fix: merge datanode bm25 error after reload growing segment with no d…

…ata (#37154) Segment with numrow 0 don't init bm25 stats, cause flush with bm25 stats failed. relate: #37150 Signed-off-by: aoiasd <[email protected]>
milvus-io · Oct 25, 2024 · fd72151 · fd72151
1 parent 05f8807
commit fd72151
Show file tree

Hide file tree

Showing 2 changed files with 13 additions and 1 deletion.
diff --git a/internal/flushcommon/metacache/actions.go b/internal/flushcommon/metacache/actions.go
@@ -161,6 +161,9 @@ func RollStats(newStats ...*storage.PrimaryKeyStats) SegmentAction {
 
 func MergeBm25Stats(newStats map[int64]*storage.BM25Stats) SegmentAction {
 	return func(info *SegmentInfo) {
+		if info.bm25stats == nil {
+			info.bm25stats = NewEmptySegmentBM25Stats()
+		}
 		info.bm25stats.Merge(newStats)
 	}
 }

diff --git a/internal/flushcommon/syncmgr/storage_serializer.go b/internal/flushcommon/syncmgr/storage_serializer.go
@@ -131,7 +131,7 @@ func (s *storageV1Serializer) EncodeBuffer(ctx context.Context, pack *SyncPack)
 			}
 			task.mergedStatsBlob = mergedStatsBlob
 
-			if len(pack.bm25Stats) > 0 {
+			if hasBM25Function(s.schema) {
 				mergedBM25Blob, err := s.serializeMergedBM25Stats(pack)
 				if err != nil {
 					log.Warn("failed to serialize merged bm25 stats log", zap.Error(err))
@@ -311,3 +311,12 @@ func (s *storageV1Serializer) serializeDeltalog(pack *SyncPack) (*storage.Blob,
 	writer.Close()
 	return finalizer()
 }
+
+func hasBM25Function(schema *schemapb.CollectionSchema) bool {
+	for _, function := range schema.GetFunctions() {
+		if function.GetType() == schemapb.FunctionType_BM25 {
+			return true
+		}
+	}
+	return false
+}