feat(v2.0.0): tsdb (#1162)

* feat: tsdb

* fix(ci): remove --parseGoList=false from swag init to fix dependency resolution

* fix(ci): fix swag init directory and temporary remove s390x support due to cgo issues

* fix(ci): fix swag init output directory to cmd/dashboard/docs

* fix(ci): set GOTOOLCHAIN=auto for gosec

* feat: add system storage maintenance for SQLite and TSDB

* shit

* feat: add s390x support and improve service monitoring

* ci: upgrade goreleaser-cross image to v1.25

* ci: add libzstd-dev:s390x for cross-compilation

* ci: build libzstd for s390x from source

* ci: add libzstd_linux_s390x.go for gozstd linking

* ci: use vendor mode for s390x gozstd build

* ci: clone zstd source for s390x build

* refactor(tsdb): rename MaxDiskUsageGB to MinFreeDiskSpaceGB and optimize queries

- Rename config to accurately reflect VictoriaMetrics behavior: minimum free disk space threshold
- Add QueryServiceHistoryByServerID for batch query optimization
- Fix hasStatus to avoid false status counting when only delay data exists
- Fix service aggregation boundary: use successCount*2 >= count
- Fix serviceID parsing with strconv.ParseUint error handling
- Add TagFiltersCacheSize for better query performance

* feat(api): add server metrics endpoint and simplify service history response

- Add /server/:id/metrics API for querying TSDB server metrics
- Simplify getServiceHistory by removing redundant data conversion
- Change AvgDelay type from float32 to float64
- Remove generated swagger docs (to be regenerated)
- Update TSDB query, writer and tests

* chore: 临时禁用不支持前端

* ci: cache zstd build for s390x to speed up CI

* fix(tsdb): fix race conditions, data correctness and optimize performance

- Fix TOCTOU race between IsClosed() and write/query by holding RLock
- Fix delay=0 excluded from stats by using hasDelay flag instead of value > 0
- Fix fmt.Sscanf -> strconv.ParseUint for server_id parsing with error logging
- Fix buffer unbounded growth by flushing inside lock when over maxSize
- Split makeMetricRow into makeServerMetricRow/makeServiceMetricRow
- Extract InitGlobalSettings() from Open() for VictoriaMetrics globals
- Remove redundant instance/GetInstance/SetInstance singleton
- Add error logging for silently skipped block decode errors
- Optimize WriteBatch* to build all rows in single write call
- Optimize downsample to use linear scan instead of map for sorted data
- Optimize query slice reuse across block iterations

* 服务添加DisplayIndex (#1166)

* 服务添加DisplayIndex

* 根据ai建议修改

---------

Co-authored-by: huYang <306061454@qq.com>

* fix(tsdb): restore SQLite fallback and monthly status reload on restart

- Restore ServiceHistory model and SQLite write fallback when TSDB is disabled
- Reload monthlyStatus (30-day) and serviceStatusToday from TSDB/SQLite on startup
- Add SQLite fallback query for /service/:id/history and /server/:id/service
- Remove breaking GET /service/:id endpoint, keep /service/:id/history only
- Add QueryServiceDailyStats to TSDB for per-day aggregation
- Add tests for monthly status and today stats loading from both TSDB and SQLite
- Migrate ServiceHistory table only when TSDB is disabled

* ci: exclude false-positive gosec rules G117, G703, G704

* feat(api): expose tsdb_enabled in setting response

* ci: restore G115 exclusion accidentally dropped in previous commit

* fix: update version numbers for OfficialAdmin and Official templates

* chore: upgrade frontend

* chore: upgrade frontend

---------

Co-authored-by: 胡说丷刂 <34758853+laosan-xx@users.noreply.github.com>
Co-authored-by: huYang <306061454@qq.com>

service/singleton/singleton.go

@@ -87,12 +87,13 @@ func InitDBFromPath(path string) error {
 	}
 	err = DB.AutoMigrate(model.Server{}, model.User{}, model.ServerGroup{}, model.NotificationGroup{},
 		model.Notification{}, model.AlertRule{}, model.Service{}, model.NotificationGroupNotification{},
-		model.ServiceHistory{}, model.Cron{}, model.Transfer{}, model.ServerGroupServer{},
+		model.Cron{}, model.Transfer{}, model.ServerGroupServer{},
 		model.NAT{}, model.DDNSProfile{}, model.NotificationGroupNotification{},
 		model.WAF{}, model.Oauth2Bind{})
 	if err != nil {
 		return err
 	}
+
 	return nil
 }
 
@@ -130,14 +131,9 @@ func RecordTransferHourlyUsage(servers ...*model.Server) {
 	log.Printf("NEZHA>> Saved traffic metrics to database. Affected %d row(s), Error: %v", len(txs), DB.Create(txs).Error)
 }
 
-// CleanServiceHistory 清理无效或过时的 监控记录 和 流量记录
-func CleanServiceHistory() {
-	// 清理已被删除的服务器的监控记录与流量记录
-	DB.Unscoped().Delete(&model.ServiceHistory{}, "created_at < ? OR service_id NOT IN (SELECT `id` FROM services)", time.Now().AddDate(0, 0, -30))
-	// 由于网络监控记录的数据较多，并且前端仅使用了 1 天的数据
-	// 考虑到 sqlite 数据量问题，仅保留一天数据，
-	// server_id = 0 的数据会用于/service页面的可用性展示
-	DB.Unscoped().Delete(&model.ServiceHistory{}, "(created_at < ? AND server_id != 0) OR service_id NOT IN (SELECT `id` FROM services)", time.Now().AddDate(0, 0, -1))
+// CleanMonitorHistory 清理流量记录（TSDB 有自己的保留策略）
+func CleanMonitorHistory() {
+	// 清理已被删除的服务器的流量记录
 	DB.Unscoped().Delete(&model.Transfer{}, "server_id NOT IN (SELECT `id` FROM servers)")
 	// 计算可清理流量记录的时长
 	var allServerKeep time.Time
@@ -179,6 +175,28 @@ func CleanServiceHistory() {
 	}
 }
 
+// PerformMaintenance 执行系统维护（SQLite VACUUM 和 TSDB 维护）
+func PerformMaintenance() {
+	log.Println("NEZHA>> Starting system maintenance...")
+
+	// 1. SQLite 维护
+	if DB != nil {
+		log.Println("NEZHA>> SQLite: Starting VACUUM...")
+		if err := DB.Exec("VACUUM").Error; err != nil {
+			log.Printf("NEZHA>> SQLite: VACUUM failed: %v", err)
+		} else {
+			log.Println("NEZHA>> SQLite: VACUUM completed")
+		}
+	}
+
+	// 2. TSDB 维护
+	if TSDBEnabled() {
+		TSDBShared.Maintenance()
+	}
+
+	log.Println("NEZHA>> System maintenance completed")
+}
+
 // IPDesensitize 根据设置选择是否对IP进行打码处理 返回处理后的IP(关闭打码则返回原IP)
 func IPDesensitize(ip string) string {
 	if Conf.EnablePlainIPInNotification {