Fix TIndexBuildInfo::AddParent access beyond map end (#20527) (#20529)

Author: vitalif

ydb/core/tx/schemeshard/schemeshard_info_types.cpp

@@ -2221,7 +2221,9 @@ void TIndexBuildInfo::AddParent(const TSerializedTableRange& range, TShardIdx sh
     const auto [parentFrom, parentTo] = KMeans.Parent == 0
         ? std::pair<NTableIndex::TClusterId, NTableIndex::TClusterId>{0, 0}
         : KMeans.RangeToBorders(range);
-    // TODO(mbkkt) We can make it more granular
+    // TODO Make it more granular - now we just merge all intersecting ranges.
+    // So if all shards have ranges like [1 2] [2 3] [3 4] and so on then we scan
+    // all shards for each cluster even though we could only scan 2 shards for it.
 
     // the new range does not intersect with other ranges, just add it with 1 shard
     auto itFrom = Cluster2Shards.lower_bound(parentFrom);
@@ -2233,7 +2235,7 @@ void TIndexBuildInfo::AddParent(const TSerializedTableRange& range, TShardIdx sh
     // otherwise, this range has multiple shards and we need to merge all intersecting ranges
     auto itTo = parentTo < itFrom->first ? itFrom : Cluster2Shards.lower_bound(parentTo);
     if (itTo == Cluster2Shards.end()) {
-        itTo = Cluster2Shards.rbegin().base();
+        itTo--;
     }
     if (itTo->first < parentTo) {
         const bool needsToReplaceFrom = itFrom == itTo;

ydb/core/tx/schemeshard/schemeshard_info_types.h

@@ -3596,7 +3596,7 @@ struct TIndexBuildInfo: public TSimpleRefCount<TIndexBuildInfo> {
                     indexInfo->KMeans.Rounds = NTableIndex::NTableVectorKmeansTreeIndex::DefaultKMeansRounds;
                     TString createError;
                     indexInfo->Clusters = NKikimr::NKMeans::CreateClusters(desc.settings().settings(), indexInfo->KMeans.Rounds, createError);
-                    Y_ENSURE(indexInfo->Clusters && createError == "");
+                    Y_ENSURE(indexInfo->Clusters, createError);
                     indexInfo->SpecializedIndexDescription = std::move(desc);
                 } break;
                 case NKikimrSchemeOp::TIndexCreationConfig::SPECIALIZEDINDEXDESCRIPTION_NOT_SET:

ydb/core/tx/schemeshard/ut_base/ut_info_types.cpp

@@ -239,4 +239,67 @@ ColumnFamilies {
         "{ 0 -> 0, 1 -> 1 }");
     }
 
+    Y_UNIT_TEST(IndexBuildInfoAddParent) {
+        TIndexBuildInfo info;
+
+        info.KMeans.ParentBegin = 1;
+        info.KMeans.Parent = 1;
+        info.KMeans.ChildBegin = 201;
+        info.KMeans.Child = 201;
+
+        TVector<TCell> from = {TCell::Make((ui64)1), TCell::Make(123)};
+        TVector<TCell> to = {TCell::Make((ui64)3), TCell::Make(123)};
+        auto range = TSerializedTableRange(from, true, to, true);
+        info.AddParent(range, TShardIdx(1, 1));
+
+        UNIT_ASSERT_VALUES_EQUAL(info.Cluster2Shards.size(), 1);
+        UNIT_ASSERT_VALUES_EQUAL(info.Cluster2Shards.at(3).From, 1);
+        UNIT_ASSERT_VALUES_EQUAL(info.Cluster2Shards.at(3).Shards.size(), 1);
+
+        // add an intersecting shard. it even crashed here until 02.07.2025 :)
+        from = {TCell::Make((ui64)3), TCell::Make(123)};
+        to = {TCell::Make((ui64)4), TCell::Make(123)};
+        range = TSerializedTableRange(from, true, to, true);
+        info.AddParent(range, TShardIdx(1, 2));
+
+        // Cluster2Shards is a rather stupid thing - for now, it just merges all intersecting
+        // shard ranges into a single item, thus losing the range info for individual cluster IDs
+        UNIT_ASSERT_VALUES_EQUAL(info.Cluster2Shards.size(), 1);
+        UNIT_ASSERT_VALUES_EQUAL(info.Cluster2Shards.at(4).From, 1);
+        UNIT_ASSERT_VALUES_EQUAL(info.Cluster2Shards.at(4).Shards.size(), 2);
+
+        // add a non-intersecting shard
+        from = {TCell::Make((ui64)5), TCell::Make(123)};
+        to = {TCell::Make((ui64)6), TCell::Make(123)};
+        range = TSerializedTableRange(from, true, to, true);
+        info.AddParent(range, TShardIdx(1, 3));
+
+        UNIT_ASSERT_VALUES_EQUAL(info.Cluster2Shards.size(), 2);
+        UNIT_ASSERT_VALUES_EQUAL(info.Cluster2Shards.at(4).From, 1);
+        UNIT_ASSERT_VALUES_EQUAL(info.Cluster2Shards.at(4).Shards.size(), 2);
+        UNIT_ASSERT_VALUES_EQUAL(info.Cluster2Shards.at(6).From, 5);
+        UNIT_ASSERT_VALUES_EQUAL(info.Cluster2Shards.at(6).Shards.size(), 1);
+
+        // incomplete range is "from (99, +infinity)", thus equal to "from [100]"
+        from = {TCell::Make((ui64)99)};
+        to = {TCell::Make((ui64)100), TCell::Make(123)};
+        range = TSerializedTableRange(from, true, to, true);
+        info.AddParent(range, TShardIdx(1, 4));
+
+        UNIT_ASSERT_VALUES_EQUAL(info.Cluster2Shards.size(), 3);
+        UNIT_ASSERT_VALUES_EQUAL(info.Cluster2Shards.at(100).From, 100);
+        UNIT_ASSERT_VALUES_EQUAL(info.Cluster2Shards.at(100).Shards.size(), 1);
+
+        // insert and check something before an existing item
+        from = {TCell::Make((ui64)50), TCell::Make(123)};
+        to = {TCell::Make((ui64)51), TCell::Make(123)};
+        range = TSerializedTableRange(from, true, to, true);
+        info.AddParent(range, TShardIdx(1, 5));
+
+        UNIT_ASSERT_VALUES_EQUAL(info.Cluster2Shards.size(), 4);
+        UNIT_ASSERT_VALUES_EQUAL(info.Cluster2Shards.at(51).From, 50);
+        UNIT_ASSERT_VALUES_EQUAL(info.Cluster2Shards.at(51).Shards.size(), 1);
+
+    }
+
 }