feat: impl StructArray -- support create index for vector array (embedding list) and search on it (#43726)

Ref https://github.com/milvus-io/milvus/issues/42148

This PR supports create index for vector array (now, only for
`DataType.FLOAT_VECTOR`) and search on it.
The index type supported in this PR is `EMB_LIST_HNSW` and the metric
type is `MAX_SIM` only.

The way to use it:
```python
milvus_client = MilvusClient("xxx:19530")
schema = milvus_client.create_schema(enable_dynamic_field=True, auto_id=True)
...
struct_schema = milvus_client.create_struct_array_field_schema("struct_array_field")
...
struct_schema.add_field("struct_float_vec", DataType.ARRAY_OF_VECTOR, element_type=DataType.FLOAT_VECTOR, dim=128, max_capacity=1000)
...
schema.add_struct_array_field(struct_schema)
index_params = milvus_client.prepare_index_params()
index_params.add_index(field_name="struct_float_vec", index_type="EMB_LIST_HNSW", metric_type="MAX_SIM", index_params={"nlist": 128})
...
milvus_client.create_index(COLLECTION_NAME, schema=schema, index_params=index_params)
```

Note: This PR uses `Lims` to convey offsets of the vector array to
knowhere where vectors of multiple vector arrays are concatenated and we
need offsets to specify which vectors belong to which vector array.

---------

Signed-off-by: SpadeA <tangchenjie1210@gmail.com>
Signed-off-by: SpadeA-Tang <tangchenjie1210@gmail.com>

internal/core/src/common/Chunk.h

@@ -395,6 +395,14 @@ class VectorArrayChunk : public Chunk {
           dim_(dim),
           element_type_(element_type) {
         offsets_lens_ = reinterpret_cast<uint32_t*>(data);
+
+        auto offset = 0;
+        lims_.reserve(row_nums_ + 1);
+        lims_.push_back(offset);
+        for (int64_t i = 0; i < row_nums_; i++) {
+            offset += offsets_lens_[i * 2 + 1];
+            lims_.push_back(offset);
+        }
     }
 
     VectorArrayView
@@ -424,10 +432,23 @@ class VectorArrayChunk : public Chunk {
                   "VectorArrayChunk::ValueAt is not supported");
     }
 
+    const char*
+    Data() const override {
+        return data_ + offsets_lens_[0];
+    }
+
+    const size_t*
+    Lims() const {
+        return lims_.data();
+    }
+
  private:
     int64_t dim_;
     uint32_t* offsets_lens_;
     milvus::DataType element_type_;
+    // The name 'Lims' is consistent with knowhere::DataSet::SetLims which describes the number of vectors
+    // in each vector array (embedding list). This is needed as vectors are flattened in the chunk.
+    std::vector<size_t> lims_;
 };
 
 class SparseFloatVectorChunk : public Chunk {

internal/core/src/common/FieldData.h

@@ -92,6 +92,15 @@ class FieldData<VectorArray> : public FieldDataVectorArrayImpl {
         ThrowInfo(Unsupported,
                   "Call get_dim on FieldData<VectorArray> is not supported");
     }
+
+    const VectorArray*
+    value_at(ssize_t offset) const {
+        AssertInfo(offset < get_num_rows(),
+                   "field data subscript out of range");
+        AssertInfo(offset < length(),
+                   "subscript position don't has valid value");
+        return &data_[offset];
+    }
 };
 
 template <>

internal/core/src/common/TypeTraits.h

@@ -47,6 +47,7 @@ constexpr bool IsVariableType =
     IsSparse<T> || std::is_same_v<T, VectorArray> ||
     std::is_same_v<T, VectorArrayView>;
 
+// todo(SpadeA): support vector array
 template <typename T>
 constexpr bool IsVariableTypeSupportInChunk =
     std::is_same_v<T, std::string> || std::is_same_v<T, Array> ||

internal/core/src/common/Types.h

@@ -493,7 +493,8 @@ IsFloatVectorMetricType(const MetricType& metric_type) {
     return metric_type == knowhere::metric::L2 ||
            metric_type == knowhere::metric::IP ||
            metric_type == knowhere::metric::COSINE ||
-           metric_type == knowhere::metric::BM25;
+           metric_type == knowhere::metric::BM25 ||
+           metric_type == knowhere::metric::MAX_SIM;
 }
 
 inline bool

internal/core/src/common/VectorTrait.h

@@ -28,20 +28,23 @@
 
 namespace milvus {
 
-#define GET_ELEM_TYPE_FOR_VECTOR_TRAIT                             \
-    using elem_type = std::conditional_t<                          \
-        std::is_same_v<TraitType, milvus::FloatVector>,            \
-        milvus::FloatVector::embedded_type,                        \
-        std::conditional_t<                                        \
-            std::is_same_v<TraitType, milvus::Float16Vector>,      \
-            milvus::Float16Vector::embedded_type,                  \
-            std::conditional_t<                                    \
-                std::is_same_v<TraitType, milvus::BFloat16Vector>, \
-                milvus::BFloat16Vector::embedded_type,             \
-                std::conditional_t<                                \
-                    std::is_same_v<TraitType, milvus::Int8Vector>, \
-                    milvus::Int8Vector::embedded_type,             \
-                    milvus::BinaryVector::embedded_type>>>>;
+#define GET_ELEM_TYPE_FOR_VECTOR_TRAIT                                 \
+    using elem_type = std::conditional_t<                              \
+        std::is_same_v<TraitType, milvus::EmbListFloatVector>,         \
+        milvus::EmbListFloatVector::embedded_type,                     \
+        std::conditional_t<                                            \
+            std::is_same_v<TraitType, milvus::FloatVector>,            \
+            milvus::FloatVector::embedded_type,                        \
+            std::conditional_t<                                        \
+                std::is_same_v<TraitType, milvus::Float16Vector>,      \
+                milvus::Float16Vector::embedded_type,                  \
+                std::conditional_t<                                    \
+                    std::is_same_v<TraitType, milvus::BFloat16Vector>, \
+                    milvus::BFloat16Vector::embedded_type,             \
+                    std::conditional_t<                                \
+                        std::is_same_v<TraitType, milvus::Int8Vector>, \
+                        milvus::Int8Vector::embedded_type,             \
+                        milvus::BinaryVector::embedded_type>>>>>;
 
 #define GET_SCHEMA_DATA_TYPE_FOR_VECTOR_TRAIT               \
     auto schema_data_type =                                 \
@@ -55,7 +58,13 @@ namespace milvus {
             ? milvus::Int8Vector::schema_data_type          \
             : milvus::BinaryVector::schema_data_type;
 
-class VectorTrait {};
+class VectorTrait {
+ public:
+    static constexpr bool
+    is_embedding_list() {
+        return false;
+    }
+};
 
 class FloatVector : public VectorTrait {
  public:
@@ -136,6 +145,25 @@ class Int8Vector : public VectorTrait {
         proto::common::PlaceholderType::Int8Vector;
 };
 
+class EmbListFloatVector : public VectorTrait {
+ public:
+    using embedded_type = float;
+    static constexpr int32_t dim_factor = 1;
+    static constexpr auto data_type = DataType::VECTOR_ARRAY;
+    static constexpr auto c_data_type = CDataType::VectorArray;
+    static constexpr auto schema_data_type =
+        proto::schema::DataType::ArrayOfVector;
+    static constexpr auto vector_type =
+        proto::plan::VectorType::EmbListFloatVector;
+    static constexpr auto placeholder_type =
+        proto::common::PlaceholderType::EmbListFloatVector;
+
+    static constexpr bool
+    is_embedding_list() {
+        return true;
+    }
+};
+
 struct FundamentalTag {};
 struct StringTag {};
 

internal/core/src/common/type_c.h

@@ -55,6 +55,7 @@ enum CDataType {
     BFloat16Vector = 103,
     SparseFloatVector = 104,
     Int8Vector = 105,
+    VectorArray = 106,
 };
 typedef enum CDataType CDataType;
 

internal/core/src/exec/operator/VectorSearchNode.cpp

@@ -69,6 +69,7 @@ PhyVectorSearchNode::GetOutput() {
 
     auto& ph = placeholder_group_->at(0);
     auto src_data = ph.get_blob();
+    auto src_lims = ph.get_lims();
     auto num_queries = ph.num_of_queries_;
     milvus::SearchResult search_result;
 
@@ -85,6 +86,7 @@ PhyVectorSearchNode::GetOutput() {
                                   col_input->size());
     segment_->vector_search(search_info_,
                             src_data,
+                            src_lims,
                             num_queries,
                             query_timestamp_,
                             final_view,

internal/core/src/index/IndexFactory.cpp

@@ -98,13 +98,18 @@ IndexFactory::CreatePrimitiveScalarIndex<std::string>(
 LoadResourceRequest
 IndexFactory::IndexLoadResource(
     DataType field_type,
+    DataType element_type,
     IndexVersion index_version,
     float index_size,
     const std::map<std::string, std::string>& index_params,
     bool mmap_enable) {
     if (milvus::IsVectorDataType(field_type)) {
-        return VecIndexLoadResource(
-            field_type, index_version, index_size, index_params, mmap_enable);
+        return VecIndexLoadResource(field_type,
+                                    element_type,
+                                    index_version,
+                                    index_size,
+                                    index_params,
+                                    mmap_enable);
     } else {
         return ScalarIndexLoadResource(
             field_type, index_version, index_size, index_params, mmap_enable);
@@ -114,6 +119,7 @@ IndexFactory::IndexLoadResource(
 LoadResourceRequest
 IndexFactory::VecIndexLoadResource(
     DataType field_type,
+    DataType element_type,
     IndexVersion index_version,
     float index_size,
     const std::map<std::string, std::string>& index_params,
@@ -198,6 +204,29 @@ IndexFactory::VecIndexLoadResource(
                 knowhere::IndexStaticFaced<knowhere::int8>::HasRawData(
                     index_type, index_version, config);
             break;
+        case milvus::DataType::VECTOR_ARRAY: {
+            switch (element_type) {
+                case milvus::DataType::VECTOR_FLOAT:
+                    resource = knowhere::IndexStaticFaced<
+                        knowhere::fp32>::EstimateLoadResource(index_type,
+                                                              index_version,
+                                                              index_size_gb,
+                                                              config);
+                    has_raw_data =
+                        knowhere::IndexStaticFaced<knowhere::fp32>::HasRawData(
+                            index_type, index_version, config);
+                    break;
+
+                default:
+                    LOG_ERROR(
+                        "invalid data type to estimate index load resource: "
+                        "field_type {}, element_type {}",
+                        field_type,
+                        element_type);
+                    return LoadResourceRequest{0, 0, 0, 0, true};
+            }
+            break;
+        }
         default:
             LOG_ERROR("invalid data type to estimate index load resource: {}",
                       field_type);
@@ -491,8 +520,14 @@ IndexFactory::CreateVectorIndex(
                 return std::make_unique<VectorDiskAnnIndex<float>>(
                     index_type, metric_type, version, file_manager_context);
             }
+            case DataType::VECTOR_ARRAY: {
+                ThrowInfo(Unsupported,
+                          "VECTOR_ARRAY for DiskAnnIndex is not supported");
+            }
             case DataType::VECTOR_INT8: {
                 // TODO caiyd, not support yet
+                ThrowInfo(Unsupported,
+                          "VECTOR_INT8 for DiskAnnIndex is not supported");
             }
             default:
                 ThrowInfo(
@@ -505,6 +540,7 @@ IndexFactory::CreateVectorIndex(
             case DataType::VECTOR_FLOAT:
             case DataType::VECTOR_SPARSE_FLOAT: {
                 return std::make_unique<VectorMemIndex<float>>(
+                    DataType::NONE,
                     index_type,
                     metric_type,
                     version,
@@ -513,6 +549,7 @@ IndexFactory::CreateVectorIndex(
             }
             case DataType::VECTOR_BINARY: {
                 return std::make_unique<VectorMemIndex<bin1>>(
+                    DataType::NONE,
                     index_type,
                     metric_type,
                     version,
@@ -521,6 +558,7 @@ IndexFactory::CreateVectorIndex(
             }
             case DataType::VECTOR_FLOAT16: {
                 return std::make_unique<VectorMemIndex<float16>>(
+                    DataType::NONE,
                     index_type,
                     metric_type,
                     version,
@@ -529,6 +567,7 @@ IndexFactory::CreateVectorIndex(
             }
             case DataType::VECTOR_BFLOAT16: {
                 return std::make_unique<VectorMemIndex<bfloat16>>(
+                    DataType::NONE,
                     index_type,
                     metric_type,
                     version,
@@ -537,12 +576,33 @@ IndexFactory::CreateVectorIndex(
             }
             case DataType::VECTOR_INT8: {
                 return std::make_unique<VectorMemIndex<int8>>(
+                    DataType::NONE,
                     index_type,
                     metric_type,
                     version,
                     use_knowhere_build_pool,
                     file_manager_context);
             }
+            case DataType::VECTOR_ARRAY: {
+                auto element_type =
+                    static_cast<DataType>(file_manager_context.fieldDataMeta
+                                              .field_schema.element_type());
+                switch (element_type) {
+                    case DataType::VECTOR_FLOAT:
+                        return std::make_unique<VectorMemIndex<float>>(
+                            element_type,
+                            index_type,
+                            metric_type,
+                            version,
+                            use_knowhere_build_pool,
+                            file_manager_context);
+                    default:
+                        ThrowInfo(NotImplemented,
+                                  fmt::format("not implemented data type to "
+                                              "build mem index: {}",
+                                              data_type));
+                }
+            }
             default:
                 ThrowInfo(
                     DataTypeInvalid,

internal/core/src/index/IndexFactory.h

@@ -56,6 +56,7 @@ class IndexFactory {
 
     LoadResourceRequest
     IndexLoadResource(DataType field_type,
+                      DataType element_type,
                       IndexVersion index_version,
                       float index_size,
                       const std::map<std::string, std::string>& index_params,
@@ -63,6 +64,7 @@ class IndexFactory {
 
     LoadResourceRequest
     VecIndexLoadResource(DataType field_type,
+                         DataType element_type,
                          IndexVersion index_version,
                          float index_size,
                          const std::map<std::string, std::string>& index_params,

internal/core/src/index/Utils.h

@@ -229,4 +229,21 @@ void inline SetBitsetGrowing(void* bitset,
     }
 }
 
+inline size_t
+vector_element_size(const DataType data_type) {
+    switch (data_type) {
+        case DataType::VECTOR_FLOAT:
+            return sizeof(float);
+        case DataType::VECTOR_FLOAT16:
+            return sizeof(float16);
+        case DataType::VECTOR_BFLOAT16:
+            return sizeof(bfloat16);
+        case DataType::VECTOR_INT8:
+            return sizeof(int8);
+        default:
+            ThrowInfo(UnexpectedError,
+                      fmt::format("invalid data type: {}", data_type));
+    }
+}
+
 }  // namespace milvus::index

internal/core/src/index/VectorDiskIndex.cpp

@@ -245,7 +245,7 @@ VectorDiskAnnIndex<T>::Query(const DatasetPtr dataset,
                              SearchResult& search_result) const {
     AssertInfo(GetMetricType() == search_info.metric_type_,
                "Metric type of field index isn't the same with search info");
-    auto num_queries = dataset->GetRows();
+    auto num_rows = dataset->GetRows();
     auto topk = search_info.topk_;
 
     knowhere::Json search_config = PrepareSearchParams(search_info);
@@ -277,7 +277,7 @@ VectorDiskAnnIndex<T>::Query(const DatasetPtr dataset,
                                       res.what()));
             }
             return ReGenRangeSearchResult(
-                res.value(), topk, num_queries, GetMetricType());
+                res.value(), topk, num_rows, GetMetricType());
         } else {
             auto res = index_.Search(dataset, search_config, bitset);
             if (!res.has_value()) {
@@ -291,6 +291,8 @@ VectorDiskAnnIndex<T>::Query(const DatasetPtr dataset,
     }();
 
     auto ids = final->GetIds();
+    // In embedding list query, final->GetRows() can be different from dataset->GetRows().
+    auto num_queries = final->GetRows();
     float* distances = const_cast<float*>(final->GetDistance());
     final->SetIsOwner(true);
 

internal/core/src/index/VectorMemIndex.cpp

@@ -59,12 +59,14 @@ namespace milvus::index {
 
 template <typename T>
 VectorMemIndex<T>::VectorMemIndex(
+    DataType elem_type,
     const IndexType& index_type,
     const MetricType& metric_type,
     const IndexVersion& version,
     bool use_knowhere_build_pool,
     const storage::FileManagerContext& file_manager_context)
     : VectorIndex(index_type, metric_type),
+      elem_type_(elem_type),
       use_knowhere_build_pool_(use_knowhere_build_pool) {
     CheckMetricTypeSupport<T>(metric_type);
     AssertInfo(!is_unsupported(index_type, metric_type),
@@ -89,12 +91,14 @@ VectorMemIndex<T>::VectorMemIndex(
 }
 
 template <typename T>
-VectorMemIndex<T>::VectorMemIndex(const IndexType& index_type,
+VectorMemIndex<T>::VectorMemIndex(DataType elem_type,
+                                  const IndexType& index_type,
                                   const MetricType& metric_type,
                                   const IndexVersion& version,
                                   const knowhere::ViewDataOp view_data,
                                   bool use_knowhere_build_pool)
     : VectorIndex(index_type, metric_type),
+      elem_type_(elem_type),
       use_knowhere_build_pool_(use_knowhere_build_pool) {
     CheckMetricTypeSupport<T>(metric_type);
     AssertInfo(!is_unsupported(index_type, metric_type),
@@ -304,6 +308,11 @@ VectorMemIndex<T>::BuildWithDataset(const DatasetPtr& dataset,
     SetDim(index_.Dim());
 }
 
+bool
+is_embedding_list_index(const IndexType& index_type) {
+    return index_type == knowhere::IndexEnum::INDEX_EMB_LIST_HNSW;
+}
+
 template <typename T>
 void
 VectorMemIndex<T>::Build(const Config& config) {
@@ -331,23 +340,74 @@ VectorMemIndex<T>::Build(const Config& config) {
             total_num_rows += data->get_num_rows();
             AssertInfo(dim == 0 || dim == data->get_dim(),
                        "inconsistent dim value between field datas!");
-            dim = data->get_dim();
+
+            // todo(SapdeA): now, vector arrays (embedding list) are serialized
+            // to parquet by using binary format which does not provide dim
+            // information so we use this temporary solution.
+            if (is_embedding_list_index(index_type_)) {
+                AssertInfo(elem_type_ != DataType::NONE,
+                           "embedding list index must have elem_type");
+                dim = config[DIM_KEY].get<int64_t>();
+            } else {
+                dim = data->get_dim();
+            }
         }
 
         auto buf = std::shared_ptr<uint8_t[]>(new uint8_t[total_size]);
+
+        size_t lim_offset = 0;
+        std::vector<size_t> lims;
+        lims.reserve(total_num_rows + 1);
+        lims.push_back(lim_offset);
+
         int64_t offset = 0;
-        // TODO: avoid copying
-        for (auto data : field_datas) {
-            std::memcpy(buf.get() + offset, data->Data(), data->Size());
-            offset += data->Size();
-            data.reset();
+        if (!is_embedding_list_index(index_type_)) {
+            // TODO: avoid copying
+            for (auto data : field_datas) {
+                std::memcpy(buf.get() + offset, data->Data(), data->Size());
+                offset += data->Size();
+                data.reset();
+            }
+        } else {
+            auto elem_size = vector_element_size(elem_type_);
+            for (auto data : field_datas) {
+                auto vec_array_data =
+                    dynamic_cast<FieldData<VectorArray>*>(data.get());
+                AssertInfo(vec_array_data != nullptr,
+                           "failed to cast field data to vector array");
+
+                auto rows = vec_array_data->get_num_rows();
+                for (auto i = 0; i < rows; ++i) {
+                    auto size = vec_array_data->DataSize(i);
+                    assert(size % (dim * elem_size) == 0);
+                    assert(dim * elem_size != 0);
+
+                    auto vec_array = vec_array_data->value_at(i);
+
+                    std::memcpy(buf.get() + offset, vec_array->data(), size);
+                    offset += size;
+
+                    lim_offset += size / (dim * elem_size);
+                    lims.push_back(lim_offset);
+                }
+
+                assert(data->Size() == offset);
+
+                data.reset();
+            }
+
+            total_num_rows = lim_offset;
         }
+
         field_datas.clear();
 
         auto dataset = GenDataset(total_num_rows, dim, buf.get());
         if (!scalar_info.empty()) {
             dataset->Set(knowhere::meta::SCALAR_INFO, std::move(scalar_info));
         }
+        if (!lims.empty()) {
+            dataset->SetLims(lims.data());
+        }
         BuildWithDataset(dataset, build_config);
     } else {
         // sparse
@@ -409,7 +469,7 @@ VectorMemIndex<T>::Query(const DatasetPtr dataset,
     //    AssertInfo(GetMetricType() == search_info.metric_type_,
     //               "Metric type of field index isn't the same with search info");
 
-    auto num_queries = dataset->GetRows();
+    auto num_vectors = dataset->GetRows();
     knowhere::Json search_conf = PrepareSearchParams(search_info);
     auto topk = search_info.topk_;
     // TODO :: check dim of search data
@@ -427,7 +487,7 @@ VectorMemIndex<T>::Query(const DatasetPtr dataset,
                           res.what());
             }
             auto result = ReGenRangeSearchResult(
-                res.value(), topk, num_queries, GetMetricType());
+                res.value(), topk, num_vectors, GetMetricType());
             milvus::tracer::AddEvent("finish_ReGenRangeSearchResult");
             return result;
         } else {
@@ -448,6 +508,8 @@ VectorMemIndex<T>::Query(const DatasetPtr dataset,
     }();
 
     auto ids = final->GetIds();
+    // In embedding list query, final->GetRows() can be different from dataset->GetRows().
+    auto num_queries = final->GetRows();
     float* distances = const_cast<float*>(final->GetDistance());
     final->SetIsOwner(true);
     auto round_decimal = search_info.round_decimal_;

internal/core/src/index/VectorMemIndex.h

@@ -35,6 +35,7 @@ template <typename T>
 class VectorMemIndex : public VectorIndex {
  public:
     explicit VectorMemIndex(
+        DataType elem_type /* used for embedding list only */,
         const IndexType& index_type,
         const MetricType& metric_type,
         const IndexVersion& version,
@@ -43,7 +44,8 @@ class VectorMemIndex : public VectorIndex {
             storage::FileManagerContext());
 
     // knowhere data view index special constucter for intermin index, no need to hold file_manager_ to upload or download files
-    VectorMemIndex(const IndexType& index_type,
+    VectorMemIndex(DataType elem_type /* used for embedding list only */,
+                   const IndexType& index_type,
                    const MetricType& metric_type,
                    const IndexVersion& version,
                    const knowhere::ViewDataOp view_data,
@@ -108,6 +110,8 @@ class VectorMemIndex : public VectorIndex {
     Config config_;
     knowhere::Index<knowhere::IndexNode> index_;
     std::shared_ptr<storage::MemFileManagerImpl> file_manager_;
+    // used for embedding list only
+    DataType elem_type_;
 
     CreateIndexInfo create_index_info_;
     bool use_knowhere_build_pool_;

internal/core/src/indexbuilder/IndexFactory.h

@@ -70,11 +70,8 @@ class IndexFactory {
             case DataType::VECTOR_BINARY:
             case DataType::VECTOR_SPARSE_FLOAT:
             case DataType::VECTOR_INT8:
-                return std::make_unique<VecIndexCreator>(type, config, context);
-
             case DataType::VECTOR_ARRAY:
-                ThrowInfo(DataTypeInvalid,
-                          fmt::format("VECTOR_ARRAY is not implemented"));
+                return std::make_unique<VecIndexCreator>(type, config, context);
 
             default:
                 ThrowInfo(DataTypeInvalid,

internal/core/src/indexbuilder/VecIndexCreator.cpp

@@ -34,6 +34,13 @@ VecIndexCreator::VecIndexCreator(
     Config& config,
     const storage::FileManagerContext& file_manager_context)
     : config_(config), data_type_(data_type) {
+    if (data_type == DataType::VECTOR_ARRAY) {
+        // TODO(SpadeA): record dim in config as there's the dim cannot be inferred in
+        // parquet due to the serialize method of vector array.
+        // This should be a temp solution.
+        config_[DIM_KEY] = file_manager_context.indexMeta.dim;
+    }
+
     index::CreateIndexInfo index_info;
     index_info.field_type = data_type_;
     index_info.index_type = index::GetIndexTypeFromConfig(config_);

internal/core/src/mmap/ChunkedColumn.h

@@ -273,6 +273,13 @@ class ChunkedColumnBase : public ChunkedColumnInterface {
             "VectorArrayViews only supported for ChunkedVectorArrayColumn");
     }
 
+    virtual PinWrapper<const size_t*>
+    VectorArrayLims(int64_t chunk_id) const override {
+        ThrowInfo(
+            ErrorCode::Unsupported,
+            "VectorArrayLims only supported for ChunkedVectorArrayColumn");
+    }
+
     PinWrapper<std::pair<std::vector<std::string_view>, FixedVector<bool>>>
     StringViewsByOffsets(int64_t chunk_id,
                          const FixedVector<int32_t>& offsets) const override {
@@ -621,6 +628,15 @@ class ChunkedVectorArrayColumn : public ChunkedColumnBase {
         return PinWrapper<std::vector<VectorArrayView>>(
             ca, static_cast<VectorArrayChunk*>(chunk)->Views());
     }
+
+    PinWrapper<const size_t*>
+    VectorArrayLims(int64_t chunk_id) const override {
+        auto ca =
+            SemiInlineGet(slot_->PinCells({static_cast<cid_t>(chunk_id)}));
+        auto chunk = ca->get_cell_of(chunk_id);
+        return PinWrapper<const size_t*>(
+            ca, static_cast<VectorArrayChunk*>(chunk)->Lims());
+    }
 };
 
 inline std::shared_ptr<ChunkedColumnInterface>

internal/core/src/mmap/ChunkedColumnGroup.h

@@ -319,6 +319,19 @@ class ProxyChunkColumn : public ChunkedColumnInterface {
             static_cast<VectorArrayChunk*>(chunk.get())->Views());
     }
 
+    PinWrapper<const size_t*>
+    VectorArrayLims(int64_t chunk_id) const override {
+        if (!IsChunkedVectorArrayColumnDataType(data_type_)) {
+            ThrowInfo(ErrorCode::Unsupported,
+                      "VectorArrayLims only supported for "
+                      "ChunkedVectorArrayColumn");
+        }
+        auto chunk_wrapper = group_->GetGroupChunk(chunk_id);
+        auto chunk = chunk_wrapper.get()->GetChunk(field_id_);
+        return PinWrapper<const size_t*>(
+            chunk_wrapper, static_cast<VectorArrayChunk*>(chunk.get())->Lims());
+    }
+
     PinWrapper<std::pair<std::vector<std::string_view>, FixedVector<bool>>>
     StringViewsByOffsets(int64_t chunk_id,
                          const FixedVector<int32_t>& offsets) const override {

internal/core/src/mmap/ChunkedColumnInterface.h

@@ -84,6 +84,9 @@ class ChunkedColumnInterface {
     virtual PinWrapper<std::vector<VectorArrayView>>
     VectorArrayViews(int64_t chunk_id) const = 0;
 
+    virtual PinWrapper<const size_t*>
+    VectorArrayLims(int64_t chunk_id) const = 0;
+
     virtual PinWrapper<
         std::pair<std::vector<std::string_view>, FixedVector<bool>>>
     StringViewsByOffsets(int64_t chunk_id,

internal/core/src/query/ExecPlanNodeVisitor.cpp

@@ -242,4 +242,9 @@ ExecPlanNodeVisitor::visit(Int8VectorANNS& node) {
     VectorVisitorImpl<Int8Vector>(node);
 }
 
+void
+ExecPlanNodeVisitor::visit(EmbListFloatVectorANNS& node) {
+    VectorVisitorImpl<EmbListFloatVector>(node);
+}
+
 }  // namespace milvus::query

internal/core/src/query/ExecPlanNodeVisitor.h

@@ -43,6 +43,9 @@ class ExecPlanNodeVisitor : public PlanNodeVisitor {
     void
     visit(RetrievePlanNode& node) override;
 
+    void
+    visit(EmbListFloatVectorANNS& node) override;
+
  public:
     ExecPlanNodeVisitor(const segcore::SegmentInterface& segment,
                         Timestamp timestamp,

internal/core/src/query/Plan.cpp

@@ -30,6 +30,17 @@ ParsePlaceholderGroup(const Plan* plan,
         placeholder_group_blob.size());
 }
 
+bool
+check_data_type(const FieldMeta& field_meta,
+                const milvus::proto::common::PlaceholderType type) {
+    if (field_meta.get_data_type() == DataType::VECTOR_ARRAY) {
+        return type ==
+               milvus::proto::common::PlaceholderType::EmbListFloatVector;
+    }
+    return static_cast<int>(field_meta.get_data_type()) ==
+           static_cast<int>(type);
+}
+
 std::unique_ptr<PlaceholderGroup>
 ParsePlaceholderGroup(const Plan* plan,
                       const uint8_t* blob,
@@ -44,8 +55,7 @@ ParsePlaceholderGroup(const Plan* plan,
         Assert(plan->tag2field_.count(element.tag_));
         auto field_id = plan->tag2field_.at(element.tag_);
         auto& field_meta = plan->schema_->operator[](field_id);
-        AssertInfo(static_cast<int>(field_meta.get_data_type()) ==
-                       static_cast<int>(info.type()),
+        AssertInfo(check_data_type(field_meta, info.type()),
                    "vector type must be the same, field {} - type {}, search "
                    "info type {}",
                    field_meta.get_name().get(),
@@ -59,23 +69,47 @@ ParsePlaceholderGroup(const Plan* plan,
                 SparseBytesToRows(info.values(), /*validate=*/true);
         } else {
             auto line_size = info.values().Get(0).size();
-            if (field_meta.get_sizeof() != line_size) {
-                ThrowInfo(
-                    DimNotMatch,
-                    fmt::format("vector dimension mismatch, expected vector "
-                                "size(byte) {}, actual {}.",
-                                field_meta.get_sizeof(),
-                                line_size));
-            }
             auto& target = element.blob_;
-            target.reserve(line_size * element.num_of_queries_);
-            for (auto& line : info.values()) {
-                AssertInfo(line_size == line.size(),
-                           "vector dimension mismatch, expected vector "
-                           "size(byte) {}, actual {}.",
-                           line_size,
-                           line.size());
-                target.insert(target.end(), line.begin(), line.end());
+
+            if (field_meta.get_data_type() != DataType::VECTOR_ARRAY) {
+                if (field_meta.get_sizeof() != line_size) {
+                    ThrowInfo(DimNotMatch,
+                              fmt::format(
+                                  "vector dimension mismatch, expected vector "
+                                  "size(byte) {}, actual {}.",
+                                  field_meta.get_sizeof(),
+                                  line_size));
+                }
+                target.reserve(line_size * element.num_of_queries_);
+                for (auto& line : info.values()) {
+                    AssertInfo(line_size == line.size(),
+                               "vector dimension mismatch, expected vector "
+                               "size(byte) {}, actual {}.",
+                               line_size,
+                               line.size());
+                    target.insert(target.end(), line.begin(), line.end());
+                }
+            } else {
+                target.reserve(line_size * element.num_of_queries_);
+                auto dim = field_meta.get_dim();
+
+                // If the vector is embedding list, line contains multiple vectors.
+                // And we should record the offsets so that we can identify each
+                // embedding list in a flattened vectors.
+                auto& lims = element.lims_;
+                lims.reserve(element.num_of_queries_ + 1);
+                size_t offset = 0;
+                lims.push_back(offset);
+
+                auto elem_size = milvus::index::vector_element_size(
+                    field_meta.get_element_type());
+                for (auto& line : info.values()) {
+                    target.insert(target.end(), line.begin(), line.end());
+
+                    Assert(line.size() % (dim * elem_size) == 0);
+                    offset += line.size() / (dim * elem_size);
+                    lims.push_back(offset);
+                }
             }
         }
         result->emplace_back(std::move(element));

internal/core/src/query/PlanImpl.h

@@ -68,6 +68,9 @@ struct Plan {
 
 struct Placeholder {
     std::string tag_;
+    // note: for embedding list search, num_of_queries_ stands for the number of vectors.
+    // lims_ records the offsets of embedding list in the flattened vector and
+    // hence lims_.size() - 1 is the number of queries in embedding list search.
     int64_t num_of_queries_;
     // TODO(SPARSE): add a dim_ field here, use the dim passed in search request
     // instead of the dim in schema, since the dim of sparse float column is
@@ -78,6 +81,8 @@ struct Placeholder {
     // dense vector search and sparse_matrix_ is for sparse vector search.
     aligned_vector<char> blob_;
     std::unique_ptr<knowhere::sparse::SparseRow<float>[]> sparse_matrix_;
+    // offsets for embedding list
+    aligned_vector<size_t> lims_;
 
     const void*
     get_blob() const {
@@ -94,6 +99,16 @@ struct Placeholder {
         }
         return blob_.data();
     }
+
+    const size_t*
+    get_lims() const {
+        return lims_.data();
+    }
+
+    size_t*
+    get_lims() {
+        return lims_.data();
+    }
 };
 
 struct RetrievePlan {

internal/core/src/query/PlanNode.cpp

@@ -50,4 +50,9 @@ RetrievePlanNode::accept(PlanNodeVisitor& visitor) {
     visitor.visit(*this);
 }
 
+void
+EmbListFloatVectorANNS::accept(PlanNodeVisitor& visitor) {
+    visitor.visit(*this);
+}
+
 }  // namespace milvus::query

internal/core/src/query/PlanNode.h

@@ -77,6 +77,12 @@ struct Int8VectorANNS : VectorPlanNode {
     accept(PlanNodeVisitor&) override;
 };
 
+struct EmbListFloatVectorANNS : VectorPlanNode {
+ public:
+    void
+    accept(PlanNodeVisitor&) override;
+};
+
 struct RetrievePlanNode : PlanNode {
  public:
     void

internal/core/src/query/PlanNodeVisitor.h

@@ -39,5 +39,8 @@ class PlanNodeVisitor {
 
     virtual void
     visit(RetrievePlanNode&) = 0;
+
+    virtual void
+    visit(EmbListFloatVectorANNS&) = 0;
 };
 }  // namespace milvus::query

internal/core/src/query/PlanProto.cpp

@@ -127,6 +127,9 @@ ProtoParser::PlanNodeFromProto(const planpb::PlanNode& plan_node_proto) {
         } else if (anns_proto.vector_type() ==
                    milvus::proto::plan::VectorType::Int8Vector) {
             return std::make_unique<Int8VectorANNS>();
+        } else if (anns_proto.vector_type() ==
+                   milvus::proto::plan::VectorType::EmbListFloatVector) {
+            return std::make_unique<EmbListFloatVectorANNS>();
         } else {
             return std::make_unique<FloatVectorANNS>();
         }

internal/core/src/query/SearchBruteForce.cpp

@@ -89,8 +89,23 @@ PrepareBFDataSet(const dataset::SearchDataset& query_ds,
                  DataType data_type) {
     auto base_dataset =
         knowhere::GenDataSet(raw_ds.num_raw_data, raw_ds.dim, raw_ds.raw_data);
+    if (raw_ds.raw_data_lims != nullptr) {
+        // knowhere::DataSet count vectors in a flattened manner where as the num_raw_data here is the number
+        // of embedding lists where each embedding list contains multiple vectors. So we should use the last element
+        // in lims which equals to the total number of vectors.
+        base_dataset->SetLims(raw_ds.raw_data_lims);
+        // the length of lims equals to the number of embedding lists + 1
+        base_dataset->SetRows(raw_ds.raw_data_lims[raw_ds.num_raw_data]);
+    }
+
     auto query_dataset = knowhere::GenDataSet(
         query_ds.num_queries, query_ds.dim, query_ds.query_data);
+    if (query_ds.query_lims != nullptr) {
+        // ditto
+        query_dataset->SetLims(query_ds.query_lims);
+        query_dataset->SetRows(query_ds.query_lims[query_ds.num_queries]);
+    }
+
     if (data_type == DataType::VECTOR_SPARSE_FLOAT) {
         base_dataset->SetIsSparse(true);
         query_dataset->SetIsSparse(true);
@@ -105,7 +120,8 @@ BruteForceSearch(const dataset::SearchDataset& query_ds,
                  const SearchInfo& search_info,
                  const std::map<std::string, std::string>& index_info,
                  const BitsetView& bitset,
-                 DataType data_type) {
+                 DataType data_type,
+                 DataType element_type) {
     SubSearchResult sub_result(query_ds.num_queries,
                                query_ds.topk,
                                query_ds.metric_type,
@@ -122,7 +138,18 @@ BruteForceSearch(const dataset::SearchDataset& query_ds,
     sub_result.mutable_seg_offsets().resize(nq * topk);
     sub_result.mutable_distances().resize(nq * topk);
 
+    // For vector array (embedding list), element type is used to determine how to operate search.
+    if (data_type == DataType::VECTOR_ARRAY) {
+        AssertInfo(element_type != DataType::NONE,
+                   "Element type is not specified for vector array");
+        data_type = element_type;
+    }
+
     if (search_cfg.contains(RADIUS)) {
+        AssertInfo(data_type != DataType::VECTOR_ARRAY,
+                   "Vector array(embedding list) is not supported for range "
+                   "search");
+
         if (search_cfg.contains(RANGE_FILTER)) {
             CheckRangeSearchParam(search_cfg[RADIUS],
                                   search_cfg[RANGE_FILTER],
@@ -238,7 +265,10 @@ DispatchBruteForceIteratorByDataType(const knowhere::DataSetPtr& base_dataset,
                                      const knowhere::DataSetPtr& query_dataset,
                                      const knowhere::Json& config,
                                      const BitsetView& bitset,
-                                     const milvus::DataType& data_type) {
+                                     milvus::DataType data_type) {
+    AssertInfo(data_type != DataType::VECTOR_ARRAY,
+               "VECTOR_ARRAY is not supported for brute force iterator");
+
     switch (data_type) {
         case DataType::VECTOR_FLOAT:
             return knowhere::BruteForce::AnnIterator<float>(

internal/core/src/query/SearchBruteForce.h

@@ -29,7 +29,8 @@ BruteForceSearch(const dataset::SearchDataset& query_ds,
                  const SearchInfo& search_info,
                  const std::map<std::string, std::string>& index_info,
                  const BitsetView& bitset,
-                 DataType data_type);
+                 DataType data_type,
+                 DataType element_type);
 
 knowhere::expected<std::vector<knowhere::IndexNode::IteratorPtr>>
 GetBruteForceSearchIterators(

internal/core/src/query/SearchOnGrowing.cpp

@@ -71,6 +71,7 @@ void
 SearchOnGrowing(const segcore::SegmentGrowingImpl& segment,
                 const SearchInfo& info,
                 const void* query_data,
+                const size_t* query_lims,
                 int64_t num_queries,
                 Timestamp timestamp,
                 const BitsetView& bitset,
@@ -87,6 +88,7 @@ SearchOnGrowing(const segcore::SegmentGrowingImpl& segment,
     CheckBruteForceSearchParam(field, info);
 
     auto data_type = field.get_data_type();
+    auto element_type = field.get_element_type();
     AssertInfo(IsVectorDataType(data_type),
                "[SearchOnGrowing]Data type isn't vector type");
 
@@ -96,6 +98,11 @@ SearchOnGrowing(const segcore::SegmentGrowingImpl& segment,
 
     // step 2: small indexing search
     if (segment.get_indexing_record().SyncDataWithIndex(field.get_id())) {
+        AssertInfo(
+            data_type != DataType::VECTOR_ARRAY,
+            "vector array(embedding list) is not supported for growing segment "
+            "indexing search");
+
         FloatSegmentIndexSearch(
             segment, info, query_data, num_queries, bitset, search_result);
     } else {
@@ -103,6 +110,10 @@ SearchOnGrowing(const segcore::SegmentGrowingImpl& segment,
             segment.get_chunk_mutex());
         // check SyncDataWithIndex() again, in case the vector chunks has been removed.
         if (segment.get_indexing_record().SyncDataWithIndex(field.get_id())) {
+            AssertInfo(data_type != DataType::VECTOR_ARRAY,
+                       "vector array(embedding list) is not supported for "
+                       "growing segment indexing search");
+
             return FloatSegmentIndexSearch(
                 segment, info, query_data, num_queries, bitset, search_result);
         }
@@ -111,8 +122,13 @@ SearchOnGrowing(const segcore::SegmentGrowingImpl& segment,
         auto dim = field.get_data_type() == DataType::VECTOR_SPARSE_FLOAT
                        ? 0
                        : field.get_dim();
-        dataset::SearchDataset search_dataset{
-            metric_type, num_queries, topk, round_decimal, dim, query_data};
+        dataset::SearchDataset search_dataset{metric_type,
+                                              num_queries,
+                                              topk,
+                                              round_decimal,
+                                              dim,
+                                              query_data,
+                                              query_lims};
         int32_t current_chunk_id = 0;
 
         // get K1 and B from index for bm25 brute force
@@ -127,6 +143,10 @@ SearchOnGrowing(const segcore::SegmentGrowingImpl& segment,
         auto vec_ptr = record.get_data_base(vecfield_id);
 
         if (info.iterator_v2_info_.has_value()) {
+            AssertInfo(data_type != DataType::VECTOR_ARRAY,
+                       "vector array(embedding list) is not supported for "
+                       "vector iterator");
+
             CachedSearchIterator cached_iter(search_dataset,
                                              vec_ptr,
                                              active_count,
@@ -150,9 +170,54 @@ SearchOnGrowing(const segcore::SegmentGrowingImpl& segment,
                 std::min(active_count, (chunk_id + 1) * vec_size_per_chunk);
             auto size_per_chunk = element_end - element_begin;
 
-            auto sub_data = query::dataset::RawDataset{
-                element_begin, dim, size_per_chunk, chunk_data};
+            query::dataset::RawDataset sub_data;
+            std::unique_ptr<uint8_t[]> buf = nullptr;
+            std::vector<size_t> offsets;
+            if (data_type != DataType::VECTOR_ARRAY) {
+                sub_data = query::dataset::RawDataset{
+                    element_begin, dim, size_per_chunk, chunk_data};
+            } else {
+                // TODO(SpadeA): For VectorArray(Embedding List), data is
+                // discreted stored in FixedVector which means we will copy the
+                // data to a contiguous memory buffer. This is inefficient and
+                // will be optimized in the future.
+                auto vec_ptr = reinterpret_cast<const VectorArray*>(chunk_data);
+                auto size = 0;
+                for (int i = 0; i < size_per_chunk; ++i) {
+                    size += vec_ptr[i].byte_size();
+                }
+
+                buf = std::make_unique<uint8_t[]>(size);
+                offsets.reserve(size_per_chunk + 1);
+                offsets.push_back(0);
+
+                auto offset = 0;
+                auto ptr = buf.get();
+                for (int i = 0; i < size_per_chunk; ++i) {
+                    memcpy(ptr, vec_ptr[i].data(), vec_ptr[i].byte_size());
+                    ptr += vec_ptr[i].byte_size();
+
+                    offset += vec_ptr[i].length();
+                    offsets.push_back(offset);
+                }
+                sub_data = query::dataset::RawDataset{element_begin,
+                                                      dim,
+                                                      size_per_chunk,
+                                                      buf.get(),
+                                                      offsets.data()};
+            }
+
+            if (data_type == DataType::VECTOR_ARRAY) {
+                AssertInfo(
+                    query_lims != nullptr,
+                    "query_lims is nullptr, but data_type is vector array");
+            }
+
             if (milvus::exec::UseVectorIterator(info)) {
+                AssertInfo(data_type != DataType::VECTOR_ARRAY,
+                           "vector array(embedding list) is not supported for "
+                           "vector iterator");
+
                 auto sub_qr =
                     PackBruteForceSearchIteratorsIntoSubResult(search_dataset,
                                                                sub_data,
@@ -167,7 +232,8 @@ SearchOnGrowing(const segcore::SegmentGrowingImpl& segment,
                                                info,
                                                index_info,
                                                bitset,
-                                               data_type);
+                                               data_type,
+                                               element_type);
                 final_qr.merge(sub_qr);
             }
         }

internal/core/src/query/SearchOnGrowing.h

@@ -20,6 +20,7 @@ void
 SearchOnGrowing(const segcore::SegmentGrowingImpl& segment,
                 const SearchInfo& info,
                 const void* query_data,
+                const size_t* query_lims,
                 int64_t num_queries,
                 Timestamp timestamp,
                 const BitsetView& bitset,

internal/core/src/query/SearchOnSealed.cpp

@@ -31,6 +31,7 @@ SearchOnSealedIndex(const Schema& schema,
                     const segcore::SealedIndexingRecord& record,
                     const SearchInfo& search_info,
                     const void* query_data,
+                    const size_t* query_lims,
                     int64_t num_queries,
                     const BitsetView& bitset,
                     SearchResult& search_result) {
@@ -52,7 +53,18 @@ SearchOnSealedIndex(const Schema& schema,
                field_indexing->metric_type_,
                search_info.metric_type_);
 
-    auto dataset = knowhere::GenDataSet(num_queries, dim, query_data);
+    knowhere::DataSetPtr dataset;
+    if (query_lims == nullptr) {
+        dataset = knowhere::GenDataSet(num_queries, dim, query_data);
+    } else {
+        // Rather than non-embedding list search where num_queries equals to the number of vectors,
+        // in embedding list search, multiple vectors form an embedding list and the last element of query_lims
+        // stands for the total number of vectors.
+        auto num_vectors = query_lims[num_queries];
+        dataset = knowhere::GenDataSet(num_vectors, dim, query_data);
+        dataset->SetLims(query_lims);
+    }
+
     dataset->SetIsSparse(is_sparse);
     auto accessor = SemiInlineGet(field_indexing->indexing_->PinCells({0}));
     auto vec_index =
@@ -92,6 +104,7 @@ SearchOnSealedColumn(const Schema& schema,
                      const SearchInfo& search_info,
                      const std::map<std::string, std::string>& index_info,
                      const void* query_data,
+                     const size_t* query_lims,
                      int64_t num_queries,
                      int64_t row_count,
                      const BitsetView& bitview,
@@ -99,22 +112,26 @@ SearchOnSealedColumn(const Schema& schema,
     auto field_id = search_info.field_id_;
     auto& field = schema[field_id];
 
+    auto data_type = field.get_data_type();
+    auto element_type = field.get_element_type();
     // TODO(SPARSE): see todo in PlanImpl.h::PlaceHolder.
-    auto dim = field.get_data_type() == DataType::VECTOR_SPARSE_FLOAT
-                   ? 0
-                   : field.get_dim();
+    auto dim = data_type == DataType::VECTOR_SPARSE_FLOAT ? 0 : field.get_dim();
 
     query::dataset::SearchDataset query_dataset{search_info.metric_type_,
                                                 num_queries,
                                                 search_info.topk_,
                                                 search_info.round_decimal_,
                                                 dim,
-                                                query_data};
+                                                query_data,
+                                                query_lims};
 
-    auto data_type = field.get_data_type();
     CheckBruteForceSearchParam(field, search_info);
 
     if (search_info.iterator_v2_info_.has_value()) {
+        AssertInfo(data_type != DataType::VECTOR_ARRAY,
+                   "vector array(embedding list) is not supported for "
+                   "vector iterator");
+
         CachedSearchIterator cached_iter(
             column, query_dataset, search_info, index_info, bitview, data_type);
         cached_iter.NextBatch(search_info, result);
@@ -135,7 +152,20 @@ SearchOnSealedColumn(const Schema& schema,
         auto chunk_size = column->chunk_row_nums(i);
         auto raw_dataset =
             query::dataset::RawDataset{offset, dim, chunk_size, vec_data};
+
+        PinWrapper<const size_t*> lims_pw;
+        if (data_type == DataType::VECTOR_ARRAY) {
+            AssertInfo(query_lims != nullptr,
+                       "query_lims is nullptr, but data_type is vector array");
+
+            lims_pw = column->VectorArrayLims(i);
+            raw_dataset.raw_data_lims = lims_pw.get();
+        }
+
         if (milvus::exec::UseVectorIterator(search_info)) {
+            AssertInfo(data_type != DataType::VECTOR_ARRAY,
+                       "vector array(embedding list) is not supported for "
+                       "vector iterator");
             auto sub_qr =
                 PackBruteForceSearchIteratorsIntoSubResult(query_dataset,
                                                            raw_dataset,
@@ -150,7 +180,8 @@ SearchOnSealedColumn(const Schema& schema,
                                            search_info,
                                            index_info,
                                            bitview,
-                                           data_type);
+                                           data_type,
+                                           element_type);
             final_qr.merge(sub_qr);
         }
         offset += chunk_size;

internal/core/src/query/SearchOnSealed.h

@@ -23,6 +23,7 @@ SearchOnSealedIndex(const Schema& schema,
                     const segcore::SealedIndexingRecord& record,
                     const SearchInfo& search_info,
                     const void* query_data,
+                    const size_t* query_lims,
                     int64_t num_queries,
                     const BitsetView& view,
                     SearchResult& search_result);
@@ -33,6 +34,7 @@ SearchOnSealedColumn(const Schema& schema,
                      const SearchInfo& search_info,
                      const std::map<std::string, std::string>& index_info,
                      const void* query_data,
+                     const size_t* query_lims,
                      int64_t num_queries,
                      int64_t row_count,
                      const BitsetView& bitset,

internal/core/src/query/helper.h

@@ -24,6 +24,7 @@ struct RawDataset {
     int64_t dim;
     int64_t num_raw_data;
     const void* raw_data;
+    const size_t* raw_data_lims = nullptr;
 };
 struct SearchDataset {
     knowhere::MetricType metric_type;
@@ -32,6 +33,8 @@ struct SearchDataset {
     int64_t round_decimal;
     int64_t dim;
     const void* query_data;
+    // used for embedding list query
+    const size_t* query_lims = nullptr;
 };
 
 }  // namespace dataset

internal/core/src/segcore/ChunkedSegmentSealedImpl.cpp

@@ -95,10 +95,6 @@ ChunkedSegmentSealedImpl::LoadIndex(const LoadIndexInfo& info) {
     auto field_id = FieldId(info.field_id);
     auto& field_meta = schema_->operator[](field_id);
 
-    if (field_meta.get_data_type() == DataType::VECTOR_ARRAY) {
-        ThrowInfo(DataTypeInvalid, "VECTOR_ARRAY is not implemented");
-    }
-
     if (field_meta.is_vector()) {
         LoadVecIndex(info);
     } else {
@@ -127,6 +123,7 @@ ChunkedSegmentSealedImpl::LoadVecIndex(const LoadIndexInfo& info) {
     LoadResourceRequest request =
         milvus::index::IndexFactory::GetInstance().VecIndexLoadResource(
             field_meta.get_data_type(),
+            info.element_type,
             info.index_engine_version,
             info.index_size,
             info.index_params,
@@ -498,10 +495,6 @@ int64_t
 ChunkedSegmentSealedImpl::num_chunk_index(FieldId field_id) const {
     auto& field_meta = schema_->operator[](field_id);
 
-    if (field_meta.get_data_type() == DataType::VECTOR_ARRAY) {
-        ThrowInfo(DataTypeInvalid, "VECTOR_ARRAY is not implemented");
-    }
-
     if (field_meta.is_vector()) {
         return int64_t(vector_indexings_.is_ready(field_id));
     }
@@ -720,6 +713,7 @@ ChunkedSegmentSealedImpl::mask_with_delete(BitsetTypeView& bitset,
 void
 ChunkedSegmentSealedImpl::vector_search(SearchInfo& search_info,
                                         const void* query_data,
+                                        const size_t* query_lims,
                                         int64_t query_count,
                                         Timestamp timestamp,
                                         const BitsetView& bitset,
@@ -745,6 +739,7 @@ ChunkedSegmentSealedImpl::vector_search(SearchInfo& search_info,
                                    vector_indexings_,
                                    binlog_search_info,
                                    query_data,
+                                   query_lims,
                                    query_count,
                                    bitset,
                                    output);
@@ -758,6 +753,7 @@ ChunkedSegmentSealedImpl::vector_search(SearchInfo& search_info,
                                    vector_indexings_,
                                    search_info,
                                    query_data,
+                                   query_lims,
                                    query_count,
                                    bitset,
                                    output);
@@ -782,6 +778,7 @@ ChunkedSegmentSealedImpl::vector_search(SearchInfo& search_info,
                                     search_info,
                                     index_info,
                                     query_data,
+                                    query_lims,
                                     query_count,
                                     row_count,
                                     bitset,

internal/core/src/segcore/ChunkedSegmentSealedImpl.h

@@ -395,6 +395,7 @@ class ChunkedSegmentSealedImpl : public SegmentSealed {
     void
     vector_search(SearchInfo& search_info,
                   const void* query_data,
+                  const size_t* query_lims,
                   int64_t query_count,
                   Timestamp timestamp,
                   const BitsetView& bitset,

internal/core/src/segcore/FieldIndexing.cpp

@@ -47,6 +47,7 @@ VectorFieldIndexing::recreate_index(DataType data_type,
                                     const VectorBase* field_raw_data) {
     if (IsSparseFloatVectorDataType(data_type)) {
         index_ = std::make_unique<index::VectorMemIndex<float>>(
+            DataType::NONE,
             config_->GetIndexType(),
             config_->GetMetricType(),
             knowhere::Version::GetCurrentVersion().VersionNumber());
@@ -62,6 +63,7 @@ VectorFieldIndexing::recreate_index(DataType data_type,
             return (const void*)field_raw_data_ptr->get_element(id);
         };
         index_ = std::make_unique<index::VectorMemIndex<float>>(
+            DataType::NONE,
             config_->GetIndexType(),
             config_->GetMetricType(),
             knowhere::Version::GetCurrentVersion().VersionNumber(),
@@ -78,6 +80,7 @@ VectorFieldIndexing::recreate_index(DataType data_type,
             return (const void*)field_raw_data_ptr->get_element(id);
         };
         index_ = std::make_unique<index::VectorMemIndex<float16>>(
+            DataType::NONE,
             config_->GetIndexType(),
             config_->GetMetricType(),
             knowhere::Version::GetCurrentVersion().VersionNumber(),
@@ -94,6 +97,7 @@ VectorFieldIndexing::recreate_index(DataType data_type,
             return (const void*)field_raw_data_ptr->get_element(id);
         };
         index_ = std::make_unique<index::VectorMemIndex<bfloat16>>(
+            DataType::NONE,
             config_->GetIndexType(),
             config_->GetMetricType(),
             knowhere::Version::GetCurrentVersion().VersionNumber(),

internal/core/src/segcore/FieldIndexing.h

@@ -292,8 +292,9 @@ class IndexingRecord {
                     index_meta_->HasFiled(field_id)) {
                     auto vec_field_meta =
                         index_meta_->GetFieldIndexMeta(field_id);
-                    //Disable growing index for flat
-                    if (!vec_field_meta.IsFlatIndex()) {
+                    //Disable growing index for flat and embedding list
+                    if (!vec_field_meta.IsFlatIndex() &&
+                        field_meta.get_data_type() != DataType::VECTOR_ARRAY) {
                         auto field_raw_data =
                             insert_record->get_data_base(field_id);
                         field_indexings_.try_emplace(

internal/core/src/segcore/SegmentGrowingImpl.cpp

@@ -695,12 +695,19 @@ SegmentGrowingImpl::search_batch_pks(
 void
 SegmentGrowingImpl::vector_search(SearchInfo& search_info,
                                   const void* query_data,
+                                  const size_t* query_lims,
                                   int64_t query_count,
                                   Timestamp timestamp,
                                   const BitsetView& bitset,
                                   SearchResult& output) const {
-    query::SearchOnGrowing(
-        *this, search_info, query_data, query_count, timestamp, bitset, output);
+    query::SearchOnGrowing(*this,
+                           search_info,
+                           query_data,
+                           query_lims,
+                           query_count,
+                           timestamp,
+                           bitset,
+                           output);
 }
 
 std::unique_ptr<DataArray>

internal/core/src/segcore/SegmentGrowingImpl.h

@@ -326,6 +326,7 @@ class SegmentGrowingImpl : public SegmentGrowing {
     void
     vector_search(SearchInfo& search_info,
                   const void* query_data,
+                  const size_t* query_lims,
                   int64_t query_count,
                   Timestamp timestamp,
                   const BitsetView& bitset,

internal/core/src/segcore/SegmentInterface.h

@@ -394,9 +394,14 @@ class SegmentInternalInterface : public SegmentInterface {
                          const std::string& nested_path) const override;
 
  public:
+    // `query_lims` is not null only for vector array (embedding list) search
+    // where it denotes the number of vectors in each embedding list. The length
+    // of `query_lims` is the number of queries in the search plus one (the first
+    // element in query_lims is 0).
     virtual void
     vector_search(SearchInfo& search_info,
                   const void* query_data,
+                  const size_t* query_lims,
                   int64_t query_count,
                   Timestamp timestamp,
                   const BitsetView& bitset,

internal/core/src/segcore/Types.h

@@ -35,6 +35,8 @@ struct LoadIndexInfo {
     int64_t segment_id;
     int64_t field_id;
     DataType field_type;
+    // The element type of the field. It's DataType::NONE if field_type is array/vector_array.
+    DataType element_type;
     bool enable_mmap;
     std::string mmap_dir_path;
     int64_t index_id;

internal/core/src/segcore/Utils.cpp

@@ -668,7 +668,11 @@ MergeDataArray(std::vector<MergeBase>& merge_bases,
                 auto obj = vector_array->mutable_int8_vector();
                 obj->assign(data, dim * sizeof(int8));
             } else if (field_meta.get_data_type() == DataType::VECTOR_ARRAY) {
-                ThrowInfo(DataTypeInvalid, "VECTOR_ARRAY is not implemented");
+                auto data = src_field_data->vectors().vector_array();
+                auto obj = vector_array->mutable_vector_array();
+                obj->set_element_type(
+                    proto::schema::DataType(field_meta.get_element_type()));
+                obj->CopyFrom(data);
             } else {
                 ThrowInfo(DataTypeInvalid,
                           fmt::format("unsupported datatype {}", data_type));

internal/core/src/segcore/check_vec_index_c.cpp

@@ -15,7 +15,11 @@
 #include "knowhere/comp/knowhere_check.h"
 
 bool
-CheckVecIndexWithDataType(const char* index_type, enum CDataType data_type) {
+CheckVecIndexWithDataType(const char* index_type,
+                          enum CDataType data_type,
+                          bool is_emb_list_data) {
     return knowhere::KnowhereCheck::IndexTypeAndDataTypeCheck(
-        std::string(index_type), knowhere::VecType(data_type));
+        std::string(index_type),
+        knowhere::VecType(data_type),
+        is_emb_list_data);
 }

internal/core/src/segcore/check_vec_index_c.h

@@ -16,7 +16,9 @@ extern "C" {
 #endif
 #include "common/type_c.h"
 bool
-CheckVecIndexWithDataType(const char* index_type, enum CDataType data_type);
+CheckVecIndexWithDataType(const char* index_type,
+                          enum CDataType data_type,
+                          bool is_emb_list_data);
 
 #ifdef __cplusplus
 }

internal/core/src/segcore/load_index_c.cpp

@@ -98,40 +98,6 @@ AppendIndexParam(CLoadIndexInfo c_load_index_info,
     }
 }
 
-CStatus
-AppendFieldInfo(CLoadIndexInfo c_load_index_info,
-                int64_t collection_id,
-                int64_t partition_id,
-                int64_t segment_id,
-                int64_t field_id,
-                enum CDataType field_type,
-                bool enable_mmap,
-                const char* mmap_dir_path) {
-    SCOPE_CGO_CALL_METRIC();
-
-    try {
-        auto load_index_info =
-            (milvus::segcore::LoadIndexInfo*)c_load_index_info;
-        load_index_info->collection_id = collection_id;
-        load_index_info->partition_id = partition_id;
-        load_index_info->segment_id = segment_id;
-        load_index_info->field_id = field_id;
-        load_index_info->field_type = milvus::DataType(field_type);
-        load_index_info->enable_mmap = enable_mmap;
-        load_index_info->mmap_dir_path = std::string(mmap_dir_path);
-
-        auto status = CStatus();
-        status.error_code = milvus::Success;
-        status.error_msg = "";
-        return status;
-    } catch (std::exception& e) {
-        auto status = CStatus();
-        status.error_code = milvus::UnexpectedError;
-        status.error_msg = strdup(e.what());
-        return status;
-    }
-}
-
 CStatus
 appendVecIndex(CLoadIndexInfo c_load_index_info, CBinarySet c_binary_set) {
     SCOPE_CGO_CALL_METRIC();
@@ -252,6 +218,7 @@ EstimateLoadIndexResource(CLoadIndexInfo c_load_index_info) {
         auto load_index_info =
             (milvus::segcore::LoadIndexInfo*)c_load_index_info;
         auto field_type = load_index_info->field_type;
+        auto element_type = load_index_info->element_type;
         auto& index_params = load_index_info->index_params;
         bool find_index_type =
             index_params.count("index_type") > 0 ? true : false;
@@ -261,6 +228,7 @@ EstimateLoadIndexResource(CLoadIndexInfo c_load_index_info) {
         LoadResourceRequest request =
             milvus::index::IndexFactory::GetInstance().IndexLoadResource(
                 field_type,
+                element_type,
                 load_index_info->index_engine_version,
                 load_index_info->index_size,
                 index_params,
@@ -581,6 +549,8 @@ FinishLoadIndexInfo(CLoadIndexInfo c_load_index_info,
             load_index_info->field_id = info_proto->field().fieldid();
             load_index_info->field_type =
                 static_cast<milvus::DataType>(info_proto->field().data_type());
+            load_index_info->element_type = static_cast<milvus::DataType>(
+                info_proto->field().element_type());
             load_index_info->enable_mmap = info_proto->enable_mmap();
             load_index_info->mmap_dir_path = info_proto->mmap_dir_path();
             load_index_info->index_id = info_proto->indexid();

internal/core/src/segcore/load_index_c.h

@@ -38,16 +38,6 @@ AppendIndexParam(CLoadIndexInfo c_load_index_info,
                  const char* index_key,
                  const char* index_value);
 
-CStatus
-AppendFieldInfo(CLoadIndexInfo c_load_index_info,
-                int64_t collection_id,
-                int64_t partition_id,
-                int64_t segment_id,
-                int64_t field_id,
-                enum CDataType field_type,
-                bool enable_mmap,
-                const char* mmap_dir_path);
-
 LoadResourceRequest
 EstimateLoadIndexResource(CLoadIndexInfo c_load_index_info);
 

internal/core/src/segcore/reduce/Reduce.cpp

@@ -440,7 +440,10 @@ ReduceHelper::GetSearchResultDataSlice(int slice_index) {
                 ->set_element_type(
                     proto::schema::DataType(field_meta.get_element_type()));
         } else if (field_meta.get_data_type() == DataType::VECTOR_ARRAY) {
-            ThrowInfo(NotImplemented, "VECTOR_ARRAY is not implemented");
+            field_data->mutable_vectors()
+                ->mutable_vector_array()
+                ->set_element_type(
+                    proto::schema::DataType(field_meta.get_element_type()));
         }
         search_result_data->mutable_fields_data()->AddAllocated(
             field_data.release());

internal/core/src/segcore/reduce/StreamReduce.cpp

@@ -155,7 +155,10 @@ StreamReducerHelper::AssembleMergedResult() {
                     ->set_element_type(
                         proto::schema::DataType(field_meta.get_element_type()));
             } else if (field_meta.get_data_type() == DataType::VECTOR_ARRAY) {
-                ThrowInfo(NotImplemented, "VECTOR_ARRAY is not implemented");
+                field_data->mutable_vectors()
+                    ->mutable_vector_array()
+                    ->set_element_type(
+                        proto::schema::DataType(field_meta.get_element_type()));
             }
 
             new_merged_result->output_fields_data_[field_id] =
@@ -674,7 +677,10 @@ StreamReducerHelper::GetSearchResultDataSlice(int slice_index) {
                 ->set_element_type(
                     proto::schema::DataType(field_meta.get_element_type()));
         } else if (field_meta.get_data_type() == DataType::VECTOR_ARRAY) {
-            ThrowInfo(NotImplemented, "VECTOR_ARRAY is not implemented");
+            field_data->mutable_vectors()
+                ->mutable_vector_array()
+                ->set_element_type(
+                    proto::schema::DataType(field_meta.get_element_type()));
         }
         search_result_data->mutable_fields_data()->AddAllocated(
             field_data.release());

internal/core/src/segcore/storagev1translator/InterimSealedIndexTranslator.cpp

@@ -81,6 +81,7 @@ InterimSealedIndexTranslator::get_cells(
 
         if (vec_data_type_ == DataType::VECTOR_FLOAT) {
             vec_index = std::make_unique<index::VectorMemIndex<float>>(
+                DataType::NONE,
                 index_type_,
                 metric_type_,
                 knowhere::Version::GetCurrentVersion().VersionNumber(),
@@ -88,6 +89,7 @@ InterimSealedIndexTranslator::get_cells(
                 false);
         } else if (vec_data_type_ == DataType::VECTOR_FLOAT16) {
             vec_index = std::make_unique<index::VectorMemIndex<knowhere::fp16>>(
+                DataType::NONE,
                 index_type_,
                 metric_type_,
                 knowhere::Version::GetCurrentVersion().VersionNumber(),
@@ -95,6 +97,7 @@ InterimSealedIndexTranslator::get_cells(
                 false);
         } else if (vec_data_type_ == DataType::VECTOR_BFLOAT16) {
             vec_index = std::make_unique<index::VectorMemIndex<knowhere::bf16>>(
+                DataType::NONE,
                 index_type_,
                 metric_type_,
                 knowhere::Version::GetCurrentVersion().VersionNumber(),
@@ -103,6 +106,7 @@ InterimSealedIndexTranslator::get_cells(
         }
     } else {
         vec_index = std::make_unique<index::VectorMemIndex<float>>(
+            DataType::NONE,
             index_type_,
             metric_type_,
             knowhere::Version::GetCurrentVersion().VersionNumber(),

internal/core/src/segcore/storagev1translator/SealedIndexTranslator.cpp

@@ -22,6 +22,7 @@ SealedIndexTranslator::SealedIndexTranslator(
       index_load_info_({load_index_info->enable_mmap,
                         load_index_info->mmap_dir_path,
                         load_index_info->field_type,
+                        load_index_info->element_type,
                         load_index_info->index_params,
                         load_index_info->index_size,
                         load_index_info->index_engine_version,
@@ -54,6 +55,7 @@ SealedIndexTranslator::estimated_byte_size_of_cell(
     LoadResourceRequest request =
         milvus::index::IndexFactory::GetInstance().IndexLoadResource(
             index_load_info_.field_type,
+            index_load_info_.element_type,
             index_load_info_.index_engine_version,
             index_load_info_.index_size,
             index_load_info_.index_params,

internal/core/src/segcore/storagev1translator/SealedIndexTranslator.h

@@ -45,6 +45,7 @@ class SealedIndexTranslator
         bool enable_mmap;
         std::string mmap_dir_path;
         DataType field_type;
+        DataType element_type;
         std::map<std::string, std::string> index_params;
         int64_t index_size;
         int64_t index_engine_version;

internal/core/src/segcore/storagev1translator/V1SealedIndexTranslator.cpp

@@ -13,6 +13,7 @@ V1SealedIndexTranslator::V1SealedIndexTranslator(
           load_index_info->enable_mmap,
           load_index_info->mmap_dir_path,
           load_index_info->field_type,
+          load_index_info->element_type,
           load_index_info->index_params,
           load_index_info->index_files,
           load_index_info->index_size,

internal/core/src/segcore/storagev1translator/V1SealedIndexTranslator.h

@@ -44,6 +44,7 @@ class V1SealedIndexTranslator : public Translator<milvus::index::IndexBase> {
         bool enable_mmap;
         std::string mmap_dir_path;
         DataType field_type;
+        DataType element_type;
         std::map<std::string, std::string> index_params;
         std::vector<std::string> index_files;
         int64_t index_size;

internal/core/src/segcore/vector_index_c.cpp

@@ -24,6 +24,7 @@
 CStatus
 ValidateIndexParams(const char* index_type,
                     enum CDataType data_type,
+                    enum CDataType element_type,
                     const uint8_t* serialized_index_params,
                     const uint64_t length) {
     try {
@@ -44,45 +45,64 @@ ValidateIndexParams(const char* index_type,
 
         knowhere::Status status;
         std::string error_msg;
-        if (dataType == milvus::DataType::VECTOR_BINARY) {
-            status = knowhere::IndexStaticFaced<knowhere::bin1>::ConfigCheck(
-                index_type,
-                knowhere::Version::GetCurrentVersion().VersionNumber(),
-                json,
-                error_msg);
-        } else if (dataType == milvus::DataType::VECTOR_FLOAT) {
-            status = knowhere::IndexStaticFaced<knowhere::fp32>::ConfigCheck(
-                index_type,
-                knowhere::Version::GetCurrentVersion().VersionNumber(),
-                json,
-                error_msg);
-        } else if (dataType == milvus::DataType::VECTOR_BFLOAT16) {
-            status = knowhere::IndexStaticFaced<knowhere::bf16>::ConfigCheck(
-                index_type,
-                knowhere::Version::GetCurrentVersion().VersionNumber(),
-                json,
-                error_msg);
-        } else if (dataType == milvus::DataType::VECTOR_FLOAT16) {
-            status = knowhere::IndexStaticFaced<knowhere::fp16>::ConfigCheck(
-                index_type,
-                knowhere::Version::GetCurrentVersion().VersionNumber(),
-                json,
-                error_msg);
-        } else if (dataType == milvus::DataType::VECTOR_SPARSE_FLOAT) {
-            status = knowhere::IndexStaticFaced<knowhere::fp32>::ConfigCheck(
-                index_type,
-                knowhere::Version::GetCurrentVersion().VersionNumber(),
-                json,
-                error_msg);
-        } else if (dataType == milvus::DataType::VECTOR_INT8) {
-            status = knowhere::IndexStaticFaced<knowhere::int8>::ConfigCheck(
-                index_type,
-                knowhere::Version::GetCurrentVersion().VersionNumber(),
-                json,
-                error_msg);
+        auto check_leaf_type = [&index_type, &json, &error_msg, &status](
+                                   milvus::DataType dataType) {
+            if (dataType == milvus::DataType::VECTOR_BINARY) {
+                status =
+                    knowhere::IndexStaticFaced<knowhere::bin1>::ConfigCheck(
+                        index_type,
+                        knowhere::Version::GetCurrentVersion().VersionNumber(),
+                        json,
+                        error_msg);
+            } else if (dataType == milvus::DataType::VECTOR_FLOAT) {
+                status =
+                    knowhere::IndexStaticFaced<knowhere::fp32>::ConfigCheck(
+                        index_type,
+                        knowhere::Version::GetCurrentVersion().VersionNumber(),
+                        json,
+                        error_msg);
+            } else if (dataType == milvus::DataType::VECTOR_BFLOAT16) {
+                status =
+                    knowhere::IndexStaticFaced<knowhere::bf16>::ConfigCheck(
+                        index_type,
+                        knowhere::Version::GetCurrentVersion().VersionNumber(),
+                        json,
+                        error_msg);
+            } else if (dataType == milvus::DataType::VECTOR_FLOAT16) {
+                status =
+                    knowhere::IndexStaticFaced<knowhere::fp16>::ConfigCheck(
+                        index_type,
+                        knowhere::Version::GetCurrentVersion().VersionNumber(),
+                        json,
+                        error_msg);
+            } else if (dataType == milvus::DataType::VECTOR_SPARSE_FLOAT) {
+                status =
+                    knowhere::IndexStaticFaced<knowhere::fp32>::ConfigCheck(
+                        index_type,
+                        knowhere::Version::GetCurrentVersion().VersionNumber(),
+                        json,
+                        error_msg);
+            } else if (dataType == milvus::DataType::VECTOR_INT8) {
+                status =
+                    knowhere::IndexStaticFaced<knowhere::int8>::ConfigCheck(
+                        index_type,
+                        knowhere::Version::GetCurrentVersion().VersionNumber(),
+                        json,
+                        error_msg);
+            } else {
+                status = knowhere::Status::invalid_args;
+            }
+        };
+
+        if (dataType == milvus::DataType::VECTOR_ARRAY) {
+            milvus::DataType elementType(
+                static_cast<milvus::DataType>(element_type));
+
+            check_leaf_type(elementType);
         } else {
-            status = knowhere::Status::invalid_args;
+            check_leaf_type(dataType);
         }
+
         CStatus cStatus;
         if (status == knowhere::Status::success) {
             cStatus.error_code = milvus::Success;

internal/core/src/segcore/vector_index_c.h

@@ -20,6 +20,7 @@ extern "C" {
 CStatus
 ValidateIndexParams(const char* index_type,
                     enum CDataType data_type,
+                    enum CDataType element_type,
                     const uint8_t* index_params,
                     const uint64_t length);
 

internal/core/unittest/test_bf.cpp

@@ -141,7 +141,8 @@ class TestFloatSearchBruteForce : public ::testing::Test {
                                        search_info,
                                        index_info,
                                        bitset_view,
-                                       DataType::VECTOR_FLOAT);
+                                       DataType::VECTOR_FLOAT,
+                                       DataType::NONE);
         for (int i = 0; i < nq; i++) {
             auto ref = Ref(base.data(),
                            query.data() + i * dim,

internal/core/unittest/test_bf_sparse.cpp

@@ -113,7 +113,8 @@ class TestSparseFloatSearchBruteForce : public ::testing::Test {
                                               search_info,
                                               index_info,
                                               bitset_view,
-                                              DataType::VECTOR_SPARSE_FLOAT));
+                                              DataType::VECTOR_SPARSE_FLOAT,
+                                              DataType::NONE));
             return;
         }
         auto result = BruteForceSearch(query_dataset,
@@ -121,7 +122,8 @@ class TestSparseFloatSearchBruteForce : public ::testing::Test {
                                        search_info,
                                        index_info,
                                        bitset_view,
-                                       DataType::VECTOR_SPARSE_FLOAT);
+                                       DataType::VECTOR_SPARSE_FLOAT,
+                                       DataType::NONE);
         for (int i = 0; i < nq; i++) {
             auto ref = SearchRef(base.get(), *(query.get() + i), nb, topk);
             auto ans = result.get_seg_offsets() + i * topk;
@@ -135,7 +137,8 @@ class TestSparseFloatSearchBruteForce : public ::testing::Test {
                                         search_info,
                                         index_info,
                                         bitset_view,
-                                        DataType::VECTOR_SPARSE_FLOAT);
+                                        DataType::VECTOR_SPARSE_FLOAT,
+                                        DataType::NONE);
         for (int i = 0; i < nq; i++) {
             auto ref = RangeSearchRef(
                 base.get(), *(query.get() + i), nb, 0.1, 0.5, topk);

internal/core/unittest/test_c_api.cpp

@@ -1936,6 +1936,7 @@ TEST(CApiTest, LoadIndexSearch) {
     auto& index_params = load_index_info.index_params;
     index_params["index_type"] = knowhere::IndexEnum::INDEX_FAISS_IVFSQ8;
     auto index = std::make_unique<VectorMemIndex<float>>(
+        DataType::NONE,
         index_params["index_type"],
         knowhere::metric::L2,
         knowhere::Version::GetCurrentVersion().VersionNumber());

internal/core/unittest/test_chunked_segment.cpp

@@ -128,6 +128,7 @@ TEST(test_chunk_segment, TestSearchOnSealed) {
                                 search_info,
                                 index_info,
                                 query_data,
+                                nullptr,
                                 1,
                                 total_row_count,
                                 bv,
@@ -153,6 +154,7 @@ TEST(test_chunk_segment, TestSearchOnSealed) {
                                 search_info,
                                 index_info,
                                 query_data,
+                                nullptr,
                                 1,
                                 total_row_count,
                                 bv,

internal/core/unittest/test_expr.cpp

@@ -16674,6 +16674,7 @@ TEST(JsonIndexTest, TestJsonNotEqualExpr) {
     file_manager_ctx.fieldDataMeta.field_schema.set_data_type(
         milvus::proto::schema::JSON);
     file_manager_ctx.fieldDataMeta.field_schema.set_fieldid(json_fid.get());
+    file_manager_ctx.fieldDataMeta.field_id = json_fid.get();
 
     auto inv_index = index::IndexFactory::GetInstance().CreateJsonIndex(
         index::CreateIndexInfo{
@@ -16784,6 +16785,7 @@ TEST_P(JsonIndexExistsTest, TestExistsExpr) {
         milvus::proto::schema::JSON);
     file_manager_ctx.fieldDataMeta.field_schema.set_fieldid(json_fid.get());
     file_manager_ctx.fieldDataMeta.field_schema.set_nullable(true);
+    file_manager_ctx.fieldDataMeta.field_id = json_fid.get();
     auto inv_index = index::IndexFactory::GetInstance().CreateJsonIndex(
         index::CreateIndexInfo{
             .index_type = index::INVERTED_INDEX_TYPE,
@@ -16971,6 +16973,7 @@ TEST_P(JsonIndexBinaryExprTest, TestBinaryRangeExpr) {
     file_manager_ctx.fieldDataMeta.field_schema.set_data_type(
         milvus::proto::schema::JSON);
     file_manager_ctx.fieldDataMeta.field_schema.set_fieldid(json_fid.get());
+    file_manager_ctx.fieldDataMeta.field_id = json_fid.get();
 
     auto inv_index = index::IndexFactory::GetInstance().CreateJsonIndex(
         index::CreateIndexInfo{

internal/core/unittest/test_growing.cpp

@@ -541,3 +541,87 @@ TEST(GrowingTest, LoadVectorArrayData) {
         verify_float_vectors(arrow_array, expected_array);
     }
 }
+
+TEST(GrowingTest, SearchVectorArray) {
+    using namespace milvus::query;
+
+    auto schema = std::make_shared<Schema>();
+    auto metric_type = knowhere::metric::MAX_SIM;
+
+    // Add fields
+    auto int64_field = schema->AddDebugField("int64", DataType::INT64);
+    auto array_vec = schema->AddDebugVectorArrayField(
+        "array_vec", DataType::VECTOR_FLOAT, 128, metric_type);
+    schema->set_primary_field_id(int64_field);
+
+    // Configure segment
+    auto config = SegcoreConfig::default_config();
+    config.set_chunk_rows(1024);
+    config.set_enable_interim_segment_index(true);
+
+    std::map<std::string, std::string> index_params = {
+        {"index_type", knowhere::IndexEnum::INDEX_EMB_LIST_HNSW},
+        {"metric_type", metric_type},
+        {"nlist", "128"}};
+    std::map<std::string, std::string> type_params = {{"dim", "128"}};
+    FieldIndexMeta fieldIndexMeta(
+        array_vec, std::move(index_params), std::move(type_params));
+    std::map<FieldId, FieldIndexMeta> fieldMap = {{array_vec, fieldIndexMeta}};
+
+    IndexMetaPtr metaPtr =
+        std::make_shared<CollectionIndexMeta>(100000, std::move(fieldMap));
+    auto segment = CreateGrowingSegment(schema, metaPtr, 1, config);
+    auto segmentImplPtr = dynamic_cast<SegmentGrowingImpl*>(segment.get());
+
+    // Insert data
+    int64_t N = 100;
+    uint64_t seed = 42;
+    int emb_list_len = 5;  // Each row contains 5 vectors
+    auto dataset = DataGen(schema, N, seed, 0, 1, emb_list_len);
+
+    auto offset = 0;
+    segment->Insert(offset,
+                    N,
+                    dataset.row_ids_.data(),
+                    dataset.timestamps_.data(),
+                    dataset.raw_);
+
+    // Prepare search query
+    int vec_num = 10;  // Total number of query vectors
+    int dim = 128;
+    std::vector<float> query_vec = generate_float_vector(vec_num, dim);
+
+    // Create query dataset with lims for VectorArray
+    std::vector<size_t> query_vec_lims;
+    query_vec_lims.push_back(0);  // First query has 3 vectors
+    query_vec_lims.push_back(3);
+    query_vec_lims.push_back(10);  // Second query has 7 vectors
+
+    // Create search plan
+    const char* raw_plan = R"(vector_anns: <
+                                  field_id: 101
+                                  query_info: <
+                                    topk: 5
+                                    round_decimal: 3
+                                    metric_type: "MAX_SIM"
+                                    search_params: "{\"nprobe\": 10}"
+                                  >
+                                  placeholder_tag: "$0"
+      >)";
+
+    auto plan_str = translate_text_plan_to_binary_plan(raw_plan);
+    auto plan =
+        CreateSearchPlanByExpr(schema, plan_str.data(), plan_str.size());
+
+    // Use CreatePlaceholderGroupFromBlob for VectorArray
+    auto ph_group_raw = CreatePlaceholderGroupFromBlob<EmbListFloatVector>(
+        vec_num, dim, query_vec.data(), query_vec_lims);
+    auto ph_group =
+        ParsePlaceholderGroup(plan.get(), ph_group_raw.SerializeAsString());
+
+    // Execute search
+    Timestamp timestamp = 10000000;
+    auto sr = segment->Search(plan.get(), ph_group.get(), timestamp);
+    auto sr_parsed = SearchResultToJson(*sr);
+    std::cout << sr_parsed.dump(1) << std::endl;
+}

internal/core/unittest/test_growing_index.cpp

@@ -360,6 +360,7 @@ TEST_P(GrowingIndexTest, AddWithoutBuildPool) {
 
     if (data_type == DataType::VECTOR_FLOAT) {
         auto index = std::make_unique<milvus::index::VectorMemIndex<float>>(
+            DataType::NONE,
             index_type,
             metric_type,
             knowhere::Version::GetCurrentVersion().VersionNumber(),
@@ -375,6 +376,7 @@ TEST_P(GrowingIndexTest, AddWithoutBuildPool) {
         EXPECT_EQ(index->Count(), (add_cont + 1) * N);
     } else if (data_type == DataType::VECTOR_FLOAT16) {
         auto index = std::make_unique<milvus::index::VectorMemIndex<float16>>(
+            DataType::NONE,
             index_type,
             metric_type,
             knowhere::Version::GetCurrentVersion().VersionNumber(),
@@ -391,6 +393,7 @@ TEST_P(GrowingIndexTest, AddWithoutBuildPool) {
         EXPECT_EQ(index->Count(), (add_cont + 1) * N);
     } else if (data_type == DataType::VECTOR_BFLOAT16) {
         auto index = std::make_unique<milvus::index::VectorMemIndex<bfloat16>>(
+            DataType::NONE,
             index_type,
             metric_type,
             knowhere::Version::GetCurrentVersion().VersionNumber(),
@@ -407,6 +410,7 @@ TEST_P(GrowingIndexTest, AddWithoutBuildPool) {
         EXPECT_EQ(index->Count(), (add_cont + 1) * N);
     } else if (is_sparse) {
         auto index = std::make_unique<milvus::index::VectorMemIndex<float>>(
+            DataType::NONE,
             index_type,
             metric_type,
             knowhere::Version::GetCurrentVersion().VersionNumber(),

internal/core/unittest/test_indexing.cpp

@@ -184,7 +184,8 @@ TEST(Indexing, BinaryBruteForce) {
                                               search_info,
                                               index_info,
                                               nullptr,
-                                              DataType::VECTOR_BINARY);
+                                              DataType::VECTOR_BINARY,
+                                              DataType::NONE);
 
     SearchResult sr;
     sr.total_nq_ = num_queries;

internal/core/unittest/test_json_flat_index.cpp

@@ -564,6 +564,7 @@ class JsonFlatIndexExprTest : public ::testing::Test {
         file_manager_ctx.fieldDataMeta.field_schema.set_fieldid(
             json_fid_.get());
         file_manager_ctx.fieldDataMeta.field_schema.set_nullable(true);
+        file_manager_ctx.fieldDataMeta.field_id = json_fid_.get();
         auto index = index::IndexFactory::GetInstance().CreateJsonIndex(
             index::CreateIndexInfo{
                 .index_type = index::INVERTED_INDEX_TYPE,

internal/core/unittest/test_ngram_query.cpp

@@ -45,13 +45,13 @@ test_ngram_with_data(const boost::container::vector<std::string>& data,
     auto schema = std::make_shared<Schema>();
     auto field_id = schema->AddDebugField("ngram", DataType::VARCHAR);
 
-    auto field_meta = gen_field_meta(collection_id,
-                                     partition_id,
-                                     segment_id,
-                                     field_id.get(),
-                                     DataType::VARCHAR,
-                                     DataType::NONE,
-                                     false);
+    auto field_meta = milvus::segcore::gen_field_meta(collection_id,
+                                                      partition_id,
+                                                      segment_id,
+                                                      field_id.get(),
+                                                      DataType::VARCHAR,
+                                                      DataType::NONE,
+                                                      false);
     auto index_meta = gen_index_meta(
         segment_id, field_id.get(), index_build_id, index_version);
 

internal/core/unittest/test_sealed.cpp

@@ -19,6 +19,7 @@
 #include "knowhere/version.h"
 #include "storage/RemoteChunkManagerSingleton.h"
 #include "storage/Util.h"
+#include "common/VectorArray.h"
 
 #include "test_utils/cachinglayer_test_utils.h"
 #include "test_utils/DataGen.h"
@@ -2333,3 +2334,257 @@ TEST(Sealed, QueryVectorArrayAllFields) {
     EXPECT_EQ(int64_result->valid_data_size(), 0);
     EXPECT_EQ(array_float_vector_result->valid_data_size(), 0);
 }
+
+TEST(Sealed, SearchVectorArray) {
+    int64_t collection_id = 1;
+    int64_t partition_id = 2;
+    int64_t segment_id = 3;
+    int64_t index_build_id = 4000;
+    int64_t index_version = 4000;
+    int64_t index_id = 5000;
+
+    auto schema = std::make_shared<Schema>();
+    auto metric_type = knowhere::metric::L2;
+    auto int64_field = schema->AddDebugField("int64", DataType::INT64);
+    auto array_vec = schema->AddDebugVectorArrayField(
+        "array_vec", DataType::VECTOR_FLOAT, 128, metric_type);
+    schema->set_primary_field_id(int64_field);
+
+    auto field_meta = milvus::segcore::gen_field_meta(collection_id,
+                                                      partition_id,
+                                                      segment_id,
+                                                      array_vec.get(),
+                                                      DataType::VECTOR_ARRAY,
+                                                      DataType::VECTOR_FLOAT,
+                                                      false);
+    auto index_meta = gen_index_meta(
+        segment_id, array_vec.get(), index_build_id, index_version);
+
+    std::map<FieldId, FieldIndexMeta> filedMap{};
+    IndexMetaPtr metaPtr =
+        std::make_shared<CollectionIndexMeta>(100000, std::move(filedMap));
+
+    int64_t dataset_size = 1000;
+    int64_t dim = 128;
+    auto emb_list_len = 10;
+    auto dataset = DataGen(schema, dataset_size, 42, 0, 1, emb_list_len);
+
+    // create field data
+    std::string root_path = "/tmp/test-vector-array/";
+    auto storage_config = gen_local_storage_config(root_path);
+    auto cm = CreateChunkManager(storage_config);
+    auto vec_array_col = dataset.get_col<VectorFieldProto>(array_vec);
+    std::vector<milvus::VectorArray> vector_arrays;
+    for (auto& v : vec_array_col) {
+        vector_arrays.push_back(milvus::VectorArray(v));
+    }
+    auto field_data = storage::CreateFieldData(DataType::VECTOR_ARRAY, false);
+    field_data->FillFieldData(vector_arrays.data(), vector_arrays.size());
+
+    // create sealed segment
+    auto segment = CreateSealedSegment(schema);
+    auto field_data_info = PrepareSingleFieldInsertBinlog(collection_id,
+                                                          partition_id,
+                                                          segment_id,
+                                                          array_vec.get(),
+                                                          {field_data},
+                                                          cm);
+    segment->LoadFieldData(field_data_info);
+
+    // serialize bin logs
+    auto payload_reader =
+        std::make_shared<milvus::storage::PayloadReader>(field_data);
+    storage::InsertData insert_data(payload_reader);
+    insert_data.SetFieldDataMeta(field_meta);
+    insert_data.SetTimestamps(0, 100);
+
+    auto serialized_bytes = insert_data.Serialize(storage::Remote);
+
+    auto get_binlog_path = [=](int64_t log_id) {
+        return fmt::format("{}/{}/{}/{}/{}",
+                           collection_id,
+                           partition_id,
+                           segment_id,
+                           array_vec.get(),
+                           log_id);
+    };
+
+    auto log_path = get_binlog_path(0);
+
+    auto cm_w = ChunkManagerWrapper(cm);
+    cm_w.Write(log_path, serialized_bytes.data(), serialized_bytes.size());
+
+    storage::FileManagerContext ctx(field_meta, index_meta, cm);
+    std::vector<std::string> index_files;
+
+    // create index
+    milvus::index::CreateIndexInfo create_index_info;
+    create_index_info.field_type = DataType::VECTOR_ARRAY;
+    create_index_info.metric_type = knowhere::metric::MAX_SIM;
+    create_index_info.index_type = knowhere::IndexEnum::INDEX_EMB_LIST_HNSW;
+    create_index_info.index_engine_version =
+        knowhere::Version::GetCurrentVersion().VersionNumber();
+
+    auto emb_list_hnsw_index =
+        milvus::index::IndexFactory::GetInstance().CreateIndex(
+            create_index_info,
+            storage::FileManagerContext(field_meta, index_meta, cm));
+
+    // build index
+    Config config;
+    config[milvus::index::INDEX_TYPE] =
+        knowhere::IndexEnum::INDEX_EMB_LIST_HNSW;
+    config[INSERT_FILES_KEY] = std::vector<std::string>{log_path};
+    config[knowhere::meta::METRIC_TYPE] = create_index_info.metric_type;
+    config[knowhere::indexparam::M] = "16";
+    config[knowhere::indexparam::EF] = "10";
+    config[DIM_KEY] = dim;
+    emb_list_hnsw_index->Build(config);
+
+    auto vec_index =
+        dynamic_cast<milvus::index::VectorIndex*>(emb_list_hnsw_index.get());
+    EXPECT_EQ(vec_index->Count(), dataset_size * emb_list_len);
+    EXPECT_EQ(vec_index->GetDim(), dim);
+
+    // search
+    auto vec_num = 10;
+    std::vector<float> query_vec = generate_float_vector(vec_num, dim);
+    auto query_dataset = knowhere::GenDataSet(vec_num, dim, query_vec.data());
+    std::vector<size_t> query_vec_lims;
+    query_vec_lims.push_back(0);
+    query_vec_lims.push_back(3);
+    query_vec_lims.push_back(10);
+    query_dataset->SetLims(query_vec_lims.data());
+
+    auto search_conf = knowhere::Json{{knowhere::indexparam::NPROBE, 10}};
+    milvus::SearchInfo searchInfo;
+    searchInfo.topk_ = 5;
+    searchInfo.metric_type_ = knowhere::metric::L2;
+    searchInfo.search_params_ = search_conf;
+    SearchResult result;
+    vec_index->Query(query_dataset, searchInfo, nullptr, result);
+    auto ref_result = SearchResultToJson(result);
+    std::cout << ref_result.dump(1) << std::endl;
+    EXPECT_EQ(result.total_nq_, 2);
+    EXPECT_EQ(result.distances_.size(), 2 * searchInfo.topk_);
+
+    // create sealed segment
+    auto sealed_segment = CreateSealedWithFieldDataLoaded(schema, dataset);
+
+    // brute force search
+    {
+        const char* raw_plan = R"(vector_anns: <
+                                    field_id: 101
+                                    query_info: <
+                                      topk: 5
+                                      round_decimal: 3
+                                      metric_type: "MAX_SIM"
+                                      search_params: "{\"nprobe\": 10}"
+                                    >
+                                    placeholder_tag: "$0"
+        >)";
+        auto plan_str = translate_text_plan_to_binary_plan(raw_plan);
+        auto plan =
+            CreateSearchPlanByExpr(schema, plan_str.data(), plan_str.size());
+        auto ph_group_raw = CreatePlaceholderGroupFromBlob<EmbListFloatVector>(
+            vec_num, dim, query_vec.data(), query_vec_lims);
+        auto ph_group =
+            ParsePlaceholderGroup(plan.get(), ph_group_raw.SerializeAsString());
+        Timestamp timestamp = 1000000;
+        std::vector<const PlaceholderGroup*> ph_group_arr = {ph_group.get()};
+
+        auto sr = sealed_segment->Search(plan.get(), ph_group.get(), timestamp);
+        auto sr_parsed = SearchResultToJson(*sr);
+        std::cout << sr_parsed.dump(1) << std::endl;
+    }
+
+    // // brute force search with iterative filter
+    // {
+    //     auto [min, max] =
+    //         std::minmax_element(int_values.begin(), int_values.end());
+    //     auto min_val = *min;
+    //     auto max_val = *max;
+
+    //     auto raw_plan = fmt::format(R"(vector_anns: <
+    //                                 field_id: 101
+    //                                 predicates: <
+    //                                   binary_range_expr: <
+    //                                     column_info: <
+    //                                       field_id: 100
+    //                                       data_type: Int64
+    //                                     >
+    //                                     lower_inclusive: true
+    //                                     upper_inclusive: true
+    //                                     lower_value: <
+    //                                       int64_val: {}
+    //                                     >
+    //                                     upper_value: <
+    //                                       int64_val: {}
+    //                                     >
+    //                                   >
+    //                                 >
+    //                                 query_info: <
+    //                                   topk: 5
+    //                                   round_decimal: 3
+    //                                   metric_type: "MAX_SIM"
+    //                                   hints: "iterative_filter"
+    //                                   search_params: "{{\"nprobe\": 10}}"
+    //                                 >
+    //                                 placeholder_tag: "$0"
+    //                               >)",
+    //                                 min_val,
+    //                                 max_val);
+    //     auto plan_str = translate_text_plan_to_binary_plan(raw_plan.c_str());
+    //     auto plan =
+    //         CreateSearchPlanByExpr(schema, plan_str.data(), plan_str.size());
+    //     auto ph_group_raw = CreatePlaceholderGroupFromBlob<EmbListFloatVector>(
+    //         vec_num, dim, query_vec.data(), query_vec_lims);
+    //     auto ph_group =
+    //         ParsePlaceholderGroup(plan.get(), ph_group_raw.SerializeAsString());
+    //     Timestamp timestamp = 1000000;
+    //     std::vector<const PlaceholderGroup*> ph_group_arr = {ph_group.get()};
+
+    //     auto sr = sealed_segment->Search(plan.get(), ph_group.get(), timestamp);
+    //     auto sr_parsed = SearchResultToJson(*sr);
+    //     std::cout << sr_parsed.dump(1) << std::endl;
+    // }
+
+    // search with index
+    {
+        LoadIndexInfo load_info;
+        load_info.field_id = array_vec.get();
+        load_info.field_type = DataType::VECTOR_ARRAY;
+        load_info.element_type = DataType::VECTOR_FLOAT;
+        load_info.index_params = GenIndexParams(emb_list_hnsw_index.get());
+        load_info.cache_index =
+            CreateTestCacheIndex("test", std::move(emb_list_hnsw_index));
+        load_info.index_params["metric_type"] = knowhere::metric::MAX_SIM;
+
+        sealed_segment->DropFieldData(array_vec);
+        sealed_segment->LoadIndex(load_info);
+
+        const char* raw_plan = R"(vector_anns: <
+                                    field_id: 101
+                                    query_info: <
+                                      topk: 5
+                                      round_decimal: 3
+                                      metric_type: "MAX_SIM"
+                                      search_params: "{\"nprobe\": 10}"
+                                    >
+                                    placeholder_tag: "$0"
+        >)";
+        auto plan_str = translate_text_plan_to_binary_plan(raw_plan);
+        auto plan =
+            CreateSearchPlanByExpr(schema, plan_str.data(), plan_str.size());
+        auto ph_group_raw = CreatePlaceholderGroupFromBlob<EmbListFloatVector>(
+            vec_num, dim, query_vec.data(), query_vec_lims);
+        auto ph_group =
+            ParsePlaceholderGroup(plan.get(), ph_group_raw.SerializeAsString());
+        Timestamp timestamp = 1000000;
+        std::vector<const PlaceholderGroup*> ph_group_arr = {ph_group.get()};
+
+        auto sr = sealed_segment->Search(plan.get(), ph_group.get(), timestamp);
+        auto sr_parsed = SearchResultToJson(*sr);
+        std::cout << sr_parsed.dump(1) << std::endl;
+    }
+}

internal/core/unittest/test_string_expr.cpp

@@ -1562,7 +1562,8 @@ TEST(AlwaysTrueStringPlan, SearchWithOutputFields) {
                                        search_info,
                                        index_info,
                                        nullptr,
-                                       DataType::VECTOR_FLOAT);
+                                       DataType::VECTOR_FLOAT,
+                                       DataType::NONE);
 
     auto sr = segment->Search(plan.get(), ph_group.get(), MAX_TIMESTAMP);
     segment->FillPrimaryKeys(plan.get(), *sr);

internal/core/unittest/test_utils/DataGen.h

@@ -1004,7 +1004,10 @@ CreatePlaceholderGroup(int64_t num_queries, int dim, int64_t seed = 42) {
 
 template <class TraitType = milvus::FloatVector>
 inline auto
-CreatePlaceholderGroupFromBlob(int64_t num_queries, int dim, const void* src) {
+CreatePlaceholderGroupFromBlob(int64_t num_queries,
+                               int dim,
+                               const void* src,
+                               std::vector<size_t> offsets = {}) {
     if (std::is_same_v<TraitType, milvus::BinaryVector>) {
         assert(dim % 8 == 0);
     }
@@ -1017,12 +1020,27 @@ CreatePlaceholderGroupFromBlob(int64_t num_queries, int dim, const void* src) {
     value->set_type(TraitType::placeholder_type);
     int64_t src_index = 0;
 
-    for (int i = 0; i < num_queries; ++i) {
-        std::vector<elem_type> vec;
-        for (int d = 0; d < dim / TraitType::dim_factor; ++d) {
-            vec.push_back(((elem_type*)src)[src_index++]);
+    if (offsets.empty()) {
+        for (int i = 0; i < num_queries; ++i) {
+            std::vector<elem_type> vec;
+            for (int d = 0; d < dim / TraitType::dim_factor; ++d) {
+                vec.push_back(((elem_type*)src)[src_index++]);
+            }
+            value->add_values(vec.data(), vec.size() * sizeof(elem_type));
+        }
+    } else {
+        assert(offsets.back() == num_queries);
+        for (int i = 0; i < offsets.size() - 1; i++) {
+            auto start = offsets[i];
+            auto end = offsets[i + 1];
+            std::vector<elem_type> vec;
+            for (int j = start; j < end; j++) {
+                for (int d = 0; d < dim / TraitType::dim_factor; ++d) {
+                    vec.push_back(((elem_type*)src)[src_index++]);
+                }
+            }
+            value->add_values(vec.data(), vec.size() * sizeof(elem_type));
         }
-        value->add_values(vec.data(), vec.size() * sizeof(elem_type));
     }
     return raw_group;
 }
@@ -1362,6 +1380,7 @@ GenVecIndexing(int64_t N,
     milvus::storage::FileManagerContext file_manager_context(
         field_data_meta, index_meta, chunk_manager);
     auto indexing = std::make_unique<index::VectorMemIndex<float>>(
+        DataType::NONE,
         index_type,
         knowhere::metric::L2,
         knowhere::Version::GetCurrentVersion().VersionNumber(),
@@ -1631,4 +1650,29 @@ GenChunkedSegmentTestSchema(bool pk_is_string) {
     return schema;
 }
 
+inline std::vector<float>
+generate_float_vector(int64_t N, int64_t dim) {
+    auto seed = 42;
+    auto offset = 0;
+    std::vector<float> final(dim * N);
+    for (int n = 0; n < N; ++n) {
+        std::vector<float> data(dim);
+        float sum = 0;
+
+        std::default_random_engine er2(seed + n);
+        std::normal_distribution<> distr2(0, 1);
+        for (auto& x : data) {
+            x = distr2(er2) + offset++;
+            sum += x * x;
+        }
+        sum = sqrt(sum);
+        for (auto& x : data) {
+            x /= sum;
+        }
+
+        std::copy(data.begin(), data.end(), final.begin() + dim * n);
+    }
+    return final;
+};
+
 }  // namespace milvus::segcore

internal/core/unittest/test_utils/c_api_test_utils.h

@@ -31,6 +31,7 @@
 #include "segcore/segment_c.h"
 #include "futures/Future.h"
 #include "futures/future_c.h"
+#include "segcore/load_index_c.h"
 #include "DataGen.h"
 #include "PbHelper.h"
 #include "indexbuilder_test_utils.h"
@@ -38,6 +39,39 @@
 using namespace milvus;
 using namespace milvus::segcore;
 
+// Test utility function for AppendFieldInfo
+inline CStatus
+AppendFieldInfo(CLoadIndexInfo c_load_index_info,
+                int64_t collection_id,
+                int64_t partition_id,
+                int64_t segment_id,
+                int64_t field_id,
+                enum CDataType field_type,
+                bool enable_mmap,
+                const char* mmap_dir_path) {
+    try {
+        auto load_index_info =
+            (milvus::segcore::LoadIndexInfo*)c_load_index_info;
+        load_index_info->collection_id = collection_id;
+        load_index_info->partition_id = partition_id;
+        load_index_info->segment_id = segment_id;
+        load_index_info->field_id = field_id;
+        load_index_info->field_type = milvus::DataType(field_type);
+        load_index_info->enable_mmap = enable_mmap;
+        load_index_info->mmap_dir_path = std::string(mmap_dir_path);
+
+        auto status = CStatus();
+        status.error_code = milvus::Success;
+        status.error_msg = "";
+        return status;
+    } catch (std::exception& e) {
+        auto status = CStatus();
+        status.error_code = milvus::UnexpectedError;
+        status.error_msg = strdup(e.what());
+        return status;
+    }
+}
+
 namespace {
 
 std::string

internal/core/unittest/test_vector_array.cpp

@@ -109,7 +109,7 @@ TEST(VectorArray, TestConstructVectorArray) {
     field_float_vector_array.mutable_float_vector()->mutable_data()->Add(
         data.begin(), data.end());
 
-    auto float_vector_array = VectorArray(field_float_vector_array);
+    auto float_vector_array = milvus::VectorArray(field_float_vector_array);
     ASSERT_EQ(float_vector_array.length(), N);
     ASSERT_EQ(float_vector_array.dim(), dim);
     ASSERT_EQ(float_vector_array.get_element_type(), DataType::VECTOR_FLOAT);
@@ -117,16 +117,16 @@ TEST(VectorArray, TestConstructVectorArray) {
 
     ASSERT_TRUE(float_vector_array.is_same_array(field_float_vector_array));
 
-    auto float_vector_array_tmp = VectorArray(float_vector_array);
+    auto float_vector_array_tmp = milvus::VectorArray(float_vector_array);
 
     ASSERT_TRUE(float_vector_array_tmp.is_same_array(field_float_vector_array));
 
     auto float_vector_array_view =
-        VectorArrayView(const_cast<char*>(float_vector_array.data()),
-                        float_vector_array.length(),
-                        float_vector_array.dim(),
-                        float_vector_array.byte_size(),
-                        float_vector_array.get_element_type());
+        milvus::VectorArrayView(const_cast<char*>(float_vector_array.data()),
+                                float_vector_array.length(),
+                                float_vector_array.dim(),
+                                float_vector_array.byte_size(),
+                                float_vector_array.get_element_type());
 
     ASSERT_TRUE(
         float_vector_array_view.is_same_array(field_float_vector_array));

internal/datacoord/index_service.go

@@ -70,13 +70,29 @@ func (s *Server) getSchema(ctx context.Context, collID int64) (*schemapb.Collect
 	return resp.GetSchema(), nil
 }
 
-func isJsonField(schema *schemapb.CollectionSchema, fieldID int64) (bool, error) {
+func FieldExists(schema *schemapb.CollectionSchema, fieldID int64) bool {
 	for _, f := range schema.Fields {
 		if f.FieldID == fieldID {
-			return typeutil.IsJSONType(f.DataType), nil
+			return true
 		}
 	}
-	return false, merr.WrapErrFieldNotFound(fieldID)
+	for _, structField := range schema.StructArrayFields {
+		for _, f := range structField.Fields {
+			if f.FieldID == fieldID {
+				return true
+			}
+		}
+	}
+	return false
+}
+
+func isJsonField(schema *schemapb.CollectionSchema, fieldID int64) bool {
+	for _, f := range schema.Fields {
+		if f.FieldID == fieldID {
+			return typeutil.IsJSONType(f.DataType)
+		}
+	}
+	return false
 }
 
 func getIndexParam(indexParams []*commonpb.KeyValuePair, key string) (string, error) {
@@ -154,11 +170,12 @@ func (s *Server) CreateIndex(ctx context.Context, req *indexpb.CreateIndexReques
 	if err != nil {
 		return merr.Status(err), nil
 	}
-	isJson, err := isJsonField(schema, req.GetFieldID())
-	if err != nil {
-		return merr.Status(err), nil
+
+	if !FieldExists(schema, req.GetFieldID()) {
+		return merr.Status(merr.WrapErrFieldNotFound(req.GetFieldID())), nil
 	}
 
+	isJson := isJsonField(schema, req.GetFieldID())
 	if isJson {
 		// check json_path and json_cast_type exist
 		jsonPath, err := getIndexParam(req.GetIndexParams(), common.JSONPathKey)

internal/datacoord/task_index.go

@@ -250,7 +250,8 @@ func (it *indexBuildTask) prepareJobRequest(ctx context.Context, segment *Segmen
 	schema := collectionInfo.Schema
 	var field *schemapb.FieldSchema
 
-	for _, f := range schema.Fields {
+	allFields := typeutil.GetAllFieldSchemas(schema)
+	for _, f := range allFields {
 		if f.FieldID == fieldID {
 			field = f
 			break
@@ -263,7 +264,11 @@ func (it *indexBuildTask) prepareJobRequest(ctx context.Context, segment *Segmen
 
 	// Extract dim only for vector types to avoid unnecessary warnings
 	dim := -1
-	if typeutil.IsFixDimVectorType(field.GetDataType()) {
+	dataType := field.GetDataType()
+	if typeutil.IsVectorArrayType(dataType) {
+		dataType = field.GetElementType()
+	}
+	if typeutil.IsFixDimVectorType(dataType) {
 		if dimVal, err := storage.GetDimFromParams(field.GetTypeParams()); err != nil {
 			log.Warn("failed to get dim from field type params",
 				zap.String("field type", field.GetDataType().String()), zap.Error(err))

internal/parser/planparserv2/plan_parser_v2.go

@@ -180,6 +180,7 @@ func CreateSearchPlan(schema *typeutil.SchemaHelper, exprStr string, vectorField
 	// plan ok with schema, check ann field
 	fieldID := vectorField.FieldID
 	dataType := vectorField.DataType
+	elementType := vectorField.ElementType
 
 	var vectorType planpb.VectorType
 	if !typeutil.IsVectorType(dataType) {
@@ -198,6 +199,15 @@ func CreateSearchPlan(schema *typeutil.SchemaHelper, exprStr string, vectorField
 		vectorType = planpb.VectorType_SparseFloatVector
 	case schemapb.DataType_Int8Vector:
 		vectorType = planpb.VectorType_Int8Vector
+	case schemapb.DataType_ArrayOfVector:
+		switch elementType {
+		case schemapb.DataType_FloatVector:
+			vectorType = planpb.VectorType_EmbListFloatVector
+		default:
+			log.Error("Invalid elementType", zap.Any("elementType", elementType))
+			return nil, err
+		}
+
 	default:
 		log.Error("Invalid dataType", zap.Any("dataType", dataType))
 		return nil, err

internal/proxy/cgo_util.go

@@ -62,13 +62,18 @@ func GetDynamicPool() *conc.Pool[any] {
 	return dp.Load()
 }
 
-func CheckVecIndexWithDataTypeExist(name string, dType schemapb.DataType) bool {
+func CheckVecIndexWithDataTypeExist(name string, dataType schemapb.DataType, elementType schemapb.DataType) bool {
+	isEmbeddingList := dataType == schemapb.DataType_ArrayOfVector
+	if isEmbeddingList {
+		dataType = elementType
+	}
+
 	var result bool
 	GetDynamicPool().Submit(func() (any, error) {
 		cIndexName := C.CString(name)
-		cType := uint32(dType)
+		cType := uint32(dataType)
 		defer C.free(unsafe.Pointer(cIndexName))
-		result = bool(C.CheckVecIndexWithDataType(cIndexName, cType))
+		result = bool(C.CheckVecIndexWithDataType(cIndexName, cType, C.bool(isEmbeddingList)))
 		return nil, nil
 	}).Await()
 

internal/proxy/cgo_util_test.go

@@ -50,7 +50,7 @@ func Test_CheckVecIndexWithDataTypeExist(t *testing.T) {
 	}
 
 	for _, test := range cases {
-		if got := CheckVecIndexWithDataTypeExist(test.indexType, test.dataType); got != test.want {
+		if got := CheckVecIndexWithDataTypeExist(test.indexType, test.dataType, schemapb.DataType_None); got != test.want {
 			t.Errorf("CheckVecIndexWithDataTypeExist(%v, %v) = %v", test.indexType, test.dataType, test.want)
 		}
 	}

internal/proxy/search_pipeline.go

@@ -549,24 +549,24 @@ func (op *lambdaOperator) run(ctx context.Context, span trace.Span, inputs ...an
 
 type filterFieldOperator struct {
 	outputFieldNames []string
-	schema           *schemaInfo
+	fieldSchemas     []*schemapb.FieldSchema
 }
 
 func newFilterFieldOperator(t *searchTask, _ map[string]any) (operator, error) {
 	return &filterFieldOperator{
 		outputFieldNames: t.translatedOutputFields,
-		schema:           t.schema,
+		fieldSchemas:     typeutil.GetAllFieldSchemas(t.schema.CollectionSchema),
 	}, nil
 }
 
 func (op *filterFieldOperator) run(ctx context.Context, span trace.Span, inputs ...any) ([]any, error) {
 	result := inputs[0].(*milvuspb.SearchResults)
 	for _, retField := range result.Results.FieldsData {
-		for _, schemaField := range op.schema.Fields {
-			if retField != nil && retField.FieldId == schemaField.FieldID {
-				retField.FieldName = schemaField.Name
-				retField.Type = schemaField.DataType
-				retField.IsDynamic = schemaField.IsDynamic
+		for _, fieldSchema := range op.fieldSchemas {
+			if retField != nil && retField.FieldId == fieldSchema.FieldID {
+				retField.FieldName = fieldSchema.Name
+				retField.Type = fieldSchema.DataType
+				retField.IsDynamic = fieldSchema.IsDynamic
 			}
 		}
 	}

internal/proxy/search_pipeline_test.go

@@ -560,6 +560,79 @@ func (s *SearchPipelineSuite) TestHybridSearchPipe() {
 	s.Len(results.Results.Scores, 20) // 2 queries * 10 topk
 }
 
+func (s *SearchPipelineSuite) TestFilterFieldOperatorWithStructArrayFields() {
+	schema := &schemapb.CollectionSchema{
+		Fields: []*schemapb.FieldSchema{
+			{FieldID: 100, Name: "int64", DataType: schemapb.DataType_Int64, IsPrimaryKey: true},
+			{FieldID: 101, Name: "intField", DataType: schemapb.DataType_Int64},
+			{FieldID: 102, Name: "floatField", DataType: schemapb.DataType_Float},
+		},
+		StructArrayFields: []*schemapb.StructArrayFieldSchema{
+			{
+				Name: "structArray",
+				Fields: []*schemapb.FieldSchema{
+					{FieldID: 104, Name: "structArrayField", DataType: schemapb.DataType_Array, ElementType: schemapb.DataType_Int32},
+					{FieldID: 105, Name: "structVectorField", DataType: schemapb.DataType_ArrayOfVector, ElementType: schemapb.DataType_FloatVector},
+				},
+			},
+		},
+	}
+
+	task := &searchTask{
+		schema: &schemaInfo{
+			CollectionSchema: schema,
+		},
+		translatedOutputFields: []string{"intField", "floatField", "structArrayField", "structVectorField"},
+	}
+
+	op, err := newFilterFieldOperator(task, nil)
+	s.NoError(err)
+
+	// Create mock search results with fields including struct array fields
+	searchResults := &milvuspb.SearchResults{
+		Results: &schemapb.SearchResultData{
+			FieldsData: []*schemapb.FieldData{
+				{FieldId: 101}, // intField
+				{FieldId: 102}, // floatField
+				{FieldId: 104}, // structArrayField
+				{FieldId: 105}, // structVectorField
+			},
+		},
+	}
+
+	results, err := op.run(context.Background(), s.span, searchResults)
+	s.NoError(err)
+	s.NotNil(results)
+
+	resultData := results[0].(*milvuspb.SearchResults)
+	s.NotNil(resultData.Results.FieldsData)
+	s.Len(resultData.Results.FieldsData, 4)
+
+	// Verify all fields including struct array fields got their names and types set
+	for _, field := range resultData.Results.FieldsData {
+		switch field.FieldId {
+		case 101:
+			s.Equal("intField", field.FieldName)
+			s.Equal(schemapb.DataType_Int64, field.Type)
+			s.False(field.IsDynamic)
+		case 102:
+			s.Equal("floatField", field.FieldName)
+			s.Equal(schemapb.DataType_Float, field.Type)
+			s.False(field.IsDynamic)
+		case 104:
+			// Struct array field should be handled by GetAllFieldSchemas
+			s.Equal("structArrayField", field.FieldName)
+			s.Equal(schemapb.DataType_Array, field.Type)
+			s.False(field.IsDynamic)
+		case 105:
+			// Struct array vector field should be handled by GetAllFieldSchemas
+			s.Equal("structVectorField", field.FieldName)
+			s.Equal(schemapb.DataType_ArrayOfVector, field.Type)
+			s.False(field.IsDynamic)
+		}
+	}
+}
+
 func (s *SearchPipelineSuite) TestHybridSearchWithRequeryPipe() {
 	task := getHybridSearchTask("test_collection", [][]string{
 		{"1", "2"},

internal/proxy/search_util.go

@@ -281,6 +281,28 @@ func parseSearchInfo(searchParamsPair []*commonpb.KeyValuePair, schema *schemapb
 		return nil, fmt.Errorf("parse iterator v2 info failed: %w", err)
 	}
 
+	// 7. check search for embedding list
+	annsFieldName, _ := funcutil.GetAttrByKeyFromRepeatedKV(AnnsFieldKey, searchParamsPair)
+	if annsFieldName != "" {
+		annField := typeutil.GetFieldByName(schema, annsFieldName)
+		if annField != nil && annField.GetDataType() == schemapb.DataType_ArrayOfVector {
+			if strings.Contains(searchParamStr, radiusKey) {
+				return nil, merr.WrapErrParameterInvalid("", "",
+					"range search is not supported for vector array (embedding list) fields, fieldName: %s", annsFieldName)
+			}
+
+			if groupByFieldId > 0 {
+				return nil, merr.WrapErrParameterInvalid("", "",
+					"group by search is not supported for vector array (embedding list) fields, fieldName: %s", annsFieldName)
+			}
+
+			if isIterator {
+				return nil, merr.WrapErrParameterInvalid("", "",
+					"search iterator is not supported for vector array (embedding list) fields, fieldName: %s", annsFieldName)
+			}
+		}
+	}
+
 	return &SearchInfo{
 		planInfo: &planpb.QueryInfo{
 			Topk:                 queryTopK,

internal/proxy/task.go

@@ -68,6 +68,8 @@ const (
 	RoundDecimalKey      = "round_decimal"
 	OffsetKey            = "offset"
 	LimitKey             = "limit"
+	// offsets for embedding list search
+	LimsKey = "lims"
 
 	SearchIterV2Key        = "search_iter_v2"
 	SearchIterBatchSizeKey = "search_iter_batch_size"
@@ -2047,7 +2049,8 @@ func (t *loadCollectionTask) Execute(ctx context.Context) (err error) {
 
 	loadFieldsSet := typeutil.NewSet(loadFields...)
 	unindexedVecFields := make([]string, 0)
-	for _, field := range collSchema.GetFields() {
+	allFields := typeutil.GetAllFieldSchemas(collSchema.CollectionSchema)
+	for _, field := range allFields {
 		if typeutil.IsVectorType(field.GetDataType()) && loadFieldsSet.Contain(field.GetFieldID()) {
 			if _, ok := fieldIndexIDs[field.GetFieldID()]; !ok {
 				unindexedVecFields = append(unindexedVecFields, field.GetName())
@@ -2055,8 +2058,6 @@ func (t *loadCollectionTask) Execute(ctx context.Context) (err error) {
 		}
 	}
 
-	// todo(SpadeA): check vector field in StructArrayField when index is implemented
-
 	if len(unindexedVecFields) != 0 {
 		errMsg := fmt.Sprintf("there is no vector index on field: %v, please create index firstly", unindexedVecFields)
 		log.Debug(errMsg)
@@ -2305,7 +2306,8 @@ func (t *loadPartitionsTask) Execute(ctx context.Context) error {
 
 	loadFieldsSet := typeutil.NewSet(loadFields...)
 	unindexedVecFields := make([]string, 0)
-	for _, field := range collSchema.GetFields() {
+	allFields := typeutil.GetAllFieldSchemas(collSchema.CollectionSchema)
+	for _, field := range allFields {
 		if typeutil.IsVectorType(field.GetDataType()) && loadFieldsSet.Contain(field.GetFieldID()) {
 			if _, ok := fieldIndexIDs[field.GetFieldID()]; !ok {
 				unindexedVecFields = append(unindexedVecFields, field.GetName())
@@ -2313,8 +2315,6 @@ func (t *loadPartitionsTask) Execute(ctx context.Context) error {
 		}
 	}
 
-	// todo(SpadeA): check vector field in StructArrayField when index is implemented
-
 	if len(unindexedVecFields) != 0 {
 		errMsg := fmt.Sprintf("there is no vector index on field: %v, please create index firstly", unindexedVecFields)
 		log.Ctx(ctx).Debug(errMsg)

internal/proxy/task_index.go

@@ -202,10 +202,12 @@ func (cit *createIndexTask) parseIndexParams(ctx context.Context) error {
 
 	specifyIndexType, exist := indexParamsMap[common.IndexTypeKey]
 	if exist && specifyIndexType != "" {
+		// todo(SpadeA): mmap check for struct array index
 		if err := indexparamcheck.ValidateMmapIndexParams(specifyIndexType, indexParamsMap); err != nil {
 			log.Ctx(ctx).Warn("Invalid mmap type params", zap.String(common.IndexTypeKey, specifyIndexType), zap.Error(err))
 			return merr.WrapErrParameterInvalidMsg("invalid mmap type params: %s", err.Error())
 		}
+		// todo(SpadeA): check for struct array index
 		checker, err := indexparamcheck.GetIndexCheckerMgrInstance().GetChecker(specifyIndexType)
 		// not enable hybrid index for user, used in milvus internally
 		if err != nil || indexparamcheck.IsHYBRIDChecker(checker) {
@@ -327,16 +329,20 @@ func (cit *createIndexTask) parseIndexParams(ctx context.Context) error {
 			}
 
 			var config map[string]string
-			if typeutil.IsDenseFloatVectorType(cit.fieldSchema.DataType) {
+			if typeutil.IsDenseFloatVectorType(cit.fieldSchema.DataType) ||
+				(typeutil.IsArrayOfVectorType(cit.fieldSchema.DataType) && typeutil.IsDenseFloatVectorType(cit.fieldSchema.ElementType)) {
 				// override float vector index params by autoindex
 				config = Params.AutoIndexConfig.IndexParams.GetAsJSONMap()
-			} else if typeutil.IsSparseFloatVectorType(cit.fieldSchema.DataType) {
+			} else if typeutil.IsSparseFloatVectorType(cit.fieldSchema.DataType) ||
+				(typeutil.IsArrayOfVectorType(cit.fieldSchema.DataType) && typeutil.IsSparseFloatVectorType(cit.fieldSchema.ElementType)) {
 				// override sparse float vector index params by autoindex
 				config = Params.AutoIndexConfig.SparseIndexParams.GetAsJSONMap()
-			} else if typeutil.IsBinaryVectorType(cit.fieldSchema.DataType) {
+			} else if typeutil.IsBinaryVectorType(cit.fieldSchema.DataType) ||
+				(typeutil.IsArrayOfVectorType(cit.fieldSchema.DataType) && typeutil.IsBinaryVectorType(cit.fieldSchema.ElementType)) {
 				// override binary vector index params by autoindex
 				config = Params.AutoIndexConfig.BinaryIndexParams.GetAsJSONMap()
-			} else if typeutil.IsIntVectorType(cit.fieldSchema.DataType) {
+			} else if typeutil.IsIntVectorType(cit.fieldSchema.DataType) ||
+				(typeutil.IsArrayOfVectorType(cit.fieldSchema.DataType) && typeutil.IsIntVectorType(cit.fieldSchema.ElementType)) {
 				// override int vector index params by autoindex
 				config = Params.AutoIndexConfig.IndexParams.GetAsJSONMap()
 			}
@@ -397,6 +403,12 @@ func (cit *createIndexTask) parseIndexParams(ctx context.Context) error {
 			if !funcutil.SliceContain(indexparamcheck.IntVectorMetrics, metricType) {
 				return merr.WrapErrParameterInvalid("valid index params", "invalid index params", "int vector index does not support metric type: "+metricType)
 			}
+		} else if typeutil.IsArrayOfVectorType(cit.fieldSchema.DataType) {
+			// TODO(SpadeA): adjust it when more metric types are supported. Especially, when different metric types
+			// are supported for different element types.
+			if !funcutil.SliceContain(indexparamcheck.EmbListMetrics, metricType) {
+				return merr.WrapErrParameterInvalid("valid index params", "invalid index params", "array of vector index does not support metric type: "+metricType)
+			}
 		}
 	}
 
@@ -500,7 +512,7 @@ func checkTrain(ctx context.Context, field *schemapb.FieldSchema, indexParams ma
 	}
 
 	if typeutil.IsVectorType(field.DataType) && indexType != indexparamcheck.AutoIndex {
-		exist := CheckVecIndexWithDataTypeExist(indexType, field.DataType)
+		exist := CheckVecIndexWithDataTypeExist(indexType, field.DataType, field.ElementType)
 		if !exist {
 			return fmt.Errorf("data type %s can't build with this index %s", schemapb.DataType_name[int32(field.GetDataType())], indexType)
 		}
@@ -519,7 +531,7 @@ func checkTrain(ctx context.Context, field *schemapb.FieldSchema, indexParams ma
 		return err
 	}
 
-	if err := checker.CheckTrain(field.DataType, indexParams); err != nil {
+	if err := checker.CheckTrain(field.DataType, field.ElementType, indexParams); err != nil {
 		log.Ctx(ctx).Info("create index with invalid parameters", zap.Error(err))
 		return err
 	}

internal/proxy/task_index_test.go

@@ -21,6 +21,7 @@ import (
 	"os"
 	"sort"
 	"strconv"
+	"strings"
 	"testing"
 
 	"github.com/cockroachdb/errors"
@@ -1164,6 +1165,131 @@ func Test_parseIndexParams(t *testing.T) {
 	})
 }
 
+func Test_checkEmbeddingListIndex(t *testing.T) {
+	t.Run("check embedding list index", func(t *testing.T) {
+		cit := &createIndexTask{
+			Condition: nil,
+			req: &milvuspb.CreateIndexRequest{
+				ExtraParams: []*commonpb.KeyValuePair{
+					{
+						Key:   common.IndexTypeKey,
+						Value: "EMB_LIST_HNSW",
+					},
+					{
+						Key:   common.MetricTypeKey,
+						Value: metric.MaxSim,
+					},
+				},
+				IndexName: "",
+			},
+			fieldSchema: &schemapb.FieldSchema{
+				FieldID:      101,
+				Name:         "EmbListFloat",
+				IsPrimaryKey: false,
+				DataType:     schemapb.DataType_ArrayOfVector,
+				ElementType:  schemapb.DataType_FloatVector,
+				TypeParams: []*commonpb.KeyValuePair{
+					{Key: common.DimKey, Value: "128"},
+				},
+			},
+		}
+		err := cit.parseIndexParams(context.TODO())
+		assert.NoError(t, err)
+	})
+
+	t.Run("metrics wrong for embedding list index", func(t *testing.T) {
+		cit := &createIndexTask{
+			Condition: nil,
+			req: &milvuspb.CreateIndexRequest{
+				ExtraParams: []*commonpb.KeyValuePair{
+					{
+						Key:   common.IndexTypeKey,
+						Value: "EMB_LIST_HNSW",
+					},
+					{
+						Key:   common.MetricTypeKey,
+						Value: metric.L2,
+					},
+				},
+				IndexName: "",
+			},
+			fieldSchema: &schemapb.FieldSchema{
+				FieldID:      101,
+				Name:         "EmbListFloat",
+				IsPrimaryKey: false,
+				DataType:     schemapb.DataType_ArrayOfVector,
+				ElementType:  schemapb.DataType_FloatVector,
+				TypeParams: []*commonpb.KeyValuePair{
+					{Key: common.DimKey, Value: "128"},
+				},
+			},
+		}
+		err := cit.parseIndexParams(context.TODO())
+		assert.True(t, strings.Contains(err.Error(), "array of vector index does not support metric type: L2"))
+	})
+
+	t.Run("metric type wrong", func(t *testing.T) {
+		cit := &createIndexTask{
+			Condition: nil,
+			req: &milvuspb.CreateIndexRequest{
+				ExtraParams: []*commonpb.KeyValuePair{
+					{
+						Key:   common.IndexTypeKey,
+						Value: "HNSW",
+					},
+					{
+						Key:   common.MetricTypeKey,
+						Value: metric.MaxSim,
+					},
+				},
+				IndexName: "",
+			},
+			fieldSchema: &schemapb.FieldSchema{
+				FieldID:      101,
+				Name:         "FieldFloat",
+				IsPrimaryKey: false,
+				DataType:     schemapb.DataType_FloatVector,
+				TypeParams: []*commonpb.KeyValuePair{
+					{Key: common.DimKey, Value: "128"},
+				},
+			},
+		}
+		err := cit.parseIndexParams(context.TODO())
+		assert.True(t, strings.Contains(err.Error(), "float vector index does not support metric type: MAX_SIM"))
+	})
+
+	t.Run("data type wrong", func(t *testing.T) {
+		cit := &createIndexTask{
+			Condition: nil,
+			req: &milvuspb.CreateIndexRequest{
+				ExtraParams: []*commonpb.KeyValuePair{
+					{
+						Key:   common.IndexTypeKey,
+						Value: "EMB_LIST_HNSW",
+					},
+					{
+						Key:   common.MetricTypeKey,
+						Value: metric.L2,
+					},
+				},
+				IndexName: "",
+			},
+			fieldSchema: &schemapb.FieldSchema{
+				FieldID:      101,
+				Name:         "FieldFloat",
+				IsPrimaryKey: false,
+				DataType:     schemapb.DataType_FloatVector,
+				TypeParams: []*commonpb.KeyValuePair{
+					{Key: common.DimKey, Value: "128"},
+				},
+			},
+		}
+
+		err := cit.parseIndexParams(context.TODO())
+		assert.True(t, strings.Contains(err.Error(), "data type FloatVector can't build with this index EMB_LIST_HNSW"))
+	})
+}
+
 func Test_ngram_parseIndexParams(t *testing.T) {
 	t.Run("valid ngram index params", func(t *testing.T) {
 		cit := &createIndexTask{

internal/proxy/task_insert.go

@@ -217,11 +217,7 @@ func (it *insertTask) PreExecute(ctx context.Context) error {
 		return err
 	}
 
-	allFields := make([]*schemapb.FieldSchema, 0, len(it.schema.Fields)+5)
-	allFields = append(allFields, it.schema.Fields...)
-	for _, structField := range it.schema.GetStructArrayFields() {
-		allFields = append(allFields, structField.GetFields()...)
-	}
+	allFields := typeutil.GetAllFieldSchemas(it.schema)
 
 	// check primaryFieldData whether autoID is true or not
 	// set rowIDs as primary data if autoID == true

internal/proxy/task_query.go

@@ -650,7 +650,9 @@ func (t *queryTask) PostExecute(ctx context.Context) error {
 		return err
 	}
 	t.result.OutputFields = t.userOutputFields
-	reconstructStructFieldData(t.result, t.schema.CollectionSchema)
+	if !t.reQuery {
+		reconstructStructFieldDataForQuery(t.result, t.schema.CollectionSchema)
+	}
 
 	primaryFieldSchema, err := t.schema.GetPkField()
 	if err != nil {

internal/proxy/task_query_test.go

@@ -1292,470 +1292,3 @@ func TestQueryTask_CanSkipAllocTimestamp(t *testing.T) {
 		assert.True(t, skip)
 	})
 }
-
-func Test_reconstructStructFieldData(t *testing.T) {
-	t.Run("count(*) query - should return early", func(t *testing.T) {
-		results := &milvuspb.QueryResults{
-			OutputFields: []string{"count(*)"},
-			FieldsData: []*schemapb.FieldData{
-				{
-					FieldName: "count(*)",
-					FieldId:   0,
-					Type:      schemapb.DataType_Int64,
-				},
-			},
-		}
-
-		schema := &schemapb.CollectionSchema{
-			StructArrayFields: []*schemapb.StructArrayFieldSchema{
-				{
-					FieldID: 102,
-					Name:    "test_struct",
-					Fields: []*schemapb.FieldSchema{
-						{
-							FieldID:     1021,
-							Name:        "sub_field",
-							DataType:    schemapb.DataType_Array,
-							ElementType: schemapb.DataType_Int32,
-						},
-					},
-				},
-			},
-		}
-
-		originalFieldsData := make([]*schemapb.FieldData, len(results.FieldsData))
-		copy(originalFieldsData, results.FieldsData)
-		originalOutputFields := make([]string, len(results.OutputFields))
-		copy(originalOutputFields, results.OutputFields)
-
-		reconstructStructFieldData(results, schema)
-
-		// Should not modify anything for count(*) query
-		assert.Equal(t, originalFieldsData, results.FieldsData)
-		assert.Equal(t, originalOutputFields, results.OutputFields)
-	})
-
-	t.Run("no struct array fields - should return early", func(t *testing.T) {
-		results := &milvuspb.QueryResults{
-			OutputFields: []string{"field1", "field2"},
-			FieldsData: []*schemapb.FieldData{
-				{
-					FieldName: "field1",
-					FieldId:   100,
-					Type:      schemapb.DataType_Int64,
-				},
-				{
-					FieldName: "field2",
-					FieldId:   101,
-					Type:      schemapb.DataType_VarChar,
-				},
-			},
-		}
-
-		schema := &schemapb.CollectionSchema{
-			StructArrayFields: []*schemapb.StructArrayFieldSchema{},
-		}
-
-		originalFieldsData := make([]*schemapb.FieldData, len(results.FieldsData))
-		copy(originalFieldsData, results.FieldsData)
-		originalOutputFields := make([]string, len(results.OutputFields))
-		copy(originalOutputFields, results.OutputFields)
-
-		reconstructStructFieldData(results, schema)
-
-		// Should not modify anything when no struct array fields
-		assert.Equal(t, originalFieldsData, results.FieldsData)
-		assert.Equal(t, originalOutputFields, results.OutputFields)
-	})
-
-	t.Run("reconstruct single struct field", func(t *testing.T) {
-		// Create mock data
-		subField1Data := &schemapb.FieldData{
-			FieldName: "sub_int_array",
-			FieldId:   1021,
-			Type:      schemapb.DataType_Array,
-			Field: &schemapb.FieldData_Scalars{
-				Scalars: &schemapb.ScalarField{
-					Data: &schemapb.ScalarField_ArrayData{
-						ArrayData: &schemapb.ArrayArray{
-							ElementType: schemapb.DataType_Int32,
-							Data: []*schemapb.ScalarField{
-								{
-									Data: &schemapb.ScalarField_IntData{
-										IntData: &schemapb.IntArray{Data: []int32{1, 2, 3}},
-									},
-								},
-							},
-						},
-					},
-				},
-			},
-		}
-
-		subField2Data := &schemapb.FieldData{
-			FieldName: "sub_text_array",
-			FieldId:   1022,
-			Type:      schemapb.DataType_Array,
-			Field: &schemapb.FieldData_Scalars{
-				Scalars: &schemapb.ScalarField{
-					Data: &schemapb.ScalarField_ArrayData{
-						ArrayData: &schemapb.ArrayArray{
-							ElementType: schemapb.DataType_VarChar,
-							Data: []*schemapb.ScalarField{
-								{
-									Data: &schemapb.ScalarField_StringData{
-										StringData: &schemapb.StringArray{Data: []string{"hello", "world"}},
-									},
-								},
-							},
-						},
-					},
-				},
-			},
-		}
-
-		results := &milvuspb.QueryResults{
-			OutputFields: []string{"sub_int_array", "sub_text_array"},
-			FieldsData:   []*schemapb.FieldData{subField1Data, subField2Data},
-		}
-
-		schema := &schemapb.CollectionSchema{
-			Fields: []*schemapb.FieldSchema{
-				{
-					FieldID:      100,
-					Name:         "pk",
-					IsPrimaryKey: true,
-					DataType:     schemapb.DataType_Int64,
-				},
-			},
-			StructArrayFields: []*schemapb.StructArrayFieldSchema{
-				{
-					FieldID: 102,
-					Name:    "test_struct",
-					Fields: []*schemapb.FieldSchema{
-						{
-							FieldID:     1021,
-							Name:        "sub_int_array",
-							DataType:    schemapb.DataType_Array,
-							ElementType: schemapb.DataType_Int32,
-						},
-						{
-							FieldID:     1022,
-							Name:        "sub_text_array",
-							DataType:    schemapb.DataType_Array,
-							ElementType: schemapb.DataType_VarChar,
-						},
-					},
-				},
-			},
-		}
-
-		reconstructStructFieldData(results, schema)
-
-		// Check result
-		assert.Len(t, results.FieldsData, 1, "Should only have one reconstructed struct field")
-		assert.Len(t, results.OutputFields, 1, "Output fields should only have one")
-
-		structField := results.FieldsData[0]
-		assert.Equal(t, "test_struct", structField.FieldName)
-		assert.Equal(t, int64(102), structField.FieldId)
-		assert.Equal(t, schemapb.DataType_ArrayOfStruct, structField.Type)
-		assert.Equal(t, "test_struct", results.OutputFields[0])
-
-		// Check fields inside struct
-		structArrays := structField.GetStructArrays()
-		assert.NotNil(t, structArrays)
-		assert.Len(t, structArrays.Fields, 2, "Struct should contain 2 sub fields")
-
-		// Check sub fields
-		var foundIntField, foundTextField bool
-		for _, field := range structArrays.Fields {
-			switch field.FieldId {
-			case 1021:
-				assert.Equal(t, "sub_int_array", field.FieldName)
-				assert.Equal(t, schemapb.DataType_Array, field.Type)
-				foundIntField = true
-			case 1022:
-				assert.Equal(t, "sub_text_array", field.FieldName)
-				assert.Equal(t, schemapb.DataType_Array, field.Type)
-				foundTextField = true
-			}
-		}
-		assert.True(t, foundIntField, "Should find int array field")
-		assert.True(t, foundTextField, "Should find text array field")
-	})
-
-	t.Run("mixed regular and struct fields", func(t *testing.T) {
-		// Create regular field data
-		regularField := &schemapb.FieldData{
-			FieldName: "regular_field",
-			FieldId:   100,
-			Type:      schemapb.DataType_Int64,
-			Field: &schemapb.FieldData_Scalars{
-				Scalars: &schemapb.ScalarField{
-					Data: &schemapb.ScalarField_LongData{
-						LongData: &schemapb.LongArray{Data: []int64{1, 2, 3}},
-					},
-				},
-			},
-		}
-
-		// Create struct sub field data
-		subFieldData := &schemapb.FieldData{
-			FieldName: "sub_field",
-			FieldId:   1021,
-			Type:      schemapb.DataType_Array,
-			Field: &schemapb.FieldData_Scalars{
-				Scalars: &schemapb.ScalarField{
-					Data: &schemapb.ScalarField_ArrayData{
-						ArrayData: &schemapb.ArrayArray{
-							ElementType: schemapb.DataType_Int32,
-							Data: []*schemapb.ScalarField{
-								{
-									Data: &schemapb.ScalarField_IntData{
-										IntData: &schemapb.IntArray{Data: []int32{10, 20}},
-									},
-								},
-							},
-						},
-					},
-				},
-			},
-		}
-
-		results := &milvuspb.QueryResults{
-			OutputFields: []string{"regular_field", "sub_field"},
-			FieldsData:   []*schemapb.FieldData{regularField, subFieldData},
-		}
-
-		schema := &schemapb.CollectionSchema{
-			Fields: []*schemapb.FieldSchema{
-				{
-					FieldID:  100,
-					Name:     "regular_field",
-					DataType: schemapb.DataType_Int64,
-				},
-			},
-			StructArrayFields: []*schemapb.StructArrayFieldSchema{
-				{
-					FieldID: 102,
-					Name:    "test_struct",
-					Fields: []*schemapb.FieldSchema{
-						{
-							FieldID:     1021,
-							Name:        "sub_field",
-							DataType:    schemapb.DataType_Array,
-							ElementType: schemapb.DataType_Int32,
-						},
-					},
-				},
-			},
-		}
-
-		reconstructStructFieldData(results, schema)
-
-		// Check result: should have 2 fields (1 regular + 1 reconstructed struct)
-		assert.Len(t, results.FieldsData, 2)
-		assert.Len(t, results.OutputFields, 2)
-
-		// Check regular and struct fields both exist
-		var foundRegularField, foundStructField bool
-		for i, field := range results.FieldsData {
-			switch field.FieldId {
-			case 100:
-				assert.Equal(t, "regular_field", field.FieldName)
-				assert.Equal(t, schemapb.DataType_Int64, field.Type)
-				assert.Equal(t, "regular_field", results.OutputFields[i])
-				foundRegularField = true
-			case 102:
-				assert.Equal(t, "test_struct", field.FieldName)
-				assert.Equal(t, schemapb.DataType_ArrayOfStruct, field.Type)
-				assert.Equal(t, "test_struct", results.OutputFields[i])
-				foundStructField = true
-			}
-		}
-		assert.True(t, foundRegularField, "Should find regular field")
-		assert.True(t, foundStructField, "Should find reconstructed struct field")
-	})
-
-	t.Run("multiple struct fields", func(t *testing.T) {
-		// Create sub field for first struct
-		struct1SubField := &schemapb.FieldData{
-			FieldName: "struct1_sub",
-			FieldId:   1021,
-			Type:      schemapb.DataType_Array,
-		}
-
-		// Create sub fields for second struct
-		struct2SubField1 := &schemapb.FieldData{
-			FieldName: "struct2_sub1",
-			FieldId:   1031,
-			Type:      schemapb.DataType_Array,
-		}
-
-		struct2SubField2 := &schemapb.FieldData{
-			FieldName: "struct2_sub2",
-			FieldId:   1032,
-			Type:      schemapb.DataType_Array,
-		}
-
-		results := &milvuspb.QueryResults{
-			OutputFields: []string{"struct1_sub", "struct2_sub1", "struct2_sub2"},
-			FieldsData:   []*schemapb.FieldData{struct1SubField, struct2SubField1, struct2SubField2},
-		}
-
-		schema := &schemapb.CollectionSchema{
-			StructArrayFields: []*schemapb.StructArrayFieldSchema{
-				{
-					FieldID: 102,
-					Name:    "struct1",
-					Fields: []*schemapb.FieldSchema{
-						{
-							FieldID:     1021,
-							Name:        "struct1_sub",
-							DataType:    schemapb.DataType_Array,
-							ElementType: schemapb.DataType_Int32,
-						},
-					},
-				},
-				{
-					FieldID: 103,
-					Name:    "struct2",
-					Fields: []*schemapb.FieldSchema{
-						{
-							FieldID:     1031,
-							Name:        "struct2_sub1",
-							DataType:    schemapb.DataType_Array,
-							ElementType: schemapb.DataType_VarChar,
-						},
-						{
-							FieldID:     1032,
-							Name:        "struct2_sub2",
-							DataType:    schemapb.DataType_Array,
-							ElementType: schemapb.DataType_Float,
-						},
-					},
-				},
-			},
-		}
-
-		reconstructStructFieldData(results, schema)
-
-		// Check result: should have 2 reconstructed struct fields
-		assert.Len(t, results.FieldsData, 2)
-		assert.Len(t, results.OutputFields, 2)
-
-		// Check both struct fields are reconstructed correctly
-		foundStruct1, foundStruct2 := false, false
-		for i, field := range results.FieldsData {
-			switch field.FieldId {
-			case 102:
-				assert.Equal(t, "struct1", field.FieldName)
-				assert.Equal(t, schemapb.DataType_ArrayOfStruct, field.Type)
-				assert.Equal(t, "struct1", results.OutputFields[i])
-
-				structArrays := field.GetStructArrays()
-				assert.NotNil(t, structArrays)
-				assert.Len(t, structArrays.Fields, 1)
-				assert.Equal(t, int64(1021), structArrays.Fields[0].FieldId)
-				foundStruct1 = true
-
-			case 103:
-				assert.Equal(t, "struct2", field.FieldName)
-				assert.Equal(t, schemapb.DataType_ArrayOfStruct, field.Type)
-				assert.Equal(t, "struct2", results.OutputFields[i])
-
-				structArrays := field.GetStructArrays()
-				assert.NotNil(t, structArrays)
-				assert.Len(t, structArrays.Fields, 2)
-
-				// Check struct2 contains two sub fields
-				subFieldIds := make([]int64, 0, 2)
-				for _, subField := range structArrays.Fields {
-					subFieldIds = append(subFieldIds, subField.FieldId)
-				}
-				assert.Contains(t, subFieldIds, int64(1031))
-				assert.Contains(t, subFieldIds, int64(1032))
-				foundStruct2 = true
-			}
-		}
-		assert.True(t, foundStruct1, "Should find struct1")
-		assert.True(t, foundStruct2, "Should find struct2")
-	})
-
-	t.Run("empty fields data", func(t *testing.T) {
-		results := &milvuspb.QueryResults{
-			OutputFields: []string{},
-			FieldsData:   []*schemapb.FieldData{},
-		}
-
-		schema := &schemapb.CollectionSchema{
-			StructArrayFields: []*schemapb.StructArrayFieldSchema{
-				{
-					FieldID: 102,
-					Name:    "test_struct",
-					Fields: []*schemapb.FieldSchema{
-						{
-							FieldID:     1021,
-							Name:        "sub_field",
-							DataType:    schemapb.DataType_Array,
-							ElementType: schemapb.DataType_Int32,
-						},
-					},
-				},
-			},
-		}
-
-		reconstructStructFieldData(results, schema)
-
-		// Empty data should remain unchanged
-		assert.Len(t, results.FieldsData, 0)
-		assert.Len(t, results.OutputFields, 0)
-	})
-
-	t.Run("no matching sub fields", func(t *testing.T) {
-		// Field data does not match any struct definition
-		regularField := &schemapb.FieldData{
-			FieldName: "regular_field",
-			FieldId:   200, // Not in any struct
-			Type:      schemapb.DataType_Int64,
-		}
-
-		results := &milvuspb.QueryResults{
-			OutputFields: []string{"regular_field"},
-			FieldsData:   []*schemapb.FieldData{regularField},
-		}
-
-		schema := &schemapb.CollectionSchema{
-			Fields: []*schemapb.FieldSchema{
-				{
-					FieldID:  200,
-					Name:     "regular_field",
-					DataType: schemapb.DataType_Int64,
-				},
-			},
-			StructArrayFields: []*schemapb.StructArrayFieldSchema{
-				{
-					FieldID: 102,
-					Name:    "test_struct",
-					Fields: []*schemapb.FieldSchema{
-						{
-							FieldID:     1021,
-							Name:        "sub_field",
-							DataType:    schemapb.DataType_Array,
-							ElementType: schemapb.DataType_Int32,
-						},
-					},
-				},
-			},
-		}
-
-		reconstructStructFieldData(results, schema)
-
-		// Should only keep the regular field, no struct field
-		assert.Len(t, results.FieldsData, 1)
-		assert.Len(t, results.OutputFields, 1)
-		assert.Equal(t, int64(200), results.FieldsData[0].FieldId)
-		assert.Equal(t, "regular_field", results.OutputFields[0])
-	})
-}

internal/proxy/task_search.go

@@ -544,7 +544,8 @@ func (t *searchTask) initSearchRequest(ctx context.Context) error {
 		}
 	}
 
-	vectorOutputFields := lo.Filter(t.schema.GetFields(), func(field *schemapb.FieldSchema, _ int) bool {
+	allFields := typeutil.GetAllFieldSchemas(t.schema.CollectionSchema)
+	vectorOutputFields := lo.Filter(allFields, func(field *schemapb.FieldSchema, _ int) bool {
 		return lo.Contains(t.translatedOutputFields, field.GetName()) && typeutil.IsVectorType(field.GetDataType())
 	})
 	t.needRequery = len(vectorOutputFields) > 0
@@ -765,6 +766,7 @@ func (t *searchTask) PostExecute(ctx context.Context) error {
 	}
 	t.fillResult()
 	t.result.Results.OutputFields = t.userOutputFields
+	reconstructStructFieldDataForSearch(t.result, t.schema.CollectionSchema)
 	t.result.CollectionName = t.request.GetCollectionName()
 
 	primaryFieldSchema, _ := t.schema.GetPkField()

internal/proxy/task_search_test.go

@@ -3548,6 +3548,278 @@ func TestTaskSearch_parseSearchInfo(t *testing.T) {
 			assert.ErrorContains(t, err, "failed to parse input last bound")
 		})
 	})
+
+	t.Run("check vector array unsupported features", func(t *testing.T) {
+		// Helper function to create a schema with vector array field
+		createSchemaWithVectorArray := func(annsFieldName string) *schemapb.CollectionSchema {
+			return &schemapb.CollectionSchema{
+				Fields: []*schemapb.FieldSchema{
+					{
+						FieldID:      100,
+						Name:         "id",
+						DataType:     schemapb.DataType_Int64,
+						IsPrimaryKey: true,
+					},
+					{
+						FieldID:  101,
+						Name:     "normal_vector",
+						DataType: schemapb.DataType_FloatVector,
+						TypeParams: []*commonpb.KeyValuePair{
+							{Key: common.DimKey, Value: "128"},
+						},
+					},
+					{
+						FieldID:  102,
+						Name:     "group_field",
+						DataType: schemapb.DataType_VarChar,
+					},
+				},
+				StructArrayFields: []*schemapb.StructArrayFieldSchema{
+					{
+						FieldID: 103,
+						Name:    "struct_array_field",
+						Fields: []*schemapb.FieldSchema{
+							{
+								FieldID:     104,
+								Name:        annsFieldName,
+								DataType:    schemapb.DataType_ArrayOfVector,
+								ElementType: schemapb.DataType_FloatVector,
+								TypeParams: []*commonpb.KeyValuePair{
+									{Key: common.DimKey, Value: "128"},
+								},
+							},
+						},
+					},
+				},
+			}
+		}
+
+		// Helper function to create search params with anns field
+		createSearchParams := func(annsFieldName string) []*commonpb.KeyValuePair {
+			return []*commonpb.KeyValuePair{
+				{
+					Key:   AnnsFieldKey,
+					Value: annsFieldName,
+				},
+				{
+					Key:   TopKKey,
+					Value: "10",
+				},
+				{
+					Key:   common.MetricTypeKey,
+					Value: metric.MaxSim,
+				},
+				{
+					Key:   ParamsKey,
+					Value: `{"nprobe": 10}`,
+				},
+				{
+					Key:   RoundDecimalKey,
+					Value: "-1",
+				},
+				{
+					Key:   IgnoreGrowingKey,
+					Value: "false",
+				},
+			}
+		}
+
+		t.Run("vector array with range search", func(t *testing.T) {
+			schema := createSchemaWithVectorArray("embeddings_list")
+			params := createSearchParams("embeddings_list")
+
+			// Add radius parameter for range search
+			resetSearchParamsValue(params, ParamsKey, `{"nprobe": 10, "radius": 0.2}`)
+
+			searchInfo, err := parseSearchInfo(params, schema, nil)
+			assert.Error(t, err)
+			assert.Nil(t, searchInfo)
+			assert.ErrorIs(t, err, merr.ErrParameterInvalid)
+			fmt.Println(err.Error())
+			assert.Contains(t, err.Error(), "range search is not supported for vector array (embedding list) fields")
+		})
+
+		t.Run("vector array with group by", func(t *testing.T) {
+			schema := createSchemaWithVectorArray("embeddings_list")
+			params := createSearchParams("embeddings_list")
+
+			// Add group by parameter
+			params = append(params, &commonpb.KeyValuePair{
+				Key:   GroupByFieldKey,
+				Value: "group_field",
+			})
+
+			searchInfo, err := parseSearchInfo(params, schema, nil)
+			assert.Error(t, err)
+			assert.Nil(t, searchInfo)
+			assert.ErrorIs(t, err, merr.ErrParameterInvalid)
+			assert.Contains(t, err.Error(), "group by search is not supported for vector array (embedding list) fields")
+			assert.Contains(t, err.Error(), "embeddings_list")
+		})
+
+		t.Run("vector array with iterator", func(t *testing.T) {
+			schema := createSchemaWithVectorArray("embeddings_list")
+			params := createSearchParams("embeddings_list")
+
+			// Add iterator parameter
+			params = append(params, &commonpb.KeyValuePair{
+				Key:   IteratorField,
+				Value: "True",
+			})
+
+			searchInfo, err := parseSearchInfo(params, schema, nil)
+			assert.Error(t, err)
+			assert.Nil(t, searchInfo)
+			assert.ErrorIs(t, err, merr.ErrParameterInvalid)
+			assert.Contains(t, err.Error(), "search iterator is not supported for vector array (embedding list) fields")
+			assert.Contains(t, err.Error(), "embeddings_list")
+		})
+
+		t.Run("vector array with iterator v2", func(t *testing.T) {
+			schema := createSchemaWithVectorArray("embeddings_list")
+			params := createSearchParams("embeddings_list")
+
+			// Add iterator v2 parameters
+			params = append(params,
+				&commonpb.KeyValuePair{
+					Key:   SearchIterV2Key,
+					Value: "True",
+				},
+				&commonpb.KeyValuePair{
+					Key:   IteratorField,
+					Value: "True",
+				},
+				&commonpb.KeyValuePair{
+					Key:   SearchIterBatchSizeKey,
+					Value: "10",
+				},
+			)
+
+			searchInfo, err := parseSearchInfo(params, schema, nil)
+			assert.Error(t, err)
+			assert.Nil(t, searchInfo)
+			assert.ErrorIs(t, err, merr.ErrParameterInvalid)
+			assert.Contains(t, err.Error(), "search iterator is not supported for vector array (embedding list) fields")
+			assert.Contains(t, err.Error(), "embeddings_list")
+		})
+
+		t.Run("normal search on vector array should succeed", func(t *testing.T) {
+			schema := createSchemaWithVectorArray("embeddings_list")
+			params := createSearchParams("embeddings_list")
+
+			searchInfo, err := parseSearchInfo(params, schema, nil)
+			assert.NoError(t, err)
+			assert.NotNil(t, searchInfo)
+			assert.NotNil(t, searchInfo.planInfo)
+		})
+
+		t.Run("normal vector field with range search should succeed", func(t *testing.T) {
+			schema := createSchemaWithVectorArray("embeddings_list")
+			params := createSearchParams("normal_vector")
+
+			// Add radius parameter for range search
+			resetSearchParamsValue(params, ParamsKey, `{"nprobe": 10, "radius": 0.2}`)
+
+			searchInfo, err := parseSearchInfo(params, schema, nil)
+			assert.NoError(t, err)
+			assert.NotNil(t, searchInfo)
+			assert.NotNil(t, searchInfo.planInfo)
+		})
+
+		t.Run("normal vector field with group by should succeed", func(t *testing.T) {
+			schema := createSchemaWithVectorArray("embeddings_list")
+			params := createSearchParams("normal_vector")
+
+			// Add group by parameter
+			params = append(params, &commonpb.KeyValuePair{
+				Key:   GroupByFieldKey,
+				Value: "group_field",
+			})
+
+			searchInfo, err := parseSearchInfo(params, schema, nil)
+			assert.NoError(t, err)
+			assert.NotNil(t, searchInfo)
+			assert.NotNil(t, searchInfo.planInfo)
+			assert.Equal(t, int64(102), searchInfo.planInfo.GroupByFieldId)
+		})
+
+		t.Run("normal vector field with iterator should succeed", func(t *testing.T) {
+			schema := createSchemaWithVectorArray("embeddings_list")
+			params := createSearchParams("normal_vector")
+
+			// Add iterator parameter
+			params = append(params, &commonpb.KeyValuePair{
+				Key:   IteratorField,
+				Value: "True",
+			})
+
+			searchInfo, err := parseSearchInfo(params, schema, nil)
+			assert.NoError(t, err)
+			assert.NotNil(t, searchInfo)
+			assert.NotNil(t, searchInfo.planInfo)
+		})
+
+		t.Run("vector array with range search", func(t *testing.T) {
+			schema := createSchemaWithVectorArray("embeddings_list")
+			params := createSearchParams("embeddings_list")
+			resetSearchParamsValue(params, ParamsKey, `{"nprobe": 10, "radius": 0.2}`)
+
+			searchInfo, err := parseSearchInfo(params, schema, nil)
+			assert.Error(t, err)
+			assert.Nil(t, searchInfo)
+			// Should fail on range search first
+			assert.Contains(t, err.Error(), "range search is not supported for vector array (embedding list) fields")
+		})
+
+		t.Run("no anns field specified", func(t *testing.T) {
+			schema := createSchemaWithVectorArray("embeddings_list")
+			params := getValidSearchParams()
+			// Don't specify anns field
+
+			searchInfo, err := parseSearchInfo(params, schema, nil)
+			assert.NoError(t, err)
+			assert.NotNil(t, searchInfo)
+			// Should not trigger vector array validation without anns field
+		})
+
+		t.Run("non-existent anns field", func(t *testing.T) {
+			schema := createSchemaWithVectorArray("embeddings_list")
+			params := createSearchParams("non_existent_field")
+
+			searchInfo, err := parseSearchInfo(params, schema, nil)
+			assert.NoError(t, err)
+			assert.NotNil(t, searchInfo)
+			// Should not trigger vector array validation for non-existent field
+		})
+
+		t.Run("hybrid search with outer group by on vector array", func(t *testing.T) {
+			schema := createSchemaWithVectorArray("embeddings_list")
+
+			// Create rank params with group by
+			rankParams := getValidSearchParams()
+			rankParams = append(rankParams,
+				&commonpb.KeyValuePair{
+					Key:   GroupByFieldKey,
+					Value: "group_field",
+				},
+				&commonpb.KeyValuePair{
+					Key:   LimitKey,
+					Value: "100",
+				},
+			)
+
+			parsedRankParams, err := parseRankParams(rankParams, schema)
+			assert.NoError(t, err)
+
+			searchParams := createSearchParams("embeddings_list")
+			// Parse search info with rank params
+			searchInfo, err := parseSearchInfo(searchParams, schema, parsedRankParams)
+			assert.Error(t, err)
+			assert.Nil(t, searchInfo)
+			assert.ErrorIs(t, err, merr.ErrParameterInvalid)
+			assert.Contains(t, err.Error(), "group by search is not supported for vector array (embedding list) fields")
+		})
+	})
 }
 
 func getSearchResultData(nq, topk int64) *schemapb.SearchResultData {
@@ -4301,3 +4573,110 @@ func genTestSearchResultData(nq int64, topk int64, dType schemapb.DataType, fiel
 	}
 	return result
 }
+
+func TestSearchTask_InitSearchRequestWithStructArrayFields(t *testing.T) {
+	ctx := context.Background()
+
+	schema := &schemapb.CollectionSchema{
+		Name: "test_collection",
+		Fields: []*schemapb.FieldSchema{
+			{FieldID: 100, Name: "pk", DataType: schemapb.DataType_Int64, IsPrimaryKey: true},
+			{FieldID: 101, Name: "regular_vec", DataType: schemapb.DataType_FloatVector, TypeParams: []*commonpb.KeyValuePair{{Key: common.DimKey, Value: "128"}}},
+			{FieldID: 102, Name: "regular_scalar", DataType: schemapb.DataType_Int32},
+		},
+		StructArrayFields: []*schemapb.StructArrayFieldSchema{
+			{
+				Name: "structArray",
+				Fields: []*schemapb.FieldSchema{
+					{FieldID: 104, Name: "struct_vec_array", DataType: schemapb.DataType_ArrayOfVector, ElementType: schemapb.DataType_FloatVector, TypeParams: []*commonpb.KeyValuePair{{Key: common.DimKey, Value: "64"}}},
+					{FieldID: 105, Name: "struct_scalar_array", DataType: schemapb.DataType_Array, ElementType: schemapb.DataType_Int32},
+				},
+			},
+		},
+	}
+
+	schemaInfo := newSchemaInfo(schema)
+
+	tests := []struct {
+		name            string
+		outputFields    []string
+		expectedRequery bool
+		description     string
+	}{
+		{
+			name:            "regular_vector_field",
+			outputFields:    []string{"pk", "regular_vec"},
+			expectedRequery: true,
+			description:     "Should require requery when regular vector field in output",
+		},
+		{
+			name:            "struct_array_vector_field",
+			outputFields:    []string{"pk", "struct_vec_array"},
+			expectedRequery: true,
+			description:     "Should require requery when struct array vector field in output (tests GetAllFieldSchemas)",
+		},
+		{
+			name:            "both_vector_fields",
+			outputFields:    []string{"pk", "regular_vec", "struct_vec_array"},
+			expectedRequery: true,
+			description:     "Should require requery when both regular and struct array vector fields in output",
+		},
+		{
+			name:            "struct_scalar_array_only",
+			outputFields:    []string{"pk", "struct_scalar_array"},
+			expectedRequery: false,
+			description:     "Should not require requery when only struct scalar array field in output",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			task := &searchTask{
+				ctx:            ctx,
+				collectionName: "test_collection",
+				SearchRequest: &internalpb.SearchRequest{
+					CollectionID:     1,
+					PartitionIDs:     []int64{1},
+					Dsl:              "",
+					PlaceholderGroup: nil,
+					DslType:          commonpb.DslType_BoolExprV1,
+					OutputFieldsId:   []int64{},
+				},
+				request: &milvuspb.SearchRequest{
+					CollectionName: "test_collection",
+					OutputFields:   tt.outputFields,
+					Dsl:            "",
+					SearchParams: []*commonpb.KeyValuePair{
+						{Key: AnnsFieldKey, Value: "regular_vec"},
+						{Key: TopKKey, Value: "10"},
+						{Key: common.MetricTypeKey, Value: metric.L2},
+						{Key: ParamsKey, Value: `{"nprobe": 10}`},
+					},
+					PlaceholderGroup: nil,
+					ConsistencyLevel: commonpb.ConsistencyLevel_Session,
+				},
+				schema:                 schemaInfo,
+				translatedOutputFields: tt.outputFields,
+				tr:                     timerecord.NewTimeRecorder("test"),
+				queryInfos:             []*planpb.QueryInfo{{}},
+			}
+
+			// Set translated output field IDs based on the schema
+			outputFieldIDs := []int64{}
+			allFields := typeutil.GetAllFieldSchemas(schema)
+			for _, fieldName := range tt.outputFields {
+				for _, field := range allFields {
+					if field.Name == fieldName {
+						outputFieldIDs = append(outputFieldIDs, field.FieldID)
+						break
+					}
+				}
+			}
+			task.SearchRequest.OutputFieldsId = outputFieldIDs
+
+			err := task.initSearchRequest(ctx)
+			assert.NoError(t, err)
+			assert.Equal(t, tt.expectedRequery, task.needRequery, tt.description)
+		})
+	}
+}

internal/proxy/task_upsert.go

@@ -585,11 +585,7 @@ func (it *upsertTask) insertPreExecute(ctx context.Context) error {
 		return err
 	}
 
-	allFields := make([]*schemapb.FieldSchema, 0, len(it.schema.Fields)+5)
-	allFields = append(allFields, it.schema.Fields...)
-	for _, structField := range it.schema.GetStructArrayFields() {
-		allFields = append(allFields, structField.GetFields()...)
-	}
+	allFields := typeutil.GetAllFieldSchemas(it.schema.CollectionSchema)
 
 	// use the passed pk as new pk when autoID == false
 	// automatic generate pk as new pk wehen autoID == true

internal/proxy/util.go

@@ -619,9 +619,6 @@ func ValidateFieldsInStruct(field *schemapb.FieldSchema, schema *schemapb.Collec
 	if field.GetNullable() {
 		return fmt.Errorf("nullable is not supported for fields in struct array now, fieldName = %s", field.Name)
 	}
-
-	// todo(SpadeA): add more check when index is enabled
-
 	return nil
 }
 
@@ -2547,3 +2544,88 @@ func getCollectionTTL(pairs []*commonpb.KeyValuePair) uint64 {
 
 	return 0
 }
+
+// reconstructStructFieldDataCommon reconstructs struct fields from flattened sub-fields
+// It works with both QueryResults and SearchResults by operating on the common data structures
+func reconstructStructFieldDataCommon(
+	fieldsData []*schemapb.FieldData,
+	outputFields []string,
+	schema *schemapb.CollectionSchema,
+) ([]*schemapb.FieldData, []string) {
+	if len(outputFields) == 1 && outputFields[0] == "count(*)" {
+		return fieldsData, outputFields
+	}
+
+	if len(schema.StructArrayFields) == 0 {
+		return fieldsData, outputFields
+	}
+
+	regularFieldIDs := make(map[int64]interface{})
+	subFieldToStructMap := make(map[int64]int64)
+	groupedStructFields := make(map[int64][]*schemapb.FieldData)
+	structFieldNames := make(map[int64]string)
+	reconstructedOutputFields := make([]string, 0, len(fieldsData))
+
+	// record all regular field IDs
+	for _, field := range schema.Fields {
+		regularFieldIDs[field.GetFieldID()] = nil
+	}
+
+	// build the mapping from sub-field ID to struct field ID
+	for _, structField := range schema.StructArrayFields {
+		for _, subField := range structField.GetFields() {
+			subFieldToStructMap[subField.GetFieldID()] = structField.GetFieldID()
+		}
+		structFieldNames[structField.GetFieldID()] = structField.GetName()
+	}
+
+	newFieldsData := make([]*schemapb.FieldData, 0, len(fieldsData))
+	for _, field := range fieldsData {
+		fieldID := field.GetFieldId()
+		if _, ok := regularFieldIDs[fieldID]; ok {
+			newFieldsData = append(newFieldsData, field)
+			reconstructedOutputFields = append(reconstructedOutputFields, field.GetFieldName())
+		} else {
+			structFieldID := subFieldToStructMap[fieldID]
+			groupedStructFields[structFieldID] = append(groupedStructFields[structFieldID], field)
+		}
+	}
+
+	for structFieldID, fields := range groupedStructFields {
+		fieldData := &schemapb.FieldData{
+			FieldName: structFieldNames[structFieldID],
+			FieldId:   structFieldID,
+			Type:      schemapb.DataType_ArrayOfStruct,
+			Field:     &schemapb.FieldData_StructArrays{StructArrays: &schemapb.StructArrayField{Fields: fields}},
+		}
+		newFieldsData = append(newFieldsData, fieldData)
+		reconstructedOutputFields = append(reconstructedOutputFields, structFieldNames[structFieldID])
+	}
+
+	return newFieldsData, reconstructedOutputFields
+}
+
+// Wrapper for QueryResults
+func reconstructStructFieldDataForQuery(results *milvuspb.QueryResults, schema *schemapb.CollectionSchema) {
+	fieldsData, outputFields := reconstructStructFieldDataCommon(
+		results.FieldsData,
+		results.OutputFields,
+		schema,
+	)
+	results.FieldsData = fieldsData
+	results.OutputFields = outputFields
+}
+
+// New wrapper for SearchResults
+func reconstructStructFieldDataForSearch(results *milvuspb.SearchResults, schema *schemapb.CollectionSchema) {
+	if results.Results == nil {
+		return
+	}
+	fieldsData, outputFields := reconstructStructFieldDataCommon(
+		results.Results.FieldsData,
+		results.Results.OutputFields,
+		schema,
+	)
+	results.Results.FieldsData = fieldsData
+	results.Results.OutputFields = outputFields
+}

internal/proxy/util_test.go

@@ -3825,5 +3825,644 @@ func TestCheckAndFlattenStructFieldData(t *testing.T) {
 }
 
 func TestValidateFieldsInStruct(t *testing.T) {
-	// todo(SpadeA): add test cases
+	schema := &schemapb.CollectionSchema{
+		Name: "test_collection",
+	}
+
+	t.Run("valid array field", func(t *testing.T) {
+		field := &schemapb.FieldSchema{
+			Name:        "valid_array",
+			DataType:    schemapb.DataType_Array,
+			ElementType: schemapb.DataType_Int32,
+		}
+		err := ValidateFieldsInStruct(field, schema)
+		assert.NoError(t, err)
+	})
+
+	t.Run("valid array of vector field", func(t *testing.T) {
+		field := &schemapb.FieldSchema{
+			Name:        "valid_array_vector",
+			DataType:    schemapb.DataType_ArrayOfVector,
+			ElementType: schemapb.DataType_FloatVector,
+			TypeParams: []*commonpb.KeyValuePair{
+				{Key: common.DimKey, Value: "128"},
+			},
+		}
+		err := ValidateFieldsInStruct(field, schema)
+		assert.NoError(t, err)
+	})
+
+	t.Run("invalid field name", func(t *testing.T) {
+		testCases := []struct {
+			name     string
+			expected string
+		}{
+			{"", "field name should not be empty"},
+			{"123abc", "The first character of a field name must be an underscore or letter"},
+			{"abc-def", "Field name can only contain numbers, letters, and underscores"},
+		}
+
+		for _, tc := range testCases {
+			field := &schemapb.FieldSchema{
+				Name:        tc.name,
+				DataType:    schemapb.DataType_Array,
+				ElementType: schemapb.DataType_Int32,
+			}
+			err := ValidateFieldsInStruct(field, schema)
+			assert.Error(t, err)
+			assert.Contains(t, err.Error(), tc.expected)
+		}
+	})
+
+	t.Run("invalid data type", func(t *testing.T) {
+		field := &schemapb.FieldSchema{
+			Name:     "invalid_type",
+			DataType: schemapb.DataType_Int32, // Not array or array of vector
+		}
+		err := ValidateFieldsInStruct(field, schema)
+		assert.Error(t, err)
+		assert.Contains(t, err.Error(), "Fields in StructArrayField can only be array or array of struct")
+	})
+
+	t.Run("nested array not supported", func(t *testing.T) {
+		testCases := []struct {
+			elementType schemapb.DataType
+		}{
+			{schemapb.DataType_ArrayOfStruct},
+			{schemapb.DataType_ArrayOfVector},
+			{schemapb.DataType_Array},
+		}
+
+		for _, tc := range testCases {
+			field := &schemapb.FieldSchema{
+				Name:        "nested_array",
+				DataType:    schemapb.DataType_Array,
+				ElementType: tc.elementType,
+			}
+			err := ValidateFieldsInStruct(field, schema)
+			assert.Error(t, err)
+			assert.Contains(t, err.Error(), "Nested array is not supported")
+		}
+	})
+
+	t.Run("array field with vector element type", func(t *testing.T) {
+		field := &schemapb.FieldSchema{
+			Name:        "array_with_vector",
+			DataType:    schemapb.DataType_Array,
+			ElementType: schemapb.DataType_FloatVector,
+		}
+		err := ValidateFieldsInStruct(field, schema)
+		assert.Error(t, err)
+		assert.Contains(t, err.Error(), "element type of array field array_with_vector is a vector type")
+	})
+
+	t.Run("array of vector field with non-vector element type", func(t *testing.T) {
+		field := &schemapb.FieldSchema{
+			Name:        "array_vector_with_scalar",
+			DataType:    schemapb.DataType_ArrayOfVector,
+			ElementType: schemapb.DataType_Int32,
+		}
+		err := ValidateFieldsInStruct(field, schema)
+		assert.Error(t, err)
+		assert.Contains(t, err.Error(), "element type of array field array_vector_with_scalar is not a vector type")
+	})
+
+	t.Run("array of vector missing dimension", func(t *testing.T) {
+		field := &schemapb.FieldSchema{
+			Name:        "array_vector_no_dim",
+			DataType:    schemapb.DataType_ArrayOfVector,
+			ElementType: schemapb.DataType_FloatVector,
+			TypeParams:  []*commonpb.KeyValuePair{}, // No dimension specified
+		}
+		err := ValidateFieldsInStruct(field, schema)
+		assert.Error(t, err)
+		assert.Contains(t, err.Error(), "dimension is not defined in field")
+	})
+
+	t.Run("array of vector with invalid dimension", func(t *testing.T) {
+		field := &schemapb.FieldSchema{
+			Name:        "array_vector_invalid_dim",
+			DataType:    schemapb.DataType_ArrayOfVector,
+			ElementType: schemapb.DataType_FloatVector,
+			TypeParams: []*commonpb.KeyValuePair{
+				{Key: common.DimKey, Value: "not_a_number"},
+			},
+		}
+		err := ValidateFieldsInStruct(field, schema)
+		assert.Error(t, err)
+	})
+
+	t.Run("varchar array without max_length", func(t *testing.T) {
+		field := &schemapb.FieldSchema{
+			Name:        "varchar_array_no_max_length",
+			DataType:    schemapb.DataType_Array,
+			ElementType: schemapb.DataType_VarChar,
+			TypeParams:  []*commonpb.KeyValuePair{}, // No max_length specified
+		}
+		err := ValidateFieldsInStruct(field, schema)
+		assert.Error(t, err)
+		assert.Contains(t, err.Error(), "type param(max_length) should be specified")
+	})
+
+	t.Run("varchar array with valid max_length", func(t *testing.T) {
+		field := &schemapb.FieldSchema{
+			Name:        "varchar_array_valid",
+			DataType:    schemapb.DataType_Array,
+			ElementType: schemapb.DataType_VarChar,
+			TypeParams: []*commonpb.KeyValuePair{
+				{Key: common.MaxLengthKey, Value: "100"},
+			},
+		}
+		err := ValidateFieldsInStruct(field, schema)
+		assert.NoError(t, err)
+	})
+
+	t.Run("varchar array with invalid max_length", func(t *testing.T) {
+		// Test with max_length exceeding limit
+		field := &schemapb.FieldSchema{
+			Name:        "varchar_array_invalid_length",
+			DataType:    schemapb.DataType_Array,
+			ElementType: schemapb.DataType_VarChar,
+			TypeParams: []*commonpb.KeyValuePair{
+				{Key: common.MaxLengthKey, Value: "99999999"}, // Exceeds limit
+			},
+		}
+		err := ValidateFieldsInStruct(field, schema)
+		assert.Error(t, err)
+		assert.Contains(t, err.Error(), "the maximum length specified for the field")
+	})
+
+	t.Run("nullable field not supported", func(t *testing.T) {
+		field := &schemapb.FieldSchema{
+			Name:        "nullable_field",
+			DataType:    schemapb.DataType_Array,
+			ElementType: schemapb.DataType_Int32,
+			Nullable:    true,
+		}
+		err := ValidateFieldsInStruct(field, schema)
+		assert.Error(t, err)
+		assert.Contains(t, err.Error(), "nullable is not supported for fields in struct array now")
+	})
+
+	t.Run("sparse float vector in array of vector", func(t *testing.T) {
+		// Note: ArrayOfVector with sparse vector element type still requires dimension
+		// because validateDimension checks the field's DataType (ArrayOfVector), not ElementType
+		field := &schemapb.FieldSchema{
+			Name:        "sparse_vector_array",
+			DataType:    schemapb.DataType_ArrayOfVector,
+			ElementType: schemapb.DataType_SparseFloatVector,
+			TypeParams:  []*commonpb.KeyValuePair{},
+		}
+		err := ValidateFieldsInStruct(field, schema)
+		assert.Error(t, err)
+		assert.Contains(t, err.Error(), "dimension is not defined")
+	})
+
+	t.Run("array with various scalar element types", func(t *testing.T) {
+		validScalarTypes := []schemapb.DataType{
+			schemapb.DataType_Bool,
+			schemapb.DataType_Int8,
+			schemapb.DataType_Int16,
+			schemapb.DataType_Int32,
+			schemapb.DataType_Int64,
+			schemapb.DataType_Float,
+			schemapb.DataType_Double,
+			schemapb.DataType_String,
+		}
+
+		for _, dt := range validScalarTypes {
+			field := &schemapb.FieldSchema{
+				Name:        "array_" + dt.String(),
+				DataType:    schemapb.DataType_Array,
+				ElementType: dt,
+			}
+			err := ValidateFieldsInStruct(field, schema)
+			assert.NoError(t, err)
+		}
+	})
+
+	t.Run("array of vector with various vector types", func(t *testing.T) {
+		validVectorTypes := []schemapb.DataType{
+			schemapb.DataType_FloatVector,
+			schemapb.DataType_BinaryVector,
+			schemapb.DataType_Float16Vector,
+			schemapb.DataType_BFloat16Vector,
+			// Note: SparseFloatVector is excluded because validateDimension checks
+			// the field's DataType (ArrayOfVector), not ElementType, so it still requires dimension
+		}
+
+		for _, vt := range validVectorTypes {
+			field := &schemapb.FieldSchema{
+				Name:        "vector_array_" + vt.String(),
+				DataType:    schemapb.DataType_ArrayOfVector,
+				ElementType: vt,
+				TypeParams: []*commonpb.KeyValuePair{
+					{Key: common.DimKey, Value: "128"},
+				},
+			}
+			err := ValidateFieldsInStruct(field, schema)
+			assert.NoError(t, err)
+		}
+	})
+}
+
+func Test_reconstructStructFieldDataCommon(t *testing.T) {
+	t.Run("count(*) query - should return early", func(t *testing.T) {
+		fieldsData := []*schemapb.FieldData{
+			{
+				FieldName: "count(*)",
+				FieldId:   0,
+				Type:      schemapb.DataType_Int64,
+			},
+		}
+		outputFields := []string{"count(*)"}
+
+		schema := &schemapb.CollectionSchema{
+			StructArrayFields: []*schemapb.StructArrayFieldSchema{
+				{
+					FieldID: 102,
+					Name:    "test_struct",
+					Fields: []*schemapb.FieldSchema{
+						{
+							FieldID:     1021,
+							Name:        "sub_field",
+							DataType:    schemapb.DataType_Array,
+							ElementType: schemapb.DataType_Int32,
+						},
+					},
+				},
+			},
+		}
+
+		originalFieldsData := make([]*schemapb.FieldData, len(fieldsData))
+		copy(originalFieldsData, fieldsData)
+		originalOutputFields := make([]string, len(outputFields))
+		copy(originalOutputFields, outputFields)
+
+		resultFieldsData, resultOutputFields := reconstructStructFieldDataCommon(fieldsData, outputFields, schema)
+
+		// Should not modify anything for count(*) query
+		assert.Equal(t, originalFieldsData, resultFieldsData)
+		assert.Equal(t, originalOutputFields, resultOutputFields)
+	})
+
+	t.Run("no struct array fields - should return early", func(t *testing.T) {
+		fieldsData := []*schemapb.FieldData{
+			{
+				FieldName: "field1",
+				FieldId:   100,
+				Type:      schemapb.DataType_Int64,
+			},
+			{
+				FieldName: "field2",
+				FieldId:   101,
+				Type:      schemapb.DataType_VarChar,
+			},
+		}
+		outputFields := []string{"field1", "field2"}
+
+		schema := &schemapb.CollectionSchema{
+			StructArrayFields: []*schemapb.StructArrayFieldSchema{},
+		}
+
+		originalFieldsData := make([]*schemapb.FieldData, len(fieldsData))
+		copy(originalFieldsData, fieldsData)
+		originalOutputFields := make([]string, len(outputFields))
+		copy(originalOutputFields, outputFields)
+
+		resultFieldsData, resultOutputFields := reconstructStructFieldDataCommon(fieldsData, outputFields, schema)
+
+		// Should not modify anything when no struct array fields
+		assert.Equal(t, originalFieldsData, resultFieldsData)
+		assert.Equal(t, originalOutputFields, resultOutputFields)
+	})
+
+	t.Run("reconstruct single struct field", func(t *testing.T) {
+		// Create mock data
+		subField1Data := &schemapb.FieldData{
+			FieldName: "sub_int_array",
+			FieldId:   1021,
+			Type:      schemapb.DataType_Array,
+			Field: &schemapb.FieldData_Scalars{
+				Scalars: &schemapb.ScalarField{
+					Data: &schemapb.ScalarField_ArrayData{
+						ArrayData: &schemapb.ArrayArray{
+							ElementType: schemapb.DataType_Int32,
+							Data: []*schemapb.ScalarField{
+								{
+									Data: &schemapb.ScalarField_IntData{
+										IntData: &schemapb.IntArray{Data: []int32{1, 2, 3}},
+									},
+								},
+							},
+						},
+					},
+				},
+			},
+		}
+
+		subField2Data := &schemapb.FieldData{
+			FieldName: "sub_text_array",
+			FieldId:   1022,
+			Type:      schemapb.DataType_Array,
+			Field: &schemapb.FieldData_Scalars{
+				Scalars: &schemapb.ScalarField{
+					Data: &schemapb.ScalarField_ArrayData{
+						ArrayData: &schemapb.ArrayArray{
+							ElementType: schemapb.DataType_VarChar,
+							Data: []*schemapb.ScalarField{
+								{
+									Data: &schemapb.ScalarField_StringData{
+										StringData: &schemapb.StringArray{Data: []string{"hello", "world"}},
+									},
+								},
+							},
+						},
+					},
+				},
+			},
+		}
+
+		fieldsData := []*schemapb.FieldData{subField1Data, subField2Data}
+		outputFields := []string{"sub_int_array", "sub_text_array"}
+
+		schema := &schemapb.CollectionSchema{
+			Fields: []*schemapb.FieldSchema{
+				{
+					FieldID:      100,
+					Name:         "pk",
+					IsPrimaryKey: true,
+					DataType:     schemapb.DataType_Int64,
+				},
+			},
+			StructArrayFields: []*schemapb.StructArrayFieldSchema{
+				{
+					FieldID: 102,
+					Name:    "test_struct",
+					Fields: []*schemapb.FieldSchema{
+						{
+							FieldID:     1021,
+							Name:        "sub_int_array",
+							DataType:    schemapb.DataType_Array,
+							ElementType: schemapb.DataType_Int32,
+						},
+						{
+							FieldID:     1022,
+							Name:        "sub_text_array",
+							DataType:    schemapb.DataType_Array,
+							ElementType: schemapb.DataType_VarChar,
+						},
+					},
+				},
+			},
+		}
+
+		resultFieldsData, resultOutputFields := reconstructStructFieldDataCommon(fieldsData, outputFields, schema)
+
+		// Check result
+		assert.Len(t, resultFieldsData, 1, "Should only have one reconstructed struct field")
+		assert.Len(t, resultOutputFields, 1, "Output fields should only have one")
+
+		structField := resultFieldsData[0]
+		assert.Equal(t, "test_struct", structField.FieldName)
+		assert.Equal(t, int64(102), structField.FieldId)
+		assert.Equal(t, schemapb.DataType_ArrayOfStruct, structField.Type)
+		assert.Equal(t, "test_struct", resultOutputFields[0])
+
+		// Check fields inside struct
+		structArrays := structField.GetStructArrays()
+		assert.NotNil(t, structArrays)
+		assert.Len(t, structArrays.Fields, 2, "Struct should contain 2 sub fields")
+
+		// Check sub fields
+		var foundIntField, foundTextField bool
+		for _, field := range structArrays.Fields {
+			switch field.FieldId {
+			case 1021:
+				assert.Equal(t, "sub_int_array", field.FieldName)
+				assert.Equal(t, schemapb.DataType_Array, field.Type)
+				foundIntField = true
+			case 1022:
+				assert.Equal(t, "sub_text_array", field.FieldName)
+				assert.Equal(t, schemapb.DataType_Array, field.Type)
+				foundTextField = true
+			}
+		}
+		assert.True(t, foundIntField, "Should find int array field")
+		assert.True(t, foundTextField, "Should find text array field")
+	})
+
+	t.Run("mixed regular and struct fields", func(t *testing.T) {
+		// Create regular field data
+		regularField := &schemapb.FieldData{
+			FieldName: "regular_field",
+			FieldId:   100,
+			Type:      schemapb.DataType_Int64,
+			Field: &schemapb.FieldData_Scalars{
+				Scalars: &schemapb.ScalarField{
+					Data: &schemapb.ScalarField_LongData{
+						LongData: &schemapb.LongArray{Data: []int64{1, 2, 3}},
+					},
+				},
+			},
+		}
+
+		// Create struct sub field data
+		subFieldData := &schemapb.FieldData{
+			FieldName: "sub_field",
+			FieldId:   1021,
+			Type:      schemapb.DataType_Array,
+			Field: &schemapb.FieldData_Scalars{
+				Scalars: &schemapb.ScalarField{
+					Data: &schemapb.ScalarField_ArrayData{
+						ArrayData: &schemapb.ArrayArray{
+							ElementType: schemapb.DataType_Int32,
+							Data: []*schemapb.ScalarField{
+								{
+									Data: &schemapb.ScalarField_IntData{
+										IntData: &schemapb.IntArray{Data: []int32{10, 20}},
+									},
+								},
+							},
+						},
+					},
+				},
+			},
+		}
+
+		fieldsData := []*schemapb.FieldData{regularField, subFieldData}
+		outputFields := []string{"regular_field", "sub_field"}
+
+		schema := &schemapb.CollectionSchema{
+			Fields: []*schemapb.FieldSchema{
+				{
+					FieldID:  100,
+					Name:     "regular_field",
+					DataType: schemapb.DataType_Int64,
+				},
+			},
+			StructArrayFields: []*schemapb.StructArrayFieldSchema{
+				{
+					FieldID: 102,
+					Name:    "test_struct",
+					Fields: []*schemapb.FieldSchema{
+						{
+							FieldID:     1021,
+							Name:        "sub_field",
+							DataType:    schemapb.DataType_Array,
+							ElementType: schemapb.DataType_Int32,
+						},
+					},
+				},
+			},
+		}
+
+		resultFieldsData, resultOutputFields := reconstructStructFieldDataCommon(fieldsData, outputFields, schema)
+
+		// Check result: should have 2 fields (1 regular + 1 reconstructed struct)
+		assert.Len(t, resultFieldsData, 2)
+		assert.Len(t, resultOutputFields, 2)
+
+		// Check regular and struct fields both exist
+		var foundRegularField, foundStructField bool
+		for i, field := range resultFieldsData {
+			switch field.FieldId {
+			case 100:
+				assert.Equal(t, "regular_field", field.FieldName)
+				assert.Equal(t, schemapb.DataType_Int64, field.Type)
+				assert.Equal(t, "regular_field", resultOutputFields[i])
+				foundRegularField = true
+			case 102:
+				assert.Equal(t, "test_struct", field.FieldName)
+				assert.Equal(t, schemapb.DataType_ArrayOfStruct, field.Type)
+				assert.Equal(t, "test_struct", resultOutputFields[i])
+				foundStructField = true
+			}
+		}
+		assert.True(t, foundRegularField, "Should find regular field")
+		assert.True(t, foundStructField, "Should find reconstructed struct field")
+	})
+
+	t.Run("multiple struct fields", func(t *testing.T) {
+		// Create sub field for first struct
+		struct1SubField := &schemapb.FieldData{
+			FieldName: "struct1_sub",
+			FieldId:   1021,
+			Type:      schemapb.DataType_Array,
+			Field: &schemapb.FieldData_Scalars{
+				Scalars: &schemapb.ScalarField{
+					Data: &schemapb.ScalarField_ArrayData{
+						ArrayData: &schemapb.ArrayArray{
+							ElementType: schemapb.DataType_Int32,
+							Data:        []*schemapb.ScalarField{},
+						},
+					},
+				},
+			},
+		}
+
+		// Create sub fields for second struct
+		struct2SubField1 := &schemapb.FieldData{
+			FieldName: "struct2_sub1",
+			FieldId:   1031,
+			Type:      schemapb.DataType_Array,
+			Field: &schemapb.FieldData_Scalars{
+				Scalars: &schemapb.ScalarField{
+					Data: &schemapb.ScalarField_ArrayData{
+						ArrayData: &schemapb.ArrayArray{
+							ElementType: schemapb.DataType_Int32,
+							Data:        []*schemapb.ScalarField{},
+						},
+					},
+				},
+			},
+		}
+
+		struct2SubField2 := &schemapb.FieldData{
+			FieldName: "struct2_sub2",
+			FieldId:   1032,
+			Type:      schemapb.DataType_VarChar,
+			Field: &schemapb.FieldData_Scalars{
+				Scalars: &schemapb.ScalarField{
+					Data: &schemapb.ScalarField_StringData{
+						StringData: &schemapb.StringArray{Data: []string{"test"}},
+					},
+				},
+			},
+		}
+
+		fieldsData := []*schemapb.FieldData{struct1SubField, struct2SubField1, struct2SubField2}
+		outputFields := []string{"struct1_sub", "struct2_sub1", "struct2_sub2"}
+
+		schema := &schemapb.CollectionSchema{
+			Fields: []*schemapb.FieldSchema{
+				{
+					FieldID:      100,
+					Name:         "pk",
+					IsPrimaryKey: true,
+					DataType:     schemapb.DataType_Int64,
+				},
+			},
+			StructArrayFields: []*schemapb.StructArrayFieldSchema{
+				{
+					FieldID: 102,
+					Name:    "struct1",
+					Fields: []*schemapb.FieldSchema{
+						{
+							FieldID:     1021,
+							Name:        "struct1_sub",
+							DataType:    schemapb.DataType_Array,
+							ElementType: schemapb.DataType_Int32,
+						},
+					},
+				},
+				{
+					FieldID: 103,
+					Name:    "struct2",
+					Fields: []*schemapb.FieldSchema{
+						{
+							FieldID:     1031,
+							Name:        "struct2_sub1",
+							DataType:    schemapb.DataType_Array,
+							ElementType: schemapb.DataType_Int32,
+						},
+						{
+							FieldID:  1032,
+							Name:     "struct2_sub2",
+							DataType: schemapb.DataType_VarChar,
+						},
+					},
+				},
+			},
+		}
+
+		resultFieldsData, resultOutputFields := reconstructStructFieldDataCommon(fieldsData, outputFields, schema)
+
+		// Check result: should have 2 struct fields
+		assert.Len(t, resultFieldsData, 2)
+		assert.Len(t, resultOutputFields, 2)
+
+		// Check both struct fields
+		var foundStruct1, foundStruct2 bool
+		for _, field := range resultFieldsData {
+			switch field.FieldId {
+			case 102:
+				assert.Equal(t, "struct1", field.FieldName)
+				assert.Equal(t, schemapb.DataType_ArrayOfStruct, field.Type)
+				foundStruct1 = true
+				structArrays := field.GetStructArrays()
+				assert.NotNil(t, structArrays)
+				assert.Len(t, structArrays.Fields, 1)
+			case 103:
+				assert.Equal(t, "struct2", field.FieldName)
+				assert.Equal(t, schemapb.DataType_ArrayOfStruct, field.Type)
+				foundStruct2 = true
+				structArrays := field.GetStructArrays()
+				assert.NotNil(t, structArrays)
+				assert.Len(t, structArrays.Fields, 2)
+			}
+		}
+		assert.True(t, foundStruct1, "Should find struct1")
+		assert.True(t, foundStruct2, "Should find struct2")
+	})
 }

internal/querynodev2/segments/load_index_info.go

@@ -33,7 +33,6 @@ import (
 
 	"google.golang.org/protobuf/proto"
 
-	"github.com/milvus-io/milvus-proto/go-api/v2/schemapb"
 	"github.com/milvus-io/milvus/pkg/v2/metrics"
 	"github.com/milvus-io/milvus/pkg/v2/proto/cgopb"
 	"github.com/milvus-io/milvus/pkg/v2/util/paramtable"
@@ -115,25 +114,6 @@ func (li *LoadIndexInfo) appendIndexFile(ctx context.Context, filePath string) e
 	return HandleCStatus(ctx, &status, "AppendIndexIFile failed")
 }
 
-// appendFieldInfo appends fieldID & fieldType to index
-func (li *LoadIndexInfo) appendFieldInfo(ctx context.Context, collectionID int64, partitionID int64, segmentID int64, fieldID int64, fieldType schemapb.DataType, enableMmap bool, mmapDirPath string) error {
-	var status C.CStatus
-	GetDynamicPool().Submit(func() (any, error) {
-		cColID := C.int64_t(collectionID)
-		cParID := C.int64_t(partitionID)
-		cSegID := C.int64_t(segmentID)
-		cFieldID := C.int64_t(fieldID)
-		cintDType := uint32(fieldType)
-		cEnableMmap := C.bool(enableMmap)
-		cMmapDirPath := C.CString(mmapDirPath)
-		defer C.free(unsafe.Pointer(cMmapDirPath))
-		status = C.AppendFieldInfo(li.cLoadIndexInfo, cColID, cParID, cSegID, cFieldID, cintDType, cEnableMmap, cMmapDirPath)
-		return nil, nil
-	}).Await()
-
-	return HandleCStatus(ctx, &status, "AppendFieldInfo failed")
-}
-
 func (li *LoadIndexInfo) appendStorageInfo(uri string, version int64) {
 	GetDynamicPool().Submit(func() (any, error) {
 		cURI := C.CString(uri)

internal/rootcoord/util.go

@@ -476,9 +476,16 @@ func checkFieldSchema(fieldSchemas []*schemapb.FieldSchema) error {
 
 func checkStructArrayFieldSchema(schemas []*schemapb.StructArrayFieldSchema) error {
 	for _, schema := range schemas {
-		// todo(SpadeA): check struct array field schema
+		if len(schema.GetFields()) == 0 {
+			return merr.WrapErrParameterInvalidMsg("empty fields in StructArrayField is not allowed")
+		}
 
 		for _, field := range schema.GetFields() {
+			if field.GetDataType() != schemapb.DataType_Array && field.GetDataType() != schemapb.DataType_ArrayOfVector {
+				msg := fmt.Sprintf("Fields in StructArrayField can only be array or array of vector, but field %s is %s", field.Name, field.DataType.String())
+				return merr.WrapErrParameterInvalidMsg(msg)
+			}
+
 			if field.IsPartitionKey || field.IsPrimaryKey {
 				msg := fmt.Sprintf("partition key or primary key can not be in struct array field. data type:%s, element type:%s, name:%s",
 					field.DataType.String(), field.ElementType.String(), field.Name)

internal/storage/data_sorter.go

@@ -19,6 +19,7 @@ package storage
 import (
 	"github.com/milvus-io/milvus-proto/go-api/v2/schemapb"
 	"github.com/milvus-io/milvus/pkg/v2/common"
+	"github.com/milvus-io/milvus/pkg/v2/util/typeutil"
 )
 
 // DataSorter sorts insert data
@@ -52,11 +53,7 @@ func (ds *DataSorter) Len() int {
 // Swap swaps each field's i-th and j-th element
 func (ds *DataSorter) Swap(i, j int) {
 	if ds.AllFields == nil {
-		allFields := ds.InsertCodec.Schema.Schema.Fields
-		for _, field := range ds.InsertCodec.Schema.Schema.StructArrayFields {
-			allFields = append(allFields, field.Fields...)
-		}
-		ds.AllFields = allFields
+		ds.AllFields = typeutil.GetAllFieldSchemas(ds.InsertCodec.Schema.Schema)
 	}
 	for _, field := range ds.AllFields {
 		singleData, has := ds.InsertData.Data[field.FieldID]

internal/storage/serde_events.go

@@ -282,9 +282,9 @@ func ValueDeserializerWithSchema(r Record, v []*Value, schema *schemapb.Collecti
 	return valueDeserializer(r, v, allFields, shouldCopy)
 }
 
-func valueDeserializer(r Record, v []*Value, fieldSchema []*schemapb.FieldSchema, shouldCopy bool) error {
+func valueDeserializer(r Record, v []*Value, fields []*schemapb.FieldSchema, shouldCopy bool) error {
 	pkField := func() *schemapb.FieldSchema {
-		for _, field := range fieldSchema {
+		for _, field := range fields {
 			if field.GetIsPrimaryKey() {
 				return field
 			}
@@ -299,12 +299,12 @@ func valueDeserializer(r Record, v []*Value, fieldSchema []*schemapb.FieldSchema
 		value := v[i]
 		if value == nil {
 			value = &Value{}
-			value.Value = make(map[FieldID]interface{}, len(fieldSchema))
+			value.Value = make(map[FieldID]interface{}, len(fields))
 			v[i] = value
 		}
 
 		m := value.Value.(map[FieldID]interface{})
-		for _, f := range fieldSchema {
+		for _, f := range fields {
 			j := f.FieldID
 			dt := f.DataType
 			if r.Column(j).IsNull(i) {

internal/storage/utils.go

@@ -1532,7 +1532,9 @@ func GetDefaultValue(fieldSchema *schemapb.FieldSchema) interface{} {
 func fillMissingFields(schema *schemapb.CollectionSchema, insertData *InsertData) error {
 	batchRows := int64(insertData.GetRowNum())
 
-	for _, field := range schema.Fields {
+	allFields := typeutil.GetAllFieldSchemas(schema)
+
+	for _, field := range allFields {
 		// Skip function output fields and system fields
 		if field.GetIsFunctionOutput() || field.GetFieldID() < 100 {
 			continue

internal/util/indexparamcheck/auto_index_checker.go

@@ -9,7 +9,7 @@ type AUTOINDEXChecker struct {
 	baseChecker
 }
 
-func (c *AUTOINDEXChecker) CheckTrain(dataType schemapb.DataType, params map[string]string) error {
+func (c *AUTOINDEXChecker) CheckTrain(dataType schemapb.DataType, elementType schemapb.DataType, params map[string]string) error {
 	return nil
 }
 

internal/util/indexparamcheck/base_checker.go

@@ -24,7 +24,7 @@ import (
 
 type baseChecker struct{}
 
-func (c baseChecker) CheckTrain(dataType schemapb.DataType, params map[string]string) error {
+func (c baseChecker) CheckTrain(_ schemapb.DataType, _ schemapb.DataType, _ map[string]string) error {
 	return nil
 }
 
@@ -35,7 +35,7 @@ func (c baseChecker) CheckValidDataType(indexType IndexType, field *schemapb.Fie
 
 func (c baseChecker) SetDefaultMetricTypeIfNotExist(dType schemapb.DataType, m map[string]string) {}
 
-func (c baseChecker) StaticCheck(dataType schemapb.DataType, params map[string]string) error {
+func (c baseChecker) StaticCheck(dataType schemapb.DataType, elementType schemapb.DataType, params map[string]string) error {
 	return errors.New("unsupported index type")
 }
 

internal/util/indexparamcheck/base_checker_test.go

@@ -47,9 +47,9 @@ func Test_baseChecker_CheckTrain(t *testing.T) {
 		test.params[common.IndexTypeKey] = "HNSW"
 		var err error
 		if test.params[common.IsSparseKey] == "True" {
-			err = c.CheckTrain(schemapb.DataType_SparseFloatVector, test.params)
+			err = c.CheckTrain(schemapb.DataType_SparseFloatVector, schemapb.DataType_None, test.params)
 		} else {
-			err = c.CheckTrain(schemapb.DataType_FloatVector, test.params)
+			err = c.CheckTrain(schemapb.DataType_FloatVector, schemapb.DataType_None, test.params)
 		}
 		if test.errIsNil {
 			assert.NoError(t, err)
@@ -132,5 +132,5 @@ func Test_baseChecker_CheckValidDataType(t *testing.T) {
 
 func Test_baseChecker_StaticCheck(t *testing.T) {
 	// TODO
-	assert.Error(t, newBaseChecker().StaticCheck(schemapb.DataType_FloatVector, nil))
+	assert.Error(t, newBaseChecker().StaticCheck(schemapb.DataType_FloatVector, schemapb.DataType_None, nil))
 }

internal/util/indexparamcheck/bin_flat_checker_test.go

@@ -68,7 +68,7 @@ func Test_binFlatChecker_CheckTrain(t *testing.T) {
 	c, _ := GetIndexCheckerMgrInstance().GetChecker("BINFLAT")
 	for _, test := range cases {
 		test.params[common.IndexTypeKey] = "BINFLAT"
-		err := c.CheckTrain(schemapb.DataType_BinaryVector, test.params)
+		err := c.CheckTrain(schemapb.DataType_BinaryVector, schemapb.DataType_None, test.params)
 		if test.errIsNil {
 			assert.NoError(t, err)
 		} else {

internal/util/indexparamcheck/bin_ivf_flat_checker_test.go

@@ -119,7 +119,7 @@ func Test_binIVFFlatChecker_CheckTrain(t *testing.T) {
 	c, _ := GetIndexCheckerMgrInstance().GetChecker("BIN_IVF_FLAT")
 	for _, test := range cases {
 		test.params[common.IndexTypeKey] = "BIN_IVF_FLAT"
-		err := c.CheckTrain(schemapb.DataType_BinaryVector, test.params)
+		err := c.CheckTrain(schemapb.DataType_BinaryVector, schemapb.DataType_None, test.params)
 		if test.errIsNil {
 			assert.NoError(t, err)
 		} else {