Merge pull request #2090 from touwaeriol/feat/ops-retention-zero

feat(ops): allow retention days = 0 to wipe table on each scheduled cleanup

backend/internal/service/ops_cleanup_service.go

@@ -184,6 +184,25 @@ func (c opsCleanupDeletedCounts) String() string {
 	)
 }
 
+// opsCleanupPlan 把"保留天数"翻译成具体的清理动作。
+//   - days < 0  → 跳过该项清理（ok=false），保留兼容老数据
+//   - days == 0 → TRUNCATE TABLE（O(1) 全清），truncate=true
+//   - days > 0  → 批量 DELETE 早于 now-N天 的行，cutoff = now - N 天
+//
+// 之所以 days==0 走 TRUNCATE 而非"now+24h cutoff + DELETE"：
+//   - 速度从 O(N) 降到 O(1)，对百万行级表毫秒完成
+//   - 无 WAL 写入、无后续 VACUUM 压力
+//   - 这些 ops 表只有 cleanup 任务自己写，TRUNCATE 的 ACCESS EXCLUSIVE 锁影响可忽略
+func opsCleanupPlan(now time.Time, days int) (cutoff time.Time, truncate, ok bool) {
+	if days < 0 {
+		return time.Time{}, false, false
+	}
+	if days == 0 {
+		return time.Time{}, true, true
+	}
+	return now.AddDate(0, 0, -days), false, true
+}
+
 func (s *OpsCleanupService) runCleanupOnce(ctx context.Context) (opsCleanupDeletedCounts, error) {
 	out := opsCleanupDeletedCounts{}
 	if s == nil || s.db == nil || s.cfg == nil {
@@ -194,34 +213,42 @@ func (s *OpsCleanupService) runCleanupOnce(ctx context.Context) (opsCleanupDelet
 
 	now := time.Now().UTC()
 
-	// Error-like tables: error logs / retry attempts / alert events.
-	if days := s.cfg.Ops.Cleanup.ErrorLogRetentionDays; days > 0 {
-		cutoff := now.AddDate(0, 0, -days)
-		n, err := deleteOldRowsByID(ctx, s.db, "ops_error_logs", "created_at", cutoff, batchSize, false)
+	// runOne 把"truncate? cutoff? batched delete?"封装到一处，
+	// 让三组清理（错误日志类 / 分钟指标 / 小时+日预聚合）调用方只关心表名和列名。
+	runOne := func(truncate bool, cutoff time.Time, table, timeCol string, castDate bool) (int64, error) {
+		if truncate {
+			return truncateOpsTable(ctx, s.db, table)
+		}
+		return deleteOldRowsByID(ctx, s.db, table, timeCol, cutoff, batchSize, castDate)
+	}
+
+	// Error-like tables: error logs / retry attempts / alert events / system logs / cleanup audits.
+	if cutoff, truncate, ok := opsCleanupPlan(now, s.cfg.Ops.Cleanup.ErrorLogRetentionDays); ok {
+		n, err := runOne(truncate, cutoff, "ops_error_logs", "created_at", false)
 		if err != nil {
 			return out, err
 		}
 		out.errorLogs = n
 
-		n, err = deleteOldRowsByID(ctx, s.db, "ops_retry_attempts", "created_at", cutoff, batchSize, false)
+		n, err = runOne(truncate, cutoff, "ops_retry_attempts", "created_at", false)
 		if err != nil {
 			return out, err
 		}
 		out.retryAttempts = n
 
-		n, err = deleteOldRowsByID(ctx, s.db, "ops_alert_events", "created_at", cutoff, batchSize, false)
+		n, err = runOne(truncate, cutoff, "ops_alert_events", "created_at", false)
 		if err != nil {
 			return out, err
 		}
 		out.alertEvents = n
 
-		n, err = deleteOldRowsByID(ctx, s.db, "ops_system_logs", "created_at", cutoff, batchSize, false)
+		n, err = runOne(truncate, cutoff, "ops_system_logs", "created_at", false)
 		if err != nil {
 			return out, err
 		}
 		out.systemLogs = n
 
-		n, err = deleteOldRowsByID(ctx, s.db, "ops_system_log_cleanup_audits", "created_at", cutoff, batchSize, false)
+		n, err = runOne(truncate, cutoff, "ops_system_log_cleanup_audits", "created_at", false)
 		if err != nil {
 			return out, err
 		}
@@ -229,9 +256,8 @@ func (s *OpsCleanupService) runCleanupOnce(ctx context.Context) (opsCleanupDelet
 	}
 
 	// Minute-level metrics snapshots.
-	if days := s.cfg.Ops.Cleanup.MinuteMetricsRetentionDays; days > 0 {
-		cutoff := now.AddDate(0, 0, -days)
-		n, err := deleteOldRowsByID(ctx, s.db, "ops_system_metrics", "created_at", cutoff, batchSize, false)
+	if cutoff, truncate, ok := opsCleanupPlan(now, s.cfg.Ops.Cleanup.MinuteMetricsRetentionDays); ok {
+		n, err := runOne(truncate, cutoff, "ops_system_metrics", "created_at", false)
 		if err != nil {
 			return out, err
 		}
@@ -239,15 +265,14 @@ func (s *OpsCleanupService) runCleanupOnce(ctx context.Context) (opsCleanupDelet
 	}
 
 	// Pre-aggregation tables (hourly/daily).
-	if days := s.cfg.Ops.Cleanup.HourlyMetricsRetentionDays; days > 0 {
-		cutoff := now.AddDate(0, 0, -days)
-		n, err := deleteOldRowsByID(ctx, s.db, "ops_metrics_hourly", "bucket_start", cutoff, batchSize, false)
+	if cutoff, truncate, ok := opsCleanupPlan(now, s.cfg.Ops.Cleanup.HourlyMetricsRetentionDays); ok {
+		n, err := runOne(truncate, cutoff, "ops_metrics_hourly", "bucket_start", false)
 		if err != nil {
 			return out, err
 		}
 		out.hourlyPreagg = n
 
-		n, err = deleteOldRowsByID(ctx, s.db, "ops_metrics_daily", "bucket_date", cutoff, batchSize, true)
+		n, err = runOne(truncate, cutoff, "ops_metrics_daily", "bucket_date", true)
 		if err != nil {
 			return out, err
 		}
@@ -303,7 +328,7 @@ WHERE id IN (SELECT id FROM batch)
 		res, err := db.ExecContext(ctx, q, cutoff, batchSize)
 		if err != nil {
 			// If ops tables aren't present yet (partial deployments), treat as no-op.
-			if strings.Contains(strings.ToLower(err.Error()), "does not exist") && strings.Contains(strings.ToLower(err.Error()), "relation") {
+			if isMissingRelationError(err) {
 				return total, nil
 			}
 			return total, err
@@ -320,6 +345,46 @@ WHERE id IN (SELECT id FROM batch)
 	return total, nil
 }
 
+// truncateOpsTable 用 TRUNCATE TABLE 清空指定表，先 SELECT COUNT(*) 取得清空前行数用于 heartbeat。
+//
+// 与 deleteOldRowsByID 的差异：
+//   - 不可指定 WHERE 条件，仅用于 days==0 的"清空全部"语义
+//   - O(1) 释放表的物理存储页，毫秒级完成，无 WAL 写入、无 VACUUM 压力
+//   - 需要 ACCESS EXCLUSIVE 锁，但 ops 表只有清理任务自己写入，瞬间锁影响可忽略
+//
+// 表不存在（部分部署）静默返回 0，与 deleteOldRowsByID 保持一致。
+func truncateOpsTable(ctx context.Context, db *sql.DB, table string) (int64, error) {
+	if db == nil {
+		return 0, nil
+	}
+	var count int64
+	if err := db.QueryRowContext(ctx, fmt.Sprintf("SELECT COUNT(*) FROM %s", table)).Scan(&count); err != nil {
+		if isMissingRelationError(err) {
+			return 0, nil
+		}
+		return 0, fmt.Errorf("count %s: %w", table, err)
+	}
+	if count == 0 {
+		return 0, nil
+	}
+	if _, err := db.ExecContext(ctx, fmt.Sprintf("TRUNCATE TABLE %s", table)); err != nil {
+		if isMissingRelationError(err) {
+			return 0, nil
+		}
+		return 0, fmt.Errorf("truncate %s: %w", table, err)
+	}
+	return count, nil
+}
+
+// isMissingRelationError 判断 PG 报错是否为"表不存在"，用于让清理任务在部分部署场景静默跳过。
+func isMissingRelationError(err error) bool {
+	if err == nil {
+		return false
+	}
+	s := strings.ToLower(err.Error())
+	return strings.Contains(s, "does not exist") && strings.Contains(s, "relation")
+}
+
 func (s *OpsCleanupService) tryAcquireLeaderLock(ctx context.Context) (func(), bool) {
 	if s == nil {
 		return nil, false

backend/internal/service/ops_cleanup_service_test.go

+package service
+
+import (
+	"testing"
+	"time"
+)
+
+func TestOpsCleanupPlan(t *testing.T) {
+	now := time.Date(2026, 4, 29, 12, 0, 0, 0, time.UTC)
+
+	cases := []struct {
+		name         string
+		days         int
+		wantOK       bool
+		wantTruncate bool
+		wantCutoff   time.Time
+	}{
+		{name: "negative skips", days: -1, wantOK: false},
+		{name: "zero truncates", days: 0, wantOK: true, wantTruncate: true},
+		{name: "positive yields past cutoff", days: 7, wantOK: true, wantCutoff: now.AddDate(0, 0, -7)},
+	}
+
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			cutoff, truncate, ok := opsCleanupPlan(now, tc.days)
+			if ok != tc.wantOK {
+				t.Fatalf("ok = %v, want %v", ok, tc.wantOK)
+			}
+			if !ok {
+				return
+			}
+			if truncate != tc.wantTruncate {
+				t.Fatalf("truncate = %v, want %v", truncate, tc.wantTruncate)
+			}
+			if !tc.wantTruncate && !cutoff.Equal(tc.wantCutoff) {
+				t.Fatalf("cutoff = %v, want %v", cutoff, tc.wantCutoff)
+			}
+		})
+	}
+}
+
+func TestIsMissingRelationError(t *testing.T) {
+	cases := []struct {
+		name string
+		err  error
+		want bool
+	}{
+		{name: "nil is not missing", err: nil, want: false},
+		{name: "match relation does not exist", err: fakeErr(`pq: relation "ops_error_logs" does not exist`), want: true},
+		{name: "match case-insensitive", err: fakeErr(`ERROR: Relation "x" Does Not Exist`), want: true},
+		{name: "non-matching error", err: fakeErr("connection refused"), want: false},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			if got := isMissingRelationError(tc.err); got != tc.want {
+				t.Fatalf("got %v, want %v", got, tc.want)
+			}
+		})
+	}
+}
+
+type fakeErr string
+
+func (e fakeErr) Error() string { return string(e) }

backend/internal/service/ops_settings.go

@@ -387,13 +387,15 @@ func normalizeOpsAdvancedSettings(cfg *OpsAdvancedSettings) {
 	if cfg.DataRetention.CleanupSchedule == "" {
 		cfg.DataRetention.CleanupSchedule = "0 2 * * *"
 	}
-	if cfg.DataRetention.ErrorLogRetentionDays <= 0 {
+	// 保留天数：0 表示每次定时清理全部（清空所有），> 0 表示按天数保留；
+	// 仅在拿到非法的负数时回填默认值，避免覆盖用户主动设的 0。
+	if cfg.DataRetention.ErrorLogRetentionDays < 0 {
 		cfg.DataRetention.ErrorLogRetentionDays = 30
 	}
-	if cfg.DataRetention.MinuteMetricsRetentionDays <= 0 {
+	if cfg.DataRetention.MinuteMetricsRetentionDays < 0 {
 		cfg.DataRetention.MinuteMetricsRetentionDays = 30
 	}
-	if cfg.DataRetention.HourlyMetricsRetentionDays <= 0 {
+	if cfg.DataRetention.HourlyMetricsRetentionDays < 0 {
 		cfg.DataRetention.HourlyMetricsRetentionDays = 30
 	}
 	// Normalize auto refresh interval (default 30 seconds)
@@ -406,14 +408,15 @@ func validateOpsAdvancedSettings(cfg *OpsAdvancedSettings) error {
 	if cfg == nil {
 		return errors.New("invalid config")
 	}
-	if cfg.DataRetention.ErrorLogRetentionDays < 1 || cfg.DataRetention.ErrorLogRetentionDays > 365 {
-		return errors.New("error_log_retention_days must be between 1 and 365")
+	// 保留天数：0 表示每次清理全部，1-365 表示按天数保留。
+	if cfg.DataRetention.ErrorLogRetentionDays < 0 || cfg.DataRetention.ErrorLogRetentionDays > 365 {
+		return errors.New("error_log_retention_days must be between 0 and 365")
 	}
-	if cfg.DataRetention.MinuteMetricsRetentionDays < 1 || cfg.DataRetention.MinuteMetricsRetentionDays > 365 {
-		return errors.New("minute_metrics_retention_days must be between 1 and 365")
+	if cfg.DataRetention.MinuteMetricsRetentionDays < 0 || cfg.DataRetention.MinuteMetricsRetentionDays > 365 {
+		return errors.New("minute_metrics_retention_days must be between 0 and 365")
 	}
-	if cfg.DataRetention.HourlyMetricsRetentionDays < 1 || cfg.DataRetention.HourlyMetricsRetentionDays > 365 {
-		return errors.New("hourly_metrics_retention_days must be between 1 and 365")
+	if cfg.DataRetention.HourlyMetricsRetentionDays < 0 || cfg.DataRetention.HourlyMetricsRetentionDays > 365 {
+		return errors.New("hourly_metrics_retention_days must be between 0 and 365")
 	}
 	if cfg.AutoRefreshIntervalSec < 15 || cfg.AutoRefreshIntervalSec > 300 {
 		return errors.New("auto_refresh_interval_seconds must be between 15 and 300")

frontend/src/i18n/locales/en.ts

@@ -4648,7 +4648,7 @@ export default {
         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',
+        retentionDaysHint: 'Recommended 7-90 days; longer periods consume more storage. Set to 0 to wipe all history on every scheduled cleanup',
         aggregation: 'Pre-aggregation Tasks',
         enableAggregation: 'Enable Pre-aggregation',
         aggregationHint: 'Pre-aggregation improves query performance for long time windows',
@@ -4678,7 +4678,7 @@ export default {
         autoRefreshCountdown: 'Auto refresh: {seconds}s',
         validation: {
           title: 'Please fix the following issues',
-          retentionDaysRange: 'Retention days must be between 1-365 days',
+          retentionDaysRange: 'Retention days must be between 0 and 365 (0 = wipe all on every cleanup)',
           slaMinPercentRange: 'SLA minimum percentage must be between 0 and 100',
           ttftP99MaxRange: 'TTFT P99 maximum must be a number ≥ 0',
           requestErrorRateMaxRange: 'Request error rate maximum must be between 0 and 100',

frontend/src/i18n/locales/zh.ts

@@ -4810,7 +4810,7 @@ export default {
         errorLogRetentionDays: '错误日志保留天数',
         minuteMetricsRetentionDays: '分钟指标保留天数',
         hourlyMetricsRetentionDays: '小时指标保留天数',
-        retentionDaysHint: '建议保留7-90天，过长会占用存储空间',
+        retentionDaysHint: '建议保留 7-90 天，过长会占用存储空间；填 0 表示每次定时清理时清空所有历史',
         aggregation: '预聚合任务',
         enableAggregation: '启用预聚合任务',
         aggregationHint: '预聚合可提升长时间窗口查询性能',
@@ -4841,7 +4841,7 @@ export default {
         autoRefreshCountdown: '自动刷新：{seconds}s',
         validation: {
           title: '请先修正以下问题',
-          retentionDaysRange: '保留天数必须在1-365天之间',
+          retentionDaysRange: '保留天数必须在 0-365 天之间（0 = 每次清理时清空所有）',
           slaMinPercentRange: 'SLA最低百分比必须在0-100之间',
           ttftP99MaxRange: 'TTFT P99最大值必须大于等于0',
           requestErrorRateMaxRange: '请求错误率最大值必须在0-100之间',

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

@@ -136,13 +136,13 @@ const validation = computed(() => {
   // 验证高级设置
   if (advancedSettings.value) {
     const { error_log_retention_days, minute_metrics_retention_days, hourly_metrics_retention_days } = advancedSettings.value.data_retention
-    if (error_log_retention_days < 1 || error_log_retention_days > 365) {
+    if (error_log_retention_days < 0 || error_log_retention_days > 365) {
       errors.push(t('admin.ops.settings.validation.retentionDaysRange'))
     }
-    if (minute_metrics_retention_days < 1 || minute_metrics_retention_days > 365) {
+    if (minute_metrics_retention_days < 0 || minute_metrics_retention_days > 365) {
       errors.push(t('admin.ops.settings.validation.retentionDaysRange'))
     }
-    if (hourly_metrics_retention_days < 1 || hourly_metrics_retention_days > 365) {
+    if (hourly_metrics_retention_days < 0 || hourly_metrics_retention_days > 365) {
       errors.push(t('admin.ops.settings.validation.retentionDaysRange'))
     }
   }
@@ -431,7 +431,7 @@ async function saveAllSettings() {
                 <input
                   v-model.number="advancedSettings.data_retention.error_log_retention_days"
                   type="number"
-                  min="1"
+                  min="0"
                   max="365"
                   class="input"
                 />
@@ -441,7 +441,7 @@ async function saveAllSettings() {
                 <input
                   v-model.number="advancedSettings.data_retention.minute_metrics_retention_days"
                   type="number"
-                  min="1"
+                  min="0"
                   max="365"
                   class="input"
                 />
@@ -451,7 +451,7 @@ async function saveAllSettings() {
                 <input
                   v-model.number="advancedSettings.data_retention.hourly_metrics_retention_days"
                   type="number"
-                  min="1"
+                  min="0"
                   max="365"
                   class="input"
                 />