',
         cpu: 'CPU usage percentage, showing system processor load.',
         memory: 'Memory usage, including used and total available memory.',
         db: 'Database connection pool status, including active, idle, and waiting connections.',
@@ -2349,6 +2403,7 @@ export default {
         tokens: 'Total number of tokens processed in the current time window.',
         sla: 'Service Level Agreement success rate, excluding business limits (e.g., insufficient balance, quota exceeded).',
         errors: 'Error statistics, including total errors, error rate, and upstream error rate.',
+        upstreamErrors: 'Upstream error statistics, excluding rate limit errors (429/529).',
         latency: 'Request latency statistics, including p50, p90, p95, p99 percentiles.',
         ttft: 'Time To First Token, measuring the speed of first byte return in streaming responses.',
         health: 'System health score (0-100), considering SLA, error rate, and resource usage.'

frontend/src/i18n/locales/zh.ts

@@ -2022,7 +2022,7 @@ export default {
       ready: '就绪',
       requestsTotal: '请求（总计）',
       slaScope: 'SLA 范围：',
-      tokens: 'Token',
+      tokens: 'Token数',
       tps: 'TPS',
       current: '当前',
       peak: '峰值',
@@ -2051,7 +2051,8 @@ export default {
       avg: 'avg',
       max: 'max',
       qps: 'QPS',
-      requests: '请求',
+      requests: '请求数',
+      requestsTitle: '请求',
       upstream: '上游',
       client: '客户端',
       system: '系统',
@@ -2469,6 +2470,18 @@ export default {
         reportRecipients: '评估报告接收邮箱',
         dailySummary: '每日摘要',
         weeklySummary: '每周摘要',
+        metricThresholds: '指标阈值配置',
+        metricThresholdsHint: '配置各项指标的告警阈值，超出阈值时将以红色显示',
+        slaMinPercent: 'SLA最低百分比',
+        slaMinPercentHint: 'SLA低于此值时显示为红色（默认：99.5%）',
+        latencyP99MaxMs: '延迟P99最大值（毫秒）',
+        latencyP99MaxMsHint: '延迟P99高于此值时显示为红色（默认：2000ms）',
+        ttftP99MaxMs: 'TTFT P99最大值（毫秒）',
+        ttftP99MaxMsHint: 'TTFT P99高于此值时显示为红色（默认：500ms）',
+        requestErrorRateMaxPercent: '请求错误率最大值（%）',
+        requestErrorRateMaxPercentHint: '请求错误率高于此值时显示为红色（默认：5%）',
+        upstreamErrorRateMaxPercent: '上游错误率最大值（%）',
+        upstreamErrorRateMaxPercentHint: '上游错误率高于此值时显示为红色（默认：5%）',
         advancedSettings: '高级设置',
         dataRetention: '数据保留策略',
         enableCleanup: '启用数据清理',

frontend/src/views/admin/ops/OpsDashboard.vue

@@ -13,17 +13,13 @@
       <OpsDashboardHeader
         v-else-if="opsEnabled"
         :overview="overview"
-        :ws-status="wsStatus"
-        :ws-reconnect-in-ms="wsReconnectInMs"
-        :ws-has-data="wsHasData"
-        :real-time-qps="realTimeQPS"
-        :real-time-tps="realTimeTPS"
         :platform="platform"
         :group-id="groupId"
         :time-range="timeRange"
         :query-mode="queryMode"
         :loading="loading"
         :last-updated="lastUpdated"
+        :thresholds="metricThresholds"
         @update:time-range="onTimeRangeChange"
         @update:platform="onPlatformChange"
         @update:group="onGroupChange"
@@ -75,7 +71,7 @@
       <OpsAlertEventsCard v-if="opsEnabled && !(loading && !hasLoadedOnce)" />
 
       <!-- Settings Dialog -->
-      <OpsSettingsDialog :show="showSettingsDialog" @close="showSettingsDialog = false" @saved="fetchData" />
+      <OpsSettingsDialog :show="showSettingsDialog" @close="showSettingsDialog = false" @saved="onSettingsSaved" />
 
       <!-- Alert Rules Dialog -->
       <BaseDialog :show="showAlertRulesCard" :title="t('admin.ops.alertRules.title')" width="extra-wide" @close="showAlertRulesCard = false">
@@ -115,13 +111,12 @@ import AppLayout from '@/components/layout/AppLayout.vue'
 import BaseDialog from '@/components/common/BaseDialog.vue'
 import {
   opsAPI,
-  OPS_WS_CLOSE_CODES,
-  type OpsWSStatus,
   type OpsDashboardOverview,
   type OpsErrorDistributionResponse,
   type OpsErrorTrendResponse,
   type OpsLatencyHistogramResponse,
-  type OpsThroughputTrendResponse
+  type OpsThroughputTrendResponse,
+  type OpsMetricThresholds
 } from '@/api/admin/ops'
 import { useAdminSettingsStore, useAppStore } from '@/stores'
 import OpsDashboardHeader from './components/OpsDashboardHeader.vue'
@@ -172,14 +167,6 @@ const QUERY_KEYS = {
 const isApplyingRouteQuery = ref(false)
 const isSyncingRouteQuery = ref(false)
 
-// WebSocket for realtime QPS/TPS
-const realTimeQPS = ref(0)
-const realTimeTPS = ref(0)
-const wsStatus = ref<OpsWSStatus>('closed')
-const wsReconnectInMs = ref<number | null>(null)
-const wsHasData = ref(false)
-let unsubscribeQPS: (() => void) | null = null
-
 let dashboardFetchController: AbortController | null = null
 let dashboardFetchSeq = 0
 
@@ -199,50 +186,6 @@ function abortDashboardFetch() {
   }
 }
 
-function stopQPSSubscription(options?: { resetMetrics?: boolean }) {
-  wsStatus.value = 'closed'
-  wsReconnectInMs.value = null
-  if (unsubscribeQPS) unsubscribeQPS()
-  unsubscribeQPS = null
-
-  if (options?.resetMetrics) {
-    realTimeQPS.value = 0
-    realTimeTPS.value = 0
-    wsHasData.value = false
-  }
-}
-
-function startQPSSubscription() {
-  stopQPSSubscription()
-  unsubscribeQPS = opsAPI.subscribeQPS(
-    (payload) => {
-      if (payload && typeof payload === 'object' && payload.type === 'qps_update' && payload.data) {
-        realTimeQPS.value = payload.data.qps || 0
-        realTimeTPS.value = payload.data.tps || 0
-        wsHasData.value = true
-      }
-    },
-    {
-      onStatusChange: (status) => {
-        wsStatus.value = status
-        if (status === 'connected') wsReconnectInMs.value = null
-      },
-      onReconnectScheduled: ({ delayMs }) => {
-        wsReconnectInMs.value = delayMs
-      },
-      onFatalClose: (event) => {
-        // Server-side feature flag says realtime is disabled; keep UI consistent and avoid reconnect loops.
-        if (event && event.code === OPS_WS_CLOSE_CODES.REALTIME_DISABLED) {
-          adminSettingsStore.setOpsRealtimeMonitoringEnabledLocal(false)
-          stopQPSSubscription({ resetMetrics: true })
-        }
-      },
-      // QPS updates may be sparse in idle periods; keep the timeout conservative.
-      staleTimeoutMs: 180_000
-    }
-  )
-}
-
 const readQueryString = (key: string): string => {
   const value = route.query[key]
   if (typeof value === 'string') return value
@@ -314,6 +257,7 @@ const syncQueryToRoute = useDebounceFn(async () => {
 }, 250)
 
 const overview = ref<OpsDashboardOverview | null>(null)
+const metricThresholds = ref<OpsMetricThresholds | null>(null)
 
 const throughputTrend = ref<OpsThroughputTrendResponse | null>(null)
 const loadingTrend = ref(false)
@@ -376,6 +320,11 @@ function onTimeRangeChange(v: string | number | boolean | null) {
   timeRange.value = v as TimeRange
 }
 
+function onSettingsSaved() {
+  loadThresholds()
+  fetchData()
+}
+
 function onPlatformChange(v: string | number | boolean | null) {
   platform.value = typeof v === 'string' ? v : ''
 }
@@ -615,31 +564,25 @@ onMounted(async () => {
     return
   }
 
-  if (adminSettingsStore.opsRealtimeMonitoringEnabled) {
-    startQPSSubscription()
-  } else {
-    stopQPSSubscription({ resetMetrics: true })
-  }
+  // Load thresholds configuration
+  loadThresholds()
 
   if (opsEnabled.value) {
     await fetchData()
   }
 })
 
+async function loadThresholds() {
+  try {
+    const settings = await opsAPI.getAlertRuntimeSettings()
+    metricThresholds.value = settings.thresholds || null
+  } catch (err) {
+    console.warn('[OpsDashboard] Failed to load thresholds', err)
+    metricThresholds.value = null
+  }
+}
+
 onUnmounted(() => {
-  stopQPSSubscription()
   abortDashboardFetch()
 })
-
-watch(
-  () => adminSettingsStore.opsRealtimeMonitoringEnabled,
-  (enabled) => {
-    if (!opsEnabled.value) return
-    if (enabled) {
-      startQPSSubscription()
-    } else {
-      stopQPSSubscription({ resetMetrics: true })
-    }
-  }
-)
 </script>

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

@@ -1,29 +1,28 @@
 <script setup lang="ts">
-import { computed, onMounted, ref, watch } from 'vue'
+import { computed, onMounted, onUnmounted, ref, watch } from 'vue'
+import { useIntervalFn } from '@vueuse/core'
 import { useI18n } from 'vue-i18n'
 import Select from '@/components/common/Select.vue'
 import HelpTooltip from '@/components/common/HelpTooltip.vue'
 import BaseDialog from '@/components/common/BaseDialog.vue'
+import Icon from '@/components/icons/Icon.vue'
 import { adminAPI } from '@/api'
-import type { OpsDashboardOverview, OpsWSStatus } from '@/api/admin/ops'
+import { opsAPI, type OpsDashboardOverview, type OpsMetricThresholds, type OpsRealtimeTrafficSummary } from '@/api/admin/ops'
 import type { OpsRequestDetailsPreset } from './OpsRequestDetailsModal.vue'
+import { useAdminSettingsStore } from '@/stores'
 import { formatNumber } from '@/utils/format'
 
 type RealtimeWindow = '1min' | '5min' | '30min' | '1h'
 
 interface Props {
   overview?: OpsDashboardOverview | null
-  wsStatus: OpsWSStatus
-  wsReconnectInMs?: number | null
-  wsHasData?: boolean
-  realTimeQps: number
-  realTimeTps: number
   platform: string
   groupId: number | null
   timeRange: string
   queryMode: string
   loading: boolean
   lastUpdated: Date | null
+  thresholds?: OpsMetricThresholds | null // 阈值配置
 }
 
 interface Emits {
@@ -42,12 +41,43 @@ const props = defineProps<Props>()
 const emit = defineEmits<Emits>()
 
 const { t } = useI18n()
+const adminSettingsStore = useAdminSettingsStore()
 
 const realtimeWindow = ref<RealtimeWindow>('1min')
 
 const overview = computed(() => props.overview ?? null)
 const systemMetrics = computed(() => overview.value?.system_metrics ?? null)
 
+const REALTIME_WINDOW_MINUTES: Record<RealtimeWindow, number> = {
+  '1min': 1,
+  '5min': 5,
+  '30min': 30,
+  '1h': 60
+}
+
+const TOOLBAR_RANGE_MINUTES: Record<string, number> = {
+  '5m': 5,
+  '30m': 30,
+  '1h': 60,
+  '6h': 6 * 60,
+  '24h': 24 * 60
+}
+
+const availableRealtimeWindows = computed(() => {
+  const toolbarMinutes = TOOLBAR_RANGE_MINUTES[props.timeRange] ?? 60
+  return (['1min', '5min', '30min', '1h'] as const).filter((w) => REALTIME_WINDOW_MINUTES[w] <= toolbarMinutes)
+})
+
+watch(
+  () => props.timeRange,
+  () => {
+    // The realtime window must be inside the toolbar window; reset to keep UX predictable.
+    realtimeWindow.value = '1min'
+    // Keep realtime traffic consistent with toolbar changes even when the window is already 1min.
+    loadRealtimeTrafficSummary()
+  }
+)
+
 // --- Filters ---
 
 const groups = ref<Array<{ id: number; name: string; platform: string }>>([])
@@ -143,56 +173,143 @@ function getLatencyColor(ms: number | null | undefined): string {
   return 'text-red-600 dark:text-red-400'
 }
 
+// --- Threshold checking helpers ---
+function isSLABelowThreshold(slaPercent: number | null): boolean {
+  if (slaPercent == null) return false
+  const threshold = props.thresholds?.sla_percent_min
+  if (threshold == null) return false
+  return slaPercent < threshold
+}
+
+function isLatencyAboveThreshold(latencyP99Ms: number | null): boolean {
+  if (latencyP99Ms == null) return false
+  const threshold = props.thresholds?.latency_p99_ms_max
+  if (threshold == null) return false
+  return latencyP99Ms > threshold
+}
+
+function isTTFTAboveThreshold(ttftP99Ms: number | null): boolean {
+  if (ttftP99Ms == null) return false
+  const threshold = props.thresholds?.ttft_p99_ms_max
+  if (threshold == null) return false
+  return ttftP99Ms > threshold
+}
+
+function isRequestErrorRateAboveThreshold(errorRatePercent: number | null): boolean {
+  if (errorRatePercent == null) return false
+  const threshold = props.thresholds?.request_error_rate_percent_max
+  if (threshold == null) return false
+  return errorRatePercent > threshold
+}
+
+function isUpstreamErrorRateAboveThreshold(upstreamErrorRatePercent: number | null): boolean {
+  if (upstreamErrorRatePercent == null) return false
+  const threshold = props.thresholds?.upstream_error_rate_percent_max
+  if (threshold == null) return false
+  return upstreamErrorRatePercent > threshold
+}
+
 // --- Realtime / Overview labels ---
 
 const totalRequestsLabel = computed(() => formatNumber(overview.value?.request_count_total ?? 0))
 const totalTokensLabel = computed(() => formatNumber(overview.value?.token_consumed ?? 0))
 
+const realtimeTrafficSummary = ref<OpsRealtimeTrafficSummary | null>(null)
+const realtimeTrafficLoading = ref(false)
+
+function makeZeroRealtimeTrafficSummary(): OpsRealtimeTrafficSummary {
+  const now = new Date().toISOString()
+  return {
+    window: realtimeWindow.value,
+    start_time: now,
+    end_time: now,
+    platform: props.platform,
+    group_id: props.groupId,
+    qps: { current: 0, peak: 0, avg: 0 },
+    tps: { current: 0, peak: 0, avg: 0 }
+  }
+}
+
+async function loadRealtimeTrafficSummary() {
+  if (realtimeTrafficLoading.value) return
+  if (!adminSettingsStore.opsRealtimeMonitoringEnabled) {
+    realtimeTrafficSummary.value = makeZeroRealtimeTrafficSummary()
+    return
+  }
+  realtimeTrafficLoading.value = true
+  try {
+    const res = await opsAPI.getRealtimeTrafficSummary(realtimeWindow.value, props.platform, props.groupId)
+    if (res && res.enabled === false) {
+      adminSettingsStore.setOpsRealtimeMonitoringEnabledLocal(false)
+    }
+    realtimeTrafficSummary.value = res?.summary ?? null
+  } catch (err) {
+    console.error('[OpsDashboardHeader] Failed to load realtime traffic summary', err)
+    realtimeTrafficSummary.value = null
+  } finally {
+    realtimeTrafficLoading.value = false
+  }
+}
+
+watch(
+  () => [realtimeWindow.value, props.platform, props.groupId] as const,
+  () => {
+    loadRealtimeTrafficSummary()
+  },
+  { immediate: true }
+)
+
+const { pause: pauseRealtimeTrafficRefresh, resume: resumeRealtimeTrafficRefresh } = useIntervalFn(
+  () => {
+    loadRealtimeTrafficSummary()
+  },
+  5000,
+  { immediate: false }
+)
+
+watch(
+  () => adminSettingsStore.opsRealtimeMonitoringEnabled,
+  (enabled) => {
+    if (enabled) {
+      resumeRealtimeTrafficRefresh()
+    } else {
+      pauseRealtimeTrafficRefresh()
+      // Keep UI stable when realtime monitoring is turned off.
+      realtimeTrafficSummary.value = makeZeroRealtimeTrafficSummary()
+    }
+  },
+  { immediate: true }
+)
+
+onUnmounted(() => {
+  pauseRealtimeTrafficRefresh()
+})
+
 const displayRealTimeQps = computed(() => {
-  const ov = overview.value
-  if (!ov) return 0
-  const useRealtime = props.wsStatus === 'connected' && !!props.wsHasData
-  const v = useRealtime ? props.realTimeQps : ov.qps?.current
+  const v = realtimeTrafficSummary.value?.qps?.current
   return typeof v === 'number' && Number.isFinite(v) ? v : 0
 })
 
 const displayRealTimeTps = computed(() => {
-  const ov = overview.value
-  if (!ov) return 0
-  const useRealtime = props.wsStatus === 'connected' && !!props.wsHasData
-  const v = useRealtime ? props.realTimeTps : ov.tps?.current
+  const v = realtimeTrafficSummary.value?.tps?.current
   return typeof v === 'number' && Number.isFinite(v) ? v : 0
 })
 
-// Sparkline history (keep last 60 data points)
-const qpsHistory = ref<number[]>([])
-const tpsHistory = ref<number[]>([])
-const MAX_HISTORY_POINTS = 60
-
-watch([displayRealTimeQps, displayRealTimeTps], ([newQps, newTps]) => {
-  // Add new data points
-  qpsHistory.value.push(newQps)
-  tpsHistory.value.push(newTps)
-
-  // Keep only last N points
-  if (qpsHistory.value.length > MAX_HISTORY_POINTS) {
-    qpsHistory.value.shift()
-  }
-  if (tpsHistory.value.length > MAX_HISTORY_POINTS) {
-    tpsHistory.value.shift()
-  }
+const realtimeQpsPeakLabel = computed(() => {
+  const v = realtimeTrafficSummary.value?.qps?.peak
+  return typeof v === 'number' && Number.isFinite(v) ? v.toFixed(1) : '-'
 })
-
-const qpsPeakLabel = computed(() => {
-  const v = overview.value?.qps?.peak
-  if (typeof v !== 'number') return '-'
-  return v.toFixed(1)
+const realtimeTpsPeakLabel = computed(() => {
+  const v = realtimeTrafficSummary.value?.tps?.peak
+  return typeof v === 'number' && Number.isFinite(v) ? v.toFixed(1) : '-'
 })
-
-const tpsPeakLabel = computed(() => {
-  const v = overview.value?.tps?.peak
-  if (typeof v !== 'number') return '-'
-  return v.toFixed(1)
+const realtimeQpsAvgLabel = computed(() => {
+  const v = realtimeTrafficSummary.value?.qps?.avg
+  return typeof v === 'number' && Number.isFinite(v) ? v.toFixed(1) : '-'
+})
+const realtimeTpsAvgLabel = computed(() => {
+  const v = realtimeTrafficSummary.value?.tps?.avg
+  return typeof v === 'number' && Number.isFinite(v) ? v.toFixed(1) : '-'
 })
 
 const qpsAvgLabel = computed(() => {
@@ -244,7 +361,7 @@ const ttftMaxMs = computed(() => overview.value?.ttft?.max_ms ?? null)
 const isSystemIdle = computed(() => {
   const ov = overview.value
   if (!ov) return true
-  const qps = props.wsStatus === 'connected' && props.wsHasData ? props.realTimeQps : ov.qps?.current
+  const qps = ov.qps?.current
   const errorRate = ov.error_rate ?? 0
   return (qps ?? 0) === 0 && errorRate === 0
 })
@@ -687,6 +804,11 @@ const showJobsDetails = ref(false)
 function openJobsDetails() {
   showJobsDetails.value = true
 }
+
+function handleToolbarRefresh() {
+  loadRealtimeTrafficSummary()
+  emit('refresh')
+}
 </script>
 
 <template>
@@ -764,7 +886,7 @@ function openJobsDetails() {
           class="flex h-8 w-8 items-center justify-center rounded-lg bg-gray-100 text-gray-500 transition-colors hover:bg-gray-200 dark:bg-dark-700 dark:text-gray-400 dark:hover:bg-dark-600"
           :disabled="loading"
           :title="t('common.refresh')"
-          @click="emit('refresh')"
+          @click="handleToolbarRefresh"
         >
           <svg class="h-4 w-4" :class="{ 'animate-spin': loading }" fill="none" viewBox="0 0 24 24" stroke="currentColor">
             <path
@@ -818,8 +940,9 @@ function openJobsDetails() {
               class="pointer-events-none absolute left-1/2 top-full z-50 mt-2 w-72 -translate-x-1/2 opacity-0 transition-opacity duration-200 group-hover:pointer-events-auto group-hover:opacity-100 md:left-full md:top-0 md:ml-2 md:mt-0 md:translate-x-0"
             >
               <div class="rounded-xl bg-white p-4 shadow-xl ring-1 ring-black/5 dark:bg-gray-800 dark:ring-white/10">
-                <h4 class="mb-3 border-b border-gray-100 pb-2 text-sm font-bold text-gray-900 dark:border-gray-700 dark:text-white">
-                  🧠 {{ t('admin.ops.diagnosis.title') }}
+                <h4 class="mb-3 border-b border-gray-100 pb-2 text-sm font-bold text-gray-900 dark:border-gray-700 dark:text-white flex items-center gap-2">
+                  <Icon name="brain" size="sm" class="text-blue-500" />
+                  {{ t('admin.ops.diagnosis.title') }}
                 </h4>
 
                 <div class="space-y-3">
@@ -850,8 +973,9 @@ function openJobsDetails() {
                     <div class="flex-1">
                       <div class="text-xs font-semibold text-gray-900 dark:text-white">{{ item.message }}</div>
                       <div class="mt-0.5 text-[11px] text-gray-500 dark:text-gray-400">{{ item.impact }}</div>
-                      <div v-if="item.action" class="mt-1 text-[11px] text-blue-600 dark:text-blue-400">
-                        💡 {{ item.action }}
+                      <div v-if="item.action" class="mt-1 text-[11px] text-blue-600 dark:text-blue-400 flex items-center gap-1">
+                        <Icon name="lightbulb" size="xs" />
+                        {{ item.action }}
                       </div>
                     </div>
                   </div>
@@ -928,7 +1052,7 @@ function openJobsDetails() {
               <!-- Time Window Selector -->
               <div class="flex flex-wrap gap-1">
                 <button
-                  v-for="window in (['1min', '5min', '30min', '1h'] as RealtimeWindow[])"
+                  v-for="window in availableRealtimeWindows"
                   :key="window"
                   type="button"
                   class="rounded px-1.5 py-0.5 text-[9px] font-bold transition-colors sm:px-2 sm:text-[10px]"
@@ -965,11 +1089,11 @@ function openJobsDetails() {
                   <div class="text-[10px] font-bold uppercase text-gray-400">{{ t('admin.ops.peak') }}</div>
                   <div class="mt-1 space-y-0.5 text-sm font-medium text-gray-600 dark:text-gray-400">
                     <div class="flex items-baseline gap-1.5">
-                      <span class="font-black text-gray-900 dark:text-white">{{ qpsPeakLabel }}</span>
+                      <span class="font-black text-gray-900 dark:text-white">{{ realtimeQpsPeakLabel }}</span>
                       <span class="text-xs">QPS</span>
                     </div>
                     <div class="flex items-baseline gap-1.5">
-                      <span class="font-black text-gray-900 dark:text-white">{{ tpsPeakLabel }}</span>
+                      <span class="font-black text-gray-900 dark:text-white">{{ realtimeTpsPeakLabel }}</span>
                       <span class="text-xs">TPS</span>
                     </div>
                   </div>
@@ -980,11 +1104,11 @@ function openJobsDetails() {
                   <div class="text-[10px] font-bold uppercase text-gray-400">{{ t('admin.ops.average') }}</div>
                   <div class="mt-1 space-y-0.5 text-sm font-medium text-gray-600 dark:text-gray-400">
                     <div class="flex items-baseline gap-1.5">
-                      <span class="font-black text-gray-900 dark:text-white">{{ qpsAvgLabel }}</span>
+                      <span class="font-black text-gray-900 dark:text-white">{{ realtimeQpsAvgLabel }}</span>
                       <span class="text-xs">QPS</span>
                     </div>
                     <div class="flex items-baseline gap-1.5">
-                      <span class="font-black text-gray-900 dark:text-white">{{ tpsAvgLabel }}</span>
+                      <span class="font-black text-gray-900 dark:text-white">{{ realtimeTpsAvgLabel }}</span>
                       <span class="text-xs">TPS</span>
                     </div>
                   </div>
@@ -1024,7 +1148,7 @@ function openJobsDetails() {
         <div class="rounded-2xl bg-gray-50 p-4 dark:bg-dark-900">
           <div class="flex items-center justify-between">
             <div class="flex items-center gap-1">
-              <span class="text-[10px] font-bold uppercase text-gray-400">{{ t('admin.ops.requests') }}</span>
+              <span class="text-[10px] font-bold uppercase text-gray-400">{{ t('admin.ops.requestsTitle') }}</span>
               <HelpTooltip :content="t('admin.ops.tooltips.totalRequests')" />
             </div>
             <button
@@ -1061,21 +1185,21 @@ function openJobsDetails() {
             <div class="flex items-center gap-2">
               <span class="text-[10px] font-bold uppercase text-gray-400">SLA</span>
               <HelpTooltip :content="t('admin.ops.tooltips.sla')" />
-              <span class="h-1.5 w-1.5 rounded-full" :class="(slaPercent ?? 0) >= 99.5 ? 'bg-green-500' : 'bg-yellow-500'"></span>
+              <span class="h-1.5 w-1.5 rounded-full" :class="isSLABelowThreshold(slaPercent) ? 'bg-red-500' : (slaPercent ?? 0) >= 99.5 ? 'bg-green-500' : 'bg-yellow-500'"></span>
             </div>
             <button
               class="text-[10px] font-bold text-blue-500 hover:underline"
               type="button"
-              @click="openDetails({ title: t('admin.ops.requestDetails.title') })"
+              @click="openDetails({ title: t('admin.ops.requestDetails.title'), kind: 'error' })"
             >
               {{ t('admin.ops.requestDetails.details') }}
             </button>
           </div>
-          <div class="mt-2 text-3xl font-black text-gray-900 dark:text-white">
+          <div class="mt-2 text-3xl font-black" :class="isSLABelowThreshold(slaPercent) ? 'text-red-600 dark:text-red-400' : 'text-gray-900 dark:text-white'">
             {{ slaPercent == null ? '-' : `${slaPercent.toFixed(3)}%` }}
           </div>
           <div class="mt-3 h-2 w-full overflow-hidden rounded-full bg-gray-200 dark:bg-dark-700">
-            <div class="h-full bg-green-500 transition-all" :style="{ width: `${Math.max((slaPercent ?? 0) - 90, 0) * 10}%` }"></div>
+            <div class="h-full transition-all" :class="isSLABelowThreshold(slaPercent) ? 'bg-red-500' : 'bg-green-500'" :style="{ width: `${Math.max((slaPercent ?? 0) - 90, 0) * 10}%` }"></div>
           </div>
           <div class="mt-3 text-xs">
             <div class="flex justify-between">
@@ -1101,7 +1225,7 @@ function openJobsDetails() {
             </button>
           </div>
           <div class="mt-2 flex items-baseline gap-2">
-            <div class="text-3xl font-black" :class="getLatencyColor(durationP99Ms)">
+            <div class="text-3xl font-black" :class="isLatencyAboveThreshold(durationP99Ms) ? 'text-red-600 dark:text-red-400' : getLatencyColor(durationP99Ms)">
               {{ durationP99Ms ?? '-' }}
             </div>
             <span class="text-xs font-bold text-gray-400">ms (P99)</span>
@@ -1145,13 +1269,13 @@ function openJobsDetails() {
             <button
               class="text-[10px] font-bold text-blue-500 hover:underline"
               type="button"
-              @click="openDetails({ title: 'TTFT' })"
+              @click="openDetails({ title: 'TTFT', sort: 'duration_desc' })"
             >
               {{ t('admin.ops.requestDetails.details') }}
             </button>
           </div>
           <div class="mt-2 flex items-baseline gap-2">
-            <div class="text-3xl font-black" :class="getLatencyColor(ttftP99Ms)">
+            <div class="text-3xl font-black" :class="isTTFTAboveThreshold(ttftP99Ms) ? 'text-red-600 dark:text-red-400' : getLatencyColor(ttftP99Ms)">
               {{ ttftP99Ms ?? '-' }}
             </div>
             <span class="text-xs font-bold text-gray-400">ms (P99)</span>
@@ -1196,7 +1320,7 @@ function openJobsDetails() {
               {{ t('admin.ops.requestDetails.details') }}
             </button>
           </div>
-          <div class="mt-2 text-3xl font-black" :class="(errorRatePercent ?? 0) > 5 ? 'text-red-500' : 'text-gray-900 dark:text-white'">
+          <div class="mt-2 text-3xl font-black" :class="isRequestErrorRateAboveThreshold(errorRatePercent) ? 'text-red-600 dark:text-red-400' : (errorRatePercent ?? 0) > 5 ? 'text-red-500' : 'text-gray-900 dark:text-white'">
             {{ errorRatePercent == null ? '-' : `${errorRatePercent.toFixed(2)}%` }}
           </div>
           <div class="mt-3 space-y-1 text-xs">
@@ -1222,7 +1346,7 @@ function openJobsDetails() {
               {{ t('admin.ops.requestDetails.details') }}
             </button>
           </div>
-          <div class="mt-2 text-3xl font-black" :class="(upstreamErrorRatePercent ?? 0) > 5 ? 'text-red-500' : 'text-gray-900 dark:text-white'">
+          <div class="mt-2 text-3xl font-black" :class="isUpstreamErrorRateAboveThreshold(upstreamErrorRatePercent) ? 'text-red-600 dark:text-red-400' : (upstreamErrorRatePercent ?? 0) > 5 ? 'text-red-500' : 'text-gray-900 dark:text-white'">
             {{ upstreamErrorRatePercent == null ? '-' : `${upstreamErrorRatePercent.toFixed(2)}%` }}
           </div>
           <div class="mt-3 space-y-1 text-xs">

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

@@ -174,69 +174,75 @@ watch(
 
 <template>
   <BaseDialog :show="show" :title="modalTitle" width="full" @close="close">
-    <!-- Filters -->
-    <div class="border-b border-gray-200 pb-4 mb-4 dark:border-dark-700">
-      <div class="grid grid-cols-1 gap-4 lg:grid-cols-12">
-        <div class="lg:col-span-5">
-          <div class="relative group">
-            <div class="pointer-events-none absolute inset-y-0 left-0 flex items-center pl-3.5">
-              <svg
-                class="h-4 w-4 text-gray-400 transition-colors group-focus-within:text-blue-500"
-                fill="none"
-                viewBox="0 0 24 24"
-                stroke="currentColor"
-              >
-                <path stroke-linecap="round" stroke-linejoin="round" stroke-width="2.5" d="M21 21l-6-6m2-5a7 7 0 11-14 0 7 7 0 0114 0z" />
-              </svg>
+    <div class="flex h-full min-h-0 flex-col">
+      <!-- Filters -->
+      <div class="mb-4 flex-shrink-0 border-b border-gray-200 pb-4 dark:border-dark-700">
+        <div class="grid grid-cols-1 gap-4 lg:grid-cols-12">
+          <div class="lg:col-span-5">
+            <div class="relative group">
+              <div class="pointer-events-none absolute inset-y-0 left-0 flex items-center pl-3.5">
+                <svg
+                  class="h-4 w-4 text-gray-400 transition-colors group-focus-within:text-blue-500"
+                  fill="none"
+                  viewBox="0 0 24 24"
+                  stroke="currentColor"
+                >
+                  <path stroke-linecap="round" stroke-linejoin="round" stroke-width="2.5" d="M21 21l-6-6m2-5a7 7 0 11-14 0 7 7 0 0114 0z" />
+                </svg>
+              </div>
+              <input
+                v-model="q"
+                type="text"
+                class="w-full rounded-2xl border-gray-200 bg-gray-50/50 py-2 pl-10 pr-4 text-sm font-medium text-gray-700 transition-all focus:border-blue-500 focus:bg-white focus:ring-4 focus:ring-blue-500/10 dark:border-dark-700 dark:bg-dark-900 dark:text-gray-300 dark:focus:bg-dark-800"
+                :placeholder="t('admin.ops.errorDetails.searchPlaceholder')"
+              />
             </div>
+          </div>
+
+          <div class="lg:col-span-2">
+            <Select :model-value="statusCode" :options="statusCodeSelectOptions" class="w-full" @update:model-value="statusCode = $event as any" />
+          </div>
+
+          <div class="lg:col-span-2">
+            <Select :model-value="phase" :options="phaseSelectOptions" class="w-full" @update:model-value="phase = String($event ?? '')" />
+          </div>
+
+          <div class="lg:col-span-2">
             <input
-              v-model="q"
+              v-model="accountIdInput"
               type="text"
-              class="w-full rounded-2xl border-gray-200 bg-gray-50/50 py-2 pl-10 pr-4 text-sm font-medium text-gray-700 transition-all focus:border-blue-500 focus:bg-white focus:ring-4 focus:ring-blue-500/10 dark:border-dark-700 dark:bg-dark-900 dark:text-gray-300 dark:focus:bg-dark-800"
-              :placeholder="t('admin.ops.errorDetails.searchPlaceholder')"
+              inputmode="numeric"
+              class="input w-full text-sm"
+              :placeholder="t('admin.ops.errorDetails.accountIdPlaceholder')"
             />
           </div>
-        </div>
 
-        <div class="lg:col-span-2">
-          <Select :model-value="statusCode" :options="statusCodeSelectOptions" class="w-full" @update:model-value="statusCode = $event as any" />
-        </div>
-
-        <div class="lg:col-span-2">
-          <Select :model-value="phase" :options="phaseSelectOptions" class="w-full" @update:model-value="phase = String($event ?? '')" />
-        </div>
-
-        <div class="lg:col-span-2">
-          <input
-            v-model="accountIdInput"
-            type="text"
-            inputmode="numeric"
-            class="input w-full text-sm"
-            :placeholder="t('admin.ops.errorDetails.accountIdPlaceholder')"
-          />
-        </div>
-
-        <div class="lg:col-span-1 flex items-center justify-end">
-          <button type="button" class="btn btn-secondary btn-sm" @click="resetFilters">
-            {{ t('common.reset') }}
-          </button>
+          <div class="lg:col-span-1 flex items-center justify-end">
+            <button type="button" class="btn btn-secondary btn-sm" @click="resetFilters">
+              {{ t('common.reset') }}
+            </button>
+          </div>
         </div>
       </div>
-    </div>
 
-    <!-- Body -->
-    <div class="text-xs text-gray-500 dark:text-gray-400 mb-2">
-      {{ t('admin.ops.errorDetails.total') }} {{ total }}
+      <!-- Body -->
+      <div class="flex min-h-0 flex-1 flex-col">
+        <div class="mb-2 flex-shrink-0 text-xs text-gray-500 dark:text-gray-400">
+          {{ t('admin.ops.errorDetails.total') }} {{ total }}
+        </div>
+
+        <OpsErrorLogTable
+          class="min-h-0 flex-1"
+          :rows="rows"
+          :total="total"
+          :loading="loading"
+          :page="page"
+          :page-size="pageSize"
+          @openErrorDetail="emit('openErrorDetail', $event)"
+          @update:page="page = $event"
+          @update:pageSize="pageSize = $event"
+        />
+      </div>
     </div>
-    <OpsErrorLogTable
-      :rows="rows"
-      :total="total"
-      :loading="loading"
-      :page="page"
-      :page-size="pageSize"
-      @openErrorDetail="emit('openErrorDetail', $event)"
-      @update:page="page = $event"
-      @update:pageSize="pageSize = $event"
-    />
   </BaseDialog>
 </template>

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

@@ -1,176 +1,178 @@
 <template>
-  <div>
-    <div v-if="loading" class="flex items-center justify-center py-10">
+  <div class="flex h-full min-h-0 flex-col">
+    <div v-if="loading" class="flex flex-1 items-center justify-center py-10">
       <div class="h-8 w-8 animate-spin rounded-full border-b-2 border-primary-600"></div>
     </div>
 
-    <div v-else class="overflow-x-auto">
-      <table class="min-w-full divide-y divide-gray-200 dark:divide-dark-700">
-        <thead class="sticky top-0 z-10 bg-gray-50/50 dark:bg-dark-800/50">
-          <tr>
-            <th
-              scope="col"
-              class="whitespace-nowrap px-6 py-4 text-left text-xs font-bold uppercase tracking-wider text-gray-500 dark:text-dark-400"
-            >
-              {{ t('admin.ops.errorLog.timeId') }}
-            </th>
-            <th
-              scope="col"
-              class="whitespace-nowrap px-6 py-4 text-left text-xs font-bold uppercase tracking-wider text-gray-500 dark:text-dark-400"
-            >
-              {{ t('admin.ops.errorLog.context') }}
-            </th>
-            <th
-              scope="col"
-              class="whitespace-nowrap px-6 py-4 text-left text-xs font-bold uppercase tracking-wider text-gray-500 dark:text-dark-400"
-            >
-              {{ t('admin.ops.errorLog.status') }}
-            </th>
-            <th
-              scope="col"
-              class="px-6 py-4 text-left text-xs font-bold uppercase tracking-wider text-gray-500 dark:text-dark-400"
-            >
-              {{ t('admin.ops.errorLog.message') }}
-            </th>
-            <th
-              scope="col"
-              class="whitespace-nowrap px-6 py-4 text-right text-xs font-bold uppercase tracking-wider text-gray-500 dark:text-dark-400"
-            >
-              {{ t('admin.ops.errorLog.latency') }}
-            </th>
-            <th
-              scope="col"
-              class="whitespace-nowrap px-6 py-4 text-right text-xs font-bold uppercase tracking-wider text-gray-500 dark:text-dark-400"
-            >
-              {{ t('admin.ops.errorLog.action') }}
-            </th>
-          </tr>
-        </thead>
-        <tbody class="divide-y divide-gray-100 dark:divide-dark-700">
-          <tr v-if="rows.length === 0" class="bg-white dark:bg-dark-900">
-            <td colspan="6" class="py-16 text-center text-sm text-gray-400 dark:text-dark-500">
-              {{ t('admin.ops.errorLog.noErrors') }}
-            </td>
-          </tr>
+    <div v-else class="flex min-h-0 flex-1 flex-col">
+      <div class="min-h-0 flex-1 overflow-auto">
+        <table class="min-w-full divide-y divide-gray-200 dark:divide-dark-700">
+          <thead class="sticky top-0 z-10 bg-gray-50/50 dark:bg-dark-800/50">
+            <tr>
+              <th
+                scope="col"
+                class="whitespace-nowrap px-6 py-4 text-left text-xs font-bold uppercase tracking-wider text-gray-500 dark:text-dark-400"
+              >
+                {{ t('admin.ops.errorLog.timeId') }}
+              </th>
+              <th
+                scope="col"
+                class="whitespace-nowrap px-6 py-4 text-left text-xs font-bold uppercase tracking-wider text-gray-500 dark:text-dark-400"
+              >
+                {{ t('admin.ops.errorLog.context') }}
+              </th>
+              <th
+                scope="col"
+                class="whitespace-nowrap px-6 py-4 text-left text-xs font-bold uppercase tracking-wider text-gray-500 dark:text-dark-400"
+              >
+                {{ t('admin.ops.errorLog.status') }}
+              </th>
+              <th
+                scope="col"
+                class="px-6 py-4 text-left text-xs font-bold uppercase tracking-wider text-gray-500 dark:text-dark-400"
+              >
+                {{ t('admin.ops.errorLog.message') }}
+              </th>
+              <th
+                scope="col"
+                class="whitespace-nowrap px-6 py-4 text-right text-xs font-bold uppercase tracking-wider text-gray-500 dark:text-dark-400"
+              >
+                {{ t('admin.ops.errorLog.latency') }}
+              </th>
+              <th
+                scope="col"
+                class="whitespace-nowrap px-6 py-4 text-right text-xs font-bold uppercase tracking-wider text-gray-500 dark:text-dark-400"
+              >
+                {{ t('admin.ops.errorLog.action') }}
+              </th>
+            </tr>
+          </thead>
+          <tbody class="divide-y divide-gray-100 dark:divide-dark-700">
+            <tr v-if="rows.length === 0" class="bg-white dark:bg-dark-900">
+              <td colspan="6" class="py-16 text-center text-sm text-gray-400 dark:text-dark-500">
+                {{ t('admin.ops.errorLog.noErrors') }}
+              </td>
+            </tr>
 
-          <tr
-            v-for="log in rows"
-            :key="log.id"
-            class="group cursor-pointer transition-all duration-200 hover:bg-gray-50/80 focus:outline-none focus:ring-2 focus:ring-primary-500 focus:ring-offset-2 dark:hover:bg-dark-800/50 dark:focus:ring-offset-dark-900"
-            tabindex="0"
-            role="button"
-            @click="emit('openErrorDetail', log.id)"
-            @keydown.enter.prevent="emit('openErrorDetail', log.id)"
-            @keydown.space.prevent="emit('openErrorDetail', log.id)"
-          >
-            <!-- Time & ID -->
-            <td class="px-6 py-4">
-              <div class="flex flex-col gap-0.5">
-                <span class="font-mono text-xs font-bold text-gray-900 dark:text-gray-200">
-                  {{ formatDateTime(log.created_at).split(' ')[1] }}
-                </span>
-                <span
-                  class="font-mono text-[10px] text-gray-400 transition-colors group-hover:text-primary-600 dark:group-hover:text-primary-400"
-                  :title="log.request_id || log.client_request_id"
-                >
-                  {{ (log.request_id || log.client_request_id || '').substring(0, 12) }}
-                </span>
-              </div>
-            </td>
+            <tr
+              v-for="log in rows"
+              :key="log.id"
+              class="group cursor-pointer transition-all duration-200 hover:bg-gray-50/80 focus:outline-none focus:ring-2 focus:ring-primary-500 focus:ring-offset-2 dark:hover:bg-dark-800/50 dark:focus:ring-offset-dark-900"
+              tabindex="0"
+              role="button"
+              @click="emit('openErrorDetail', log.id)"
+              @keydown.enter.prevent="emit('openErrorDetail', log.id)"
+              @keydown.space.prevent="emit('openErrorDetail', log.id)"
+            >
+              <!-- Time & ID -->
+              <td class="px-6 py-4">
+                <div class="flex flex-col gap-0.5">
+                  <span class="font-mono text-xs font-bold text-gray-900 dark:text-gray-200">
+                    {{ formatDateTime(log.created_at).split(' ')[1] }}
+                  </span>
+                  <span
+                    class="font-mono text-[10px] text-gray-400 transition-colors group-hover:text-primary-600 dark:group-hover:text-primary-400"
+                    :title="log.request_id || log.client_request_id"
+                  >
+                    {{ (log.request_id || log.client_request_id || '').substring(0, 12) }}
+                  </span>
+                </div>
+              </td>
 
-	            <!-- Context (Platform/Model) -->
-	            <td class="px-6 py-4">
-	              <div class="flex flex-col items-start gap-1.5">
-	                <span
-	                  class="inline-flex items-center rounded-md bg-gray-100 px-2 py-0.5 text-[10px] font-bold uppercase tracking-tight text-gray-600 dark:bg-dark-700 dark:text-gray-300"
-	                >
-	                  {{ log.platform || '-' }}
-	                </span>
-	                <span
-	                  v-if="log.model"
-	                  class="max-w-[160px] truncate font-mono text-[10px] text-gray-500 dark:text-dark-400"
-	                  :title="log.model"
-	                >
-	                  {{ log.model }}
-	                </span>
-	                <div
-	                  v-if="log.group_id || log.account_id"
-	                  class="flex flex-wrap items-center gap-2 font-mono text-[10px] font-semibold text-gray-400 dark:text-dark-500"
-	                >
-	                  <span v-if="log.group_id">{{ t('admin.ops.errorLog.grp') }} {{ log.group_id }}</span>
-	                  <span v-if="log.account_id">{{ t('admin.ops.errorLog.acc') }} {{ log.account_id }}</span>
-	                </div>
-	              </div>
-	            </td>
-
-            <!-- Status & Severity -->
-            <td class="px-6 py-4">
-              <div class="flex flex-wrap items-center gap-2">
-                <span
-                  :class="[
-                    'inline-flex items-center rounded-lg px-2 py-1 text-xs font-black ring-1 ring-inset shadow-sm',
-                    getStatusClass(log.status_code)
-                  ]"
-                >
-                  {{ log.status_code }}
-                </span>
-                <span
-                  v-if="log.severity"
-                  :class="['rounded-md px-2 py-0.5 text-[10px] font-black shadow-sm', getSeverityClass(log.severity)]"
-                >
-                  {{ log.severity }}
-                </span>
-              </div>
-            </td>
-
-            <!-- Message -->
-            <td class="px-6 py-4">
-              <div class="max-w-md lg:max-w-2xl">
-                <p class="truncate text-xs font-semibold text-gray-700 dark:text-gray-300" :title="log.message">
-                  {{ formatSmartMessage(log.message) || '-' }}
-                </p>
-                <div class="mt-1.5 flex flex-wrap gap-x-3 gap-y-1">
-                  <div v-if="log.phase" class="flex items-center gap-1">
-                    <span class="h-1 w-1 rounded-full bg-gray-300"></span>
-                    <span class="text-[9px] font-black uppercase tracking-tighter text-gray-400">{{ log.phase }}</span>
-                  </div>
-                  <div v-if="log.client_ip" class="flex items-center gap-1">
-                    <span class="h-1 w-1 rounded-full bg-gray-300"></span>
-                    <span class="text-[9px] font-mono font-bold text-gray-400">{{ log.client_ip }}</span>
+              <!-- Context (Platform/Model) -->
+              <td class="px-6 py-4">
+                <div class="flex flex-col items-start gap-1.5">
+                  <span
+                    class="inline-flex items-center rounded-md bg-gray-100 px-2 py-0.5 text-[10px] font-bold uppercase tracking-tight text-gray-600 dark:bg-dark-700 dark:text-gray-300"
+                  >
+                    {{ log.platform || '-' }}
+                  </span>
+                  <span
+                    v-if="log.model"
+                    class="max-w-[160px] truncate font-mono text-[10px] text-gray-500 dark:text-dark-400"
+                    :title="log.model"
+                  >
+                    {{ log.model }}
+                  </span>
+                  <div
+                    v-if="log.group_id || log.account_id"
+                    class="flex flex-wrap items-center gap-2 font-mono text-[10px] font-semibold text-gray-400 dark:text-dark-500"
+                  >
+                    <span v-if="log.group_id">{{ t('admin.ops.errorLog.grp') }} {{ log.group_id }}</span>
+                    <span v-if="log.account_id">{{ t('admin.ops.errorLog.acc') }} {{ log.account_id }}</span>
                   </div>
                 </div>
-              </div>
-            </td>
+              </td>
 
-            <!-- Latency -->
-            <td class="px-6 py-4 text-right">
-              <div class="flex flex-col items-end">
-                <span class="font-mono text-xs font-black" :class="getLatencyClass(log.latency_ms ?? null)">
-                  {{ log.latency_ms != null ? Math.round(log.latency_ms) + 'ms' : '--' }}
-                </span>
-              </div>
-            </td>
+              <!-- Status & Severity -->
+              <td class="px-6 py-4">
+                <div class="flex flex-wrap items-center gap-2">
+                  <span
+                    :class="[
+                      'inline-flex items-center rounded-lg px-2 py-1 text-xs font-black ring-1 ring-inset shadow-sm',
+                      getStatusClass(log.status_code)
+                    ]"
+                  >
+                    {{ log.status_code }}
+                  </span>
+                  <span
+                    v-if="log.severity"
+                    :class="['rounded-md px-2 py-0.5 text-[10px] font-black shadow-sm', getSeverityClass(log.severity)]"
+                  >
+                    {{ log.severity }}
+                  </span>
+                </div>
+              </td>
 
-            <!-- Actions -->
-            <td class="px-6 py-4 text-right" @click.stop>
-              <button type="button" class="btn btn-secondary btn-sm" @click="emit('openErrorDetail', log.id)">
-                {{ t('admin.ops.errorLog.details') }}
-              </button>
-            </td>
-          </tr>
-        </tbody>
-      </table>
+              <!-- Message -->
+              <td class="px-6 py-4">
+                <div class="max-w-md lg:max-w-2xl">
+                  <p class="truncate text-xs font-semibold text-gray-700 dark:text-gray-300" :title="log.message">
+                    {{ formatSmartMessage(log.message) || '-' }}
+                  </p>
+                  <div class="mt-1.5 flex flex-wrap gap-x-3 gap-y-1">
+                    <div v-if="log.phase" class="flex items-center gap-1">
+                      <span class="h-1 w-1 rounded-full bg-gray-300"></span>
+                      <span class="text-[9px] font-black uppercase tracking-tighter text-gray-400">{{ log.phase }}</span>
+                    </div>
+                    <div v-if="log.client_ip" class="flex items-center gap-1">
+                      <span class="h-1 w-1 rounded-full bg-gray-300"></span>
+                      <span class="text-[9px] font-mono font-bold text-gray-400">{{ log.client_ip }}</span>
+                    </div>
+                  </div>
+                </div>
+              </td>
+
+              <!-- Latency -->
+              <td class="px-6 py-4 text-right">
+                <div class="flex flex-col items-end">
+                  <span class="font-mono text-xs font-black" :class="getLatencyClass(log.latency_ms ?? null)">
+                    {{ log.latency_ms != null ? Math.round(log.latency_ms) + 'ms' : '--' }}
+                  </span>
+                </div>
+              </td>
+
+              <!-- Actions -->
+              <td class="px-6 py-4 text-right" @click.stop>
+                <button type="button" class="btn btn-secondary btn-sm" @click="emit('openErrorDetail', log.id)">
+                  {{ t('admin.ops.errorLog.details') }}
+                </button>
+              </td>
+            </tr>
+          </tbody>
+        </table>
+      </div>
+
+      <Pagination
+        v-if="total > 0"
+        :total="total"
+        :page="page"
+        :page-size="pageSize"
+        :page-size-options="[10, 20, 50, 100, 200, 500]"
+        @update:page="emit('update:page', $event)"
+        @update:pageSize="emit('update:pageSize', $event)"
+      />
     </div>
-
-    <Pagination
-      v-if="total > 0"
-      :total="total"
-      :page="page"
-      :page-size="pageSize"
-      :page-size-options="[10, 20, 50, 100, 200, 500]"
-      @update:page="emit('update:page', $event)"
-      @update:pageSize="emit('update:pageSize', $event)"
-    />
   </div>
 </template>
 

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

@@ -95,6 +95,7 @@ watch(
   (open) => {
     if (open) {
       page.value = 1
+      pageSize.value = 20
       fetchData()
     }
   }
@@ -150,45 +151,46 @@ const kindBadgeClass = (kind: string) => {
 <template>
   <BaseDialog :show="modelValue" :title="props.preset.title || t('admin.ops.requestDetails.title')" width="full" @close="close">
     <template #default>
-      <div class="flex items-center justify-between mb-4">
-        <div class="text-xs text-gray-500 dark:text-gray-400">
-          {{ t('admin.ops.requestDetails.rangeLabel', { range: rangeLabel }) }}
-        </div>
-        <button
-          type="button"
-          class="btn btn-secondary btn-sm"
-          @click="fetchData"
-        >
-          {{ t('common.refresh') }}
-        </button>
-      </div>
-
-      <!-- Loading -->
-      <div v-if="loading" class="flex items-center justify-center py-16">
-        <div class="flex flex-col items-center gap-3">
-          <svg class="h-8 w-8 animate-spin text-blue-500" fill="none" viewBox="0 0 24 24">
-            <circle class="opacity-25" cx="12" cy="12" r="10" stroke="currentColor" stroke-width="4"></circle>
-            <path
-              class="opacity-75"
-              fill="currentColor"
-              d="M4 12a8 8 0 018-8V0C5.373 0 0 5.373 0 12h4zm2 5.291A7.962 7.962 0 014 12H0c0 3.042 1.135 5.824 3 7.938l3-2.647z"
-            ></path>
-          </svg>
-          <span class="text-sm font-medium text-gray-500 dark:text-gray-400">{{ t('common.loading') }}</span>
-        </div>
-      </div>
-
-      <!-- Table -->
-      <div v-else>
-        <div v-if="items.length === 0" class="rounded-xl border border-dashed border-gray-200 p-10 text-center dark:border-dark-700">
-          <div class="text-sm font-medium text-gray-600 dark:text-gray-300">{{ t('admin.ops.requestDetails.empty') }}</div>
-          <div class="mt-1 text-xs text-gray-400">{{ t('admin.ops.requestDetails.emptyHint') }}</div>
+      <div class="flex h-full min-h-0 flex-col">
+        <div class="mb-4 flex flex-shrink-0 items-center justify-between">
+          <div class="text-xs text-gray-500 dark:text-gray-400">
+            {{ t('admin.ops.requestDetails.rangeLabel', { range: rangeLabel }) }}
+          </div>
+          <button
+            type="button"
+            class="btn btn-secondary btn-sm"
+            @click="fetchData"
+          >
+            {{ t('common.refresh') }}
+          </button>
         </div>
 
-        <div v-else class="overflow-hidden rounded-xl border border-gray-200 dark:border-dark-700">
-          <div class="overflow-x-auto">
-            <table class="min-w-full divide-y divide-gray-200 dark:divide-dark-700">
-              <thead class="bg-gray-50 dark:bg-dark-900">
+        <!-- Loading -->
+        <div v-if="loading" class="flex flex-1 items-center justify-center py-16">
+          <div class="flex flex-col items-center gap-3">
+            <svg class="h-8 w-8 animate-spin text-blue-500" fill="none" viewBox="0 0 24 24">
+              <circle class="opacity-25" cx="12" cy="12" r="10" stroke="currentColor" stroke-width="4"></circle>
+              <path
+                class="opacity-75"
+                fill="currentColor"
+                d="M4 12a8 8 0 018-8V0C5.373 0 0 5.373 0 12h4zm2 5.291A7.962 7.962 0 014 12H0c0 3.042 1.135 5.824 3 7.938l3-2.647z"
+              ></path>
+            </svg>
+            <span class="text-sm font-medium text-gray-500 dark:text-gray-400">{{ t('common.loading') }}</span>
+          </div>
+        </div>
+
+        <!-- Table -->
+        <div v-else class="flex min-h-0 flex-1 flex-col">
+          <div v-if="items.length === 0" class="rounded-xl border border-dashed border-gray-200 p-10 text-center dark:border-dark-700">
+            <div class="text-sm font-medium text-gray-600 dark:text-gray-300">{{ t('admin.ops.requestDetails.empty') }}</div>
+            <div class="mt-1 text-xs text-gray-400">{{ t('admin.ops.requestDetails.emptyHint') }}</div>
+          </div>
+
+          <div v-else class="flex min-h-0 flex-1 flex-col overflow-hidden rounded-xl border border-gray-200 dark:border-dark-700">
+            <div class="min-h-0 flex-1 overflow-auto">
+              <table class="min-w-full divide-y divide-gray-200 dark:divide-dark-700">
+                <thead class="sticky top-0 z-10 bg-gray-50 dark:bg-dark-900">
                 <tr>
                   <th class="px-4 py-3 text-left text-[11px] font-bold uppercase tracking-wider text-gray-500 dark:text-gray-400">
                     {{ t('admin.ops.requestDetails.table.time') }}
@@ -265,15 +267,16 @@ const kindBadgeClass = (kind: string) => {
                 </tr>
               </tbody>
             </table>
-          </div>
+            </div>
 
-          <Pagination
-            :total="total"
-            :page="page"
-            :page-size="pageSize"
-            @update:page="handlePageChange"
-            @update:pageSize="handlePageSizeChange"
-          />
+            <Pagination
+              :total="total"
+              :page="page"
+              :page-size="pageSize"
+              @update:page="handlePageChange"
+              @update:pageSize="handlePageSizeChange"
+            />
+          </div>
         </div>
       </div>
     </template>

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

@@ -45,6 +45,36 @@ function validateRuntimeSettings(settings: OpsAlertRuntimeSettings): ValidationR
     errors.push(t('admin.ops.runtime.validation.evalIntervalRange'))
   }
 
+  // Thresholds validation
+  const thresholds = settings.thresholds
+  if (thresholds) {
+    if (thresholds.sla_percent_min != null) {
+      if (!Number.isFinite(thresholds.sla_percent_min) || thresholds.sla_percent_min < 0 || thresholds.sla_percent_min > 100) {
+        errors.push('SLA 最低值必须在 0-100 之间')
+      }
+    }
+    if (thresholds.latency_p99_ms_max != null) {
+      if (!Number.isFinite(thresholds.latency_p99_ms_max) || thresholds.latency_p99_ms_max < 0) {
+        errors.push('延迟 P99 最大值必须大于或等于 0')
+      }
+    }
+    if (thresholds.ttft_p99_ms_max != null) {
+      if (!Number.isFinite(thresholds.ttft_p99_ms_max) || thresholds.ttft_p99_ms_max < 0) {
+        errors.push('TTFT P99 最大值必须大于或等于 0')
+      }
+    }
+    if (thresholds.request_error_rate_percent_max != null) {
+      if (!Number.isFinite(thresholds.request_error_rate_percent_max) || thresholds.request_error_rate_percent_max < 0 || thresholds.request_error_rate_percent_max > 100) {
+        errors.push('请求错误率最大值必须在 0-100 之间')
+      }
+    }
+    if (thresholds.upstream_error_rate_percent_max != null) {
+      if (!Number.isFinite(thresholds.upstream_error_rate_percent_max) || thresholds.upstream_error_rate_percent_max < 0 || thresholds.upstream_error_rate_percent_max > 100) {
+        errors.push('上游错误率最大值必须在 0-100 之间')
+      }
+    }
+  }
+
   const lock = settings.distributed_lock
   if (lock?.enabled) {
     if (!lock.key || lock.key.trim().length < 3) {
@@ -130,6 +160,15 @@ function openAlertEditor() {
     if (!Array.isArray(draftAlert.value.silencing.entries)) {
       draftAlert.value.silencing.entries = []
     }
+    if (!draftAlert.value.thresholds) {
+      draftAlert.value.thresholds = {
+        sla_percent_min: 99.5,
+        latency_p99_ms_max: 2000,
+        ttft_p99_ms_max: 500,
+        request_error_rate_percent_max: 5,
+        upstream_error_rate_percent_max: 5
+      }
+    }
   }
 
   showAlertEditor.value = true
@@ -295,6 +334,81 @@ onMounted(() => {
         <p class="mt-1 text-xs text-gray-500">{{ t('admin.ops.runtime.evalIntervalHint') }}</p>
       </div>
 
+      <div class="rounded-2xl bg-gray-50 p-4 dark:bg-dark-700/50">
+        <div class="mb-2 text-sm font-semibold text-gray-900 dark:text-white">指标阈值配置</div>
+        <p class="mb-4 text-xs text-gray-500 dark:text-gray-400">配置各项指标的告警阈值。超出阈值的指标将在看板上以红色显示。</p>
+
+        <div class="grid grid-cols-1 gap-4 md:grid-cols-2">
+          <div>
+            <div class="mb-1 text-xs font-medium text-gray-600 dark:text-gray-300">SLA 最低值 (%)</div>
+            <input
+              v-model.number="draftAlert.thresholds.sla_percent_min"
+              type="number"
+              min="0"
+              max="100"
+              step="0.1"
+              class="input"
+              placeholder="99.5"
+            />
+            <p class="mt-1 text-xs text-gray-500 dark:text-gray-400">SLA 低于此值时将显示为红色</p>
+          </div>
+
+          <div>
+            <div class="mb-1 text-xs font-medium text-gray-600 dark:text-gray-300">延迟 P99 最大值 (ms)</div>
+            <input
+              v-model.number="draftAlert.thresholds.latency_p99_ms_max"
+              type="number"
+              min="0"
+              step="100"
+              class="input"
+              placeholder="2000"
+            />
+            <p class="mt-1 text-xs text-gray-500 dark:text-gray-400">延迟 P99 高于此值时将显示为红色</p>
+          </div>
+
+          <div>
+            <div class="mb-1 text-xs font-medium text-gray-600 dark:text-gray-300">TTFT P99 最大值 (ms)</div>
+            <input
+              v-model.number="draftAlert.thresholds.ttft_p99_ms_max"
+              type="number"
+              min="0"
+              step="100"
+              class="input"
+              placeholder="500"
+            />
+            <p class="mt-1 text-xs text-gray-500 dark:text-gray-400">TTFT P99 高于此值时将显示为红色</p>
+          </div>
+
+          <div>
+            <div class="mb-1 text-xs font-medium text-gray-600 dark:text-gray-300">请求错误率最大值 (%)</div>
+            <input
+              v-model.number="draftAlert.thresholds.request_error_rate_percent_max"
+              type="number"
+              min="0"
+              max="100"
+              step="0.1"
+              class="input"
+              placeholder="5"
+            />
+            <p class="mt-1 text-xs text-gray-500 dark:text-gray-400">请求错误率高于此值时将显示为红色</p>
+          </div>
+
+          <div>
+            <div class="mb-1 text-xs font-medium text-gray-600 dark:text-gray-300">上游错误率最大值 (%)</div>
+            <input
+              v-model.number="draftAlert.thresholds.upstream_error_rate_percent_max"
+              type="number"
+              min="0"
+              max="100"
+              step="0.1"
+              class="input"
+              placeholder="5"
+            />
+            <p class="mt-1 text-xs text-gray-500 dark:text-gray-400">上游错误率高于此值时将显示为红色</p>
+          </div>
+        </div>
+      </div>
+
       <div class="rounded-2xl bg-gray-50 p-4 dark:bg-dark-700/50">
         <div class="mb-2 text-sm font-semibold text-gray-900 dark:text-white">{{ t('admin.ops.runtime.silencing.title') }}</div>
 

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

@@ -6,7 +6,7 @@ import { opsAPI } from '@/api/admin/ops'
 import BaseDialog from '@/components/common/BaseDialog.vue'
 import Select from '@/components/common/Select.vue'
 import Toggle from '@/components/common/Toggle.vue'
-import type { OpsAlertRuntimeSettings, EmailNotificationConfig, AlertSeverity, OpsAdvancedSettings } from '../types'
+import type { OpsAlertRuntimeSettings, EmailNotificationConfig, AlertSeverity, OpsAdvancedSettings, OpsMetricThresholds } from '../types'
 
 const { t } = useI18n()
 const appStore = useAppStore()
@@ -29,19 +29,38 @@ const runtimeSettings = ref<OpsAlertRuntimeSettings | null>(null)
 const emailConfig = ref<EmailNotificationConfig | null>(null)
 // 高级设置
 const advancedSettings = ref<OpsAdvancedSettings | null>(null)
+// 指标阈值配置
+const metricThresholds = ref<OpsMetricThresholds>({
+  sla_percent_min: 99.5,
+  latency_p99_ms_max: 2000,
+  ttft_p99_ms_max: 500,
+  request_error_rate_percent_max: 5,
+  upstream_error_rate_percent_max: 5
+})
 
 // 加载所有配置
 async function loadAllSettings() {
   loading.value = true
   try {
-    const [runtime, email, advanced] = await Promise.all([
+    const [runtime, email, advanced, thresholds] = await Promise.all([
       opsAPI.getAlertRuntimeSettings(),
       opsAPI.getEmailNotificationConfig(),
-      opsAPI.getAdvancedSettings()
+      opsAPI.getAdvancedSettings(),
+      opsAPI.getMetricThresholds()
     ])
     runtimeSettings.value = runtime
     emailConfig.value = email
     advancedSettings.value = advanced
+    // 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 = {
+        sla_percent_min: thresholds.sla_percent_min ?? 99.5,
+        latency_p99_ms_max: thresholds.latency_p99_ms_max ?? 2000,
+        ttft_p99_ms_max: thresholds.ttft_p99_ms_max ?? 500,
+        request_error_rate_percent_max: thresholds.request_error_rate_percent_max ?? 5,
+        upstream_error_rate_percent_max: thresholds.upstream_error_rate_percent_max ?? 5
+      }
+    }
   } catch (err: any) {
     console.error('[OpsSettingsDialog] Failed to load settings', err)
     appStore.showError(err?.response?.data?.detail || t('admin.ops.settings.loadFailed'))
@@ -138,6 +157,23 @@ const validation = computed(() => {
     }
   }
 
+  // 验证指标阈值
+  if (metricThresholds.value.sla_percent_min != null && (metricThresholds.value.sla_percent_min < 0 || metricThresholds.value.sla_percent_min > 100)) {
+    errors.push('SLA最低百分比必须在0-100之间')
+  }
+  if (metricThresholds.value.latency_p99_ms_max != null && metricThresholds.value.latency_p99_ms_max < 0) {
+    errors.push('延迟P99最大值必须大于等于0')
+  }
+  if (metricThresholds.value.ttft_p99_ms_max != null && metricThresholds.value.ttft_p99_ms_max < 0) {
+    errors.push('TTFT P99最大值必须大于等于0')
+  }
+  if (metricThresholds.value.request_error_rate_percent_max != null && (metricThresholds.value.request_error_rate_percent_max < 0 || metricThresholds.value.request_error_rate_percent_max > 100)) {
+    errors.push('请求错误率最大值必须在0-100之间')
+  }
+  if (metricThresholds.value.upstream_error_rate_percent_max != null && (metricThresholds.value.upstream_error_rate_percent_max < 0 || metricThresholds.value.upstream_error_rate_percent_max > 100)) {
+    errors.push('上游错误率最大值必须在0-100之间')
+  }
+
   return { valid: errors.length === 0, errors }
 })
 
@@ -153,14 +189,15 @@ async function saveAllSettings() {
     await Promise.all([
       runtimeSettings.value ? opsAPI.updateAlertRuntimeSettings(runtimeSettings.value) : Promise.resolve(),
       emailConfig.value ? opsAPI.updateEmailNotificationConfig(emailConfig.value) : Promise.resolve(),
-      advancedSettings.value ? opsAPI.updateAdvancedSettings(advancedSettings.value) : Promise.resolve()
+      advancedSettings.value ? opsAPI.updateAdvancedSettings(advancedSettings.value) : Promise.resolve(),
+      opsAPI.updateMetricThresholds(metricThresholds.value)
     ])
     appStore.showSuccess(t('admin.ops.settings.saveSuccess'))
     emit('saved')
     emit('close')
   } catch (err: any) {
     console.error('[OpsSettingsDialog] Failed to save settings', err)
-    appStore.showError(err?.response?.data?.detail || t('admin.ops.settings.saveFailed'))
+    appStore.showError(err?.response?.data?.message || err?.response?.data?.detail || t('admin.ops.settings.saveFailed'))
   } finally {
     saving.value = false
   }
@@ -306,6 +343,77 @@ async function saveAllSettings() {
         </div>
       </div>
 
+      <!-- 指标阈值配置 -->
+      <div class="rounded-2xl bg-gray-50 p-4 dark:bg-dark-700/50">
+        <h4 class="mb-3 text-sm font-semibold text-gray-900 dark:text-white">{{ t('admin.ops.settings.metricThresholds') }}</h4>
+        <p class="mb-4 text-xs text-gray-500 dark:text-gray-400">{{ t('admin.ops.settings.metricThresholdsHint') }}</p>
+
+        <div class="space-y-4">
+          <div>
+            <label class="input-label">{{ t('admin.ops.settings.slaMinPercent') }}</label>
+            <input
+              v-model.number="metricThresholds.sla_percent_min"
+              type="number"
+              min="0"
+              max="100"
+              step="0.1"
+              class="input"
+            />
+            <p class="mt-1 text-xs text-gray-500">{{ t('admin.ops.settings.slaMinPercentHint') }}</p>
+          </div>
+
+          <div>
+            <label class="input-label">{{ t('admin.ops.settings.latencyP99MaxMs') }}</label>
+            <input
+              v-model.number="metricThresholds.latency_p99_ms_max"
+              type="number"
+              min="0"
+              step="100"
+              class="input"
+            />
+            <p class="mt-1 text-xs text-gray-500">{{ t('admin.ops.settings.latencyP99MaxMsHint') }}</p>
+          </div>
+
+          <div>
+            <label class="input-label">{{ t('admin.ops.settings.ttftP99MaxMs') }}</label>
+            <input
+              v-model.number="metricThresholds.ttft_p99_ms_max"
+              type="number"
+              min="0"
+              step="50"
+              class="input"
+            />
+            <p class="mt-1 text-xs text-gray-500">{{ t('admin.ops.settings.ttftP99MaxMsHint') }}</p>
+          </div>
+
+          <div>
+            <label class="input-label">{{ t('admin.ops.settings.requestErrorRateMaxPercent') }}</label>
+            <input
+              v-model.number="metricThresholds.request_error_rate_percent_max"
+              type="number"
+              min="0"
+              max="100"
+              step="0.1"
+              class="input"
+            />
+            <p class="mt-1 text-xs text-gray-500">{{ t('admin.ops.settings.requestErrorRateMaxPercentHint') }}</p>
+          </div>
+
+          <div>
+            <label class="input-label">{{ t('admin.ops.settings.upstreamErrorRateMaxPercent') }}</label>
+            <input
+              v-model.number="metricThresholds.upstream_error_rate_percent_max"
+              type="number"
+              min="0"
+              max="100"
+              step="0.1"
+              class="input"
+            />
+            <p class="mt-1 text-xs text-gray-500">{{ t('admin.ops.settings.upstreamErrorRateMaxPercentHint') }}</p>
+          </div>
+        </div>
+      </div>
+
       <!-- 高级设置 -->
       <details class="rounded-2xl bg-gray-50 dark:bg-dark-700/50">
         <summary class="cursor-pointer p-4 text-sm font-semibold text-gray-900 dark:text-white">

frontend/src/views/admin/ops/types.ts

@@ -14,6 +14,7 @@ export type {
   EmailNotificationConfig,
   OpsDistributedLockSettings,
   OpsAlertRuntimeSettings,
+  OpsMetricThresholds,
   OpsAdvancedSettings,
   OpsDataRetentionSettings,
   OpsAggregationSettings