Merge remote-tracking branch 'origin/master' into pr-plan-rewrite

.clang-format

@@ -89,7 +89,7 @@ PenaltyBreakFirstLessLess: 120
 PenaltyBreakString: 1000
 PenaltyExcessCharacter: 1000000
 PenaltyReturnTypeOnItsOwnLine: 60
-RemoveBracesLLVM: true
+RemoveBracesLLVM: false
 SpaceAfterCStyleCast: false
 SpaceBeforeAssignmentOperators: true
 SpaceBeforeParens: ControlStatements

.clang-tidy

@@ -138,6 +138,8 @@ Checks: [
 
   # This is a good check, but clang-tidy crashes, see https://github.com/llvm/llvm-project/issues/91872
   '-modernize-use-constraints',
+  # https://github.com/abseil/abseil-cpp/issues/1667
+  '-clang-analyzer-optin.core.EnumCastOutOfRange'
 ]
 
 WarningsAsErrors: '*'

docs/en/operations/configuration-files.md

@@ -67,6 +67,8 @@ generates merged configuration file:
 </clickhouse>
 ```
 
+### Using from_env and from_zk
+
 To specify that a value of an element should be replaced by the value of an environment variable, you can use attribute `from_env`.
 
 Example with `$MAX_QUERY_SIZE = 150000`:
@@ -93,6 +95,59 @@ which is equal to
 </clickhouse>
 ```
 
+The same is possible using `from_zk`:
+
+``` xml
+<clickhouse>
+    <postgresql_port from_zk="/zk_configs/postgresql_port"/>
+</clickhouse>
+```
+
+```
+# clickhouse-keeper-client
+/ :) touch /zk_configs
+/ :) create /zk_configs/postgresql_port "9005"
+/ :) get /zk_configs/postgresql_port
+9005
+```
+
+which is equal to
+
+
+``` xml
+<clickhouse>
+    <postgresql_port>9005</postgresql_port>
+</clickhouse>
+```
+
+#### Default values for from_env and from_zk attributes
+
+It's possible to set the default value and substitute it only if the environment variable or zookeeper node is set using `replace="1"`.
+
+With previous example, but `MAX_QUERY_SIZE` is unset:
+
+``` xml
+<clickhouse>
+    <profiles>
+        <default>
+            <max_query_size from_env="MAX_QUERY_SIZE" replace="1">150000</max_query_size>
+        </default>
+    </profiles>
+</clickhouse>
+```
+
+will take the default value
+
+``` xml
+<clickhouse>
+    <profiles>
+        <default>
+            <max_query_size>150000</max_query_size>
+        </default>
+    </profiles>
+</clickhouse>
+```
+
 ## Substituting Configuration {#substitution}
 
 The config can define substitutions. There are two types of substitutions:

src/Common/ThreadPool.cpp

@@ -5,7 +5,6 @@
 #include <Common/OpenTelemetryTraceContext.h>
 #include <Common/noexcept_scope.h>
 
-#include <cassert>
 #include <type_traits>
 
 #include <Poco/Util/Application.h>
@@ -437,6 +436,11 @@ void ThreadPoolImpl<Thread>::worker(typename std::list<Thread>::iterator thread_
             /// We don't run jobs after `shutdown` is set, but we have to properly dequeue all jobs and finish them.
             if (shutdown)
             {
+                {
+                    ALLOW_ALLOCATIONS_IN_SCOPE;
+                    /// job can contain packaged_task which can set exception during destruction
+                    job_data.reset();
+                }
                 job_is_done = true;
                 continue;
             }

src/IO/ParallelReadBuffer.cpp

@@ -73,8 +73,9 @@ bool ParallelReadBuffer::addReaderToPool()
 
     auto worker = read_workers.emplace_back(std::make_shared<ReadWorker>(input, range_start, size));
 
-    ++active_working_readers;
     schedule([this, my_worker = std::move(worker)]() mutable { readerThreadFunction(std::move(my_worker)); }, Priority{});
+    /// increase number of workers only after we are sure that the reader was scheduled
+    ++active_working_readers;
 
     return true;
 }

src/Interpreters/ActionsDAG.cpp

@@ -21,6 +21,9 @@
 #include <base/sort.h>
 #include <Common/JSONBuilder.h>
 
+#include <absl/container/flat_hash_map.h>
+#include <absl/container/inlined_vector.h>
+
 
 namespace DB
 {
@@ -708,16 +711,18 @@ static ColumnWithTypeAndName executeActionForPartialResult(const ActionsDAG::Nod
     return res_column;
 }
 
-Block ActionsDAG::updateHeader(Block header) const
+Block ActionsDAG::updateHeader(const Block & header) const
 {
     IntermediateExecutionResult node_to_column;
     std::set<size_t> pos_to_remove;
 
     {
-        std::unordered_map<std::string_view, std::list<size_t>> input_positions;
+        using inline_vector = absl::InlinedVector<size_t, 7>; // 64B, holding max 7 size_t elements inlined
+        absl::flat_hash_map<std::string_view, inline_vector> input_positions;
 
-        for (size_t pos = 0; pos < inputs.size(); ++pos)
-            input_positions[inputs[pos]->result_name].emplace_back(pos);
+        /// We insert from last to first in the inlinedVector so it's easier to pop_back matches later
+        for (size_t pos = inputs.size(); pos != 0; pos--)
+            input_positions[inputs[pos - 1]->result_name].emplace_back(pos - 1);
 
         for (size_t pos = 0; pos < header.columns(); ++pos)
         {
@@ -725,10 +730,11 @@ Block ActionsDAG::updateHeader(Block header) const
             auto it = input_positions.find(col.name);
             if (it != input_positions.end() && !it->second.empty())
             {
-                auto & list = it->second;
                 pos_to_remove.insert(pos);
-                node_to_column[inputs[list.front()]] = col;
-                list.pop_front();
+
+                auto & v = it->second;
+                node_to_column[inputs[v.back()]] = col;
+                v.pop_back();
             }
         }
     }
@@ -746,18 +752,21 @@ Block ActionsDAG::updateHeader(Block header) const
         throw;
     }
 
-    if (isInputProjected())
-        header.clear();
-    else
-        header.erase(pos_to_remove);
 
     Block res;
-
+    res.reserve(result_columns.size());
     for (auto & col : result_columns)
         res.insert(std::move(col));
 
-    for (auto && item : header)
-        res.insert(std::move(item));
+    if (isInputProjected())
+        return res;
+
+    res.reserve(header.columns() - pos_to_remove.size());
+    for (size_t i = 0; i < header.columns(); i++)
+    {
+        if (!pos_to_remove.contains(i))
+            res.insert(header.data[i]);
+    }
 
     return res;
 }

src/Interpreters/ActionsDAG.h

@@ -272,7 +272,7 @@ public:
     ///
     /// In addition, check that result constants are constants according to DAG.
     /// In case if function return constant, but arguments are not constant, materialize it.
-    Block updateHeader(Block header) const;
+    Block updateHeader(const Block & header) const;
 
     using IntermediateExecutionResult = std::unordered_map<const Node *, ColumnWithTypeAndName>;
     static ColumnsWithTypeAndName evaluatePartialResult(

src/Interpreters/HashJoin.cpp

@@ -2487,10 +2487,15 @@ HashJoin::~HashJoin()
 {
     if (!data)
     {
-        LOG_TRACE(log, "{}Join data has been already released", instance_log_id);
+        LOG_TEST(log, "{}Join data has been already released", instance_log_id);
         return;
     }
-    LOG_TRACE(log, "{}Join data is being destroyed, {} bytes and {} rows in hash table", instance_log_id, getTotalByteCount(), getTotalRowCount());
+    LOG_TEST(
+        log,
+        "{}Join data is being destroyed, {} bytes and {} rows in hash table",
+        instance_log_id,
+        getTotalByteCount(),
+        getTotalRowCount());
 }
 
 template <typename Mapped>

src/Planner/Planner.cpp

@@ -1229,8 +1229,9 @@ void Planner::buildQueryPlanIfNeeded()
     if (query_plan.isInitialized())
         return;
 
-    LOG_TRACE(getLogger("Planner"), "Query {} to stage {}{}",
-        query_tree->formatConvertedASTForErrorMessage(),
+    LOG_TRACE(
+        getLogger("Planner"),
+        "Query to stage {}{}",
         QueryProcessingStage::toString(select_query_options.to_stage),
         select_query_options.only_analyze ? " only analyze" : "");
 
@@ -1506,8 +1507,9 @@ void Planner::buildPlanForQueryNode()
     auto & mapping = join_tree_query_plan.query_node_to_plan_step_mapping;
     query_node_to_plan_step_mapping.insert(mapping.begin(), mapping.end());
 
-    LOG_TRACE(getLogger("Planner"), "Query {} from stage {} to stage {}{}",
-        query_tree->formatConvertedASTForErrorMessage(),
+    LOG_TRACE(
+        getLogger("Planner"),
+        "Query from stage {} to stage {}{}",
         QueryProcessingStage::toString(from_stage),
         QueryProcessingStage::toString(select_query_options.to_stage),
         select_query_options.only_analyze ? " only analyze" : "");

src/Processors/QueryPlan/Optimizations/filterPushDown.cpp

@@ -262,10 +262,6 @@ static size_t tryPushDownOverJoinStep(QueryPlan::Node * parent_node, QueryPlan::
         {
             const auto & left_table_key_name = join_clause.key_names_left[i];
             const auto & right_table_key_name = join_clause.key_names_right[i];
-
-            if (!join_header.has(left_table_key_name) || !join_header.has(right_table_key_name))
-                continue;
-
             const auto & left_table_column = left_stream_input_header.getByName(left_table_key_name);
             const auto & right_table_column = right_stream_input_header.getByName(right_table_key_name);
 
@@ -338,9 +334,9 @@ static size_t tryPushDownOverJoinStep(QueryPlan::Node * parent_node, QueryPlan::
     auto join_filter_push_down_actions = filter->getExpression()->splitActionsForJOINFilterPushDown(filter->getFilterColumnName(),
         filter->removesFilterColumn(),
         left_stream_available_columns_to_push_down,
-        left_stream_input_header.getColumnsWithTypeAndName(),
+        left_stream_input_header,
         right_stream_available_columns_to_push_down,
-        right_stream_input_header.getColumnsWithTypeAndName(),
+        right_stream_input_header,
         equivalent_columns_to_push_down,
         equivalent_left_stream_column_to_right_stream_column,
         equivalent_right_stream_column_to_left_stream_column);

src/Processors/QueryPlan/SourceStepWithFilter.cpp

@@ -21,7 +21,7 @@ Block SourceStepWithFilter::applyPrewhereActions(Block block, const PrewhereInfo
     {
         if (prewhere_info->row_level_filter)
         {
-            block = prewhere_info->row_level_filter->updateHeader(std::move(block));
+            block = prewhere_info->row_level_filter->updateHeader(block);
             auto & row_level_column = block.getByName(prewhere_info->row_level_column_name);
             if (!row_level_column.type->canBeUsedInBooleanContext())
             {
@@ -36,7 +36,7 @@ Block SourceStepWithFilter::applyPrewhereActions(Block block, const PrewhereInfo
 
         if (prewhere_info->prewhere_actions)
         {
-            block = prewhere_info->prewhere_actions->updateHeader(std::move(block));
+            block = prewhere_info->prewhere_actions->updateHeader(block);
 
             auto & prewhere_column = block.getByName(prewhere_info->prewhere_column_name);
             if (!prewhere_column.type->canBeUsedInBooleanContext())

src/Processors/Transforms/ExpressionTransform.cpp

@@ -3,9 +3,9 @@
 namespace DB
 {
 
-Block ExpressionTransform::transformHeader(Block header, const ActionsDAG & expression)
+Block ExpressionTransform::transformHeader(const Block & header, const ActionsDAG & expression)
 {
-    return expression.updateHeader(std::move(header));
+    return expression.updateHeader(header);
 }
 
 

src/Processors/Transforms/ExpressionTransform.h

@@ -24,7 +24,7 @@ public:
 
     String getName() const override { return "ExpressionTransform"; }
 
-    static Block transformHeader(Block header, const ActionsDAG & expression);
+    static Block transformHeader(const Block & header, const ActionsDAG & expression);
 
 protected:
     void transform(Chunk & chunk) override;

src/Processors/Transforms/FilterTransform.cpp

@@ -174,26 +174,22 @@ static std::unique_ptr<IFilterDescription> combineFilterAndIndices(
 }
 
 Block FilterTransform::transformHeader(
-    Block header,
-    const ActionsDAG * expression,
-    const String & filter_column_name,
-    bool remove_filter_column)
+    const Block & header, const ActionsDAG * expression, const String & filter_column_name, bool remove_filter_column)
 {
-    if (expression)
-        header = expression->updateHeader(std::move(header));
+    Block result = expression ? expression->updateHeader(header) : header;
 
-    auto filter_type = header.getByName(filter_column_name).type;
+    auto filter_type = result.getByName(filter_column_name).type;
     if (!filter_type->onlyNull() && !isUInt8(removeNullable(removeLowCardinality(filter_type))))
         throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_COLUMN_FOR_FILTER,
             "Illegal type {} of column {} for filter. Must be UInt8 or Nullable(UInt8).",
             filter_type->getName(), filter_column_name);
 
     if (remove_filter_column)
-        header.erase(filter_column_name);
+        result.erase(filter_column_name);
     else
-        replaceFilterToConstant(header, filter_column_name);
+        replaceFilterToConstant(result, filter_column_name);
 
-    return header;
+    return result;
 }
 
 FilterTransform::FilterTransform(

src/Processors/Transforms/FilterTransform.h

@@ -22,11 +22,8 @@ public:
         const Block & header_, ExpressionActionsPtr expression_, String filter_column_name_,
         bool remove_filter_column_, bool on_totals_ = false, std::shared_ptr<std::atomic<size_t>> rows_filtered_ = nullptr);
 
-    static Block transformHeader(
-            Block header,
-            const ActionsDAG * expression,
-            const String & filter_column_name,
-            bool remove_filter_column);
+    static Block
+    transformHeader(const Block & header, const ActionsDAG * expression, const String & filter_column_name, bool remove_filter_column);
 
     String getName() const override { return "FilterTransform"; }
 

src/Processors/Transforms/JoiningTransform.cpp

@@ -14,12 +14,12 @@ namespace ErrorCodes
 
 Block JoiningTransform::transformHeader(Block header, const JoinPtr & join)
 {
-    LOG_DEBUG(getLogger("JoiningTransform"), "Before join block: '{}'", header.dumpStructure());
+    LOG_TEST(getLogger("JoiningTransform"), "Before join block: '{}'", header.dumpStructure());
     join->checkTypesOfKeys(header);
     join->initialize(header);
     ExtraBlockPtr tmp;
     join->joinBlock(header, tmp);
-    LOG_DEBUG(getLogger("JoiningTransform"), "After join block: '{}'", header.dumpStructure());
+    LOG_TEST(getLogger("JoiningTransform"), "After join block: '{}'", header.dumpStructure());
     return header;
 }
 

src/Processors/Transforms/MergeJoinTransform.cpp

@@ -338,8 +338,6 @@ static void prepareChunk(Chunk & chunk)
 
 void MergeJoinAlgorithm::initialize(Inputs inputs)
 {
-    LOG_DEBUG(&Poco::Logger::get("XXXX"), "{}:{}: {} - '{}'", __FILE__, __LINE__, 0, inputs[0].chunk.dumpStructure());
-    LOG_DEBUG(&Poco::Logger::get("XXXX"), "{}:{}: {} - '{}'", __FILE__, __LINE__, 1, inputs[1].chunk.dumpStructure());
     if (inputs.size() != 2)
         throw Exception(ErrorCodes::LOGICAL_ERROR, "Two inputs are required, got {}", inputs.size());
 
@@ -351,8 +349,6 @@ void MergeJoinAlgorithm::initialize(Inputs inputs)
 
 void MergeJoinAlgorithm::consume(Input & input, size_t source_num)
 {
-    LOG_DEBUG(&Poco::Logger::get("XXXX"), "{}:{}: {} - '{}'", __FILE__, __LINE__, source_num, input.chunk.dumpStructure());
-
     if (input.skip_last_row)
         throw Exception(ErrorCodes::NOT_IMPLEMENTED, "skip_last_row is not supported");
 
@@ -816,15 +812,9 @@ IMergingAlgorithm::Status MergeJoinAlgorithm::merge()
     if (!cursors[1]->cursor.isValid() && !cursors[1]->fullyCompleted())
         return Status(1);
 
-    for (size_t i = 0; i < 2; ++i)
-    {
-        LOG_DEBUG(&Poco::Logger::get("XXXX"), "{}:{}: sampleColumns {} '{}'", __FILE__, __LINE__, i, cursors[i]->sampleBlock().dumpStructure());
-    }
-
 
     if (auto result = handleAllJoinState())
     {
-        LOG_DEBUG(&Poco::Logger::get("XXXX"), "{}:{}: '{}'", __FILE__, __LINE__, result ? result->chunk.dumpStructure() : "NA");
         return std::move(*result);
     }
 

src/Processors/Transforms/TotalsHavingTransform.cpp

@@ -49,7 +49,7 @@ Block TotalsHavingTransform::transformHeader(
 
     if (expression)
     {
-        block = expression->updateHeader(std::move(block));
+        block = expression->updateHeader(block);
         if (remove_filter)
             block.erase(filter_column_name);
     }

tests/clickhouse-test

@@ -2549,15 +2549,15 @@ def reportLogStats(args):
         WITH
             240 AS mins,
             (
-                SELECT (count(), sum(length(message)))
+                SELECT (count(), sum(length(toValidUTF8(message))))
                 FROM system.text_log
                 WHERE (now() - toIntervalMinute(mins)) < event_time
             ) AS total
         SELECT
             count() AS count,
             round(count / (total.1), 3) AS `count_%`,
-            formatReadableSize(sum(length(message))) AS size,
-            round(sum(length(message)) / (total.2), 3) AS `size_%`,
+            formatReadableSize(sum(length(toValidUTF8(message)))) AS size,
+            round(sum(length(toValidUTF8(message))) / (total.2), 3) AS `size_%`,
             countDistinct(logger_name) AS uniq_loggers,
             countDistinct(thread_id) AS uniq_threads,
             groupArrayDistinct(toString(level)) AS levels,
@@ -2580,8 +2580,8 @@ def reportLogStats(args):
             240 AS mins
         SELECT
             count() AS count,
-            substr(replaceRegexpAll(message, '[^A-Za-z]+', ''), 1, 32) AS pattern,
-            substr(any(message), 1, 256) as runtime_message,
+            substr(replaceRegexpAll(toValidUTF8(message), '[^A-Za-z]+', ''), 1, 32) AS pattern,
+            substr(any(toValidUTF8(message)), 1, 256) as runtime_message,
             any((extract(source_file, '/[a-zA-Z0-9_]+\\.[a-z]+'), source_line)) as line
         FROM system.text_log
         WHERE (now() - toIntervalMinute(mins)) < event_time AND message_format_string = ''
@@ -2596,7 +2596,7 @@ def reportLogStats(args):
     print("\n")
 
     query = """
-        SELECT message_format_string, count(), any(message) AS any_message
+        SELECT message_format_string, count(), any(toValidUTF8(message)) AS any_message
         FROM system.text_log
         WHERE (now() - toIntervalMinute(240)) < event_time
         AND (message NOT LIKE (replaceRegexpAll(message_format_string, '{[:.0-9dfx]*}', '%') AS s))
@@ -2631,8 +2631,8 @@ def reportLogStats(args):
               'Unknown identifier: ''{}''', 'User name is empty', 'Expected function, got: {}',
               'Attempt to read after eof', 'String size is too big ({}), maximum: {}'
         ) AS known_short_messages
-        SELECT count() AS c, message_format_string, substr(any(message), 1, 120),
-            min(if(length(regexpExtract(message, '(.*)\\([A-Z0-9_]+\\)')) as prefix_len > 0, prefix_len, length(message)) - 26 AS length_without_exception_boilerplate) AS min_length_without_exception_boilerplate
+        SELECT count() AS c, message_format_string, substr(any(toValidUTF8(message)), 1, 120),
+            min(if(length(regexpExtract(toValidUTF8(message), '(.*)\\([A-Z0-9_]+\\)')) as prefix_len > 0, prefix_len, length(toValidUTF8(message))) - 26 AS length_without_exception_boilerplate) AS min_length_without_exception_boilerplate
         FROM system.text_log
         WHERE (now() - toIntervalMinute(240)) < event_time
             AND (length(message_format_string) < 16

tests/integration/test_non_default_compression/test.py

@@ -2,7 +2,7 @@ import random
 import string
 
 import pytest
-from helpers.cluster import ClickHouseCluster
+from helpers.cluster import ClickHouseCluster, is_arm
 
 cluster = ClickHouseCluster(__file__)
 
@@ -255,6 +255,11 @@ def test_uncompressed_cache_plus_zstd_codec(start_cluster):
 
 
 def test_preconfigured_deflateqpl_codec(start_cluster):
+    if is_arm():
+        pytest.skip(
+            "Skipping test because it's special test for Intel code (doesn't work on ARM)"
+        )
+
     node6.query(
         """
     CREATE TABLE compression_codec_multiple_with_key (

tests/queries/0_stateless/00002_log_and_exception_messages_formatting.sql

@@ -194,7 +194,7 @@ select 'exceptions shorter than 30',
     (uniqExact(message_format_string) as c) <= max_messages,
     c <= max_messages ? [] : groupUniqArray(message_format_string)
     from logs
-    where message ilike '%DB::Exception%' and if(length(extract(message, '(.*)\\([A-Z0-9_]+\\)')) as pref > 0, pref, length(message)) < 30 + 26 and message_format_string not in known_short_messages;
+    where message ilike '%DB::Exception%' and if(length(extract(toValidUTF8(message), '(.*)\\([A-Z0-9_]+\\)')) as pref > 0, pref, length(toValidUTF8(message))) < 30 + 26 and message_format_string not in known_short_messages;
 
 -- Avoid too noisy messages: top 1 message frequency must be less than 30%. We should reduce the threshold
 WITH 0.30 as threshold
@@ -207,7 +207,7 @@ select
 with 0.16 as threshold
 select
     'noisy Trace messages',
-    greatest(coalesce(((select message_format_string, count() from logs where level = 'Trace' and message_format_string not in ('Access granted: {}{}', '{} -> {}', 'Query {} to stage {}{}', 'Query {} from stage {} to stage {}{}')
+    greatest(coalesce(((select message_format_string, count() from logs where level = 'Trace' and message_format_string not in ('Access granted: {}{}', '{} -> {}', 'Query to stage {}{}', 'Query from stage {} to stage {}{}')
                         group by message_format_string order by count() desc limit 1) as top_message).2, 0) / (select count() from logs), threshold) as r,
     r <= threshold ? '' : top_message.1;
 
@@ -252,7 +252,7 @@ select 'number of noisy messages',
 -- Each message matches its pattern (returns 0 rows)
 -- Note: maybe we should make it stricter ('Code:%Exception: '||s||'%'), but it's not easy because of addMessage
 select 'incorrect patterns', greatest(uniqExact(message_format_string), 15) from (
-    select message_format_string, any(message) as any_message from logs
+    select message_format_string, any(toValidUTF8(message)) as any_message from logs
     where ((rand() % 8) = 0)
     and message not like (replaceRegexpAll(message_format_string, '{[:.0-9dfx]*}', '%') as s)
     and message not like (s || ' (skipped % similar messages)')

tests/queries/0_stateless/02455_one_row_from_csv_memory_usage.sh

@@ -9,7 +9,7 @@ USER_FILES_PATH=$($CLICKHOUSE_CLIENT --query "select _path,_file from file('none
 
 cp "$CUR_DIR"/data_csv/10m_rows.csv.xz $USER_FILES_PATH/
 
-${CLICKHOUSE_CLIENT} --query="SELECT * FROM file('10m_rows.csv.xz' , 'CSVWithNames') order by identifier, number, name, surname, birthday LIMIT 1 settings max_threads=1, max_memory_usage=1000000000"
-${CLICKHOUSE_CLIENT} --query="SELECT * FROM file('10m_rows.csv.xz' , 'CSVWithNames') order by identifier, number, name, surname, birthday LIMIT 1 settings max_threads=1, max_memory_usage=100000000"
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM file('10m_rows.csv.xz' , 'CSVWithNames') order by identifier, number, name, surname, birthday LIMIT 1 settings input_format_parallel_parsing=1, max_threads=1, max_parsing_threads=16, min_chunk_bytes_for_parallel_parsing=10485760, max_memory_usage=1000000000"
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM file('10m_rows.csv.xz' , 'CSVWithNames') order by identifier, number, name, surname, birthday LIMIT 1 settings input_format_parallel_parsing=1, max_threads=1, max_parsing_threads=16, min_chunk_bytes_for_parallel_parsing=10485760, max_memory_usage=100000000"
 
 rm $USER_FILES_PATH/10m_rows.csv.xz

tests/queries/0_stateless/03152_join_filter_push_down_equivalent_columns.reference

@@ -0,0 +1,91 @@
+-- { echoOn }
+
+EXPLAIN header = 1, indexes = 1
+SELECT name FROM users INNER JOIN users2 USING name WHERE users.name ='Alice';
+Expression ((Project names + (Projection + )))
+Header: name String
+  Join (JOIN FillRightFirst)
+  Header: __table1.name String
+    Filter (( + Change column names to column identifiers))
+    Header: __table1.name String
+      ReadFromMergeTree (default.users)
+      Header: name String
+      Indexes:
+        PrimaryKey
+          Keys: 
+            name
+          Condition: (name in [\'Alice\', \'Alice\'])
+          Parts: 1/3
+          Granules: 1/3
+    Filter (( + Change column names to column identifiers))
+    Header: __table2.name String
+      ReadFromMergeTree (default.users2)
+      Header: name String
+      Indexes:
+        PrimaryKey
+          Keys: 
+            name
+          Condition: (name in [\'Alice\', \'Alice\'])
+          Parts: 1/3
+          Granules: 1/3
+SELECT '--';
+--
+EXPLAIN header = 1, indexes = 1
+SELECT name FROM users LEFT JOIN users2 USING name WHERE users.name ='Alice';
+Expression ((Project names + (Projection + )))
+Header: name String
+  Join (JOIN FillRightFirst)
+  Header: __table1.name String
+    Filter (( + Change column names to column identifiers))
+    Header: __table1.name String
+      ReadFromMergeTree (default.users)
+      Header: name String
+      Indexes:
+        PrimaryKey
+          Keys: 
+            name
+          Condition: (name in [\'Alice\', \'Alice\'])
+          Parts: 1/3
+          Granules: 1/3
+    Filter (( + Change column names to column identifiers))
+    Header: __table2.name String
+      ReadFromMergeTree (default.users2)
+      Header: name String
+      Indexes:
+        PrimaryKey
+          Keys: 
+            name
+          Condition: (name in [\'Alice\', \'Alice\'])
+          Parts: 1/3
+          Granules: 1/3
+SELECT '--';
+--
+EXPLAIN header = 1, indexes = 1
+SELECT name FROM users RIGHT JOIN users2 USING name WHERE users2.name ='Alice';
+Expression ((Project names + (Projection + )))
+Header: name String
+  Join (JOIN FillRightFirst)
+  Header: __table1.name String
+          __table2.name String
+    Filter (( + Change column names to column identifiers))
+    Header: __table1.name String
+      ReadFromMergeTree (default.users)
+      Header: name String
+      Indexes:
+        PrimaryKey
+          Keys: 
+            name
+          Condition: (name in [\'Alice\', \'Alice\'])
+          Parts: 1/3
+          Granules: 1/3
+    Filter (( + Change column names to column identifiers))
+    Header: __table2.name String
+      ReadFromMergeTree (default.users2)
+      Header: name String
+      Indexes:
+        PrimaryKey
+          Keys: 
+            name
+          Condition: (name in [\'Alice\', \'Alice\'])
+          Parts: 1/3
+          Granules: 1/3

tests/queries/0_stateless/03152_join_filter_push_down_equivalent_columns.sql

@@ -0,0 +1,35 @@
+SET allow_experimental_analyzer = 1;
+
+DROP TABLE IF EXISTS users;
+CREATE TABLE users (uid Int16, name String, age Int16) ENGINE=MergeTree order by (uid, name);
+
+INSERT INTO users VALUES (1231, 'John', 33);
+INSERT INTO users VALUES (6666, 'Ksenia', 48);
+INSERT INTO users VALUES (8888, 'Alice', 50);
+
+DROP TABLE IF EXISTS users2;
+CREATE TABLE users2 (uid Int16, name String, age2 Int16) ENGINE=MergeTree order by (uid, name);
+
+INSERT INTO users2 VALUES (1231, 'John', 33);
+INSERT INTO users2 VALUES (6666, 'Ksenia', 48);
+INSERT INTO users2 VALUES (8888, 'Alice', 50);
+
+-- { echoOn }
+
+EXPLAIN header = 1, indexes = 1
+SELECT name FROM users INNER JOIN users2 USING name WHERE users.name ='Alice';
+
+SELECT '--';
+
+EXPLAIN header = 1, indexes = 1
+SELECT name FROM users LEFT JOIN users2 USING name WHERE users.name ='Alice';
+
+SELECT '--';
+
+EXPLAIN header = 1, indexes = 1
+SELECT name FROM users RIGHT JOIN users2 USING name WHERE users2.name ='Alice';
+
+-- { echoOff }
+
+DROP TABLE users;
+DROP TABLE users2;