Merge 56ab216e63 into 7eee149487

FIx 02374_analyzer_join_using

src/Common/ZooKeeper/ZooKeeper.cpp

@@ -341,7 +341,10 @@ Coordination::Error ZooKeeper::tryGetChildren(
     const EventPtr & watch,
     Coordination::ListRequestType list_request_type)
 {
-    return tryGetChildrenWatch(path, res, stat,
+    return tryGetChildrenWatch(
+        path,
+        res,
+        stat,
         watch ? std::make_shared<Coordination::WatchCallback>(callbackForEvent(watch)) : Coordination::WatchCallbackPtr{},
         list_request_type);
 }
@@ -975,11 +978,14 @@ void ZooKeeper::removeRecursive(const std::string & path, uint32_t remove_nodes_
 
 Coordination::Error ZooKeeper::tryRemoveRecursive(const std::string & path, uint32_t remove_nodes_limit)
 {
-    if (!isFeatureEnabled(DB::KeeperFeatureFlag::REMOVE_RECURSIVE))
+    const auto fallback_method = [&]
     {
         tryRemoveChildrenRecursive(path);
         return tryRemove(path);
-    }
+    };
+
+    if (!isFeatureEnabled(DB::KeeperFeatureFlag::REMOVE_RECURSIVE))
+        return fallback_method();
 
     auto promise = std::make_shared<std::promise<Coordination::RemoveRecursiveResponse>>();
     auto future = promise->get_future();
@@ -998,6 +1004,10 @@ Coordination::Error ZooKeeper::tryRemoveRecursive(const std::string & path, uint
     }
 
     auto response = future.get();
+
+    if (response.error == Coordination::Error::ZNOTEMPTY) /// limit was too low, try without RemoveRecursive request
+        return fallback_method();
+
     return response.error;
 }
 

src/Common/ZooKeeper/ZooKeeper.h

@@ -486,13 +486,13 @@ public:
     /// Remove the node with the subtree.
     /// If Keeper supports RemoveRecursive operation then it will be performed atomically.
     /// Otherwise if someone concurrently adds or removes a node in the subtree, the result is undefined.
-    void removeRecursive(const std::string & path, uint32_t remove_nodes_limit = 100);
+    void removeRecursive(const std::string & path, uint32_t remove_nodes_limit = 1000);
 
     /// Same as removeRecursive but in case if Keeper does not supports RemoveRecursive and
     /// if someone concurrently removes a node in the subtree, this will not cause errors.
     /// For instance, you can call this method twice concurrently for the same node and the end
     /// result would be the same as for the single call.
-    Coordination::Error tryRemoveRecursive(const std::string & path, uint32_t remove_nodes_limit = 100);
+    Coordination::Error tryRemoveRecursive(const std::string & path, uint32_t remove_nodes_limit = 1000);
 
     /// Similar to removeRecursive(...) and tryRemoveRecursive(...), but does not remove path itself.
     /// Node defined as RemoveException will not be deleted.

src/Common/ZooKeeper/ZooKeeperCommon.cpp

@@ -767,6 +767,11 @@ size_t ZooKeeperMultiRequest::sizeImpl() const
 }
 
 void ZooKeeperMultiRequest::readImpl(ReadBuffer & in)
+{
+    readImpl(in, /*request_validator=*/{});
+}
+
+void ZooKeeperMultiRequest::readImpl(ReadBuffer & in, RequestValidator request_validator)
 {
     while (true)
     {
@@ -788,6 +793,8 @@ void ZooKeeperMultiRequest::readImpl(ReadBuffer & in)
 
         ZooKeeperRequestPtr request = ZooKeeperRequestFactory::instance().get(op_num);
         request->readImpl(in);
+        if (request_validator)
+            request_validator(*request);
         requests.push_back(request);
 
         if (in.eof())

src/Common/ZooKeeper/ZooKeeperCommon.h

@@ -570,6 +570,9 @@ struct ZooKeeperMultiRequest final : MultiRequest<ZooKeeperRequestPtr>, ZooKeepe
     void writeImpl(WriteBuffer & out) const override;
     size_t sizeImpl() const override;
     void readImpl(ReadBuffer & in) override;
+
+    using RequestValidator = std::function<void(const ZooKeeperRequest &)>;
+    void readImpl(ReadBuffer & in, RequestValidator request_validator);
     std::string toStringImpl(bool short_format) const override;
 
     ZooKeeperResponsePtr makeResponse() const override;

src/Coordination/KeeperContext.cpp

@@ -514,7 +514,13 @@ void KeeperContext::initializeFeatureFlags(const Poco::Util::AbstractConfigurati
                 feature_flags.disableFeatureFlag(feature_flag.value());
         }
 
+        if (feature_flags.isEnabled(KeeperFeatureFlag::MULTI_READ))
+            feature_flags.enableFeatureFlag(KeeperFeatureFlag::FILTERED_LIST);
+        else
+            system_nodes_with_data[keeper_api_version_path] = toString(static_cast<uint8_t>(KeeperApiVersion::ZOOKEEPER_COMPATIBLE));
+
         system_nodes_with_data[keeper_api_feature_flags_path] = feature_flags.getFeatureFlags();
+
     }
 
     feature_flags.logFlags(getLogger("KeeperContext"));
@@ -569,6 +575,25 @@ const CoordinationSettings & KeeperContext::getCoordinationSettings() const
     return *coordination_settings;
 }
 
+bool KeeperContext::isOperationSupported(Coordination::OpNum operation) const
+{
+    switch (operation)
+    {
+        case Coordination::OpNum::FilteredList:
+            return feature_flags.isEnabled(KeeperFeatureFlag::FILTERED_LIST);
+        case Coordination::OpNum::MultiRead:
+            return feature_flags.isEnabled(KeeperFeatureFlag::MULTI_READ);
+        case Coordination::OpNum::CreateIfNotExists:
+            return feature_flags.isEnabled(KeeperFeatureFlag::CREATE_IF_NOT_EXISTS);
+        case Coordination::OpNum::CheckNotExists:
+            return feature_flags.isEnabled(KeeperFeatureFlag::CHECK_NOT_EXISTS);
+        case Coordination::OpNum::RemoveRecursive:
+            return feature_flags.isEnabled(KeeperFeatureFlag::REMOVE_RECURSIVE);
+        default:
+            return true;
+    }
+}
+
 uint64_t KeeperContext::lastCommittedIndex() const
 {
     return last_committed_log_idx.load(std::memory_order_relaxed);

src/Coordination/KeeperContext.h

@@ -1,6 +1,7 @@
 #pragma once
 #include <Coordination/KeeperFeatureFlags.h>
 #include <Poco/Util/AbstractConfiguration.h>
+#include <Common/ZooKeeper/ZooKeeperConstants.h>
 #include <atomic>
 #include <condition_variable>
 #include <cstdint>
@@ -103,6 +104,7 @@ public:
         return precommit_sleep_probability_for_testing;
     }
 
+    bool isOperationSupported(Coordination::OpNum operation) const;
 private:
     /// local disk defined using path or disk name
     using Storage = std::variant<DiskPtr, std::string>;

src/Server/KeeperTCPHandler.cpp

@@ -1,5 +1,4 @@
 #include <Server/KeeperTCPHandler.h>
-#include "Common/ZooKeeper/ZooKeeperConstants.h"
 
 #if USE_NURAFT
 
@@ -19,6 +18,8 @@
 #    include <Common/NetException.h>
 #    include <Common/PipeFDs.h>
 #    include <Common/Stopwatch.h>
+#    include <Common/ZooKeeper/ZooKeeperCommon.h>
+#    include <Common/ZooKeeper/ZooKeeperConstants.h>
 #    include <Common/ZooKeeper/ZooKeeperIO.h>
 #    include <Common/logger_useful.h>
 #    include <Common/setThreadName.h>
@@ -63,6 +64,7 @@ namespace ErrorCodes
     extern const int LOGICAL_ERROR;
     extern const int UNEXPECTED_PACKET_FROM_CLIENT;
     extern const int TIMEOUT_EXCEEDED;
+    extern const int BAD_ARGUMENTS;
 }
 
 struct PollResult
@@ -637,7 +639,23 @@ std::pair<Coordination::OpNum, Coordination::XID> KeeperTCPHandler::receiveReque
 
     Coordination::ZooKeeperRequestPtr request = Coordination::ZooKeeperRequestFactory::instance().get(opnum);
     request->xid = xid;
-    request->readImpl(read_buffer);
+
+    auto request_validator = [&](const Coordination::ZooKeeperRequest & current_request)
+    {
+        if (!keeper_dispatcher->getKeeperContext()->isOperationSupported(current_request.getOpNum()))
+            throw Exception(ErrorCodes::BAD_ARGUMENTS, "Unsupported operation: {}", current_request.getOpNum());
+    };
+
+    if (auto * multi_request = dynamic_cast<Coordination::ZooKeeperMultiRequest *>(request.get()))
+    {
+        multi_request->readImpl(read_buffer, request_validator);
+    }
+    else
+    {
+        request->readImpl(read_buffer);
+        request_validator(*request);
+    }
+
 
     if (!keeper_dispatcher->putRequest(request, session_id, use_xid_64))
         throw Exception(ErrorCodes::TIMEOUT_EXCEEDED, "Session {} already disconnected", session_id);

src/Storages/ObjectStorage/DataLakes/DataLakeConfiguration.h

@@ -56,6 +56,14 @@ public:
         return std::nullopt;
     }
 
+    void implementPartitionPruning(const ActionsDAG & filter_dag) override
+    {
+        if (!current_metadata || !current_metadata->supportsPartitionPruning())
+            return;
+        BaseStorageConfiguration::setPaths(current_metadata->makePartitionPruning(filter_dag));
+    }
+
+
 private:
     DataLakeMetadataPtr current_metadata;
 

src/Storages/ObjectStorage/DataLakes/IDataLakeMetadata.h

@@ -2,6 +2,7 @@
 #include <boost/noncopyable.hpp>
 #include <Core/Types.h>
 #include <Core/NamesAndTypes.h>
+#include "Interpreters/ActionsDAG.h"
 #include "PartitionColumns.h"
 
 namespace DB
@@ -16,6 +17,8 @@ public:
     virtual bool operator==(const IDataLakeMetadata & other) const = 0;
     virtual const DataLakePartitionColumns & getPartitionColumns() const = 0;
     virtual const std::unordered_map<String, String> & getColumnNameToPhysicalNameMapping() const = 0;
+    virtual bool supportsPartitionPruning() { return false; }
+    virtual Strings makePartitionPruning(const ActionsDAG &) { return {}; }
 };
 using DataLakeMetadataPtr = std::unique_ptr<IDataLakeMetadata>;
 

src/Storages/ObjectStorage/DataLakes/Iceberg/IcebergMetadata.cpp

@@ -75,21 +75,6 @@ IcebergMetadata::IcebergMetadata(
 namespace
 {
 
-enum class ManifestEntryStatus : uint8_t
-{
-    EXISTING = 0,
-    ADDED = 1,
-    DELETED = 2,
-};
-
-enum class DataFileContent : uint8_t
-{
-    DATA = 0,
-    POSITION_DELETES = 1,
-    EQUALITY_DELETES = 2,
-};
-
-
 /**
  * Iceberg supports the next data types (see https://iceberg.apache.org/spec/#schemas-and-data-types):
  * - Primitive types:
@@ -309,7 +294,7 @@ parseTableSchema(const Poco::JSON::Object::Ptr & metadata_object, int format_ver
         /// Field "schemas" is optional for version 1, but after version 2 was introduced,
         /// in most cases this field is added for new tables in version 1 as well.
         if (!ignore_schema_evolution && metadata_object->has("schemas")
-            && metadata_object->get("schemas").extract<Poco::JSON::Array::Ptr>()->size() > 1§)
+            && metadata_object->get("schemas").extract<Poco::JSON::Array::Ptr>()->size() > 1)
             throw Exception(
                 ErrorCodes::UNSUPPORTED_METHOD,
                 "Cannot read Iceberg table: the table schema has been changed at least 1 time, reading tables with evolved schema is not "
@@ -645,8 +630,27 @@ Strings IcebergMetadata::getDataFiles() const
                 data_files.push_back(file_path);
             }
         }
+        ColumnPtr big_partition_column = data_file_tuple_column->getColumnPtr(data_file_tuple_type.getPositionByName("partition"));
+        if (big_partition_column->getDataType() != TypeIndex::Tuple)
+        {
+            throw Exception(
+                ErrorCodes::ILLEGAL_COLUMN,
+                "The parsed column from Avro file of `file_path` field should be Tuple type, got {}",
+                columns.at(1)->getFamilyName());
+        }
+        const auto * big_partition_tuple = assert_cast<const ColumnTuple *>(big_partition_column.get());
+        std::vector<uint8_t> partition_spec_json_bytes = avro_metadata["partition-spec"];
+        String partition_spec_json_string
+            = String(reinterpret_cast<char *>(partition_spec_json_bytes.data()), partition_spec_json_bytes.size());
+        Poco::Dynamic::Var partition_spec_json = parser.parse(partition_spec_json_string);
+        const Poco::JSON::Array::Ptr & partition_spec = partition_spec_json.extract<Poco::JSON::Array::Ptr>();
+        common_partition_infos.push_back(pruning_processor.getCommonPartitionInfo(partition_spec, big_partition_tuple));
+        specific_partition_infos.push_back(
+            pruning_processor.getSpecificPartitionInfo(common_partition_infos.back(), current_schema_id, name_and_type_by_source_id));
     }
-    return data_files;
+
+    return pruning_processor.getDataFiles(
+        common_partition_infos, specific_partition_infos, nullptr, getContext(), configuration_ptr->getPath());
 }
 
 }

src/Storages/ObjectStorage/DataLakes/Iceberg/IcebergMetadata.h

@@ -1,5 +1,6 @@
 #pragma once
 
+#include "Columns/IColumn.h"
 #include "Core/NamesAndTypes.h"
 #include "DataTypes/DataTypeNullable.h"
 #include "config.h"
@@ -15,6 +16,7 @@
 namespace DB
 {
 
+
 /**
  * Useful links:
  * - https://iceberg.apache.org/spec/
@@ -72,11 +74,28 @@ enum class PartitionTransform
     Unsupported
 };
 
+enum class ManifestEntryStatus : uint8_t
+{
+    EXISTING = 0,
+    ADDED = 1,
+    DELETED = 2,
+};
+
+enum class DataFileContent : uint8_t
+{
+    DATA = 0,
+    POSITION_DELETES = 1,
+    EQUALITY_DELETES = 2,
+};
+
+
 struct CommonPartitionInfo
 {
     std::vector<ColumnPtr> partition_columns;
     std::vector<PartitionTransform> partition_transforms;
     std::vector<Int32> partition_source_ids;
+    ColumnPtr file_path_column;
+    ColumnPtr status_column;
 };
 
 struct SpecificSchemaPartitionInfo
@@ -88,16 +107,23 @@ struct SpecificSchemaPartitionInfo
 class PartitionPruningProcessor
 {
 public:
-    CommonPartitionInfo addCommonPartitionInfo(Poco::JSON::Array::Ptr partition_specification, const ColumnTuple * big_partition_tuple);
+    CommonPartitionInfo
+    getCommonPartitionInfo(const Poco::JSON::Array::Ptr & partition_specification, const ColumnTuple * big_partition_tuple) const;
 
-    SpecificSchemaPartitionInfo getSpecificPartitionPruning(
+    SpecificSchemaPartitionInfo getSpecificPartitionInfo(
         const CommonPartitionInfo & common_info,
         Int32 schema_version,
-        const std::unordered_map<Int32, NameAndTypePair> & name_and_type_by_source_id);
+        const std::unordered_map<Int32, NameAndTypePair> & name_and_type_by_source_id) const;
 
-    std::vector<bool> getPruningMask(const SpecificSchemaPartitionInfo & specific_info, const ActionsDAG * filter_dag, ContextPtr context);
+    std::vector<bool>
+    getPruningMask(const SpecificSchemaPartitionInfo & specific_info, const ActionsDAG * filter_dag, ContextPtr context) const;
 
-    std::vector<bool> getAllFilesMask(const ActionsDAG * filter_dag, ContextPtr context);
+    Strings getDataFiles(
+        const std::vector<CommonPartitionInfo> & manifest_partitions_infos,
+        const std::vector<SpecificSchemaPartitionInfo> & specific_infos,
+        const ActionsDAG * filter_dag,
+        ContextPtr context,
+        const std::string & common_path) const;
 
 private:
     static PartitionTransform getTransform(const String & transform_name)
@@ -147,7 +173,7 @@ private:
     static Int16 getDay(Int32 value, PartitionTransform transform)
     {
         DateLUTImpl::Time got_time = getTime(value, transform);
-        LOG_DEBUG(&Poco::Logger::get("Get field"), "Time: {}", got_time);
+        // LOG_DEBUG(&Poco::Logger::get("Get field"), "Time: {}", got_time);
         return DateLUT::instance().toDayNum(got_time);
     }
 
@@ -188,11 +214,6 @@ private:
             throw Exception(ErrorCodes::BAD_ARGUMENTS, "Unsupported partition transform: {}", partition_transform);
         }
     }
-
-    std::unordered_map<String, CommonPartitionInfo> common_partition_info_by_manifest_file;
-    std::map<std::pair<String, Int32>, SpecificSchemaPartitionInfo> specific_partition_info_by_manifest_file_and_schema;
-
-    std::vector<CommonPartitionInfo> common_partition_infos;
 };
 
 
@@ -241,7 +262,18 @@ public:
 
     static DataLakeMetadataPtr create(ObjectStoragePtr object_storage, ConfigurationObserverPtr configuration, ContextPtr local_context);
 
-    Strings makePartitionPruning(const ActionsDAG & filter_dag);
+    Strings makePartitionPruning(const ActionsDAG & filter_dag) override
+    {
+        auto configuration_ptr = configuration.lock();
+        if (!configuration_ptr)
+        {
+            throw Exception(ErrorCodes::LOGICAL_ERROR, "Configuration is expired");
+        }
+        return pruning_processor.getDataFiles(
+            common_partition_infos, specific_partition_infos, &filter_dag, getContext(), configuration_ptr->getPath());
+    }
+
+    bool supportsPartitionPruning() override { return true; }
 
 private:
     size_t getVersion() const { return metadata_version; }
@@ -262,9 +294,12 @@ private:
     std::vector<std::pair<String, Int32>> manifest_files_with_start_index;
 
     mutable Strings data_files;
+    std::vector<CommonPartitionInfo> partition_infos;
     mutable Strings manifest_files;
 
     PartitionPruningProcessor pruning_processor;
+    mutable std::vector<CommonPartitionInfo> common_partition_infos;
+    mutable std::vector<SpecificSchemaPartitionInfo> specific_partition_infos;
 };
 
 }

src/Storages/ObjectStorage/DataLakes/Iceberg/PartitionPruning.cpp

@@ -35,18 +35,18 @@
 #    include <Poco/JSON/Object.h>
 #    include <Poco/JSON/Parser.h>
 
-#    include <DataFile.hh>
-
-// #    include <filesystem>
-
 #    include <Storages/ObjectStorage/DataLakes/Iceberg/IcebergMetadata.h>
+#    include <DataFile.hh>
 namespace DB
 {
 
-CommonPartitionInfo
+CommonPartitionInfo PartitionPruningProcessor::getCommonPartitionInfo(
-PartitionPruningProcessor::getCommonPartitionInfo(Poco::JSON::Array::Ptr partition_specification, const ColumnTuple * big_partition_tuple)
+    const Poco::JSON::Array::Ptr & partition_specification, const ColumnTuple * data_file_tuple_column) const
 {
     CommonPartitionInfo common_info;
+    ColumnPtr big_partition_column = data_file_tuple_column->getColumnPtr(data_file_tuple_type.getPositionByName("partition"));
+
+    common_info.file_path_column = big_partition_column->getColumnPtr(0);
     for (size_t i = 0; i != partition_specification->size(); ++i)
     {
         auto current_field = partition_specification->getObject(static_cast<UInt32>(i));
@@ -61,17 +61,17 @@ PartitionPruningProcessor::getCommonPartitionInfo(Poco::JSON::Array::Ptr partiti
         auto partition_name = current_field->getValue<String>("name");
         LOG_DEBUG(&Poco::Logger::get("Partition Spec"), "Name: {}", partition_name);
 
-        common_info.partition_columns.push_back(big_partition_tuple->getColumnPtr(i));
+        common_info.partition_columns.push_back(big_partition_column->getColumnPtr(i));
         common_info.partition_transforms.push_back(transform);
         common_info.partition_source_ids.push_back(source_id);
     }
     return common_info;
 }
 
-SpecificSchemaPartitionInfo PartitionPruningProcessor::getSpecificPartitionPruning(
+SpecificSchemaPartitionInfo PartitionPruningProcessor::getSpecificPartitionInfo(
     const CommonPartitionInfo & common_info,
     [[maybe_unused]] Int32 schema_version,
-    const std::unordered_map<Int32, NameAndTypePair> & name_and_type_by_source_id)
+    const std::unordered_map<Int32, NameAndTypePair> & name_and_type_by_source_id) const
 {
     SpecificSchemaPartitionInfo specific_info;
 
@@ -104,7 +104,7 @@ SpecificSchemaPartitionInfo PartitionPruningProcessor::getSpecificPartitionPruni
 }
 
 std::vector<bool> PartitionPruningProcessor::getPruningMask(
-    const SpecificSchemaPartitionInfo & specific_info, const ActionsDAG * filter_dag, ContextPtr context)
+    const SpecificSchemaPartitionInfo & specific_info, const ActionsDAG * filter_dag, ContextPtr context) const
 {
     std::vector<bool> pruning_mask;
     if (!specific_info.partition_names_and_types.empty())
@@ -131,6 +131,47 @@ std::vector<bool> PartitionPruningProcessor::getPruningMask(
     return pruning_mask;
 }
 
+Strings PartitionPruningProcessor::getDataFiles(
+    const std::vector<CommonPartitionInfo> & manifest_partitions_infos,
+    const std::vector<SpecificSchemaPartitionInfo> & specific_infos,
+    const ActionsDAG * filter_dag,
+    ContextPtr context,
+    const std::string & common_path) const
+{
+    Strings data_files;
+    for (size_t index = 0; index < manifest_partitions_infos.size(); ++index)
+    {
+        const auto & manifest_partition_info = manifest_partitions_infos[index];
+        const auto & specific_partition_info = specific_infos[index];
+        size_t number_of_files_in_manifest = manifest_partition_info.file_path_column->size();
+        LOG_DEBUG(&Poco::Logger::get("Partition pruning"), "Filter dag is null: {}", filter_dag == nullptr);
+        auto pruning_mask = filter_dag ? getPruningMask(specific_partition_info, filter_dag, context) : std::vector<bool>{};
+        for (size_t i = 0; i < number_of_files_in_manifest; ++i)
+        {
+            if (!filter_dag || pruning_mask[i])
+            {
+                const auto status = manifest_partition_info.status_column->getInt(i);
+                const auto data_path = std::string(manifest_partition_info.file_path_column->getDataAt(i).toView());
+                const auto pos = data_path.find(common_path);
+                if (pos == std::string::npos)
+                    throw Exception(ErrorCodes::BAD_ARGUMENTS, "Expected to find {} in data path: {}", common_path, data_path);
+
+                const auto file_path = data_path.substr(pos);
+
+                if (ManifestEntryStatus(status) == ManifestEntryStatus::DELETED)
+                {
+                    throw Exception(
+                        ErrorCodes::UNSUPPORTED_METHOD, "Cannot read Iceberg table: positional and equality deletes are not supported");
+                }
+                else
+                {
+                    data_files.push_back(file_path);
+                }
+            }
+        }
+    }
+    return data_files;
+}
 }
 
 #endif

src/Storages/ObjectStorage/StorageObjectStorage.cpp

@@ -195,7 +195,10 @@ public:
         SourceStepWithFilter::applyFilters(std::move(added_filter_nodes));
         const ActionsDAG::Node * predicate = nullptr;
         if (filter_actions_dag)
+        {
+            configuration->implementPartitionPruning(*filter_actions_dag);
             predicate = filter_actions_dag->getOutputs().at(0);
+        }
         createIterator(predicate);
     }
 

src/Storages/ObjectStorage/StorageObjectStorage.h

@@ -204,6 +204,8 @@ public:
 
     virtual bool isDataLakeConfiguration() const { return false; }
 
+    virtual void implementPartitionPruning(const ActionsDAG &) { }
+
     virtual ReadFromFormatInfo prepareReadingFromFormat(
         ObjectStoragePtr object_storage,
         const Strings & requested_columns,

src/Storages/ObjectStorage/StorageObjectStorageSource.cpp

@@ -92,6 +92,10 @@ StorageObjectStorageSource::~StorageObjectStorageSource()
 void StorageObjectStorageSource::setKeyCondition(const std::optional<ActionsDAG> & filter_actions_dag, ContextPtr context_)
 {
     setKeyConditionImpl(filter_actions_dag, context_, read_from_format_info.format_header);
+    if (filter_actions_dag.has_value())
+    {
+        configuration->implementPartitionPruning(filter_actions_dag.value());
+    }
 }
 
 std::string StorageObjectStorageSource::getUniqueStoragePathIdentifier(

tests/integration/helpers/cluster.py

@@ -7,6 +7,7 @@ import os.path as p
 import platform
 import pprint
 import pwd
+import random
 import re
 import shlex
 import shutil
@@ -1650,6 +1651,8 @@ class ClickHouseCluster:
         minio_certs_dir=None,
         minio_data_dir=None,
         use_keeper=True,
+        keeper_randomize_feature_flags=True,
+        keeper_required_feature_flags=[],
         main_config_name="config.xml",
         users_config_name="users.xml",
         copy_common_configs=True,
@@ -1682,6 +1685,8 @@ class ClickHouseCluster:
         if not env_variables:
             env_variables = {}
         self.use_keeper = use_keeper
+        self.keeper_randomize_feature_flags = keeper_randomize_feature_flags
+        self.keeper_required_feature_flags = keeper_required_feature_flags
 
         # Code coverage files will be placed in database directory
         # (affect only WITH_COVERAGE=1 build)
@@ -2828,15 +2833,51 @@ class ClickHouseCluster:
 
                 if self.use_keeper:  # TODO: remove hardcoded paths from here
                     for i in range(1, 4):
+                        current_keeper_config_dir = os.path.join(
+                            f"{self.keeper_instance_dir_prefix}{i}", "config"
+                        )
                         shutil.copy(
                             os.path.join(
                                 self.keeper_config_dir, f"keeper_config{i}.xml"
                             ),
-                            os.path.join(
+                            current_keeper_config_dir,
-                                self.keeper_instance_dir_prefix + f"{i}", "config"
-                            ),
                         )
 
+                        extra_configs_dir = os.path.join(
+                            current_keeper_config_dir, f"keeper_config{i}.d"
+                        )
+                        os.mkdir(extra_configs_dir)
+                        feature_flags_config = os.path.join(
+                            extra_configs_dir, "feature_flags.yaml"
+                        )
+
+                        indentation = 4 * " "
+
+                        def get_feature_flag_value(feature_flag):
+                            if not self.keeper_randomize_feature_flags:
+                                return 1
+
+                            if feature_flag in self.keeper_required_feature_flags:
+                                return 1
+
+                            return random.randint(0, 1)
+
+                        with open(feature_flags_config, "w") as ff_config:
+                            ff_config.write("keeper_server:\n")
+                            ff_config.write(f"{indentation}feature_flags:\n")
+                            indentation *= 2
+
+                            for feature_flag in [
+                                "filtered_list",
+                                "multi_read",
+                                "check_not_exists",
+                                "create_if_not_exists",
+                                "remove_recursive",
+                            ]:
+                                ff_config.write(
+                                    f"{indentation}{feature_flag}: {get_feature_flag_value(feature_flag)}\n"
+                                )
+
                 run_and_check(self.base_zookeeper_cmd + common_opts, env=self.env)
                 self.up_called = True
 

tests/integration/test_keeper_map_retries/test.py

@@ -13,6 +13,7 @@ node = cluster.add_instance(
     main_configs=["configs/enable_keeper_map.xml"],
     user_configs=["configs/keeper_retries.xml"],
     with_zookeeper=True,
+    keeper_required_feature_flags=["multi_read"],
     stay_alive=True,
 )
 

tests/integration/test_refreshable_mv/test.py

@@ -20,6 +20,7 @@ node1 = cluster.add_instance(
     main_configs=["configs/config.xml"],
     user_configs=["configs/users.xml"],
     with_zookeeper=True,
+    keeper_required_feature_flags=["multi_read", "create_if_not_exists"],
     macros={"shard": "shard1", "replica": "1"},
     stay_alive=True,
 )
@@ -28,6 +29,7 @@ node2 = cluster.add_instance(
     main_configs=["configs/config.xml"],
     user_configs=["configs/users.xml"],
     with_zookeeper=True,
+    keeper_required_feature_flags=["multi_read", "create_if_not_exists"],
     macros={"shard": "shard1", "replica": "2"},
 )
 nodes = [node1, node2]

tests/integration/test_storage_kafka/test.py

@@ -59,6 +59,9 @@ instance = cluster.add_instance(
     user_configs=["configs/users.xml"],
     with_kafka=True,
     with_zookeeper=True,  # For Replicated Table
+    keeper_required_feature_flags=[
+        "create_if_not_exists"
+    ],  # new Kafka doesn't work without this feature
     macros={
         "kafka_broker": "kafka1",
         "kafka_topic_old": KAFKA_TOPIC_OLD,

tests/integration/test_storage_s3_queue/test.py

@@ -99,6 +99,7 @@ def started_cluster():
             with_minio=True,
             with_azurite=True,
             with_zookeeper=True,
+            keeper_required_feature_flags=["create_if_not_exists"],
             main_configs=[
                 "configs/zookeeper.xml",
                 "configs/s3queue_log.xml",
@@ -110,6 +111,7 @@ def started_cluster():
             user_configs=["configs/users.xml"],
             with_minio=True,
             with_zookeeper=True,
+            keeper_required_feature_flags=["create_if_not_exists"],
             main_configs=[
                 "configs/s3queue_log.xml",
             ],
@@ -118,6 +120,7 @@ def started_cluster():
         cluster.add_instance(
             "old_instance",
             with_zookeeper=True,
+            keeper_required_feature_flags=["create_if_not_exists"],
             image="clickhouse/clickhouse-server",
             tag="23.12",
             stay_alive=True,
@@ -127,6 +130,7 @@ def started_cluster():
         cluster.add_instance(
             "node1",
             with_zookeeper=True,
+            keeper_required_feature_flags=["create_if_not_exists"],
             stay_alive=True,
             main_configs=[
                 "configs/zookeeper.xml",
@@ -137,6 +141,7 @@ def started_cluster():
         cluster.add_instance(
             "node2",
             with_zookeeper=True,
+            keeper_required_feature_flags=["create_if_not_exists"],
             stay_alive=True,
             main_configs=[
                 "configs/zookeeper.xml",
@@ -149,6 +154,7 @@ def started_cluster():
             user_configs=["configs/users.xml"],
             with_minio=True,
             with_zookeeper=True,
+            keeper_required_feature_flags=["create_if_not_exists"],
             main_configs=[
                 "configs/s3queue_log.xml",
                 "configs/merge_tree.xml",
@@ -158,6 +164,7 @@ def started_cluster():
         cluster.add_instance(
             "instance_24.5",
             with_zookeeper=True,
+            keeper_required_feature_flags=["create_if_not_exists"],
             image="clickhouse/clickhouse-server",
             tag="24.5",
             stay_alive=True,
@@ -170,6 +177,7 @@ def started_cluster():
         cluster.add_instance(
             "node_cloud_mode",
             with_zookeeper=True,
+            keeper_required_feature_flags=["create_if_not_exists"],
             stay_alive=True,
             main_configs=[
                 "configs/zookeeper.xml",

tests/queries/0_stateless/02374_analyzer_join_using.reference

@@ -96,13 +96,15 @@ SELECT 'First JOIN INNER second JOIN INNER';
 First JOIN INNER second JOIN INNER
 SELECT id AS using_id, toTypeName(using_id), t1.id AS t1_id, toTypeName(t1_id), t1.value AS t1_value, toTypeName(t1_value),
 t2.id AS t2_id, toTypeName(t2_id), t2.value AS t2_value, toTypeName(t2_value), t3.id AS t3_id, toTypeName(t3_id), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 INNER JOIN test_table_join_2 AS t2 USING (id) INNER JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 INNER JOIN test_table_join_2 AS t2 USING (id) INNER JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
 0	UInt64	0	UInt64	Join_1_Value_0	String	0	UInt64	Join_2_Value_0	String	0	UInt64	Join_3_Value_0	String
 1	UInt64	1	UInt64	Join_1_Value_1	String	1	UInt64	Join_2_Value_1	String	1	UInt64	Join_3_Value_1	String
 SELECT '--';
 --
 SELECT t1.value AS t1_value, toTypeName(t1_value), t2.value AS t2_value, toTypeName(t2_value), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 INNER JOIN test_table_join_2 AS t2 USING (id) INNER JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 INNER JOIN test_table_join_2 AS t2 USING (id) INNER JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
 Join_1_Value_0	String	Join_2_Value_0	String	Join_3_Value_0	String
 Join_1_Value_1	String	Join_2_Value_1	String	Join_3_Value_1	String
 SELECT '--';
@@ -115,13 +117,15 @@ SELECT 'First JOIN INNER second JOIN LEFT';
 First JOIN INNER second JOIN LEFT
 SELECT id AS using_id, toTypeName(using_id), t1.id AS t1_id, toTypeName(t1_id), t1.value AS t1_value, toTypeName(t1_value),
 t2.id AS t2_id, toTypeName(t2_id), t2.value AS t2_value, toTypeName(t2_value), t3.id AS t3_id, toTypeName(t3_id), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 INNER JOIN test_table_join_2 AS t2 USING (id) LEFT JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 INNER JOIN test_table_join_2 AS t2 USING (id) LEFT JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
 0	UInt64	0	UInt64	Join_1_Value_0	String	0	UInt64	Join_2_Value_0	String	0	UInt64	Join_3_Value_0	String
 1	UInt64	1	UInt64	Join_1_Value_1	String	1	UInt64	Join_2_Value_1	String	1	UInt64	Join_3_Value_1	String
 SELECT '--';
 --
 SELECT t1.value AS t1_value, toTypeName(t1_value), t2.value AS t2_value, toTypeName(t2_value), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 INNER JOIN test_table_join_2 AS t2 USING (id) LEFT JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 INNER JOIN test_table_join_2 AS t2 USING (id) LEFT JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
 Join_1_Value_0	String	Join_2_Value_0	String	Join_3_Value_0	String
 Join_1_Value_1	String	Join_2_Value_1	String	Join_3_Value_1	String
 SELECT '--';
@@ -134,17 +138,19 @@ SELECT 'First JOIN INNER second JOIN RIGHT';
 First JOIN INNER second JOIN RIGHT
 SELECT id AS using_id, toTypeName(using_id), t1.id AS t1_id, toTypeName(t1_id), t1.value AS t1_value, toTypeName(t1_value),
 t2.id AS t2_id, toTypeName(t2_id), t2.value AS t2_value, toTypeName(t2_value), t3.id AS t3_id, toTypeName(t3_id), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 INNER JOIN test_table_join_2 AS t2 USING (id) RIGHT JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 INNER JOIN test_table_join_2 AS t2 USING (id) RIGHT JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
 0	UInt64	0	UInt64	Join_1_Value_0	String	0	UInt64	Join_2_Value_0	String	0	UInt64	Join_3_Value_0	String
 1	UInt64	1	UInt64	Join_1_Value_1	String	1	UInt64	Join_2_Value_1	String	1	UInt64	Join_3_Value_1	String
 4	UInt64	0	UInt64		String	0	UInt64		String	4	UInt64	Join_3_Value_4	String
 SELECT '--';
 --
 SELECT t1.value AS t1_value, toTypeName(t1_value), t2.value AS t2_value, toTypeName(t2_value), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 INNER JOIN test_table_join_2 AS t2 USING (id) RIGHT JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 INNER JOIN test_table_join_2 AS t2 USING (id) RIGHT JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
+	String		String	Join_3_Value_4	String
 Join_1_Value_0	String	Join_2_Value_0	String	Join_3_Value_0	String
 Join_1_Value_1	String	Join_2_Value_1	String	Join_3_Value_1	String
-	String		String	Join_3_Value_4	String
 SELECT '--';
 --
 SELECT 1 FROM test_table_join_1 AS t1 INNER JOIN test_table_join_2 AS t2 USING (id) RIGHT JOIN test_table_join_3 AS t3 USING(id);
@@ -156,17 +162,19 @@ SELECT 'First JOIN INNER second JOIN FULL';
 First JOIN INNER second JOIN FULL
 SELECT id AS using_id, toTypeName(using_id), t1.id AS t1_id, toTypeName(t1_id), t1.value AS t1_value, toTypeName(t1_value),
 t2.id AS t2_id, toTypeName(t2_id), t2.value AS t2_value, toTypeName(t2_value), t3.id AS t3_id, toTypeName(t3_id), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 INNER JOIN test_table_join_2 AS t2 USING (id) FULL JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 INNER JOIN test_table_join_2 AS t2 USING (id) FULL JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
+0	UInt64	0	UInt64		String	0	UInt64		String	4	UInt64	Join_3_Value_4	String
 0	UInt64	0	UInt64	Join_1_Value_0	String	0	UInt64	Join_2_Value_0	String	0	UInt64	Join_3_Value_0	String
 1	UInt64	1	UInt64	Join_1_Value_1	String	1	UInt64	Join_2_Value_1	String	1	UInt64	Join_3_Value_1	String
-0	UInt64	0	UInt64		String	0	UInt64		String	4	UInt64	Join_3_Value_4	String
 SELECT '--';
 --
 SELECT t1.value AS t1_value, toTypeName(t1_value), t2.value AS t2_value, toTypeName(t2_value), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 INNER JOIN test_table_join_2 AS t2 USING (id) FULL JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 INNER JOIN test_table_join_2 AS t2 USING (id) FULL JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
+	String		String	Join_3_Value_4	String
 Join_1_Value_0	String	Join_2_Value_0	String	Join_3_Value_0	String
 Join_1_Value_1	String	Join_2_Value_1	String	Join_3_Value_1	String
-	String		String	Join_3_Value_4	String
 SELECT '--';
 --
 SELECT 1 FROM test_table_join_1 AS t1 INNER JOIN test_table_join_2 AS t2 USING (id) FULL JOIN test_table_join_3 AS t3 USING(id);
@@ -178,13 +186,15 @@ SELECT 'First JOIN LEFT second JOIN INNER';
 First JOIN LEFT second JOIN INNER
 SELECT id AS using_id, toTypeName(using_id), t1.id AS t1_id, toTypeName(t1_id), t1.value AS t1_value, toTypeName(t1_value),
 t2.id AS t2_id, toTypeName(t2_id), t2.value AS t2_value, toTypeName(t2_value), t3.id AS t3_id, toTypeName(t3_id), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 LEFT JOIN test_table_join_2 AS t2 USING (id) INNER JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 LEFT JOIN test_table_join_2 AS t2 USING (id) INNER JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
 0	UInt64	0	UInt64	Join_1_Value_0	String	0	UInt64	Join_2_Value_0	String	0	UInt64	Join_3_Value_0	String
 1	UInt64	1	UInt64	Join_1_Value_1	String	1	UInt64	Join_2_Value_1	String	1	UInt64	Join_3_Value_1	String
 SELECT '--';
 --
 SELECT t1.value AS t1_value, toTypeName(t1_value), t2.value AS t2_value, toTypeName(t2_value), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 LEFT JOIN test_table_join_2 AS t2 USING (id) INNER JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 LEFT JOIN test_table_join_2 AS t2 USING (id) INNER JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
 Join_1_Value_0	String	Join_2_Value_0	String	Join_3_Value_0	String
 Join_1_Value_1	String	Join_2_Value_1	String	Join_3_Value_1	String
 SELECT '--';
@@ -197,14 +207,16 @@ SELECT 'First JOIN LEFT second JOIN LEFT';
 First JOIN LEFT second JOIN LEFT
 SELECT id AS using_id, toTypeName(using_id), t1.id AS t1_id, toTypeName(t1_id), t1.value AS t1_value, toTypeName(t1_value),
 t2.id AS t2_id, toTypeName(t2_id), t2.value AS t2_value, toTypeName(t2_value), t3.id AS t3_id, toTypeName(t3_id), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 LEFT JOIN test_table_join_2 AS t2 USING (id) LEFT JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 LEFT JOIN test_table_join_2 AS t2 USING (id) LEFT JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
 0	UInt64	0	UInt64	Join_1_Value_0	String	0	UInt64	Join_2_Value_0	String	0	UInt64	Join_3_Value_0	String
 1	UInt64	1	UInt64	Join_1_Value_1	String	1	UInt64	Join_2_Value_1	String	1	UInt64	Join_3_Value_1	String
 2	UInt64	2	UInt64	Join_1_Value_2	String	0	UInt64		String	0	UInt64		String
 SELECT '--';
 --
 SELECT t1.value AS t1_value, toTypeName(t1_value), t2.value AS t2_value, toTypeName(t2_value), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 LEFT JOIN test_table_join_2 AS t2 USING (id) LEFT JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 LEFT JOIN test_table_join_2 AS t2 USING (id) LEFT JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
 Join_1_Value_0	String	Join_2_Value_0	String	Join_3_Value_0	String
 Join_1_Value_1	String	Join_2_Value_1	String	Join_3_Value_1	String
 Join_1_Value_2	String		String		String
@@ -219,17 +231,19 @@ SELECT 'First JOIN LEFT second JOIN RIGHT';
 First JOIN LEFT second JOIN RIGHT
 SELECT id AS using_id, toTypeName(using_id), t1.id AS t1_id, toTypeName(t1_id), t1.value AS t1_value, toTypeName(t1_value),
 t2.id AS t2_id, toTypeName(t2_id), t2.value AS t2_value, toTypeName(t2_value), t3.id AS t3_id, toTypeName(t3_id), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 LEFT JOIN test_table_join_2 AS t2 USING (id) RIGHT JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 LEFT JOIN test_table_join_2 AS t2 USING (id) RIGHT JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
 0	UInt64	0	UInt64	Join_1_Value_0	String	0	UInt64	Join_2_Value_0	String	0	UInt64	Join_3_Value_0	String
 1	UInt64	1	UInt64	Join_1_Value_1	String	1	UInt64	Join_2_Value_1	String	1	UInt64	Join_3_Value_1	String
 4	UInt64	0	UInt64		String	0	UInt64		String	4	UInt64	Join_3_Value_4	String
 SELECT '--';
 --
 SELECT t1.value AS t1_value, toTypeName(t1_value), t2.value AS t2_value, toTypeName(t2_value), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 LEFT JOIN test_table_join_2 AS t2 USING (id) RIGHT JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 LEFT JOIN test_table_join_2 AS t2 USING (id) RIGHT JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
+	String		String	Join_3_Value_4	String
 Join_1_Value_0	String	Join_2_Value_0	String	Join_3_Value_0	String
 Join_1_Value_1	String	Join_2_Value_1	String	Join_3_Value_1	String
-	String		String	Join_3_Value_4	String
 SELECT '--';
 --
 SELECT 1 FROM test_table_join_1 AS t1 LEFT JOIN test_table_join_2 AS t2 USING (id) RIGHT JOIN test_table_join_3 AS t3 USING(id);
@@ -241,19 +255,21 @@ SELECT 'First JOIN LEFT second JOIN FULL';
 First JOIN LEFT second JOIN FULL
 SELECT id AS using_id, toTypeName(using_id), t1.id AS t1_id, toTypeName(t1_id), t1.value AS t1_value, toTypeName(t1_value),
 t2.id AS t2_id, toTypeName(t2_id), t2.value AS t2_value, toTypeName(t2_value), t3.id AS t3_id, toTypeName(t3_id), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 LEFT JOIN test_table_join_2 AS t2 USING (id) FULL JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 LEFT JOIN test_table_join_2 AS t2 USING (id) FULL JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
+0	UInt64	0	UInt64		String	0	UInt64		String	4	UInt64	Join_3_Value_4	String
 0	UInt64	0	UInt64	Join_1_Value_0	String	0	UInt64	Join_2_Value_0	String	0	UInt64	Join_3_Value_0	String
 1	UInt64	1	UInt64	Join_1_Value_1	String	1	UInt64	Join_2_Value_1	String	1	UInt64	Join_3_Value_1	String
 2	UInt64	2	UInt64	Join_1_Value_2	String	0	UInt64		String	0	UInt64		String
-0	UInt64	0	UInt64		String	0	UInt64		String	4	UInt64	Join_3_Value_4	String
 SELECT '--';
 --
 SELECT t1.value AS t1_value, toTypeName(t1_value), t2.value AS t2_value, toTypeName(t2_value), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 LEFT JOIN test_table_join_2 AS t2 USING (id) FULL JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 LEFT JOIN test_table_join_2 AS t2 USING (id) FULL JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
+	String		String	Join_3_Value_4	String
 Join_1_Value_0	String	Join_2_Value_0	String	Join_3_Value_0	String
 Join_1_Value_1	String	Join_2_Value_1	String	Join_3_Value_1	String
 Join_1_Value_2	String		String		String
-	String		String	Join_3_Value_4	String
 SELECT '--';
 --
 SELECT 1 FROM test_table_join_1 AS t1 LEFT JOIN test_table_join_2 AS t2 USING (id) FULL JOIN test_table_join_3 AS t3 USING(id);
@@ -266,13 +282,15 @@ SELECT 'First JOIN RIGHT second JOIN INNER';
 First JOIN RIGHT second JOIN INNER
 SELECT id AS using_id, toTypeName(using_id), t1.id AS t1_id, toTypeName(t1_id), t1.value AS t1_value, toTypeName(t1_value),
 t2.id AS t2_id, toTypeName(t2_id), t2.value AS t2_value, toTypeName(t2_value), t3.id AS t3_id, toTypeName(t3_id), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 RIGHT JOIN test_table_join_2 AS t2 USING (id) INNER JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 RIGHT JOIN test_table_join_2 AS t2 USING (id) INNER JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
 0	UInt64	0	UInt64	Join_1_Value_0	String	0	UInt64	Join_2_Value_0	String	0	UInt64	Join_3_Value_0	String
 1	UInt64	1	UInt64	Join_1_Value_1	String	1	UInt64	Join_2_Value_1	String	1	UInt64	Join_3_Value_1	String
 SELECT '--';
 --
 SELECT t1.value AS t1_value, toTypeName(t1_value), t2.value AS t2_value, toTypeName(t2_value), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 RIGHT JOIN test_table_join_2 AS t2 USING (id) INNER JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 RIGHT JOIN test_table_join_2 AS t2 USING (id) INNER JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
 Join_1_Value_0	String	Join_2_Value_0	String	Join_3_Value_0	String
 Join_1_Value_1	String	Join_2_Value_1	String	Join_3_Value_1	String
 SELECT '--';
@@ -285,17 +303,19 @@ SELECT 'First JOIN RIGHT second JOIN LEFT';
 First JOIN RIGHT second JOIN LEFT
 SELECT id AS using_id, toTypeName(using_id), t1.id AS t1_id, toTypeName(t1_id), t1.value AS t1_value, toTypeName(t1_value),
 t2.id AS t2_id, toTypeName(t2_id), t2.value AS t2_value, toTypeName(t2_value), t3.id AS t3_id, toTypeName(t3_id), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 RIGHT JOIN test_table_join_2 AS t2 USING (id) LEFT JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 RIGHT JOIN test_table_join_2 AS t2 USING (id) LEFT JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
 0	UInt64	0	UInt64	Join_1_Value_0	String	0	UInt64	Join_2_Value_0	String	0	UInt64	Join_3_Value_0	String
 1	UInt64	1	UInt64	Join_1_Value_1	String	1	UInt64	Join_2_Value_1	String	1	UInt64	Join_3_Value_1	String
 3	UInt64	0	UInt64		String	3	UInt64	Join_2_Value_3	String	0	UInt64		String
 SELECT '--';
 --
 SELECT t1.value AS t1_value, toTypeName(t1_value), t2.value AS t2_value, toTypeName(t2_value), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 RIGHT JOIN test_table_join_2 AS t2 USING (id) LEFT JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 RIGHT JOIN test_table_join_2 AS t2 USING (id) LEFT JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
+	String	Join_2_Value_3	String		String
 Join_1_Value_0	String	Join_2_Value_0	String	Join_3_Value_0	String
 Join_1_Value_1	String	Join_2_Value_1	String	Join_3_Value_1	String
-	String	Join_2_Value_3	String		String
 SELECT '--';
 --
 SELECT 1 FROM test_table_join_1 AS t1 RIGHT JOIN test_table_join_2 AS t2 USING (id) LEFT JOIN test_table_join_3 AS t3 USING(id);
@@ -307,17 +327,19 @@ SELECT 'First JOIN RIGHT second JOIN RIGHT';
 First JOIN RIGHT second JOIN RIGHT
 SELECT id AS using_id, toTypeName(using_id), t1.id AS t1_id, toTypeName(t1_id), t1.value AS t1_value, toTypeName(t1_value),
 t2.id AS t2_id, toTypeName(t2_id), t2.value AS t2_value, toTypeName(t2_value), t3.id AS t3_id, toTypeName(t3_id), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 RIGHT JOIN test_table_join_2 AS t2 USING (id) RIGHT JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 RIGHT JOIN test_table_join_2 AS t2 USING (id) RIGHT JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
 0	UInt64	0	UInt64	Join_1_Value_0	String	0	UInt64	Join_2_Value_0	String	0	UInt64	Join_3_Value_0	String
 1	UInt64	1	UInt64	Join_1_Value_1	String	1	UInt64	Join_2_Value_1	String	1	UInt64	Join_3_Value_1	String
 4	UInt64	0	UInt64		String	0	UInt64		String	4	UInt64	Join_3_Value_4	String
 SELECT '--';
 --
 SELECT t1.value AS t1_value, toTypeName(t1_value), t2.value AS t2_value, toTypeName(t2_value), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 RIGHT JOIN test_table_join_2 AS t2 USING (id) RIGHT JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 RIGHT JOIN test_table_join_2 AS t2 USING (id) RIGHT JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
+	String		String	Join_3_Value_4	String
 Join_1_Value_0	String	Join_2_Value_0	String	Join_3_Value_0	String
 Join_1_Value_1	String	Join_2_Value_1	String	Join_3_Value_1	String
-	String		String	Join_3_Value_4	String
 SELECT '--';
 --
 SELECT 1 FROM test_table_join_1 AS t1 RIGHT JOIN test_table_join_2 AS t2 USING (id) RIGHT JOIN test_table_join_3 AS t3 USING(id);
@@ -329,19 +351,21 @@ SELECT 'First JOIN RIGHT second JOIN FULL';
 First JOIN RIGHT second JOIN FULL
 SELECT id AS using_id, toTypeName(using_id), t1.id AS t1_id, toTypeName(t1_id), t1.value AS t1_value, toTypeName(t1_value),
 t2.id AS t2_id, toTypeName(t2_id), t2.value AS t2_value, toTypeName(t2_value), t3.id AS t3_id, toTypeName(t3_id), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 RIGHT JOIN test_table_join_2 AS t2 USING (id) FULL JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 RIGHT JOIN test_table_join_2 AS t2 USING (id) FULL JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
+0	UInt64	0	UInt64		String	0	UInt64		String	4	UInt64	Join_3_Value_4	String
 0	UInt64	0	UInt64	Join_1_Value_0	String	0	UInt64	Join_2_Value_0	String	0	UInt64	Join_3_Value_0	String
 1	UInt64	1	UInt64	Join_1_Value_1	String	1	UInt64	Join_2_Value_1	String	1	UInt64	Join_3_Value_1	String
 3	UInt64	0	UInt64		String	3	UInt64	Join_2_Value_3	String	0	UInt64		String
-0	UInt64	0	UInt64		String	0	UInt64		String	4	UInt64	Join_3_Value_4	String
 SELECT '--';
 --
 SELECT t1.value AS t1_value, toTypeName(t1_value), t2.value AS t2_value, toTypeName(t2_value), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 RIGHT JOIN test_table_join_2 AS t2 USING (id) FULL JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 RIGHT JOIN test_table_join_2 AS t2 USING (id) FULL JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
+	String		String	Join_3_Value_4	String
+	String	Join_2_Value_3	String		String
 Join_1_Value_0	String	Join_2_Value_0	String	Join_3_Value_0	String
 Join_1_Value_1	String	Join_2_Value_1	String	Join_3_Value_1	String
-	String	Join_2_Value_3	String		String
-	String		String	Join_3_Value_4	String
 SELECT '--';
 --
 SELECT 1 FROM test_table_join_1 AS t1 RIGHT JOIN test_table_join_2 AS t2 USING (id) FULL JOIN test_table_join_3 AS t3 USING(id);
@@ -354,14 +378,16 @@ SELECT 'First JOIN FULL second JOIN INNER';
 First JOIN FULL second JOIN INNER
 SELECT id AS using_id, toTypeName(using_id), t1.id AS t1_id, toTypeName(t1_id), t1.value AS t1_value, toTypeName(t1_value),
 t2.id AS t2_id, toTypeName(t2_id), t2.value AS t2_value, toTypeName(t2_value), t3.id AS t3_id, toTypeName(t3_id), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 FULL JOIN test_table_join_2 AS t2 USING (id) INNER JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 FULL JOIN test_table_join_2 AS t2 USING (id) INNER JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
+0	UInt64	0	UInt64		String	3	UInt64	Join_2_Value_3	String	0	UInt64	Join_3_Value_0	String
 0	UInt64	0	UInt64	Join_1_Value_0	String	0	UInt64	Join_2_Value_0	String	0	UInt64	Join_3_Value_0	String
 1	UInt64	1	UInt64	Join_1_Value_1	String	1	UInt64	Join_2_Value_1	String	1	UInt64	Join_3_Value_1	String
-0	UInt64	0	UInt64		String	3	UInt64	Join_2_Value_3	String	0	UInt64	Join_3_Value_0	String
 SELECT '--';
 --
 SELECT t1.value AS t1_value, toTypeName(t1_value), t2.value AS t2_value, toTypeName(t2_value), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 FULL JOIN test_table_join_2 AS t2 USING (id) INNER JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 FULL JOIN test_table_join_2 AS t2 USING (id) INNER JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
 Join_1_Value_0	String	Join_2_Value_0	String	Join_3_Value_0	String
 Join_1_Value_1	String	Join_2_Value_1	String	Join_3_Value_1	String
 SELECT '--';
@@ -374,19 +400,21 @@ SELECT 'First JOIN FULL second JOIN LEFT';
 First JOIN FULL second JOIN LEFT
 SELECT id AS using_id, toTypeName(using_id), t1.id AS t1_id, toTypeName(t1_id), t1.value AS t1_value, toTypeName(t1_value),
 t2.id AS t2_id, toTypeName(t2_id), t2.value AS t2_value, toTypeName(t2_value), t3.id AS t3_id, toTypeName(t3_id), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 FULL JOIN test_table_join_2 AS t2 USING (id) LEFT JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 FULL JOIN test_table_join_2 AS t2 USING (id) LEFT JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
+0	UInt64	0	UInt64		String	3	UInt64	Join_2_Value_3	String	0	UInt64	Join_3_Value_0	String
 0	UInt64	0	UInt64	Join_1_Value_0	String	0	UInt64	Join_2_Value_0	String	0	UInt64	Join_3_Value_0	String
 1	UInt64	1	UInt64	Join_1_Value_1	String	1	UInt64	Join_2_Value_1	String	1	UInt64	Join_3_Value_1	String
 2	UInt64	2	UInt64	Join_1_Value_2	String	0	UInt64		String	0	UInt64		String
-0	UInt64	0	UInt64		String	3	UInt64	Join_2_Value_3	String	0	UInt64	Join_3_Value_0	String
 SELECT '--';
 --
 SELECT t1.value AS t1_value, toTypeName(t1_value), t2.value AS t2_value, toTypeName(t2_value), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 FULL JOIN test_table_join_2 AS t2 USING (id) LEFT JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 FULL JOIN test_table_join_2 AS t2 USING (id) LEFT JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
+	String	Join_2_Value_3	String		String
 Join_1_Value_0	String	Join_2_Value_0	String	Join_3_Value_0	String
 Join_1_Value_1	String	Join_2_Value_1	String	Join_3_Value_1	String
 Join_1_Value_2	String		String		String
-	String	Join_2_Value_3	String		String
 SELECT '--';
 --
 SELECT 1 FROM test_table_join_1 AS t1 FULL JOIN test_table_join_2 AS t2 USING (id) LEFT JOIN test_table_join_3 AS t3 USING(id);
@@ -399,18 +427,20 @@ SELECT 'First JOIN FULL second JOIN RIGHT';
 First JOIN FULL second JOIN RIGHT
 SELECT id AS using_id, toTypeName(using_id), t1.id AS t1_id, toTypeName(t1_id), t1.value AS t1_value, toTypeName(t1_value),
 t2.id AS t2_id, toTypeName(t2_id), t2.value AS t2_value, toTypeName(t2_value), t3.id AS t3_id, toTypeName(t3_id), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 FULL JOIN test_table_join_2 AS t2 USING (id) RIGHT JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 FULL JOIN test_table_join_2 AS t2 USING (id) RIGHT JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
+0	UInt64	0	UInt64		String	3	UInt64	Join_2_Value_3	String	0	UInt64	Join_3_Value_0	String
 0	UInt64	0	UInt64	Join_1_Value_0	String	0	UInt64	Join_2_Value_0	String	0	UInt64	Join_3_Value_0	String
 1	UInt64	1	UInt64	Join_1_Value_1	String	1	UInt64	Join_2_Value_1	String	1	UInt64	Join_3_Value_1	String
-0	UInt64	0	UInt64		String	3	UInt64	Join_2_Value_3	String	0	UInt64	Join_3_Value_0	String
 4	UInt64	0	UInt64		String	0	UInt64		String	4	UInt64	Join_3_Value_4	String
 SELECT '--';
 --
 SELECT t1.value AS t1_value, toTypeName(t1_value), t2.value AS t2_value, toTypeName(t2_value), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 FULL JOIN test_table_join_2 AS t2 USING (id) RIGHT JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 FULL JOIN test_table_join_2 AS t2 USING (id) RIGHT JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
+	String		String	Join_3_Value_4	String
 Join_1_Value_0	String	Join_2_Value_0	String	Join_3_Value_0	String
 Join_1_Value_1	String	Join_2_Value_1	String	Join_3_Value_1	String
-	String		String	Join_3_Value_4	String
 SELECT '--';
 --
 SELECT 1 FROM test_table_join_1 AS t1 FULL JOIN test_table_join_2 AS t2 USING (id) RIGHT JOIN test_table_join_3 AS t3 USING(id);
@@ -422,21 +452,23 @@ SELECT 'First JOIN FULL second JOIN FULL';
 First JOIN FULL second JOIN FULL
 SELECT id AS using_id, toTypeName(using_id), t1.id AS t1_id, toTypeName(t1_id), t1.value AS t1_value, toTypeName(t1_value),
 t2.id AS t2_id, toTypeName(t2_id), t2.value AS t2_value, toTypeName(t2_value), t3.id AS t3_id, toTypeName(t3_id), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 FULL JOIN test_table_join_2 AS t2 USING (id) FULL JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 FULL JOIN test_table_join_2 AS t2 USING (id) FULL JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
+0	UInt64	0	UInt64		String	0	UInt64		String	4	UInt64	Join_3_Value_4	String
+0	UInt64	0	UInt64		String	3	UInt64	Join_2_Value_3	String	0	UInt64	Join_3_Value_0	String
 0	UInt64	0	UInt64	Join_1_Value_0	String	0	UInt64	Join_2_Value_0	String	0	UInt64	Join_3_Value_0	String
 1	UInt64	1	UInt64	Join_1_Value_1	String	1	UInt64	Join_2_Value_1	String	1	UInt64	Join_3_Value_1	String
 2	UInt64	2	UInt64	Join_1_Value_2	String	0	UInt64		String	0	UInt64		String
-0	UInt64	0	UInt64		String	3	UInt64	Join_2_Value_3	String	0	UInt64	Join_3_Value_0	String
-0	UInt64	0	UInt64		String	0	UInt64		String	4	UInt64	Join_3_Value_4	String
 SELECT '--';
 --
 SELECT t1.value AS t1_value, toTypeName(t1_value), t2.value AS t2_value, toTypeName(t2_value), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 FULL JOIN test_table_join_2 AS t2 USING (id) FULL JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 FULL JOIN test_table_join_2 AS t2 USING (id) FULL JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
+	String		String	Join_3_Value_4	String
+	String	Join_2_Value_3	String		String
 Join_1_Value_0	String	Join_2_Value_0	String	Join_3_Value_0	String
 Join_1_Value_1	String	Join_2_Value_1	String	Join_3_Value_1	String
 Join_1_Value_2	String		String		String
-	String	Join_2_Value_3	String		String
-	String		String	Join_3_Value_4	String
 SELECT '--';
 --
 SELECT 1 FROM test_table_join_1 AS t1 FULL JOIN test_table_join_2 AS t2 USING (id) FULL JOIN test_table_join_3 AS t3 USING(id);

tests/queries/0_stateless/02374_analyzer_join_using.sql.j2

@@ -64,12 +64,14 @@ SELECT 'First JOIN {{ first_join_type }} second JOIN {{ second_join_type }}';
 
 SELECT id AS using_id, toTypeName(using_id), t1.id AS t1_id, toTypeName(t1_id), t1.value AS t1_value, toTypeName(t1_value),
 t2.id AS t2_id, toTypeName(t2_id), t2.value AS t2_value, toTypeName(t2_value), t3.id AS t3_id, toTypeName(t3_id), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 {{ first_join_type }} JOIN test_table_join_2 AS t2 USING (id) {{ second_join_type }} JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 {{ first_join_type }} JOIN test_table_join_2 AS t2 USING (id) {{ second_join_type }} JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
 
 SELECT '--';
 
 SELECT t1.value AS t1_value, toTypeName(t1_value), t2.value AS t2_value, toTypeName(t2_value), t3.value AS t3_value, toTypeName(t3_value)
-FROM test_table_join_1 AS t1 {{ first_join_type }} JOIN test_table_join_2 AS t2 USING (id) {{ second_join_type }} JOIN test_table_join_3 AS t3 USING(id);
+FROM test_table_join_1 AS t1 {{ first_join_type }} JOIN test_table_join_2 AS t2 USING (id) {{ second_join_type }} JOIN test_table_join_3 AS t3 USING(id)
+ORDER BY ALL;
 
 SELECT '--';