KIKIMR-19019: temporary fix before remove splitting into CS

4 files changed, 14 insertions, 2 deletions

diff --git a/ydb/core/formats/arrow/size_calcer.cpp b/ydb/core/formats/arrow/size_calcer.cpp
index 9f9e98fc72..56a8ee7c29 100644
--- a/ydb/core/formats/arrow/size_calcer.cpp
+++ b/ydb/core/formats/arrow/size_calcer.cpp
@@ -4,6 +4,7 @@
 #include "dictionary/conversion.h"
 #include <contrib/libs/apache/arrow/cpp/src/arrow/type.h>
 #include <util/system/yassert.h>
+#include <util/string/builder.h>
 
 namespace NKikimr::NArrow {
 
@@ -261,4 +262,8 @@ bool TSerializedBatch::BuildWithLimit(std::shared_ptr<arrow::RecordBatch> batch,
     return true;
 }
 
+TString TSerializedBatch::DebugString() const {
+    return TStringBuilder() << "(data_size=" << Data.size() << ";schema_data_size=" << SchemaData.size() << ";rows_count=" << RowsCount << ";raw_bytes=" << RawBytes << ";)";
+}
+
 }
diff --git a/ydb/core/formats/arrow/size_calcer.h b/ydb/core/formats/arrow/size_calcer.h
index f17566c85c..a7ca8095bf 100644
--- a/ydb/core/formats/arrow/size_calcer.h
+++ b/ydb/core/formats/arrow/size_calcer.h
@@ -56,6 +56,8 @@ public:
         return Data.size();
     }
 
+    TString DebugString() const;
+
     static bool BuildWithLimit(std::shared_ptr<arrow::RecordBatch> batch, const ui32 sizeLimit, std::vector<TSerializedBatch>& result, TString* errorMessage);
     static bool BuildWithLimit(std::shared_ptr<arrow::RecordBatch> batch, const ui32 sizeLimit, std::optional<TSerializedBatch>& sbL, std::optional<TSerializedBatch>& sbR, TString* errorMessage);
     static TSerializedBatch Build(std::shared_ptr<arrow::RecordBatch> batch);
diff --git a/ydb/core/tx/columnshard/engines/writer/indexed_blob_constructor.cpp b/ydb/core/tx/columnshard/engines/writer/indexed_blob_constructor.cpp
index 0ab5872d39..cf1037451f 100644
--- a/ydb/core/tx/columnshard/engines/writer/indexed_blob_constructor.cpp
+++ b/ydb/core/tx/columnshard/engines/writer/indexed_blob_constructor.cpp
@@ -51,6 +51,11 @@ bool TIndexedWriteController::TBlobConstructor::Init() {
         return false;
     }
     BlobsSplitted = splitResult.ReleaseResult();
+    if (BlobsSplitted.size() > 1) {
+        for (auto&& i : BlobsSplitted) {
+            AFL_DEBUG(NKikimrServices::TX_COLUMNSHARD)("event", "strange_blobs_splitting")("blob", i.DebugString())("original_size", Owner.WriteData.GetSize());
+        }
+    }
     return true;
 }
 
diff --git a/ydb/core/tx/ev_write/columnshard_splitter.h b/ydb/core/tx/ev_write/columnshard_splitter.h
index 4d6026e71d..6e41a6b2ec 100644
--- a/ydb/core/tx/ev_write/columnshard_splitter.h
+++ b/ydb/core/tx/ev_write/columnshard_splitter.h
@@ -115,7 +115,7 @@ private:
         
         TFullSplitData result(numShards);
         if (numShards == 1) {
-            NArrow::TSplitBlobResult blobsSplitted = NArrow::SplitByBlobSize(batch, NColumnShard::TLimits::GetMaxBlobSize());
+            NArrow::TSplitBlobResult blobsSplitted = NArrow::SplitByBlobSize(batch, NColumnShard::TLimits::GetMaxBlobSize() - 1024 * 1024);
             if (!blobsSplitted) {
                 return TYdbConclusionStatus::Fail(Ydb::StatusIds::SCHEME_ERROR, "cannot split batch in according to limits: " + blobsSplitted.GetErrorMessage());
             }
@@ -143,7 +143,7 @@ private:
         THashMap<ui64, TString> out;
         for (size_t i = 0; i < sharded.size(); ++i) {
             if (sharded[i]) {
-                NArrow::TSplitBlobResult blobsSplitted = NArrow::SplitByBlobSize(sharded[i], NColumnShard::TLimits::GetMaxBlobSize());
+                NArrow::TSplitBlobResult blobsSplitted = NArrow::SplitByBlobSize(sharded[i], NColumnShard::TLimits::GetMaxBlobSize() - 1024 * 1024);
                 if (!blobsSplitted) {
                     return TYdbConclusionStatus::Fail(Ydb::StatusIds::SCHEME_ERROR, "cannot split batch in according to limits: " + blobsSplitted.GetErrorMessage());
                 }