fix SS split test

Author: kunga

ydb/core/tx/datashard/datashard__stats.cpp

@@ -167,7 +167,7 @@ class TTableStatsCoroBuilder : public TActorCoroImpl, private IPages {
         LOG_DEBUG_S(GetActorContext(), NKikimrServices::TX_DATASHARD, "BuildStats result at datashard " << TabletId << ", for tableId " << TableId
             << ": RowCount " << ev->Stats.RowCount << ", DataSize " << ev->Stats.DataSize.Size << ", IndexSize " << ev->Stats.IndexSize.Size << ", PartCount " << ev->PartCount
             << (ev->PartOwners.size() > 1 || ev->PartOwners.size() == 1 && *ev->PartOwners.begin() != TabletId ? ", with borrowed parts" : "")
-            << ", LoadedSize " << PagesSize << ", " << NFmt::Do(*Spent));
+            << ", LoadedSize " << PagesSize << ", " << NFmt::Do(*Spent) << ", HistogramKeys " << ev->Stats.DataSizeHistogram.size());
 
         Send(ReplyTo, ev.Release());
 

ydb/core/tx/schemeshard/schemeshard__table_stats_histogram.cpp

@@ -31,10 +31,11 @@ static bool IsIntegerType(NScheme::TTypeInfo type) {
 }
 
 TSerializedCellVec ChooseSplitKeyByHistogram(const NKikimrTableStats::THistogram& histogram, const TConstArrayRef<NScheme::TTypeInfo> &keyColumnTypes) {
+    // Note: actually its histogram keys count, not buckets count
     ui64 bucketsCount = histogram.BucketsSize();
-    ui64 idxLo = bucketsCount * 0.33;
-    ui64 idxMed = bucketsCount * 0.5;
-    ui64 idxHi = bucketsCount * 0.66;
+    ui64 idxLo = bucketsCount / 3;
+    ui64 idxMed = bucketsCount / 2;
+    ui64 idxHi = bucketsCount * 2 / 3;
 
     TSerializedCellVec keyLo(histogram.GetBuckets(idxLo).GetKey());
     TSerializedCellVec keyMed(histogram.GetBuckets(idxMed).GetKey());
@@ -298,7 +299,8 @@ bool TTxPartitionHistogram::Execute(TTransactionContext& txc, const TActorContex
                     << " for pathId " << tableId
                     << " state '" << DatashardStateName(rec.GetShardState()).data() << "'"
                     << " dataSize " << dataSize
-                    << " rowCount " << rowCount);
+                    << " rowCount " << rowCount
+                    << " dataSizeHistogram buckets " << rec.GetTableStats().GetDataSizeHistogram().BucketsSize());
 
     if (!Self->Tables.contains(tableId))
         return true;
@@ -349,6 +351,7 @@ bool TTxPartitionHistogram::Execute(TTransactionContext& txc, const TActorContex
     } else {
         // Choose number of parts and split boundaries
         const auto& histogram = rec.GetTableStats().GetDataSizeHistogram();
+        // TODO: split with 1 key in histogram when no FlatIndex
         if (histogram.BucketsSize() < 2) {
             return true;
         }

ydb/core/tx/schemeshard/ut_split_merge/ut_split_merge.cpp

@@ -107,7 +107,7 @@ Y_UNIT_TEST_SUITE(TSchemeShardSplitBySizeTest) {
             UNIT_ASSERT_VALUES_EQUAL(err, "");
             UNIT_ASSERT_VALUES_EQUAL(status, NKikimrProto::EReplyStatus::OK);;
         };
-        for (ui64 key = 0; key < 1000; ++key) {
+        for (ui64 key = 0; key < 1200; ++key) {
             fnWriteRow(TTestTxConfig::FakeHiveTablets, key* 1'000'000);
         }