pkg/server/module_server_test.go (1)

35-38: Tracking the TODO to fix this test for PostgreSQL.

With both sqlite3 and postgres skipped, this integration test now only runs against MySQL. Consider creating a tracking issue so this doesn't remain skipped indefinitely.

Would you like me to open an issue to track re-enabling this test for PostgreSQL?

pkg/storage/unified/resource/search.go (1)

186-192: Pre-existing data race on totalBatchesIndexed when initWorkers > 1.

This isn't introduced by this PR, but totalBatchesIndexed++ (line 189) is executed concurrently inside group.Go goroutines without synchronization. If s.initWorkers > 1, this is a data race detectable by -race. Consider using atomic.AddInt64 or moving the count outside the goroutine.

♻️ Suggested fix using atomic

-	totalBatchesIndexed := 0
+	var totalBatchesIndexed atomic.Int64
 	group := errgroup.Group{}
 	group.SetLimit(s.initWorkers)
 ...
 	for _, info := range stats {
 		group.Go(func() error {
 			s.log.Debug("initializing search index", "namespace", info.Namespace, "group", info.Group, "resource", info.Resource)
-			totalBatchesIndexed++
+			totalBatchesIndexed.Add(1)
 			_, _, err = s.build(ctx, info.NamespacedResource, info.Count, info.ResourceVersion)
 			return err
 		})
 	}
 ...
-	span.AddEvent("namespaces indexed", trace.WithAttributes(attribute.Int("namespaced_indexed", totalBatchesIndexed)))
+	span.AddEvent("namespaces indexed", trace.WithAttributes(attribute.Int64("namespaced_indexed", totalBatchesIndexed.Load())))

pkg/storage/unified/search/bleve.go (2)

144-154: Pre-existing: TotalDocs iterates b.cache without holding cacheMu.

Now that BuildIndex writes to b.cache under cacheMu, reading it without a lock in TotalDocs is a data race if the two can overlap. Consider adding a read lock here.

🔒 Proposed fix

 func (b *bleveBackend) TotalDocs() int64 {
+	b.cacheMu.RLock()
+	defer b.cacheMu.RUnlock()
 	var totalDocs int64
 	for _, v := range b.cache {

---

99-101: Silenced the log and left a TODO — consider checking the existing index.

If bleve.New(dir, mapper) succeeds on a path that already has data, it creates a new index (overwriting). The TODO about checking last RV to detect changes is worth tracking.

Would you like me to open an issue to track the RV-based skip optimization for file-based indexes?