fix: fix directory confusion between indexnode and querynode

Signed-off-by: luzhang <luzhang@zilliz.com>

internal/core/src/common/Common.h

@@ -30,6 +30,32 @@ extern int64_t LOW_PRIORITY_THREAD_CORE_COEFFICIENT;
 extern int CPU_NUM;
 extern int64_t EXEC_EVAL_EXPR_BATCH_SIZE;
 
+enum class Role { QueryNode, IndexNode };
+
+const std::unordered_map<Role, std::string> RoleToStringMap = {
+    {Role::QueryNode, "querynode"}, {Role::IndexNode, "indexnode"}};
+
+// convert role to string
+inline std::string
+ToString(Role role) {
+    auto it = RoleToStringMap.find(role);
+    if (it != RoleToStringMap.end()) {
+        return it->second;
+    }
+    PanicInfo(UnexpectedError, "role {} not found", int(role));
+}
+
+// convert string to role
+inline Role
+FromString(const std::string& role_str) {
+    for (const auto& pair : RoleToStringMap) {
+        if (pair.second == role_str) {
+            return pair.first;
+        }
+    }
+    PanicInfo(UnexpectedError, "role {} not found", role_str);
+}
+
 void
 SetIndexSliceSize(const int64_t size);
 

internal/core/src/index/Index.h

@@ -71,12 +71,21 @@ class IndexBase {
         return index_type_;
     }
 
+    bool
+    IsLoadingIndex() const {
+        return is_loading_index_;
+    }
+
  protected:
     explicit IndexBase(IndexType index_type)
         : index_type_(std::move(index_type)) {
     }
 
     IndexType index_type_ = "";
+
+    // current index obj is used to build and load index in one class,
+    // this flag is used to indicate whether building index or loading index
+    bool is_loading_index_{false};
 };
 
 using IndexBasePtr = std::unique_ptr<IndexBase>;

internal/core/src/index/InvertedIndexTantivy.cpp

@@ -92,10 +92,11 @@ InvertedIndexTantivy<T>::InvertedIndexTantivy(
     const storage::FileManagerContext& ctx)
     : ScalarIndex<T>(INVERTED_INDEX_TYPE),
       schema_(ctx.fieldDataMeta.field_schema) {
+    this->is_loading_index_ = ctx.for_loading_index;
     mem_file_manager_ = std::make_shared<MemFileManager>(ctx);
     disk_file_manager_ = std::make_shared<DiskFileManager>(ctx);
     // push init wrapper to load process
-    if (ctx.for_loading_index) {
+    if (this->is_loading_index_) {
         return;
     }
     InitForBuildIndex();
@@ -107,7 +108,7 @@ InvertedIndexTantivy<T>::~InvertedIndexTantivy() {
         wrapper_->free();
     }
     auto local_chunk_manager =
-        storage::LocalChunkManagerSingleton::GetInstance().GetChunkManager();
+        storage::LocalChunkManagerFactory::GetInstance().GetChunkManager();
     auto prefix = path_;
     LOG_INFO("inverted index remove path:{}", path_);
     local_chunk_manager->RemoveDir(prefix);

internal/core/src/index/VectorDiskIndex.cpp

@@ -46,18 +46,24 @@ VectorDiskAnnIndex<T>::VectorDiskAnnIndex(
     file_manager_ =
         std::make_shared<storage::DiskFileManagerImpl>(file_manager_context);
     AssertInfo(file_manager_ != nullptr, "create file manager failed!");
-    auto local_chunk_manager =
-        storage::LocalChunkManagerSingleton::GetInstance().GetChunkManager();
-    auto local_index_path_prefix = file_manager_->GetLocalIndexObjectPrefix();
 
-    // As we have guarded dup-load in QueryNode,
-    // this assertion failed only if the Milvus rebooted in the same pod,
-    // need to remove these files then re-load the segment
-    if (local_chunk_manager->Exist(local_index_path_prefix)) {
-        local_chunk_manager->RemoveDir(local_index_path_prefix);
+    is_loading_index_ = file_manager_context.for_loading_index;
+    if (is_loading_index_) {
+        // As we have guarded dup-load in QueryNode,
+        // this assertion failed only if the Milvus rebooted in the same pod,
+        // need to remove these files then re-load the segment
+        auto local_chunk_manager =
+            storage::LocalChunkManagerFactory::GetInstance().GetChunkManager();
+        auto local_index_path_prefix =
+            file_manager_->GetLocalIndexObjectPrefix();
+
+        if (local_chunk_manager->Exist(local_index_path_prefix)) {
+            local_chunk_manager->RemoveDir(local_index_path_prefix);
+        }
+        CheckCompatible(version);
+        local_chunk_manager->CreateDir(local_index_path_prefix);
     }
-    CheckCompatible(version);
-    local_chunk_manager->CreateDir(local_index_path_prefix);
+
     auto diskann_index_pack =
         knowhere::Pack(std::shared_ptr<knowhere::FileManager>(file_manager_));
     auto get_index_obj = knowhere::IndexFactory::Instance().Create<T>(
@@ -135,7 +141,8 @@ template <typename T>
 void
 VectorDiskAnnIndex<T>::Build(const Config& config) {
     auto local_chunk_manager =
-        storage::LocalChunkManagerSingleton::GetInstance().GetChunkManager();
+        storage::LocalChunkManagerFactory::GetInstance().GetChunkManager(
+            milvus::Role::IndexNode);
     knowhere::Json build_config;
     build_config.update(config);
 
@@ -185,7 +192,8 @@ void
 VectorDiskAnnIndex<T>::BuildWithDataset(const DatasetPtr& dataset,
                                         const Config& config) {
     auto local_chunk_manager =
-        storage::LocalChunkManagerSingleton::GetInstance().GetChunkManager();
+        storage::LocalChunkManagerFactory::GetInstance().GetChunkManager(
+            milvus::Role::IndexNode);
     knowhere::Json build_config;
     build_config.update(config);
     // set data path
@@ -353,10 +361,14 @@ template <typename T>
 void
 VectorDiskAnnIndex<T>::CleanLocalData() {
     auto local_chunk_manager =
-        storage::LocalChunkManagerSingleton::GetInstance().GetChunkManager();
-    local_chunk_manager->RemoveDir(file_manager_->GetLocalIndexObjectPrefix());
-    local_chunk_manager->RemoveDir(
-        file_manager_->GetLocalRawDataObjectPrefix());
+        storage::LocalChunkManagerFactory::GetInstance().GetChunkManager();
+    if (is_loading_index_) {
+        local_chunk_manager->RemoveDir(
+            file_manager_->GetLocalIndexObjectPrefix());
+    } else {
+        local_chunk_manager->RemoveDir(
+            file_manager_->GetLocalRawDataObjectPrefix());
+    }
 }
 
 template <typename T>

internal/core/src/segcore/load_index_c.cpp

@@ -15,6 +15,7 @@
 #include "common/EasyAssert.h"
 #include "common/Types.h"
 #include "common/type_c.h"
+#include "common/Common.h"
 #include "index/Index.h"
 #include "index/IndexFactory.h"
 #include "index/Meta.h"
@@ -156,7 +157,7 @@ appendVecIndex(CLoadIndexInfo c_load_index_info, CBinarySet c_binary_set) {
         config["index_files"] = load_index_info->index_files;
 
         milvus::storage::FileManagerContext fileManagerContext(
-            field_meta, index_meta, remote_chunk_manager);
+            field_meta, index_meta, remote_chunk_manager, true);
         fileManagerContext.set_for_loading_index(true);
 
         load_index_info->index =
@@ -424,8 +425,8 @@ CleanLoadedIndex(CLoadIndexInfo c_load_index_info) {
         auto load_index_info =
             (milvus::segcore::LoadIndexInfo*)c_load_index_info;
         auto local_chunk_manager =
-            milvus::storage::LocalChunkManagerSingleton::GetInstance()
-                .GetChunkManager();
+            milvus::storage::LocalChunkManagerFactory::GetInstance()
+                .GetChunkManager(milvus::Role::QueryNode);
         auto index_file_path_prefix =
             milvus::storage::GenIndexPathPrefix(local_chunk_manager,
                                                 load_index_info->index_build_id,

internal/core/src/segcore/segment_c.cpp

@@ -466,6 +466,7 @@ LoadTextIndex(CSegmentInterface c_segment,
 
         milvus::storage::FileManagerContext ctx(
             field_meta, index_meta, remote_chunk_manager);
+        ctx.set_for_loading_index(true);
 
         auto index = std::make_unique<milvus::index::TextMatchIndex>(ctx);
         index->Load(config);

internal/core/src/storage/DiskFileManagerImpl.cpp

@@ -48,13 +48,18 @@ namespace milvus::storage {
 DiskFileManagerImpl::DiskFileManagerImpl(
     const FileManagerContext& fileManagerContext)
     : FileManagerImpl(fileManagerContext.fieldDataMeta,
-                      fileManagerContext.indexMeta) {
+                      fileManagerContext.indexMeta,
+                      fileManagerContext.for_loading_index) {
     rcm_ = fileManagerContext.chunkManagerPtr;
 }
 
 DiskFileManagerImpl::~DiskFileManagerImpl() {
     auto local_chunk_manager =
-        LocalChunkManagerSingleton::GetInstance().GetChunkManager();
+        for_loading_index_
+            ? LocalChunkManagerFactory::GetInstance().GetChunkManager(
+                  milvus::Role::QueryNode)
+            : LocalChunkManagerFactory::GetInstance().GetChunkManager(
+                  milvus::Role::IndexNode);
     local_chunk_manager->RemoveDir(GetIndexPathPrefixWithBuildID(
         local_chunk_manager, index_meta_.build_id));
 }
@@ -82,7 +87,7 @@ DiskFileManagerImpl::GetRemoteTextLogPath(const std::string& file_name,
 bool
 DiskFileManagerImpl::AddFile(const std::string& file) noexcept {
     auto local_chunk_manager =
-        LocalChunkManagerSingleton::GetInstance().GetChunkManager();
+        LocalChunkManagerFactory::GetInstance().GetChunkManager();
     FILEMANAGER_TRY
     if (!local_chunk_manager->Exist(file)) {
         LOG_ERROR("local file {} not exists", file);
@@ -134,7 +139,7 @@ DiskFileManagerImpl::AddFile(const std::string& file) noexcept {
 bool
 DiskFileManagerImpl::AddTextLog(const std::string& file) noexcept {
     auto local_chunk_manager =
-        LocalChunkManagerSingleton::GetInstance().GetChunkManager();
+        LocalChunkManagerFactory::GetInstance().GetChunkManager();
     FILEMANAGER_TRY
     if (!local_chunk_manager->Exist(file)) {
         LOG_ERROR("local file {} not exists", file);
@@ -190,7 +195,7 @@ DiskFileManagerImpl::AddBatchIndexFiles(
     const std::vector<std::string>& remote_files,
     const std::vector<int64_t>& remote_file_sizes) {
     auto local_chunk_manager =
-        LocalChunkManagerSingleton::GetInstance().GetChunkManager();
+        LocalChunkManagerFactory::GetInstance().GetChunkManager();
     auto& pool = ThreadPools::GetThreadPool(milvus::ThreadPoolPriority::HIGH);
 
     std::vector<std::future<std::shared_ptr<uint8_t[]>>> futures;
@@ -239,7 +244,7 @@ void
 DiskFileManagerImpl::CacheIndexToDisk(
     const std::vector<std::string>& remote_files) {
     auto local_chunk_manager =
-        LocalChunkManagerSingleton::GetInstance().GetChunkManager();
+        LocalChunkManagerFactory::GetInstance().GetChunkManager();
 
     std::map<std::string, std::vector<int>> index_slices;
     for (auto& file_path : remote_files) {
@@ -307,7 +312,7 @@ void
 DiskFileManagerImpl::CacheTextLogToDisk(
     const std::vector<std::string>& remote_files) {
     auto local_chunk_manager =
-        LocalChunkManagerSingleton::GetInstance().GetChunkManager();
+        LocalChunkManagerFactory::GetInstance().GetChunkManager();
 
     std::map<std::string, std::vector<int>> index_slices;
     for (auto& file_path : remote_files) {
@@ -376,7 +381,8 @@ DiskFileManagerImpl::CacheRawDataToDisk(std::vector<std::string> remote_files) {
     auto field_id = GetFieldDataMeta().field_id;
 
     auto local_chunk_manager =
-        LocalChunkManagerSingleton::GetInstance().GetChunkManager();
+        LocalChunkManagerFactory::GetInstance().GetChunkManager(
+            milvus::Role::IndexNode);
     std::string local_data_path;
     bool file_created = false;
 
@@ -619,7 +625,8 @@ DiskFileManagerImpl::CacheOptFieldToDisk(OptFieldT& fields_map) {
     auto segment_id = GetFieldDataMeta().segment_id;
     auto vec_field_id = GetFieldDataMeta().field_id;
     auto local_chunk_manager =
-        LocalChunkManagerSingleton::GetInstance().GetChunkManager();
+        LocalChunkManagerFactory::GetInstance().GetChunkManager(
+            milvus::Role::IndexNode);
     auto local_data_path = storage::GenFieldRawDataPathPrefix(
                                local_chunk_manager, segment_id, vec_field_id) +
                            std::string(VEC_OPT_FIELDS);
@@ -696,15 +703,17 @@ DiskFileManagerImpl::GetFileName(const std::string& localfile) {
 std::string
 DiskFileManagerImpl::GetLocalIndexObjectPrefix() {
     auto local_chunk_manager =
-        LocalChunkManagerSingleton::GetInstance().GetChunkManager();
+        LocalChunkManagerFactory::GetInstance().GetChunkManager(
+            milvus::Role::QueryNode);
     return GenIndexPathPrefix(
         local_chunk_manager, index_meta_.build_id, index_meta_.index_version);
 }
 
 std::string
 DiskFileManagerImpl::GetLocalTextIndexPrefix() {
     auto local_chunk_manager =
-        LocalChunkManagerSingleton::GetInstance().GetChunkManager();
+        LocalChunkManagerFactory::GetInstance().GetChunkManager(
+            milvus::Role::QueryNode);
     return GenTextIndexPathPrefix(local_chunk_manager,
                                   index_meta_.build_id,
                                   index_meta_.index_version,
@@ -729,7 +738,8 @@ DiskFileManagerImpl::GetTextIndexIdentifier() {
 std::string
 DiskFileManagerImpl::GetLocalRawDataObjectPrefix() {
     auto local_chunk_manager =
-        LocalChunkManagerSingleton::GetInstance().GetChunkManager();
+        LocalChunkManagerFactory::GetInstance().GetChunkManager(
+            milvus::Role::IndexNode);
     return GenFieldRawDataPathPrefix(
         local_chunk_manager, field_meta_.segment_id, field_meta_.field_id);
 }
@@ -744,7 +754,7 @@ std::optional<bool>
 DiskFileManagerImpl::IsExisted(const std::string& file) noexcept {
     bool isExist = false;
     auto local_chunk_manager =
-        LocalChunkManagerSingleton::GetInstance().GetChunkManager();
+        LocalChunkManagerFactory::GetInstance().GetChunkManager();
     try {
         isExist = local_chunk_manager->Exist(file);
     } catch (std::exception& e) {

internal/core/src/storage/FileManager.h

@@ -43,6 +43,16 @@ struct FileManagerContext {
           chunkManagerPtr(chunkManagerPtr) {
     }
 
+    explicit FileManagerContext(const FieldDataMeta& fieldDataMeta,
+                                const IndexMeta& indexMeta,
+                                const ChunkManagerPtr& chunkManagerPtr,
+                                bool for_loading)
+        : fieldDataMeta(fieldDataMeta),
+          indexMeta(indexMeta),
+          chunkManagerPtr(chunkManagerPtr),
+          for_loading_index(for_loading) {
+    }
+
     bool
     Valid() const {
         return chunkManagerPtr != nullptr;
@@ -74,8 +84,11 @@ struct FileManagerContext {
 class FileManagerImpl : public knowhere::FileManager {
  public:
     explicit FileManagerImpl(const FieldDataMeta& field_mata,
-                             IndexMeta index_meta)
-        : field_meta_(field_mata), index_meta_(std::move(index_meta)) {
+                             IndexMeta index_meta,
+                             bool for_loading_index)
+        : field_meta_(field_mata),
+          index_meta_(std::move(index_meta)),
+          for_loading_index_(for_loading_index) {
     }
 
  public:
@@ -176,6 +189,9 @@ class FileManagerImpl : public knowhere::FileManager {
     // index meta
     IndexMeta index_meta_;
     ChunkManagerPtr rcm_;
+
+    // indicate whether file manager is used for building index or load index
+    bool for_loading_index_{false};
 };
 
 using FileManagerImplPtr = std::shared_ptr<FileManagerImpl>;