From a566ec4e81cedb500fc2d8904a75d15c91887d04 Mon Sep 17 00:00:00 2001
From: ivanmorozov333 <ivanmorozov@ydb.tech>
Date: Sun, 7 Jul 2024 08:58:34 +0300
Subject: [PATCH] serialize special keys in separated thread (#6324)

---
 ydb/core/formats/arrow/size_calcer.cpp                     | 7 ++++---
 ydb/core/formats/arrow/size_calcer.h                       | 6 +++---
 ydb/core/tx/columnshard/background_controller.cpp          | 2 +-
 .../tx/columnshard/blobs_action/transaction/tx_write.cpp   | 2 +-
 4 files changed, 9 insertions(+), 8 deletions(-)

diff --git a/ydb/core/formats/arrow/size_calcer.cpp b/ydb/core/formats/arrow/size_calcer.cpp
index d44018faaa77..c718b7807410 100644
--- a/ydb/core/formats/arrow/size_calcer.cpp
+++ b/ydb/core/formats/arrow/size_calcer.cpp
@@ -242,11 +242,12 @@ ui64 GetArrayDataSize(const std::shared_ptr<arrow::Array>& column) {
 }
 
 NKikimr::NArrow::TSerializedBatch TSerializedBatch::Build(std::shared_ptr<arrow::RecordBatch> batch, const TBatchSplitttingContext& context) {
-    std::optional<TFirstLastSpecialKeys> specialKeys;
+    std::optional<TString> specialKeys;
     if (context.GetFieldsForSpecialKeys().size()) {
-        specialKeys = TFirstLastSpecialKeys(batch, context.GetFieldsForSpecialKeys());
+        specialKeys = TFirstLastSpecialKeys(batch, context.GetFieldsForSpecialKeys()).SerializeToString();
     }
-    return TSerializedBatch(NArrow::SerializeSchema(*batch->schema()), NArrow::SerializeBatchNoCompression(batch), batch->num_rows(), NArrow::GetBatchDataSize(batch), specialKeys);
+    return TSerializedBatch(NArrow::SerializeSchema(*batch->schema()), NArrow::SerializeBatchNoCompression(batch), batch->num_rows(), 
+        NArrow::GetBatchDataSize(batch), specialKeys);
 }
 
 TConclusionStatus TSerializedBatch::BuildWithLimit(std::shared_ptr<arrow::RecordBatch> batch, const TBatchSplitttingContext& context, std::optional<TSerializedBatch>& sbL, std::optional<TSerializedBatch>& sbR) {
diff --git a/ydb/core/formats/arrow/size_calcer.h b/ydb/core/formats/arrow/size_calcer.h
index 410d5517bc86..d260427ade3a 100644
--- a/ydb/core/formats/arrow/size_calcer.h
+++ b/ydb/core/formats/arrow/size_calcer.h
@@ -74,13 +74,13 @@ class TSerializedBatch {
     YDB_READONLY_DEF(TString, Data);
     YDB_READONLY(ui32, RowsCount, 0);
     YDB_READONLY(ui32, RawBytes, 0);
-    std::optional<TFirstLastSpecialKeys> SpecialKeys;
+    std::optional<TString> SpecialKeys;
 public:
     size_t GetSize() const {
         return Data.size();
     }
 
-    const TFirstLastSpecialKeys& GetSpecialKeysSafe() const {
+    const TString& GetSpecialKeysSafe() const {
         AFL_VERIFY(SpecialKeys);
         return *SpecialKeys;
     }
@@ -95,7 +95,7 @@ class TSerializedBatch {
     static TConclusionStatus BuildWithLimit(std::shared_ptr<arrow::RecordBatch> batch, const TBatchSplitttingContext& context, std::optional<TSerializedBatch>& sbL, std::optional<TSerializedBatch>& sbR);
     static TSerializedBatch Build(std::shared_ptr<arrow::RecordBatch> batch, const TBatchSplitttingContext& context);
 
-    TSerializedBatch(TString&& schemaData, TString&& data, const ui32 rowsCount, const ui32 rawBytes, const std::optional<TFirstLastSpecialKeys>& specialKeys)
+    TSerializedBatch(TString&& schemaData, TString&& data, const ui32 rowsCount, const ui32 rawBytes, const std::optional<TString>& specialKeys)
         : SchemaData(schemaData)
         , Data(data)
         , RowsCount(rowsCount)
diff --git a/ydb/core/tx/columnshard/background_controller.cpp b/ydb/core/tx/columnshard/background_controller.cpp
index fa6a2b32eb85..7449e7d31ff4 100644
--- a/ydb/core/tx/columnshard/background_controller.cpp
+++ b/ydb/core/tx/columnshard/background_controller.cpp
@@ -20,7 +20,7 @@ void TBackgroundController::CheckDeadlines() {
 void TBackgroundController::CheckDeadlinesIndexation() {
     for (auto&& i : ActiveIndexationTasks) {
         if (TMonotonic::Now() - i.second > NOlap::TCompactionLimits::CompactionTimeout) {
-            AFL_CRIT(NKikimrServices::TX_COLUMNSHARD)("event", "deadline_compaction")("task_id", i.first);
+            AFL_CRIT(NKikimrServices::TX_COLUMNSHARD)("event", "deadline_indexation")("task_id", i.first);
             Y_DEBUG_ABORT_UNLESS(false);
         }
     }
diff --git a/ydb/core/tx/columnshard/blobs_action/transaction/tx_write.cpp b/ydb/core/tx/columnshard/blobs_action/transaction/tx_write.cpp
index 853bcb7c5463..76e27462fa04 100644
--- a/ydb/core/tx/columnshard/blobs_action/transaction/tx_write.cpp
+++ b/ydb/core/tx/columnshard/blobs_action/transaction/tx_write.cpp
@@ -7,7 +7,7 @@ bool TTxWrite::InsertOneBlob(TTransactionContext& txc, const NOlap::TWideSeriali
     meta.SetNumRows(batch->GetRowsCount());
     meta.SetRawBytes(batch->GetRawBytes());
     meta.SetDirtyWriteTimeSeconds(batch.GetStartInstant().Seconds());
-    meta.SetSpecialKeysRawData(batch->GetSpecialKeysSafe().SerializeToString());
+    meta.SetSpecialKeysRawData(batch->GetSpecialKeysSafe());
 
     const auto& blobRange = batch.GetRange();
     Y_ABORT_UNLESS(blobRange.GetBlobId().IsValid());