Merge pull request ClickHouse#32670 from CurtizJ/fix-reading-in-order

Fix reading in order of sorting key from `Distributed` and `Merge` tables

src/Interpreters/TreeOptimizer.cpp

@@ -410,12 +410,19 @@ void optimizeDuplicateDistinct(ASTSelectQuery & select)
 /// has a single argument and not an aggregate functions.
 void optimizeMonotonousFunctionsInOrderBy(ASTSelectQuery * select_query, ContextPtr context,
                                           const TablesWithColumns & tables_with_columns,
-                                          const Names & sorting_key_columns)
+                                          const TreeRewriterResult & result)
 {
     auto order_by = select_query->orderBy();
     if (!order_by)
         return;
 
+    /// Do not apply optimization for Distributed and Merge storages,
+    /// because we can't get the sorting key of their undelying tables
+    /// and we can break the matching of the sorting key for `read_in_order`
+    /// optimization by removing monotonous functions from the prefix of key.
+    if (result.is_remote_storage || (result.storage && result.storage->getName() == "Merge"))
+        return;
+
     for (const auto & child : order_by->children)
     {
         auto * order_by_element = child->as<ASTOrderByElement>();
@@ -438,6 +445,8 @@ void optimizeMonotonousFunctionsInOrderBy(ASTSelectQuery * select_query, Context
         }
     }
 
+    auto sorting_key_columns = result.metadata_snapshot ? result.metadata_snapshot->getSortingKeyColumns() : Names{};
+
     bool is_sorting_key_prefix = true;
     for (size_t i = 0; i < order_by->children.size(); ++i)
     {
@@ -802,8 +811,7 @@ void TreeOptimizer::apply(ASTPtr & query, TreeRewriterResult & result,
 
     /// Replace monotonous functions with its argument
     if (settings.optimize_monotonous_functions_in_order_by)
-        optimizeMonotonousFunctionsInOrderBy(select_query, context, tables_with_columns,
-            result.metadata_snapshot ? result.metadata_snapshot->getSortingKeyColumns() : Names{});
+        optimizeMonotonousFunctionsInOrderBy(select_query, context, tables_with_columns, result);
 
     /// Remove duplicate items from ORDER BY.
     /// Execute it after all order by optimizations,

tests/queries/0_stateless/02147_order_by_optimizations.reference

@@ -0,0 +1,21 @@
+SELECT
+    date,
+    v
+FROM t_02147
+ORDER BY
+    toStartOfHour(date) ASC,
+    v ASC
+SELECT
+    date,
+    v
+FROM t_02147_dist
+ORDER BY
+    toStartOfHour(date) ASC,
+    v ASC
+SELECT
+    date,
+    v
+FROM t_02147_merge
+ORDER BY
+    toStartOfHour(date) ASC,
+    v ASC

tests/queries/0_stateless/02147_order_by_optimizations.sql

@@ -0,0 +1,15 @@
+DROP TABLE IF EXISTS t_02147;
+DROP TABLE IF EXISTS t_02147_dist;
+DROP TABLE IF EXISTS t_02147_merge;
+
+CREATE TABLE t_02147 (date DateTime, v UInt32)
+ENGINE = MergeTree ORDER BY toStartOfHour(date);
+
+CREATE TABLE t_02147_dist AS t_02147 ENGINE = Distributed(test_shard_localhost, currentDatabase(), t_02147);
+CREATE TABLE t_02147_merge AS t_02147 ENGINE = Merge(currentDatabase(), 't_02147');
+
+SET optimize_monotonous_functions_in_order_by = 1;
+
+EXPLAIN SYNTAX SELECT * FROM t_02147 ORDER BY toStartOfHour(date), v;
+EXPLAIN SYNTAX SELECT * FROM t_02147_dist ORDER BY toStartOfHour(date), v;
+EXPLAIN SYNTAX SELECT * FROM t_02147_merge ORDER BY toStartOfHour(date), v;