diff options
9 files changed, 141 insertions, 74 deletions
diff --git a/ydb/core/kqp/ut/indexes/kqp_indexes_prefixed_vector_ut.cpp b/ydb/core/kqp/ut/indexes/kqp_indexes_prefixed_vector_ut.cpp index 33279262c0a..7c11f618454 100644 --- a/ydb/core/kqp/ut/indexes/kqp_indexes_prefixed_vector_ut.cpp +++ b/ydb/core/kqp/ut/indexes/kqp_indexes_prefixed_vector_ut.cpp @@ -173,7 +173,7 @@ Y_UNIT_TEST_SUITE(KqpPrefixedVectorIndexes) { DoPositiveQueriesPrefixedVectorIndexOrderBy(session, "CosineSimilarity", "DESC", covered); } - TSession DoCreateTableForPrefixedVectorIndex(TTableClient& db, bool nullable) { + TSession DoCreateTableForPrefixedVectorIndex(TTableClient& db, bool nullable, bool suffixPk = false) { auto session = db.CreateSession().GetValueSync().GetSession(); { @@ -191,14 +191,25 @@ Y_UNIT_TEST_SUITE(KqpPrefixedVectorIndexes) { .AddNonNullableColumn("emb", EPrimitiveType::String) .AddNonNullableColumn("data", EPrimitiveType::String); } - tableBuilder.SetPrimaryKeyColumns({"pk"}); + if (suffixPk) { + tableBuilder.SetPrimaryKeyColumns({"pk", "user"}); + } else { + tableBuilder.SetPrimaryKeyColumns({"pk"}); + } tableBuilder.BeginPartitioningSettings() .SetMinPartitionsCount(3) - .EndPartitioningSettings(); - auto partitions = TExplicitPartitions{} - .AppendSplitPoints(TValueBuilder{}.BeginTuple().AddElement().OptionalInt64(40).EndTuple().Build()) - .AppendSplitPoints(TValueBuilder{}.BeginTuple().AddElement().OptionalInt64(60).EndTuple().Build()); - tableBuilder.SetPartitionAtKeys(partitions); + .EndPartitioningSettings(); + if (suffixPk) { + auto partitions = TExplicitPartitions{} + .AppendSplitPoints(TValueBuilder{}.BeginTuple().AddElement().OptionalInt64(40).AddElement().OptionalString("").EndTuple().Build()) + .AppendSplitPoints(TValueBuilder{}.BeginTuple().AddElement().OptionalInt64(60).AddElement().OptionalString("").EndTuple().Build()); + tableBuilder.SetPartitionAtKeys(partitions); + } else { + auto partitions = TExplicitPartitions{} + .AppendSplitPoints(TValueBuilder{}.BeginTuple().AddElement().OptionalInt64(40).EndTuple().Build()) + .AppendSplitPoints(TValueBuilder{}.BeginTuple().AddElement().OptionalInt64(60).EndTuple().Build()); + tableBuilder.SetPartitionAtKeys(partitions); + } auto result = session.CreateTable("/Root/TestTable", tableBuilder.Build()).ExtractValueSync(); UNIT_ASSERT_VALUES_EQUAL(result.IsTransportError(), false); UNIT_ASSERT_VALUES_EQUAL_C(result.GetStatus(), EStatus::SUCCESS, result.GetIssues().ToString()); @@ -488,6 +499,58 @@ Y_UNIT_TEST_SUITE(KqpPrefixedVectorIndexes) { DoPositiveQueriesPrefixedVectorIndexOrderByCosine(session, true /*covered*/); } + Y_UNIT_TEST_QUAD(CosineDistanceWithPkPrefix, Nullable, Covered) { + NKikimrConfig::TFeatureFlags featureFlags; + featureFlags.SetEnableVectorIndex(true); + auto setting = NKikimrKqp::TKqpSetting(); + auto serverSettings = TKikimrSettings() + .SetFeatureFlags(featureFlags) + .SetKqpSettings({setting}); + + TKikimrRunner kikimr(serverSettings); + kikimr.GetTestServer().GetRuntime()->SetLogPriority(NKikimrServices::BUILD_INDEX, NActors::NLog::PRI_TRACE); + kikimr.GetTestServer().GetRuntime()->SetLogPriority(NKikimrServices::FLAT_TX_SCHEMESHARD, NActors::NLog::PRI_TRACE); + + auto db = kikimr.GetTableClient(); + + auto session = DoCreateTableForPrefixedVectorIndex(db, Nullable, true); + { + const TString createIndex(Q_(Sprintf(R"( + ALTER TABLE `/Root/TestTable` + ADD INDEX index + GLOBAL USING vector_kmeans_tree + ON (user, emb) %s + WITH (distance=cosine, vector_type="uint8", vector_dimension=2, levels=2, clusters=2); + )", (Covered ? "COVER (emb, data)" : "")))); + + auto result = session.ExecuteSchemeQuery(createIndex) + .ExtractValueSync(); + + UNIT_ASSERT_C(result.IsSuccess(), result.GetIssues().ToString()); + } + { + auto result = session.DescribeTable("/Root/TestTable").ExtractValueSync(); + UNIT_ASSERT_VALUES_EQUAL(result.GetStatus(), NYdb::EStatus::SUCCESS); + const auto& indexes = result.GetTableDescription().GetIndexDescriptions(); + UNIT_ASSERT_EQUAL(indexes.size(), 1); + UNIT_ASSERT_EQUAL(indexes[0].GetIndexName(), "index"); + std::vector<std::string> indexKeyColumns{"user", "emb"}; + UNIT_ASSERT_EQUAL(indexes[0].GetIndexColumns(), indexKeyColumns); + std::vector<std::string> indexDataColumns; + if (Covered) { + indexDataColumns = {"emb", "data"}; + } + UNIT_ASSERT_EQUAL(indexes[0].GetDataColumns(), indexDataColumns); + const auto& settings = std::get<TKMeansTreeSettings>(indexes[0].GetIndexSettings()); + UNIT_ASSERT_EQUAL(settings.Settings.Metric, NYdb::NTable::TVectorIndexSettings::EMetric::CosineDistance); + UNIT_ASSERT_EQUAL(settings.Settings.VectorType, NYdb::NTable::TVectorIndexSettings::EVectorType::Uint8); + UNIT_ASSERT_EQUAL(settings.Settings.VectorDimension, 2); + UNIT_ASSERT_EQUAL(settings.Levels, 2); + UNIT_ASSERT_EQUAL(settings.Clusters, 2); + } + DoPositiveQueriesPrefixedVectorIndexOrderByCosine(session, Covered); + } + } } diff --git a/ydb/core/protos/tx_datashard.proto b/ydb/core/protos/tx_datashard.proto index 58480a322d5..ef8d32617f8 100644 --- a/ydb/core/protos/tx_datashard.proto +++ b/ydb/core/protos/tx_datashard.proto @@ -1668,6 +1668,8 @@ message TEvPrefixKMeansRequest { optional uint32 PrefixColumns = 17; optional NKikimrIndexBuilder.TIndexBuildScanSettings ScanSettings = 18; + + repeated string SourcePrimaryKeyColumns = 19; } message TEvPrefixKMeansResponse { diff --git a/ydb/core/tx/datashard/build_index/kmeans_helper.cpp b/ydb/core/tx/datashard/build_index/kmeans_helper.cpp index 383fc938d57..6fa2c4d84c6 100644 --- a/ydb/core/tx/datashard/build_index/kmeans_helper.cpp +++ b/ydb/core/tx/datashard/build_index/kmeans_helper.cpp @@ -45,51 +45,22 @@ void AddRowToLevel(TBufferData& buffer, TClusterId parent, TClusterId child, con buffer.AddRow(TSerializedCellVec{pk}, TSerializedCellVec::Serialize(data)); } -void AddRowMainToBuild(TBufferData& buffer, TClusterId parent, TArrayRef<const TCell> key, TArrayRef<const TCell> row) { - EnsureNoPostingParentFlag(parent); - - std::array<TCell, 1> cells; - cells[0] = TCell::Make(parent); - auto pk = TSerializedCellVec::Serialize(cells); - TSerializedCellVec::UnsafeAppendCells(key, pk); - buffer.AddRow(TSerializedCellVec{std::move(pk)}, TSerializedCellVec::Serialize(row), - TSerializedCellVec{key}); -} - -void AddRowMainToPosting(TBufferData& buffer, TClusterId parent, TArrayRef<const TCell> key, TArrayRef<const TCell> row, ui32 dataPos) -{ - parent = SetPostingParentFlag(parent); - - std::array<TCell, 1> cells; - cells[0] = TCell::Make(parent); - auto pk = TSerializedCellVec::Serialize(cells); - TSerializedCellVec::UnsafeAppendCells(key, pk); - buffer.AddRow(TSerializedCellVec{std::move(pk)}, TSerializedCellVec::Serialize(row.Slice(dataPos)), - TSerializedCellVec{key}); -} - -void AddRowBuildToBuild(TBufferData& buffer, TClusterId parent, TArrayRef<const TCell> key, TArrayRef<const TCell> row, ui32 prefixColumns) -{ - EnsureNoPostingParentFlag(parent); +void AddRowToData(TBufferData& buffer, TClusterId parent, TArrayRef<const TCell> sourcePk, + TArrayRef<const TCell> dataColumns, TArrayRef<const TCell> origKey, bool isPostingLevel) { + if (isPostingLevel) { + parent = SetPostingParentFlag(parent); + } else { + EnsureNoPostingParentFlag(parent); + } std::array<TCell, 1> cells; cells[0] = TCell::Make(parent); auto pk = TSerializedCellVec::Serialize(cells); - TSerializedCellVec::UnsafeAppendCells(key.Slice(prefixColumns), pk); - buffer.AddRow(TSerializedCellVec{std::move(pk)}, TSerializedCellVec::Serialize(row), - TSerializedCellVec{key}); -} - -void AddRowBuildToPosting(TBufferData& buffer, TClusterId parent, TArrayRef<const TCell> key, TArrayRef<const TCell> row, ui32 dataPos, ui32 prefixColumns) -{ - parent = SetPostingParentFlag(parent); + TSerializedCellVec::UnsafeAppendCells(sourcePk, pk); - std::array<TCell, 1> cells; - cells[0] = TCell::Make(parent); - auto pk = TSerializedCellVec::Serialize(cells); - TSerializedCellVec::UnsafeAppendCells(key.Slice(prefixColumns), pk); - buffer.AddRow(TSerializedCellVec{std::move(pk)}, TSerializedCellVec::Serialize(row.Slice(dataPos)), - TSerializedCellVec{key}); + buffer.AddRow(TSerializedCellVec{std::move(pk)}, + TSerializedCellVec::Serialize(dataColumns), + TSerializedCellVec{origKey}); } TTags MakeScanTags(const TUserTable& table, const TProtoStringType& embedding, @@ -114,12 +85,11 @@ TTags MakeScanTags(const TUserTable& table, const TProtoStringType& embedding, std::shared_ptr<NTxProxy::TUploadTypes> MakeOutputTypes(const TUserTable& table, NKikimrTxDataShard::EKMeansState uploadState, const TProtoStringType& embedding, const google::protobuf::RepeatedPtrField<TProtoStringType>& data, - ui32 prefixColumns) + const google::protobuf::RepeatedPtrField<TProtoStringType>& pkColumns) { auto types = GetAllTypes(table); auto result = std::make_shared<NTxProxy::TUploadTypes>(); - result->reserve(1 + 1 + std::min((table.KeyColumnTypes.size() - prefixColumns) + data.size(), types.size())); Ydb::Type type; type.set_type_id(NTableIndex::ClusterIdType); @@ -133,8 +103,14 @@ std::shared_ptr<NTxProxy::TUploadTypes> MakeOutputTypes(const TUserTable& table, types.erase(it); } }; - for (const auto& column : table.KeyColumnIds | std::views::drop(prefixColumns)) { - addType(table.Columns.at(column).Name); + if (pkColumns.size()) { + for (const auto& column : pkColumns) { + addType(column); + } + } else { + for (const auto& column : table.KeyColumnIds) { + addType(table.Columns.at(column).Name); + } } switch (uploadState) { case NKikimrTxDataShard::EKMeansState::UPLOAD_MAIN_TO_BUILD: diff --git a/ydb/core/tx/datashard/build_index/kmeans_helper.h b/ydb/core/tx/datashard/build_index/kmeans_helper.h index 3630050a7b5..6d2c4926a0f 100644 --- a/ydb/core/tx/datashard/build_index/kmeans_helper.h +++ b/ydb/core/tx/datashard/build_index/kmeans_helper.h @@ -140,21 +140,16 @@ struct TMaxInnerProductSimilarity : TMetric<TCoord> { void AddRowToLevel(TBufferData& buffer, TClusterId parent, TClusterId child, const TString& embedding, bool isPostingLevel); -void AddRowMainToBuild(TBufferData& buffer, TClusterId parent, TArrayRef<const TCell> key, TArrayRef<const TCell> row); +void AddRowToData(TBufferData& buffer, TClusterId parent, TArrayRef<const TCell> sourcePk, + TArrayRef<const TCell> dataColumns, TArrayRef<const TCell> origKey, bool isPostingLevel); -void AddRowMainToPosting(TBufferData& buffer, TClusterId parent, TArrayRef<const TCell> key, TArrayRef<const TCell> row, ui32 dataPos); - -void AddRowBuildToBuild(TBufferData& buffer, TClusterId parent, TArrayRef<const TCell> key, TArrayRef<const TCell> row, ui32 prefixColumns = 1); - -void AddRowBuildToPosting(TBufferData& buffer, TClusterId parent, TArrayRef<const TCell> key, TArrayRef<const TCell> row, ui32 dataPos, ui32 prefixColumns = 1); - -TTags MakeScanTags(const TUserTable& table, const TProtoStringType& embedding, +TTags MakeScanTags(const TUserTable& table, const TProtoStringType& embedding, const google::protobuf::RepeatedPtrField<TProtoStringType>& data, ui32& embeddingPos, ui32& dataPos, NTable::TTag& embeddingTag); std::shared_ptr<NTxProxy::TUploadTypes> MakeOutputTypes(const TUserTable& table, NKikimrTxDataShard::EKMeansState uploadState, const TProtoStringType& embedding, const google::protobuf::RepeatedPtrField<TProtoStringType>& data, - ui32 prefixColumns = 0); + const google::protobuf::RepeatedPtrField<TProtoStringType>& pkColumns = {}); void MakeScan(auto& record, const auto& createScan, const auto& badRequest) { diff --git a/ydb/core/tx/datashard/build_index/local_kmeans.cpp b/ydb/core/tx/datashard/build_index/local_kmeans.cpp index 6e79fec57d8..210419bce7d 100644 --- a/ydb/core/tx/datashard/build_index/local_kmeans.cpp +++ b/ydb/core/tx/datashard/build_index/local_kmeans.cpp @@ -462,28 +462,28 @@ private: void FeedMainToBuild(TArrayRef<const TCell> key, TArrayRef<const TCell> row) { if (auto pos = Clusters.FindCluster(row, EmbeddingPos); pos) { - AddRowMainToBuild(*OutputBuf, Child + *pos, key, row); + AddRowToData(*OutputBuf, Child + *pos, key, row, key, false); } } void FeedMainToPosting(TArrayRef<const TCell> key, TArrayRef<const TCell> row) { if (auto pos = Clusters.FindCluster(row, EmbeddingPos); pos) { - AddRowMainToPosting(*OutputBuf, Child + *pos, key, row, DataPos); + AddRowToData(*OutputBuf, Child + *pos, key, row.Slice(DataPos), key, true); } } void FeedBuildToBuild(TArrayRef<const TCell> key, TArrayRef<const TCell> row) { if (auto pos = Clusters.FindCluster(row, EmbeddingPos); pos) { - AddRowBuildToBuild(*OutputBuf, Child + *pos, key, row); + AddRowToData(*OutputBuf, Child + *pos, key.Slice(1), row, key, false); } } void FeedBuildToPosting(TArrayRef<const TCell> key, TArrayRef<const TCell> row) { if (auto pos = Clusters.FindCluster(row, EmbeddingPos); pos) { - AddRowBuildToPosting(*OutputBuf, Child + *pos, key, row, DataPos); + AddRowToData(*OutputBuf, Child + *pos, key.Slice(1), row.Slice(DataPos), key, true); } } diff --git a/ydb/core/tx/datashard/build_index/prefix_kmeans.cpp b/ydb/core/tx/datashard/build_index/prefix_kmeans.cpp index 469bdf2b49c..73e05b457ba 100644 --- a/ydb/core/tx/datashard/build_index/prefix_kmeans.cpp +++ b/ydb/core/tx/datashard/build_index/prefix_kmeans.cpp @@ -29,7 +29,7 @@ using namespace NKMeans; * * Request: * - The client sends TEvPrefixKMeansRequest with: -* - Child: base ID from which new cluster IDs are assigned within this request. + * - Child: base ID from which new cluster IDs are assigned within this request. * - Each prefix group processed will be assigned cluster IDs starting at Child + 1. * - For a request with K clusters per prefix, the IDs used for the first prefix group are * (Child + 1) to (Child + K), and the parent ID for these is Child. @@ -93,6 +93,9 @@ protected: // FIXME: save PrefixRows as std::vector<std::pair<TSerializedCellVec, TSerializedCellVec>> to avoid parsing const ui32 PrefixColumns; + // for PrefixKMeans, original table's primary key columns are passed separately, + // because the prefix table contains them in a different order if they are both in PK and in the prefix + const ui32 DataColumnCount; TSerializedCellVec Prefix; TBufferData PrefixRows; bool IsFirstPrefixFeed = true; @@ -126,10 +129,14 @@ public: , ResponseActorId{responseActorId} , Response{std::move(response)} , PrefixColumns{request.GetPrefixColumns()} + , DataColumnCount{(ui32)request.GetDataColumns().size()} { const auto& embedding = request.GetEmbeddingColumn(); - const auto& data = request.GetDataColumns(); - ScanTags = MakeScanTags(table, embedding, data, EmbeddingPos, DataPos, EmbeddingTag); + TVector<TString> data{request.GetDataColumns().begin(), request.GetDataColumns().end()}; + for (auto & col: request.GetSourcePrimaryKeyColumns()) { + data.push_back(col); + } + ScanTags = MakeScanTags(table, embedding, {data.begin(), data.end()}, EmbeddingPos, DataPos, EmbeddingTag); Lead.To(ScanTags, {}, NTable::ESeek::Lower); { Ydb::Type type; @@ -141,7 +148,11 @@ public: (*levelTypes)[2] = {NTableIndex::NTableVectorKmeansTreeIndex::CentroidColumn, type}; LevelBuf = Uploader.AddDestination(request.GetLevelName(), std::move(levelTypes)); } - OutputBuf = Uploader.AddDestination(request.GetOutputName(), MakeOutputTypes(table, UploadState, embedding, data, PrefixColumns)); + { + auto outputTypes = MakeOutputTypes(table, UploadState, embedding, + {data.begin(), data.begin()+request.GetDataColumns().size()}, request.GetSourcePrimaryKeyColumns()); + OutputBuf = Uploader.AddDestination(request.GetOutputName(), outputTypes); + } { auto types = GetAllTypes(table); @@ -480,14 +491,14 @@ private: void FeedBuildToBuild(TArrayRef<const TCell> key, TArrayRef<const TCell> row) { if (auto pos = Clusters.FindCluster(row, EmbeddingPos); pos) { - AddRowBuildToBuild(*OutputBuf, Child + *pos, key, row, PrefixColumns); + AddRowToData(*OutputBuf, Child + *pos, row.Slice(DataPos+DataColumnCount), row.Slice(0, DataPos+DataColumnCount), key, false); } } void FeedBuildToPosting(TArrayRef<const TCell> key, TArrayRef<const TCell> row) { if (auto pos = Clusters.FindCluster(row, EmbeddingPos); pos) { - AddRowBuildToPosting(*OutputBuf, Child + *pos, key, row, DataPos, PrefixColumns); + AddRowToData(*OutputBuf, Child + *pos, row.Slice(DataPos+DataColumnCount), row.Slice(DataPos, DataColumnCount), key, true); } } @@ -624,6 +635,14 @@ void TDataShard::HandleSafe(TEvDataShard::TEvPrefixKMeansRequest::TPtr& ev, cons if (request.GetPrefixColumns() > userTable.KeyColumnIds.size()) { badRequest(TStringBuilder() << "Should not be requested on more than " << userTable.KeyColumnIds.size() << " prefix columns"); } + if (request.GetSourcePrimaryKeyColumns().size() == 0) { + badRequest("Request should include source primary key columns"); + } + for (auto pkColumn : request.GetSourcePrimaryKeyColumns()) { + if (!tags.contains(pkColumn)) { + badRequest(TStringBuilder() << "Unknown source primary key column: " << pkColumn); + } + } if (trySendBadRequest()) { return; diff --git a/ydb/core/tx/datashard/build_index/reshuffle_kmeans.cpp b/ydb/core/tx/datashard/build_index/reshuffle_kmeans.cpp index cdf1e59bee4..7328a4e0279 100644 --- a/ydb/core/tx/datashard/build_index/reshuffle_kmeans.cpp +++ b/ydb/core/tx/datashard/build_index/reshuffle_kmeans.cpp @@ -308,28 +308,28 @@ private: void FeedMainToBuild(TArrayRef<const TCell> key, TArrayRef<const TCell> row) { if (auto pos = Clusters.FindCluster(row, EmbeddingPos); pos) { - AddRowMainToBuild(*OutputBuf, Child + *pos, key, row); + AddRowToData(*OutputBuf, Child + *pos, key, row, key, false); } } void FeedMainToPosting(TArrayRef<const TCell> key, TArrayRef<const TCell> row) { if (auto pos = Clusters.FindCluster(row, EmbeddingPos); pos) { - AddRowMainToPosting(*OutputBuf, Child + *pos, key, row, DataPos); + AddRowToData(*OutputBuf, Child + *pos, key, row.Slice(DataPos), key, true); } } void FeedBuildToBuild(TArrayRef<const TCell> key, TArrayRef<const TCell> row) { if (auto pos = Clusters.FindCluster(row, EmbeddingPos); pos) { - AddRowBuildToBuild(*OutputBuf, Child + *pos, key, row); + AddRowToData(*OutputBuf, Child + *pos, key.Slice(1), row, key, false); } } void FeedBuildToPosting(TArrayRef<const TCell> key, TArrayRef<const TCell> row) { if (auto pos = Clusters.FindCluster(row, EmbeddingPos); pos) { - AddRowBuildToPosting(*OutputBuf, Child + *pos, key, row, DataPos); + AddRowToData(*OutputBuf, Child + *pos, key.Slice(1), row.Slice(DataPos), key, true); } } }; diff --git a/ydb/core/tx/datashard/build_index/ut/ut_prefix_kmeans.cpp b/ydb/core/tx/datashard/build_index/ut/ut_prefix_kmeans.cpp index 1f31222c61a..e05f4d235b5 100644 --- a/ydb/core/tx/datashard/build_index/ut/ut_prefix_kmeans.cpp +++ b/ydb/core/tx/datashard/build_index/ut/ut_prefix_kmeans.cpp @@ -65,6 +65,9 @@ Y_UNIT_TEST_SUITE (TTxDataShardPrefixKMeansScan) { rec.SetEmbeddingColumn("embedding"); rec.SetPrefixColumns(1); + rec.AddSourcePrimaryKeyColumns("key"); + + rec.SetPrefixName(kPrefixTable); rec.SetLevelName(kLevelTable); rec.SetOutputName(kPostingTable); @@ -130,6 +133,7 @@ Y_UNIT_TEST_SUITE (TTxDataShardPrefixKMeansScan) { rec.SetEmbeddingColumn("embedding"); rec.AddDataColumns("data"); rec.SetPrefixColumns(1); + rec.AddSourcePrimaryKeyColumns("key"); rec.SetPrefixName(kPrefixTable); rec.SetLevelName(kLevelTable); @@ -252,6 +256,10 @@ Y_UNIT_TEST_SUITE (TTxDataShardPrefixKMeansScan) { }, "{ <main>: Error: Unknown table id: 0 }"); DoBadRequest(server, sender, [](NKikimrTxDataShard::TEvPrefixKMeansRequest& request) { + request.ClearSourcePrimaryKeyColumns(); + }, "{ <main>: Error: Request should include source primary key columns }"); + + DoBadRequest(server, sender, [](NKikimrTxDataShard::TEvPrefixKMeansRequest& request) { request.MutableSettings()->set_vector_type(VectorIndexSettings::VECTOR_TYPE_UNSPECIFIED); }, "{ <main>: Error: Wrong vector type }"); DoBadRequest(server, sender, [](NKikimrTxDataShard::TEvPrefixKMeansRequest& request) { diff --git a/ydb/core/tx/schemeshard/schemeshard_build_index__progress.cpp b/ydb/core/tx/schemeshard/schemeshard_build_index__progress.cpp index 0a4b7b82574..539b6016a75 100644 --- a/ydb/core/tx/schemeshard/schemeshard_build_index__progress.cpp +++ b/ydb/core/tx/schemeshard/schemeshard_build_index__progress.cpp @@ -692,6 +692,10 @@ private: *ev->Record.MutableDataColumns() = { buildInfo.DataColumns.begin(), buildInfo.DataColumns.end() }; + const auto& tableInfo = *Self->Tables.at(buildInfo.TablePathId); + for (ui32 keyPos: tableInfo.KeyColumnIds) { + ev->Record.AddSourcePrimaryKeyColumns(tableInfo.Columns.at(keyPos).Name); + } auto shardId = CommonFillRecord<false>(ev->Record, shardIdx, buildInfo); ev->Record.SetSeed(ui64(shardId)); |