feat: Vuln debug logs

server/internal/vulnsched/sched.go

@@ -51,11 +51,17 @@ func Start(ctx context.Context, deps Deps) {
 
 	dir, err := os.MkdirTemp("", "vantage-vulndb-")
 	if err != nil {
-		log.Printf("vulnsched: temp dir: %v", err)
+		// The classic form of this is "stat /tmp: no such file or directory" on
+		// the scratch runtime image. It is logged once at boot while everything
+		// else runs normally, so the only other symptom is a fleet that never
+		// reports a finding.
+		log.Printf("vulnsched: temp dir: %v (scan loop NOT started)", err)
 		return
 	}
 
 	s := &scheduler{deps: deps, dir: dir}
+	log.Printf("vulnsched: started, ref=%s tick=%s dir=%s debug=%t",
+		vulndb.Ref(), tickInterval, dir, vulndb.DebugEnabled())
 
 	go func() {
 		defer os.RemoveAll(dir)
@@ -66,6 +72,7 @@ func Start(ctx context.Context, deps Deps) {
 		for {
 			select {
 			case <-ctx.Done():
+				log.Println("vulnsched: leadership lost or shutting down, scan loop stopping")
 				return
 			case <-ticker.C:
 				s.tick(ctx)
@@ -75,16 +82,22 @@ func Start(ctx context.Context, deps Deps) {
 }
 
 func (s *scheduler) tick(ctx context.Context) {
+	started := time.Now()
+	vulndb.Debugf("vulnsched tick starting (store loaded=%t, db version=%d, pulled %s ago)",
+		s.store != nil, s.version, time.Since(s.pulled).Round(time.Second))
+
 	if err := s.ensureDB(ctx); err != nil {
 		// Keep the last good database and carry on scanning against it. A
 		// network blip must never clear findings or read as "all fixed".
 		log.Printf("vulnsched: database unavailable: %v", err)
 		s.recordDBError(ctx, err)
 		if s.store == nil {
+			log.Println("vulnsched: no database loaded at all, nothing can be scanned this tick")
 			return
 		}
 	}
 	s.scanPending(ctx)
+	vulndb.Debugf("vulnsched tick finished in %s", time.Since(started).Round(time.Millisecond))
 }
 
 // ensureDB pulls a fresh database when the local copy is stale, and marks the
@@ -93,8 +106,11 @@ func (s *scheduler) tick(ctx context.Context) {
 // next agent report.
 func (s *scheduler) ensureDB(ctx context.Context) error {
 	if s.store != nil && time.Since(s.pulled) < dbMaxAge {
+		vulndb.Debugf("database is %s old, under the %s limit; not pulling",
+			time.Since(s.pulled).Round(time.Second), dbMaxAge)
 		return nil
 	}
+	log.Printf("vulnsched: pulling database (age %s, max %s)", time.Since(s.pulled).Round(time.Second), dbMaxAge)
 
 	version, err := vulndb.Pull(ctx, s.dir)
 	if err != nil {
@@ -110,6 +126,7 @@ func (s *scheduler) ensureDB(ctx context.Context) error {
 	s.pulled = time.Now()
 
 	changed := version != s.version
+	log.Printf("vulnsched: database ready, schema %d (previous %d, changed=%t)", version, s.version, changed)
 	s.version = version
 
 	_, _ = db.Col("vulndb_meta").UpdateOne(ctx, bson.M{},
@@ -152,6 +169,12 @@ func (s *scheduler) scanPending(ctx context.Context) {
 		return
 	}
 
+	if len(pending) == 0 {
+		vulndb.Debugf("no servers pending scan")
+		return
+	}
+	log.Printf("vulnsched: %d server(s) pending scan", len(pending))
+
 	// Newly opened findings are collected across the whole tick and sent as one
 	// digest per instance. A database refresh can open several hundred findings
 	// at once; one message per finding would rate-limit the webhook or get the
@@ -182,6 +205,8 @@ func (s *scheduler) scanPending(ctx context.Context) {
 
 func (s *scheduler) scanOne(ctx context.Context, sp models.ServerPackages) []models.VulnFinding {
 	now := time.Now()
+	log.Printf("vulnsched: scanning server %s (instance %s, os %s %s, %d packages)",
+		sp.ServerID, sp.InstanceID, sp.OS.Family, sp.OS.VersionID, len(sp.Packages))
 
 	results, err := vulndb.Match(s.store, sp.OS, sp.Packages)
 	if err != nil {
@@ -193,6 +218,9 @@ func (s *scheduler) scanOne(ctx context.Context, sp models.ServerPackages) []mod
 		if !errors.Is(err, vulndb.ErrUnsupportedFamily) {
 			log.Printf("vulnsched: scan %s: %v", sp.ServerID, err)
 			status = sp.Status
+		} else {
+			log.Printf("vulnsched: server %s marked unsupported: no feed for %s %s",
+				sp.ServerID, sp.OS.Family, sp.OS.VersionID)
 		}
 		s.clearPending(ctx, sp.ID, status, now)
 		return nil
@@ -210,6 +238,10 @@ func (s *scheduler) scanOne(ctx context.Context, sp models.ServerPackages) []mod
 		return nil
 	}
 
+	log.Printf("vulnsched: server %s scanned: %d matches, %d existing, %d upserts, %d newly opened, %d reopened, %d fixed",
+		sp.ServerID, len(results), len(existing), len(diff.Upserts),
+		len(diff.NewlyOpened), len(diff.ReopenIDs), len(diff.FixedIDs))
+
 	s.clearPending(ctx, sp.ID, models.ScanStatusOK, now)
 
 	for i := range diff.NewlyOpened {