Merge pull request #12104 from paul43210/fix/search-bleve-optimize

enhancement(search): optimize bleve index after bulk reindexing

changelog/unreleased/enhancement-search-optimize-after-indexspace.md

@@ -0,0 +1,11 @@
+Enhancement: Optimize search index after bulk reindexing
+
+After an `IndexSpace` walk completes, the search engine now triggers a
+segment merge (compaction) on the bleve index. Over time, writes create
+multiple index segments that degrade query performance. The new
+`Optimize()` method calls bleve's `ForceMerge` to consolidate all
+segments into one, improving subsequent search and lookup speed. This is
+especially beneficial after bulk reindexing large spaces.
+
+https://github.com/owncloud/ocis/pull/12104
+https://github.com/owncloud/ocis/issues/12093

services/search/pkg/engine/bleve.go

@@ -2,6 +2,7 @@ package engine
 
 import (
 	"context"
+	"fmt"
 	"math"
 	"path"
 	"reflect"
@@ -15,6 +16,7 @@ import (
 	"github.com/blevesearch/bleve/v2/analysis/token/porter"
 	"github.com/blevesearch/bleve/v2/analysis/tokenizer/single"
 	"github.com/blevesearch/bleve/v2/analysis/tokenizer/unicode"
+	"github.com/blevesearch/bleve/v2/index/scorch/mergeplan"
 	"github.com/blevesearch/bleve/v2/mapping"
 	"github.com/blevesearch/bleve/v2/search/query"
 	storageProvider "github.com/cs3org/go-cs3apis/cs3/storage/provider/v1beta1"
@@ -402,6 +404,33 @@ func (b *Bleve) Purge(id string) error {
 	return bleveIndex.Delete(id)
 }
 
+// Optimize triggers a force merge of the bleve index segments into a single
+// segment, improving query performance. This is an expensive I/O operation
+// and should be called during low-usage periods (e.g., after bulk indexing).
+func (b *Bleve) Optimize(ctx context.Context) error {
+	bleveIndex, closeFn, err := b.indexGetter.GetIndex()
+	if err != nil {
+		return err
+	}
+	defer closeFn()
+
+	internal, err := bleveIndex.Advanced()
+	if err != nil {
+		return fmt.Errorf("failed to access internal index: %w", err)
+	}
+
+	type forceMerger interface {
+		ForceMerge(ctx context.Context, mo *mergeplan.MergePlanOptions) error
+	}
+
+	fm, ok := internal.(forceMerger)
+	if !ok {
+		return fmt.Errorf("index implementation does not support force merge")
+	}
+
+	return fm.ForceMerge(ctx, nil)
+}
+
 // DocCount returns the number of resources in the index.
 func (b *Bleve) DocCount() (uint64, error) {
 	bleveIndex, closeFn, err := b.indexGetter.GetIndex(bleveEngine.ReadOnly(true))

services/search/pkg/engine/engine.go

@@ -30,6 +30,12 @@ type Engine interface {
 	DocCount() (uint64, error)
 }
 
+// Optimizer is an optional interface that Engine implementations may support
+// to trigger index compaction. Callers should type-assert before use.
+type Optimizer interface {
+	Optimize(ctx context.Context) error
+}
+
 // Resource is the entity that is stored in the index.
 type Resource struct {
 	content.Document

services/search/pkg/search/service.go

@@ -486,6 +486,13 @@ func (s *Service) IndexSpace(spaceID *provider.StorageSpaceId) error {
 
 	logDocCount(s.engine, s.logger)
 
+	if opt, ok := s.engine.(engine.Optimizer); ok {
+		s.logger.Info().Msg("optimizing search index after space walk")
+		if err := opt.Optimize(ownerCtx); err != nil {
+			s.logger.Warn().Err(err).Msg("index optimization failed")
+		}
+	}
+
 	return nil
 }