fix: 初验针对修改

2026-09-09 16:45:01 +08:00
parent 38a86e1fd8
commit b0ffde3100
25 changed files with 1531 additions and 158 deletions
--- a/cmd/main/main.go
+++ b/cmd/main/main.go
@@ -1,36 +1,118 @@
-package main
-
-import (
-	"fmt"
-
-	"git.apinb.com/bsm-sdk/core/infra"
-	"git.apinb.com/bsm-sdk/core/middleware"
-	"git.apinb.com/ops/logs/internal/config"
-	"git.apinb.com/ops/logs/internal/impl"
-	"git.apinb.com/ops/logs/internal/ingest"
-	"git.apinb.com/ops/logs/internal/routers"
-	"github.com/gin-gonic/gin"
-)
-
-var ServiceKey = "Logs"
-
-func main() {
-	config.New(ServiceKey)
-	impl.NewImpl()
-
-	ingest.StartRefresher()
-	ingest.StartAlertDispatcher()
-	ingest.StartSyslogUDP()
-	ingest.StartTrapUDP()
-
-	app := gin.Default()
-	middleware.Mode(app)
-	app.Use(middleware.Cors())
-	app.Use(gin.Recovery())
-	app.HEAD("/", infra.Health)
-	routers.Register(ServiceKey, app)
-
-	if err := app.Run(fmt.Sprintf(":%s", config.Spec.Port)); err != nil {
-		panic(err)
-	}
-}
+package main
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"log"
+	"net/http"
+	"os"
+	"os/signal"
+	"sync"
+	"syscall"
+	"time"
+
+	"git.apinb.com/bsm-sdk/core/infra"
+	"git.apinb.com/bsm-sdk/core/middleware"
+	"git.apinb.com/ops/logs/internal/config"
+	"git.apinb.com/ops/logs/internal/impl"
+	"git.apinb.com/ops/logs/internal/ingest"
+	"git.apinb.com/ops/logs/internal/models"
+	"git.apinb.com/ops/logs/internal/routers"
+	settingssync "git.apinb.com/ops/logs/internal/systemsettings"
+	"github.com/gin-gonic/gin"
+)
+
+var ServiceKey = "Logs"
+
+func main() {
+	ctx, stop := signal.NotifyContext(context.Background(), os.Interrupt, syscall.SIGTERM)
+	defer stop()
+	var backgroundJobs sync.WaitGroup
+
+	config.New(ServiceKey)
+	settingsClient, err := settingssync.LoadInitial(ctx)
+	if err != nil {
+		panic(fmt.Errorf("logs: 初始化系统参数失败: %w", err))
+	}
+	impl.NewImpl()
+	if err := models.RequireSchemaVersion(impl.DBService, "logs"); err != nil {
+		panic(err)
+	}
+
+	if err := ingest.StartRefresher(ctx); err != nil {
+		shutdownAfterStartupFailure(stop, err)
+	}
+	ingest.StartAlertDispatcher(ctx)
+	if err := ingest.StartSyslogUDP(ctx); err != nil {
+		shutdownAfterStartupFailure(stop, err)
+	}
+	if err := ingest.StartTrapUDP(ctx); err != nil {
+		shutdownAfterStartupFailure(stop, err)
+	}
+	backgroundJobs.Add(1)
+	go func() {
+		defer backgroundJobs.Done()
+		settingssync.Start(ctx, settingsClient)
+	}()
+
+	app := gin.Default()
+	middleware.Mode(app)
+	app.Use(middleware.Cors())
+	app.Use(gin.Recovery())
+	app.HEAD("/", infra.Health)
+	routers.Register(ServiceKey, app)
+
+	server := &http.Server{Addr: config.Spec.Addr, Handler: app, ReadHeaderTimeout: 10 * time.Second}
+	errCh := make(chan error, 1)
+	go func() {
+		errCh <- server.ListenAndServe()
+	}()
+
+	var serveErr error
+	select {
+	case err := <-errCh:
+		if err != nil && !errors.Is(err, http.ErrServerClosed) {
+			serveErr = err
+		}
+	case <-ctx.Done():
+		shutdownCtx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
+		defer cancel()
+		if err := server.Shutdown(shutdownCtx); err != nil {
+			serveErr = err
+		}
+	}
+	stop()
+	if !ingest.Wait(20 * time.Second) {
+		serveErr = errors.Join(serveErr, fmt.Errorf("后台 worker 未在退出超时内结束"))
+	}
+	if !waitForBackgroundJobs(&backgroundJobs, 20*time.Second) {
+		serveErr = errors.Join(serveErr, fmt.Errorf("系统参数任务未在退出超时内结束"))
+	}
+	if err := errors.Join(serveErr, impl.Close()); err != nil {
+		log.Printf("logs: 服务退出失败: %v", err)
+		panic(err)
+	}
+}
+
+func waitForBackgroundJobs(wg *sync.WaitGroup, timeout time.Duration) bool {
+	done := make(chan struct{})
+	go func() {
+		wg.Wait()
+		close(done)
+	}()
+	select {
+	case <-done:
+		return true
+	case <-time.After(timeout):
+		return false
+	}
+}
+
+func shutdownAfterStartupFailure(stop context.CancelFunc, startErr error) {
+	stop()
+	if !ingest.Wait(20 * time.Second) {
+		startErr = errors.Join(startErr, fmt.Errorf("后台 worker 未在退出超时内结束"))
+	}
+	panic(errors.Join(startErr, impl.Close()))
+}