feat(ops): 增强上游错误追踪和新增定时报告服务

- 优化错误日志中间件，即使请求成功也记录上游重试/故障转移事件
- 新增OpsScheduledReportService支持定时报告功能
- 使用Redis分布式锁确保定时任务单实例执行
- 完善依赖注入配置
- 优化前端错误趋势图表展示

backend/internal/repository/ops_repo.go

@@ -149,7 +149,7 @@ SELECT
   error_phase,
   error_type,
   severity,
-  COALESCE(status_code, 0),
+  COALESCE(upstream_status_code, status_code, 0),
   COALESCE(platform, ''),
   COALESCE(model, ''),
   duration_ms,
@@ -261,7 +261,7 @@ SELECT
   error_phase,
   error_type,
   severity,
-  COALESCE(status_code, 0),
+  COALESCE(upstream_status_code, status_code, 0),
   COALESCE(platform, ''),
   COALESCE(model, ''),
   duration_ms,
@@ -605,6 +605,17 @@ func buildOpsErrorLogsWhere(filter *service.OpsErrorLogFilter) (string, []any) {
 	args := make([]any, 0, 8)
 	clauses = append(clauses, "1=1")
 
+	phaseFilter := ""
+	if filter != nil {
+		phaseFilter = strings.TrimSpace(strings.ToLower(filter.Phase))
+	}
+	// ops_error_logs primarily stores client-visible error requests (status>=400),
+	// but we also persist "recovered" upstream errors (status<400) for upstream health visibility.
+	// By default, keep list endpoints scoped to client errors unless explicitly filtering upstream phase.
+	if phaseFilter != "upstream" {
+		clauses = append(clauses, "COALESCE(status_code, 0) >= 400")
+	}
+
 	if filter.StartTime != nil && !filter.StartTime.IsZero() {
 		args = append(args, filter.StartTime.UTC())
 		clauses = append(clauses, "created_at >= $"+itoa(len(args)))
@@ -626,13 +637,13 @@ func buildOpsErrorLogsWhere(filter *service.OpsErrorLogFilter) (string, []any) {
 		args = append(args, *filter.AccountID)
 		clauses = append(clauses, "account_id = $"+itoa(len(args)))
 	}
-	if phase := strings.TrimSpace(filter.Phase); phase != "" {
+	if phase := phaseFilter; phase != "" {
 		args = append(args, phase)
 		clauses = append(clauses, "error_phase = $"+itoa(len(args)))
 	}
 	if len(filter.StatusCodes) > 0 {
 		args = append(args, pq.Array(filter.StatusCodes))
-		clauses = append(clauses, "status_code = ANY($"+itoa(len(args))+")")
+		clauses = append(clauses, "COALESCE(upstream_status_code, status_code, 0) = ANY($"+itoa(len(args))+")")
 	}
 	if q := strings.TrimSpace(filter.Query); q != "" {
 		like := "%" + q + "%"

backend/internal/repository/ops_repo_dashboard.go

@@ -815,9 +815,9 @@ func (r *opsRepository) queryErrorCounts(ctx context.Context, filter *service.Op
 
 	q := `
 SELECT
-  COALESCE(COUNT(*), 0) AS error_total,
-  COALESCE(COUNT(*) FILTER (WHERE is_business_limited), 0) AS business_limited,
-  COALESCE(COUNT(*) FILTER (WHERE NOT is_business_limited), 0) AS error_sla,
+  COALESCE(COUNT(*) FILTER (WHERE COALESCE(status_code, 0) >= 400), 0) AS error_total,
+  COALESCE(COUNT(*) FILTER (WHERE COALESCE(status_code, 0) >= 400 AND is_business_limited), 0) AS business_limited,
+  COALESCE(COUNT(*) FILTER (WHERE COALESCE(status_code, 0) >= 400 AND NOT is_business_limited), 0) AS error_sla,
   COALESCE(COUNT(*) FILTER (WHERE error_owner = 'provider' AND NOT is_business_limited AND COALESCE(upstream_status_code, status_code, 0) NOT IN (429, 529)), 0) AS upstream_excl,
   COALESCE(COUNT(*) FILTER (WHERE error_owner = 'provider' AND NOT is_business_limited AND COALESCE(upstream_status_code, status_code, 0) = 429), 0) AS upstream_429,
   COALESCE(COUNT(*) FILTER (WHERE error_owner = 'provider' AND NOT is_business_limited AND COALESCE(upstream_status_code, status_code, 0) = 529), 0) AS upstream_529
@@ -870,6 +870,7 @@ error_buckets AS (
   SELECT date_trunc('minute', created_at) AS bucket, COUNT(*) AS cnt
   FROM ops_error_logs
   ` + errorWhere + `
+    AND COALESCE(status_code, 0) >= 400
   GROUP BY 1
 ),
 combined AS (

backend/internal/repository/ops_repo_preagg.go

@@ -75,7 +75,8 @@ error_base AS (
     group_id AS group_id,
     is_business_limited AS is_business_limited,
     error_owner AS error_owner,
-    COALESCE(upstream_status_code, status_code) AS status_code
+    status_code AS client_status_code,
+    COALESCE(upstream_status_code, status_code, 0) AS effective_status_code
   FROM ops_error_logs
   WHERE created_at >= $1 AND created_at < $2
 ),
@@ -84,12 +85,12 @@ error_agg AS (
     bucket_start,
     CASE WHEN GROUPING(platform) = 1 THEN NULL ELSE platform END AS platform,
     CASE WHEN GROUPING(group_id) = 1 THEN NULL ELSE group_id END AS group_id,
-    COUNT(*) AS error_count_total,
-    COUNT(*) FILTER (WHERE is_business_limited) AS business_limited_count,
-    COUNT(*) FILTER (WHERE NOT is_business_limited) AS error_count_sla,
-    COUNT(*) FILTER (WHERE error_owner = 'provider' AND NOT is_business_limited AND COALESCE(status_code, 0) NOT IN (429, 529)) AS upstream_error_count_excl_429_529,
-    COUNT(*) FILTER (WHERE error_owner = 'provider' AND NOT is_business_limited AND COALESCE(status_code, 0) = 429) AS upstream_429_count,
-    COUNT(*) FILTER (WHERE error_owner = 'provider' AND NOT is_business_limited AND COALESCE(status_code, 0) = 529) AS upstream_529_count
+    COUNT(*) FILTER (WHERE COALESCE(client_status_code, 0) >= 400) AS error_count_total,
+    COUNT(*) FILTER (WHERE COALESCE(client_status_code, 0) >= 400 AND is_business_limited) AS business_limited_count,
+    COUNT(*) FILTER (WHERE COALESCE(client_status_code, 0) >= 400 AND NOT is_business_limited) AS error_count_sla,
+    COUNT(*) FILTER (WHERE error_owner = 'provider' AND NOT is_business_limited AND COALESCE(effective_status_code, 0) NOT IN (429, 529)) AS upstream_error_count_excl_429_529,
+    COUNT(*) FILTER (WHERE error_owner = 'provider' AND NOT is_business_limited AND COALESCE(effective_status_code, 0) = 429) AS upstream_429_count,
+    COUNT(*) FILTER (WHERE error_owner = 'provider' AND NOT is_business_limited AND COALESCE(effective_status_code, 0) = 529) AS upstream_529_count
   FROM error_base
   GROUP BY GROUPING SETS (
     (bucket_start),

backend/internal/repository/ops_repo_request_details.go

@@ -131,6 +131,7 @@ WITH combined AS (
   LEFT JOIN groups g ON g.id = o.group_id
   LEFT JOIN accounts a ON a.id = o.account_id
   WHERE o.created_at >= $1 AND o.created_at < $2
+    AND COALESCE(o.status_code, 0) >= 400
 )
 `
 

backend/internal/repository/ops_repo_trends.go

@@ -53,6 +53,7 @@ error_buckets AS (
          COUNT(*) AS error_count
   FROM ops_error_logs
   ` + errorWhere + `
+    AND COALESCE(status_code, 0) >= 400
   GROUP BY 1
 ),
 combined AS (
@@ -168,6 +169,7 @@ error_totals AS (
          COUNT(*) AS error_count
   FROM ops_error_logs
   WHERE created_at >= $1 AND created_at < $2
+    AND COALESCE(status_code, 0) >= 400
   GROUP BY 1
 ),
 combined AS (
@@ -240,6 +242,7 @@ error_totals AS (
   WHERE created_at >= $1 AND created_at < $2
     AND platform = $3
     AND group_id IS NOT NULL
+    AND COALESCE(status_code, 0) >= 400
   GROUP BY 1
 ),
 combined AS (
@@ -413,9 +416,9 @@ func (r *opsRepository) GetErrorTrend(ctx context.Context, filter *service.OpsDa
 	q := `
 SELECT
   ` + bucketExpr + ` AS bucket,
-  COUNT(*) AS error_total,
-  COUNT(*) FILTER (WHERE is_business_limited) AS business_limited,
-  COUNT(*) FILTER (WHERE NOT is_business_limited) AS error_sla,
+  COUNT(*) FILTER (WHERE COALESCE(status_code, 0) >= 400) AS error_total,
+  COUNT(*) FILTER (WHERE COALESCE(status_code, 0) >= 400 AND is_business_limited) AS business_limited,
+  COUNT(*) FILTER (WHERE COALESCE(status_code, 0) >= 400 AND NOT is_business_limited) AS error_sla,
   COUNT(*) FILTER (WHERE error_owner = 'provider' AND NOT is_business_limited AND COALESCE(upstream_status_code, status_code, 0) NOT IN (429, 529)) AS upstream_excl,
   COUNT(*) FILTER (WHERE error_owner = 'provider' AND NOT is_business_limited AND COALESCE(upstream_status_code, status_code, 0) = 429) AS upstream_429,
   COUNT(*) FILTER (WHERE error_owner = 'provider' AND NOT is_business_limited AND COALESCE(upstream_status_code, status_code, 0) = 529) AS upstream_529
@@ -524,12 +527,13 @@ func (r *opsRepository) GetErrorDistribution(ctx context.Context, filter *servic
 
 	q := `
 SELECT
-  COALESCE(status_code, 0) AS status_code,
+  COALESCE(upstream_status_code, status_code, 0) AS status_code,
   COUNT(*) AS total,
   COUNT(*) FILTER (WHERE NOT is_business_limited) AS sla,
   COUNT(*) FILTER (WHERE is_business_limited) AS business_limited
 FROM ops_error_logs
 ` + where + `
+  AND COALESCE(status_code, 0) >= 400
 GROUP BY 1
 ORDER BY total DESC
 LIMIT 20`