Merge 652e80a into 1dd2b99

Author: pashandor789

ydb/core/kqp/opt/logical/kqp_opt_log.cpp

@@ -160,10 +160,10 @@ class TKqpLogicalOptTransformer : public TOptimizeTransformerBase {
     }
 
     TMaybeNode<TExprBase> OptimizeEquiJoinWithCosts(TExprBase node, TExprContext& ctx) {
-        auto maxDPccpDPTableSize = Config->MaxDPccpDPTableSize.Get().GetOrElse(TDqSettings::TDefault::MaxDPccpDPTableSize);
+        auto maxDPhypDPTableSize = Config->MaxDPHypDPTableSize.Get().GetOrElse(TDqSettings::TDefault::MaxDPHypDPTableSize);
         auto optLevel = Config->CostBasedOptimizationLevel.Get().GetOrElse(Config->DefaultCostBasedOptimizationLevel);
         auto providerCtx = TKqpProviderContext(KqpCtx, optLevel);
-        auto opt = std::unique_ptr<IOptimizerNew>(MakeNativeOptimizerNew(providerCtx, maxDPccpDPTableSize));
+        auto opt = std::unique_ptr<IOptimizerNew>(MakeNativeOptimizerNew(providerCtx, maxDPhypDPTableSize));
         TExprBase output = DqOptimizeEquiJoinWithCosts(node, ctx, TypesCtx, optLevel,
             *opt, [](auto& rels, auto label, auto node, auto stat) {
                 rels.emplace_back(std::make_shared<TKqpRelOptimizerNode>(TString(label), *stat, node));

ydb/core/kqp/provider/yql_kikimr_settings.cpp

@@ -93,7 +93,7 @@ TKikimrConfiguration::TKikimrConfiguration() {
     REGISTER_SETTING(*this, EnableSpillingNodes)
         .Parser([](const TString& v) { return ParseEnableSpillingNodes(v); });
 
-    REGISTER_SETTING(*this, MaxDPccpDPTableSize);
+    REGISTER_SETTING(*this, MaxDPHypDPTableSize);
 
     REGISTER_SETTING(*this, MaxTasksPerStage);
     REGISTER_SETTING(*this, MaxSequentialReadsInFlight);

ydb/core/kqp/provider/yql_kikimr_settings.h

@@ -69,7 +69,7 @@ struct TKikimrSettings {
     NCommon::TConfSetting<bool, false> OptUseFinalizeByKey;
     NCommon::TConfSetting<ui32, false> CostBasedOptimizationLevel;
 
-    NCommon::TConfSetting<ui32, false> MaxDPccpDPTableSize;
+    NCommon::TConfSetting<ui32, false> MaxDPHypDPTableSize;
 
 
     NCommon::TConfSetting<ui32, false> MaxTasksPerStage;

ydb/library/yql/dq/opt/dq_opt_join_cost_based.cpp

@@ -235,9 +235,10 @@ void ComputeStatistics(const std::shared_ptr<TJoinOptimizerNode>& join, IProvide
 
 class TOptimizerNativeNew: public IOptimizerNew {
 public:
-    TOptimizerNativeNew(IProviderContext& ctx, ui32 maxDPhypDPTableSize)
+    TOptimizerNativeNew(IProviderContext& ctx, ui32 maxDPhypDPTableSize, TExprContext* exprCtx = nullptr)
         : IOptimizerNew(ctx)
-        , MaxDPhypTableSize_(maxDPhypDPTableSize)
+        , MaxDPHypTableSize_(maxDPhypDPTableSize)
+        , ExprCtx(exprCtx)
     {}
 
     std::shared_ptr<TJoinOptimizerNode> JoinSearch(
@@ -272,8 +273,17 @@ class TOptimizerNativeNew: public IOptimizerNew {
         TJoinHypergraph<TNodeSet> hypergraph = MakeJoinHypergraph<TNodeSet>(joinTree, hints);
         TDPHypSolver<TNodeSet> solver(hypergraph, this->Pctx);
 
-        if (solver.CountCC(MaxDPhypTableSize_) >= MaxDPhypTableSize_) {
+        if (solver.CountCC(MaxDPHypTableSize_) >= MaxDPHypTableSize_) {
             YQL_CLOG(TRACE, CoreDq) << "Maximum DPhyp threshold exceeded";
+            if (ExprCtx) {
+                ExprCtx->AddWarning(
+                    YqlIssue(
+                        {}, TIssuesIds::DQ_OPTIMIZE_ERROR, 
+                        "Cost Based Optimizer didn't work: "
+                        "Enumeration is too large, use PRAGMA MaxDPHypDPTableSize='4294967295' to disable the limitation"
+                    )
+                );
+            }
             ComputeStatistics(joinTree, this->Pctx);
             return joinTree;
         }
@@ -304,11 +314,12 @@ class TOptimizerNativeNew: public IOptimizerNew {
     }
 
 private:
-    ui32 MaxDPhypTableSize_;
+    ui32 MaxDPHypTableSize_;
+    TExprContext* ExprCtx;
 };
 
-IOptimizerNew* MakeNativeOptimizerNew(IProviderContext& ctx, const ui32 maxDPhypDPTableSize) {
-    return new TOptimizerNativeNew(ctx, maxDPhypDPTableSize);
+IOptimizerNew* MakeNativeOptimizerNew(IProviderContext& ctx, const ui32 maxDPhypDPTableSize, TExprContext* exprCtx = nullptr) {
+    return new TOptimizerNativeNew(ctx, maxDPhypDPTableSize, exprCtx);
 }
 
 TExprBase DqOptimizeEquiJoinWithCosts(
@@ -357,6 +368,7 @@ TExprBase DqOptimizeEquiJoinWithCosts(
     // of the EquiJoin and n-1 argument are the parameters to EquiJoin
 
     if (!DqCollectJoinRelationsWithStats(rels, typesCtx, equiJoin, providerCollect)){
+        ctx.AddWarning(YqlIssue({}, TIssuesIds::DQ_OPTIMIZE_ERROR, "Cost Based Optimizer didn't work: couldn't load statistics"));
         return node;
     }
 

ydb/library/yql/dq/opt/dq_opt_log.h

@@ -63,7 +63,7 @@ IGraphTransformer::TStatus DqWrapIO(const TExprNode::TPtr& input, TExprNode::TPt
 
 NNodes::TExprBase DqExpandMatchRecognize(NNodes::TExprBase node, TExprContext& ctx, TTypeAnnotationContext& typeAnnCtx);
 
-IOptimizerNew* MakeNativeOptimizerNew(IProviderContext& ctx, const ui32 maxDPccpDPTableSize);
+IOptimizerNew* MakeNativeOptimizerNew(IProviderContext& ctx, const ui32 maxDPhypDPTableSize);
 
 NNodes::TMaybeNode<NNodes::TExprBase> UnorderedOverDqReadWrap(NNodes::TExprBase node, TExprContext& ctx, const std::function<const TParentsMap*()>& getParents, bool enableDqReplicate, TTypeAnnotationContext& typeAnnCtx);
 

ydb/library/yql/providers/dq/common/yql_dq_settings.h

@@ -57,7 +57,7 @@ struct TDqSettings {
         static constexpr ETaskRunnerStats TaskRunnerStats = ETaskRunnerStats::Basic;
         static constexpr ESpillingEngine SpillingEngine = ESpillingEngine::Disable;
         static constexpr ui32 CostBasedOptimizationLevel = 4;
-        static constexpr ui32 MaxDPccpDPTableSize = 40000U;
+        static constexpr ui32 MaxDPHypDPTableSize = 40000U;
         static constexpr ui64 MaxAttachmentsSize = 2_GB;
         static constexpr bool SplitStageOnDqReplicate = true;
         static constexpr ui64 EnableSpillingNodes = 0;