main.go

package main

import (
	"context"
	"errors"
	"log/slog"
	"os"
	"os/signal"
	"runtime"
	"syscall"
	"time"

	"github.com/krateoplatformops/plumbing/kubeutil"
	"github.com/krateoplatformops/plumbing/server/probes"
	"github.com/krateoplatformops/resources-ingester/internal/batch"
	"github.com/krateoplatformops/resources-ingester/internal/config"
	"github.com/krateoplatformops/resources-ingester/internal/manager"
	"github.com/krateoplatformops/resources-ingester/internal/queue"
	"github.com/krateoplatformops/resources-ingester/internal/router"
	"github.com/krateoplatformops/resources-ingester/internal/storage"
	"github.com/krateoplatformops/resources-ingester/internal/storage/pgcheck"
	"github.com/krateoplatformops/resources-ingester/internal/telemetry"

	"k8s.io/client-go/dynamic"
	"k8s.io/client-go/rest"

	"github.com/krateoplatformops/plumbing/eventbus"
)

func main() {
	cfg := config.Setup()

	rootCtx, stop := signal.NotifyContext(context.Background(), os.Interrupt, syscall.SIGTERM)
	defer stop()

	metrics, shutdownMetrics, err := telemetry.Setup(rootCtx, cfg.Log, telemetry.Config{
		Enabled:        cfg.OTelEnabled,
		ServiceName:    "resources-ingester",
		ExportInterval: cfg.OTelExportIntv,
	})
	if err != nil {
		cfg.Log.Error("cannot initialize OpenTelemetry metrics", slog.Any("err", err))
		os.Exit(1)
	}
	defer func() {
		if err := shutdownMetrics(context.Background()); err != nil {
			cfg.Log.Error("OpenTelemetry metrics shutdown failed", slog.Any("err", err))
		}
	}()

	pgCtx, cancel := context.WithTimeout(rootCtx, cfg.DbReadyTimeout)
	defer cancel()

	pool, err := pgcheck.WaitForPostgres(pgCtx, cfg.Log, cfg.DbURL)
	if err != nil {
		switch {
		case errors.Is(err, pgcheck.ErrTableNotFound):
			cfg.Log.Error("krateo_resources table missing", slog.Any("err", err))
		case errors.Is(err, pgcheck.ErrSchemaMismatch):
			var detail *pgcheck.SchemaMismatchError
			errors.As(err, &detail)
			cfg.Log.Error("krateo_resources schema mismatch",
				slog.Any("missing", detail.Missing),
				slog.Any("extra", detail.Extra),
				slog.Any("mistyped", detail.Mistyped),
			)
		default:
			cfg.Log.Error("cannot connect to PostgreSQL", slog.Any("err", err))
		}
		os.Exit(1)
	}
	defer pool.Close()
	cfg.Log.Info("PostgreSQL is ready.")

	// Health probes server
	hs := probes.New(cfg.Log, pool, cfg.Port)
	hs.Start()

	// pool := &pgxpool.Pool{} // Mock database, not used yet

	restConfig, err := rest.InClusterConfig()
	if err != nil {
		cfg.Log.Error("cannot get in-cluster config", slog.Any("err", err))
		os.Exit(1)
	}

	client, err := dynamic.NewForConfig(restConfig)
	if err != nil {
		cfg.Log.Error("cannot create k8s client", slog.Any("err", err))
		os.Exit(1)
	}

	clusterName := kubeutil.DetectClusterName(restConfig)
	cfg.Log.Info("cluster name detected", slog.String("cluster", clusterName))

	// Record channel and batch worker
	recordChan := make(chan batch.InsertRecord, 100)
	batchWorker := batch.NewWorker(batch.WorkerOpts{
		Pool:       pool,
		Log:        cfg.Log,
		Input:      recordChan,
		MaxBatch:   50,
		FlushEvery: 5 * time.Second,
		Metrics:    metrics,
	})
	go batchWorker.Run(rootCtx.Done())

	// Queue with worker pool
	jobQueue := queue.NewQueue(1000, 4)
	jobQueue.Run()
	defer jobQueue.Terminate()

	// EventBus for components comunication
	eventbus := eventbus.New()

	// Storage event reader
	storageManager, err := storage.NewManager(storage.StorageOpts{
		RESTConfig:  restConfig,
		Pool:        pool,
		Eventbus:    eventbus,
		Log:         cfg.Log,
		Queue:       jobQueue,
		RecordChan:  recordChan,
		ClusterName: clusterName,
	})
	if err != nil {
		cfg.Log.Error("cannot create manager", slog.Any("err", err))
		os.Exit(1)
	}
	go storageManager.Run(rootCtx.Done())

	// Ingester
	ing, err := router.NewIngester(router.IngesterOpts{
		Log:      cfg.Log,
		EventBus: eventbus,
	})
	if err != nil {
		cfg.Log.Error("cannot create ingester", slog.Any("err", err))
		os.Exit(1)
	}

	q := router.NewSharedQueue(5 * time.Minute)
	wgpool := router.NewWorkerPool(
		q,
		10,
		ing,
		cfg.Log,
		metrics,
	)
	go wgpool.Start(rootCtx.Done())

	// Informer Manager
	infManager, err := manager.NewManager(manager.ManagerOpts{
		DynamicClient:  client,
		Namespaces:     cfg.Namespaces,
		ResyncInterval: 8 * time.Hour,   // TODO make configurable
		ThrottlePeriod: 5 * time.Minute, // TODO make configurable
		Eventbus:       eventbus,
		Log:            cfg.Log,
		Handler:        ing,
		Queue:          q,
		WgPool:         wgpool,
	})
	if err != nil {
		cfg.Log.Error("cannot create manager", slog.Any("err", err))
		os.Exit(1)
	}
	go infManager.Run(rootCtx.Done())

	// Static Routers from internal/router/ (including CRDs)
	router.NewStaticRouters(rootCtx, client, cfg.Log, ing, q, wgpool, router.MustLoad("static"))

	go func() {
		ticker := time.NewTicker(5 * time.Second)
		defer ticker.Stop()

		for {
			select {
			case <-rootCtx.Done():
				return
			case <-ticker.C:
				metrics.SetRecordChannelDepth(int64(len(recordChan)))
				metrics.SetQueueDepth(int64(jobQueue.GetJobCount()))
				metrics.SetGoRoutines(int64(runtime.NumGoroutine()))
				metrics.SetActiveInformerKinds(int64(infManager.GetInformers()))
			}
		}
	}()

	// Monitor buffer
	go func() {
		ticker := time.NewTicker(50 * time.Second)
		defer ticker.Stop()

		for {
			select {
			case <-rootCtx.Done():
				return
			case <-ticker.C:
				cfg.Log.Info("Pipeline status",
					slog.Int("recordChan", len(recordChan)),
					slog.Int("queueJobs", jobQueue.GetJobCount()),
					slog.Int("activeInformerKinds", infManager.GetInformers()),
					slog.Int("activeGoRoutines", runtime.NumGoroutine()),
				)
			}
		}
	}()

	cfg.Log.Info("Event ingester started")

	<-rootCtx.Done()
	cfg.Log.Info("Shutting down Event ingester")
	q.ShutDown()
	wgpool.Wait()

	shutdownCtx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
	defer cancel()

	if err := hs.Shutdown(shutdownCtx); err != nil {
		cfg.Log.Error("Health server shutdown failed", slog.Any("err", err))
	}

}