feat: put inverted index into ram (#35222) (#35223)

fix: https://github.com/milvus-io/milvus/issues/35224
pr: https://github.com/milvus-io/milvus/pull/35222

---------

Signed-off-by: longjiquan <jiquan.long@zilliz.com>

cmd/roles/roles.go

@@ -79,6 +79,10 @@ type component interface {
 	Stop() error
 }
 
+const (
+	TmpInvertedIndexPrefix = "/tmp/milvus/inverted-index/"
+)
+
 func cleanLocalDir(path string) {
 	_, statErr := os.Stat(path)
 	// path exist, but stat error
@@ -200,6 +204,7 @@ func (mr *MilvusRoles) runQueryNode(ctx context.Context, localMsg bool, wg *sync
 	if len(mmapDir) > 0 {
 		cleanLocalDir(mmapDir)
 	}
+	cleanLocalDir(TmpInvertedIndexPrefix)
 
 	return runComponent(ctx, localMsg, wg, components.NewQueryNode, metrics.RegisterQueryNode)
 }
@@ -224,6 +229,7 @@ func (mr *MilvusRoles) runIndexNode(ctx context.Context, localMsg bool, wg *sync
 	rootPath := paramtable.Get().LocalStorageCfg.Path.GetValue()
 	indexDataLocalPath := filepath.Join(rootPath, typeutil.IndexNodeRole)
 	cleanLocalDir(indexDataLocalPath)
+	cleanLocalDir(TmpInvertedIndexPrefix)
 
 	return runComponent(ctx, localMsg, wg, components.NewIndexNode, metrics.RegisterIndexNode)
 }

internal/core/src/index/InvertedIndexTantivy.cpp

@@ -23,6 +23,8 @@
 #include "InvertedIndexTantivy.h"
 
 namespace milvus::index {
+constexpr const char* TMP_INVERTED_INDEX_PREFIX = "/tmp/milvus/inverted-index/";
+
 inline TantivyDataType
 get_tantivy_data_type(proto::schema::DataType data_type) {
     switch (data_type) {
@@ -71,8 +73,8 @@ InvertedIndexTantivy<T>::InvertedIndexTantivy(
     disk_file_manager_ = std::make_shared<DiskFileManager>(ctx, ctx.space_);
     auto field =
         std::to_string(disk_file_manager_->GetFieldDataMeta().field_id);
-    auto prefix = disk_file_manager_->GetLocalIndexObjectPrefix();
-    path_ = prefix;
+    auto prefix = disk_file_manager_->GetIndexIdentifier();
+    path_ = std::string(TMP_INVERTED_INDEX_PREFIX) + prefix;
     boost::filesystem::create_directories(path_);
     d_type_ = get_tantivy_data_type(schema_);
     if (tantivy_index_exist(path_.c_str())) {

internal/core/src/storage/DiskFileManagerImpl.cpp

@@ -864,6 +864,12 @@ DiskFileManagerImpl::GetLocalIndexObjectPrefix() {
         local_chunk_manager, index_meta_.build_id, index_meta_.index_version);
 }
 
+std::string
+DiskFileManagerImpl::GetIndexIdentifier() {
+    return GenIndexPathIdentifier(index_meta_.build_id,
+                                  index_meta_.index_version);
+}
+
 std::string
 DiskFileManagerImpl::GetLocalRawDataObjectPrefix() {
     auto local_chunk_manager =

internal/core/src/storage/DiskFileManagerImpl.h

@@ -61,6 +61,9 @@ class DiskFileManagerImpl : public FileManagerImpl {
     std::string
     GetLocalIndexObjectPrefix();
 
+    std::string
+    GetIndexIdentifier();
+
     std::string
     GetLocalRawDataObjectPrefix();
 

internal/core/src/storage/Util.cpp

@@ -438,12 +438,17 @@ GetDimensionFromArrowArray(std::shared_ptr<arrow::Array> data,
     }
 }
 
+std::string
+GenIndexPathIdentifier(int64_t build_id, int64_t index_version) {
+    return std::to_string(build_id) + "/" + std::to_string(index_version) + "/";
+}
+
 std::string
 GenIndexPathPrefix(ChunkManagerPtr cm,
                    int64_t build_id,
                    int64_t index_version) {
     return cm->GetRootPath() + "/" + std::string(INDEX_ROOT_PATH) + "/" +
-           std::to_string(build_id) + "/" + std::to_string(index_version) + "/";
+           GenIndexPathIdentifier(build_id, index_version);
 }
 
 std::string

internal/core/src/storage/Util.h

@@ -74,6 +74,9 @@ GetDimensionFromArrowArray(std::shared_ptr<arrow::Array> array,
 std::string
 GetIndexPathPrefixWithBuildID(ChunkManagerPtr cm, int64_t build_id);
 
+std::string
+GenIndexPathIdentifier(int64_t build_id, int64_t index_version);
+
 std::string
 GenIndexPathPrefix(ChunkManagerPtr cm, int64_t build_id, int64_t index_version);