Merge 5231d00616 into 44b4bd38b9

Enable cluster table functions for DataLake Storages

docs/en/sql-reference/table-functions/deltalake.md

@@ -49,4 +49,4 @@ LIMIT 2
 **See Also**
 
 - [DeltaLake engine](/docs/en/engines/table-engines/integrations/deltalake.md)
-
+- [DeltaLake cluster table function](/docs/en/sql-reference/table-functions/deltalakeCluster.md)

docs/en/sql-reference/table-functions/deltalakeCluster.md

@@ -0,0 +1,30 @@
+---
+slug: /en/sql-reference/table-functions/deltalakeCluster
+sidebar_position: 46
+sidebar_label: deltaLakeCluster
+title: "deltaLakeCluster Table Function"
+---
+This is an extension to the [deltaLake](/docs/en/sql-reference/table-functions/deltalake.md) table function.
+
+Allows processing files from [Delta Lake](https://github.com/delta-io/delta) tables in Amazon S3 in parallel from many nodes in a specified cluster. On initiator it creates a connection to all nodes in the cluster and dispatches each file dynamically. On the worker node it asks the initiator about the next task to process and processes it. This is repeated until all tasks are finished.
+
+**Syntax**
+
+``` sql
+deltaLakeCluster(cluster_name, url [,aws_access_key_id, aws_secret_access_key] [,format] [,structure] [,compression])
+```
+
+**Arguments**
+
+- `cluster_name` — Name of a cluster that is used to build a set of addresses and connection parameters to remote and local servers.
+
+- Description of all other arguments coincides with description of arguments in equivalent [deltaLake](/docs/en/sql-reference/table-functions/deltalake.md) table function.
+
+**Returned value**
+
+A table with the specified structure for reading data from cluster in the specified Delta Lake table in S3.
+
+**See Also**
+
+- [deltaLake engine](/docs/en/engines/table-engines/integrations/deltalake.md)
+- [deltaLake table function](/docs/en/sql-reference/table-functions/deltalake.md)

docs/en/sql-reference/table-functions/hudi.md

@@ -29,4 +29,4 @@ A table with the specified structure for reading data in the specified Hudi tabl
 **See Also**
 
 - [Hudi engine](/docs/en/engines/table-engines/integrations/hudi.md)
-
+- [Hudi cluster table function](/docs/en/sql-reference/table-functions/hudiCluster.md)

docs/en/sql-reference/table-functions/hudiCluster.md

@@ -0,0 +1,30 @@
+---
+slug: /en/sql-reference/table-functions/hudiCluster
+sidebar_position: 86
+sidebar_label: hudiCluster
+title: "hudiCluster Table Function"
+---
+This is an extension to the [hudi](/docs/en/sql-reference/table-functions/hudi.md) table function.
+
+Allows processing files from Apache [Hudi](https://hudi.apache.org/) tables in Amazon S3 in parallel from many nodes in a specified cluster. On initiator it creates a connection to all nodes in the cluster and dispatches each file dynamically. On the worker node it asks the initiator about the next task to process and processes it. This is repeated until all tasks are finished.
+
+**Syntax**
+
+``` sql
+hudiCluster(cluster_name, url [,aws_access_key_id, aws_secret_access_key] [,format] [,structure] [,compression])
+```
+
+**Arguments**
+
+- `cluster_name` — Name of a cluster that is used to build a set of addresses and connection parameters to remote and local servers.
+
+- Description of all other arguments coincides with description of arguments in equivalent [hudi](/docs/en/sql-reference/table-functions/hudi.md) table function.
+
+**Returned value**
+
+A table with the specified structure for reading data from cluster in the specified Hudi table in S3.
+
+**See Also**
+
+- [Hudi engine](/docs/en/engines/table-engines/integrations/hudi.md)
+- [Hudi table function](/docs/en/sql-reference/table-functions/hudi.md)

docs/en/sql-reference/table-functions/iceberg.md

@@ -72,3 +72,4 @@ Table function `iceberg` is an alias to `icebergS3` now.
 **See Also**
 
 - [Iceberg engine](/docs/en/engines/table-engines/integrations/iceberg.md)
+- [Iceberg cluster table function](/docs/en/sql-reference/table-functions/icebergCluster.md)

docs/en/sql-reference/table-functions/icebergCluster.md

@@ -0,0 +1,43 @@
+---
+slug: /en/sql-reference/table-functions/icebergCluster
+sidebar_position: 91
+sidebar_label: icebergCluster
+title: "icebergCluster Table Function"
+---
+This is an extension to the [iceberg](/docs/en/sql-reference/table-functions/iceberg.md) table function.
+
+Allows processing files from Apache [Iceberg](https://iceberg.apache.org/) in parallel from many nodes in a specified cluster. On initiator it creates a connection to all nodes in the cluster and dispatches each file dynamically. On the worker node it asks the initiator about the next task to process and processes it. This is repeated until all tasks are finished.
+
+**Syntax**
+
+``` sql
+icebergS3Cluster(cluster_name, url [, NOSIGN | access_key_id, secret_access_key, [session_token]] [,format] [,compression_method])
+icebergS3Cluster(cluster_name, named_collection[, option=value [,..]])
+
+icebergAzureCluster(cluster_name, connection_string|storage_account_url, container_name, blobpath, [,account_name], [,account_key] [,format] [,compression_method])
+icebergAzureCluster(cluster_name, named_collection[, option=value [,..]])
+
+icebergHDFSCluster(cluster_name, path_to_table, [,format] [,compression_method])
+icebergHDFSCluster(cluster_name, named_collection[, option=value [,..]])
+```
+
+**Arguments**
+
+- `cluster_name` — Name of a cluster that is used to build a set of addresses and connection parameters to remote and local servers.
+
+- Description of all other arguments coincides with description of arguments in equivalent [iceberg](/docs/en/sql-reference/table-functions/iceberg.md) table function.
+
+**Returned value**
+
+A table with the specified structure for reading data from cluster in the specified Iceberg table.
+
+**Examples**
+
+```sql
+SELECT * FROM icebergS3Cluster('cluster_simple', 'http://test.s3.amazonaws.com/clickhouse-bucket/test_table', 'test', 'test')
+```
+
+**See Also**
+
+- [Iceberg engine](/docs/en/engines/table-engines/integrations/iceberg.md)
+- [Iceberg table function](/docs/en/sql-reference/table-functions/iceberg.md)

src/Functions/array/arrayWithConstant.cpp

@@ -62,16 +62,17 @@ public:
         for (size_t i = 0; i < num_rows; ++i)
         {
             auto array_size = col_num->getInt(i);
+            auto element_size = col_value->byteSizeAt(i);
 
             if (unlikely(array_size < 0))
                 throw Exception(ErrorCodes::TOO_LARGE_ARRAY_SIZE, "Array size {} cannot be negative: while executing function {}", array_size, getName());
 
             Int64 estimated_size = 0;
-            if (unlikely(common::mulOverflow(array_size, col_value->byteSize(), estimated_size)))
-                throw Exception(ErrorCodes::TOO_LARGE_ARRAY_SIZE, "Array size {} with element size {} bytes is too large: while executing function {}", array_size, col_value->byteSize(), getName());
+            if (unlikely(common::mulOverflow(array_size, element_size, estimated_size)))
+                throw Exception(ErrorCodes::TOO_LARGE_ARRAY_SIZE, "Array size {} with element size {} bytes is too large: while executing function {}", array_size, element_size, getName());
 
             if (unlikely(estimated_size > max_array_size_in_columns_bytes))
-                throw Exception(ErrorCodes::TOO_LARGE_ARRAY_SIZE, "Array size {} with element size {} bytes is too large: while executing function {}", array_size, col_value->byteSize(), getName());
+                throw Exception(ErrorCodes::TOO_LARGE_ARRAY_SIZE, "Array size {} with element size {} bytes is too large: while executing function {}", array_size, element_size, getName());
 
             offset += array_size;
 

src/Storages/MergeTree/MergeTreeDataWriter.cpp

@@ -756,12 +756,11 @@ MergeTreeDataWriter::TemporaryPart MergeTreeDataWriter::writeProjectionPartImpl(
     TemporaryPart temp_part;
     const auto & metadata_snapshot = projection.metadata;
 
-    MergeTreeDataPartType part_type;
     /// Size of part would not be greater than block.bytes() + epsilon
     size_t expected_size = block.bytes();
     // just check if there is enough space on parent volume
     MergeTreeData::reserveSpace(expected_size, parent_part->getDataPartStorage());
-    part_type = data.choosePartFormatOnDisk(expected_size, block.rows()).part_type;
+    MergeTreeDataPartType part_type = data.choosePartFormatOnDisk(expected_size, block.rows()).part_type;
 
     auto new_data_part = parent_part->getProjectionPartBuilder(part_name, is_temp).withPartType(part_type).build();
     auto projection_part_storage = new_data_part->getDataPartStoragePtr();

src/Storages/MergeTree/MutateTask.cpp

@@ -1008,7 +1008,6 @@ void finalizeMutatedPart(
 
     new_data_part->default_codec = codec;
 }
-
 }
 
 struct MutationContext
@@ -1023,7 +1022,9 @@ struct MutationContext
 
     FutureMergedMutatedPartPtr future_part;
     MergeTreeData::DataPartPtr source_part;
+    MergeTreeData::DataPartPtr projection_part;
     StorageMetadataPtr metadata_snapshot;
+    StorageMetadataPtr projection_metadata_snapshot;
 
     MutationCommandsConstPtr commands;
     time_t time_of_mutation;
@@ -1040,6 +1041,11 @@ struct MutationContext
     ProgressCallback progress_callback;
     Block updated_header;
 
+    /// source here is projection part instead of table part
+    QueryPipelineBuilder projection_mutating_pipeline_builder;
+    QueryPipeline projection_mutating_pipeline;
+    std::unique_ptr<PullingPipelineExecutor> projection_mutating_executor;
+
     std::unique_ptr<MutationsInterpreter> interpreter;
     UInt64 watch_prev_elapsed = 0;
     std::unique_ptr<MergeStageProgress> stage_progress;
@@ -1056,8 +1062,14 @@ struct MutationContext
     MergeTreeData::MutableDataPartPtr new_data_part;
     IMergedBlockOutputStreamPtr out;
 
+    MergeTreeData::MutableDataPartPtr new_projection_part;
+    IMergedBlockOutputStreamPtr projection_out;
+
     String mrk_extension;
 
+    /// Used in lightweight delete to bit mask the projection if possible,
+    /// preference is higher than rebuild as more performant.
+    std::vector<ProjectionDescriptionRawPtr> projections_to_mask;
     std::vector<ProjectionDescriptionRawPtr> projections_to_build;
     IMergeTreeDataPart::MinMaxIndexPtr minmax_idx;
 
@@ -1151,6 +1163,14 @@ public:
                 if (iterateThroughAllProjections())
                     return true;
 
+                state = State::NEED_MASK_PROJECTION_PARTS;
+                return true;
+            }
+            case State::NEED_MASK_PROJECTION_PARTS:
+            {
+                if (iterateThroughAllProjectionsToMask())
+                    return true;
+
                 state = State::SUCCESS;
                 return true;
             }
@@ -1170,6 +1190,7 @@ private:
     void finalizeTempProjections();
     bool iterateThroughAllProjections();
     void constructTaskForProjectionPartsMerge();
+    bool iterateThroughAllProjectionsToMask();
     void finalize();
 
     enum class State : uint8_t
@@ -1177,7 +1198,7 @@ private:
         NEED_PREPARE,
         NEED_MUTATE_ORIGINAL_PART,
         NEED_MERGE_PROJECTION_PARTS,
-
+        NEED_MASK_PROJECTION_PARTS,
         SUCCESS
     };
 
@@ -1343,6 +1364,14 @@ bool PartMergerWriter::iterateThroughAllProjections()
     return true;
 }
 
+bool PartMergerWriter::iterateThroughAllProjectionsToMask()
+{
+    for (size_t i = 0, size = ctx->projections_to_mask.size(); i < size; ++i)
+    {
+    }
+    return false;
+}
+
 void PartMergerWriter::finalize()
 {
     if (ctx->count_lightweight_deleted_rows)
@@ -1660,6 +1689,30 @@ private:
         ctx->mutating_pipeline.disableProfileEventUpdate();
         ctx->mutating_executor = std::make_unique<PullingPipelineExecutor>(ctx->mutating_pipeline);
 
+        chassert (ctx->projection_mutating_pipeline_builder.initialized());
+        builder = std::make_unique<QueryPipelineBuilder>(std::move(ctx->projection_mutating_pipeline_builder));
+
+        const auto & projections_name_and_part = ctx->source_part->getProjectionParts();
+        auto part_name = fmt::format("{}", projections_name_and_part.begin()->first);
+        auto new_data_part = ctx->new_data_part->getProjectionPartBuilder(part_name, true).withPartType(ctx->projection_part->getType()).build();
+
+        ctx->projection_out = std::make_shared<MergedBlockOutputStream>(
+            new_data_part,
+            ctx->projection_metadata_snapshot,
+            ctx->projection_metadata_snapshot->getColumns().getAllPhysical(),
+            MergeTreeIndices{},
+            ColumnsStatistics{},
+            ctx->data->getContext()->chooseCompressionCodec(0, 0),
+            Tx::PrehistoricTID,
+            /*reset_columns=*/ false,
+            /*save_marks_in_cache=*/ false,
+            /*blocks_are_granules_size=*/ false,
+            ctx->data->getContext()->getWriteSettings());
+
+        ctx->projection_mutating_pipeline = QueryPipelineBuilder::getPipeline(std::move(*builder));
+        ctx->projection_mutating_pipeline.disableProfileEventUpdate();
+        ctx->projection_mutating_executor = std::make_unique<PullingPipelineExecutor>(ctx->projection_mutating_pipeline);
+
         part_merger_writer_task = std::make_unique<PartMergerWriter>(ctx);
     }
 
@@ -1895,6 +1948,26 @@ private:
 
             ctx->projections_to_build = std::vector<ProjectionDescriptionRawPtr>{ctx->projections_to_recalc.begin(), ctx->projections_to_recalc.end()};
 
+            chassert (ctx->projection_mutating_pipeline_builder.initialized());
+            builder = std::make_unique<QueryPipelineBuilder>(std::move(ctx->projection_mutating_pipeline_builder));
+
+            // ctx->projection_out = std::make_shared<MergedColumnOnlyOutputStream>(
+            //     ctx->new_projection_part,
+            //     ctx->metadata_snapshot,
+            //     ctx->updated_header.getNamesAndTypesList(),
+            //     ctx->compression_codec,
+            //     std::vector<MergeTreeIndexPtr>(),
+            //     std::vector<ColumnStatisticsPartPtr>(),
+            //     nullptr,
+            //     /*save_marks_in_cache=*/ false,
+            //     ctx->projection_part->index_granularity,
+            //     &ctx->projection_part->index_granularity_info
+            // );
+
+            ctx->projection_mutating_pipeline = QueryPipelineBuilder::getPipeline(std::move(*builder));
+            ctx->projection_mutating_pipeline.disableProfileEventUpdate();
+            ctx->projection_mutating_executor = std::make_unique<PullingPipelineExecutor>(ctx->projection_mutating_pipeline);
+
             part_merger_writer_task = std::make_unique<PartMergerWriter>(ctx);
         }
     }
@@ -2308,6 +2381,32 @@ bool MutateTask::prepare()
         ctx->updated_header = ctx->interpreter->getUpdatedHeader();
         ctx->progress_callback = MergeProgressCallback((*ctx->mutate_entry)->ptr(), ctx->watch_prev_elapsed, *ctx->stage_progress);
 
+        const auto & proj_desc = *(ctx->metadata_snapshot->getProjections().begin());
+        ctx->projection_metadata_snapshot = proj_desc.metadata;
+        const auto & projections_name_and_part = ctx->source_part->getProjectionParts();
+        ctx->projection_part = projections_name_and_part.begin()->second;
+
+        MutationCommands projection_commands;
+
+        MutationHelpers::splitAndModifyMutationCommands(
+            ctx->projection_part,
+            proj_desc.metadata,
+            alter_conversions,
+            ctx->commands_for_part,
+            projection_commands,
+            ctx->for_file_renames,
+            false,
+            ctx->log);
+
+        chassert(!projection_commands.empty());
+
+        auto projection_interpreter = std::make_unique<MutationsInterpreter>(
+            *ctx->data, ctx->projection_part, alter_conversions,
+            proj_desc.metadata, projection_commands,
+            proj_desc.metadata->getColumns().getNamesOfPhysical(), context_for_reading, settings);
+
+        ctx->projection_mutating_pipeline_builder = projection_interpreter->execute();
+
         lightweight_delete_mode = ctx->updated_header.has(RowExistsColumn::name);
         /// If under the condition of lightweight delete mode with rebuild option, add projections again here as we can only know
         /// the condition as early as from here.

src/Storages/ProjectionsDescription.cpp

@@ -334,6 +334,17 @@ Block ProjectionDescription::calculate(const Block & block, ContextPtr context)
     return ret;
 }
 
+void ProjectionDescription::mask(const Block & block[[maybe_unused]], ContextPtr context[[maybe_unused]]) const
+{
+    // auto mut_context = Context::createCopy(context);
+    // query_ast_copy = query_ast->clone();
+    // auto builder = InterpreterAlterQuery(query_ast_copy, mut_context,
+    //                    Pipe(std::make_shared<SourceFromSingleChunk>(block)))
+    //                    .buildQueryPipeline();
+    // auto pipeline = QueryPipelineBuilder::getPipeline(std::move(builder));
+    // PullingPipelineExecutor executor(pipeline);
+    // executor.pull(ret);
+}
 
 String ProjectionsDescription::toString() const
 {

src/Storages/ProjectionsDescription.h

@@ -93,6 +93,8 @@ struct ProjectionDescription
 
     Block calculate(const Block & block, ContextPtr context) const;
 
+    void mask(const Block & block, ContextPtr context) const;
+
     String getDirectoryName() const { return name + ".proj"; }
 };
 

src/TableFunctions/TableFunctionObjectStorage.cpp

@@ -226,6 +226,26 @@ template class TableFunctionObjectStorage<HDFSClusterDefinition, StorageHDFSConf
 #endif
 template class TableFunctionObjectStorage<LocalDefinition, StorageLocalConfiguration>;
 
+#if USE_AVRO && USE_AWS_S3
+template class TableFunctionObjectStorage<IcebergS3ClusterDefinition, StorageS3IcebergConfiguration>;
+#endif
+
+#if USE_AVRO && USE_AZURE_BLOB_STORAGE
+template class TableFunctionObjectStorage<IcebergAzureClusterDefinition, StorageAzureIcebergConfiguration>;
+#endif
+
+#if USE_AVRO && USE_HDFS
+template class TableFunctionObjectStorage<IcebergHDFSClusterDefinition, StorageHDFSIcebergConfiguration>;
+#endif
+
+#if USE_PARQUET && USE_AWS_S3
+template class TableFunctionObjectStorage<DeltaLakeClusterDefinition, StorageS3DeltaLakeConfiguration>;
+#endif
+
+#if USE_AWS_S3
+template class TableFunctionObjectStorage<HudiClusterDefinition, StorageS3HudiConfiguration>;
+#endif
+
 #if USE_AVRO
 void registerTableFunctionIceberg(TableFunctionFactory & factory)
 {

src/TableFunctions/TableFunctionObjectStorageCluster.cpp

@@ -96,7 +96,7 @@ void registerTableFunctionObjectStorageCluster(TableFunctionFactory & factory)
     {
         .documentation = {
             .description=R"(The table function can be used to read the data stored on HDFS in parallel for many nodes in a specified cluster.)",
-            .examples{{"HDFSCluster", "SELECT * FROM HDFSCluster(cluster_name, uri, format)", ""}}},
+            .examples{{"HDFSCluster", "SELECT * FROM HDFSCluster(cluster, uri, format)", ""}}},
             .allow_readonly = false
         }
     );
@@ -105,15 +105,77 @@ void registerTableFunctionObjectStorageCluster(TableFunctionFactory & factory)
     UNUSED(factory);
 }
 
+
+#if USE_AVRO
+void registerTableFunctionIcebergCluster(TableFunctionFactory & factory)
+{
+    UNUSED(factory);
+
 #if USE_AWS_S3
-template class TableFunctionObjectStorageCluster<S3ClusterDefinition, StorageS3Configuration>;
+    factory.registerFunction<TableFunctionIcebergS3Cluster>(
+        {.documentation
+         = {.description = R"(The table function can be used to read the Iceberg table stored on S3 object store in parallel for many nodes in a specified cluster.)",
+            .examples{{"icebergS3Cluster", "SELECT * FROM icebergS3Cluster(cluster, url, [, NOSIGN | access_key_id, secret_access_key, [session_token]], format, [,compression])", ""}},
+            .categories{"DataLake"}},
+         .allow_readonly = false});
 #endif
 
 #if USE_AZURE_BLOB_STORAGE
-template class TableFunctionObjectStorageCluster<AzureClusterDefinition, StorageAzureConfiguration>;
+    factory.registerFunction<TableFunctionIcebergAzureCluster>(
+        {.documentation
+         = {.description = R"(The table function can be used to read the Iceberg table stored on Azure object store in parallel for many nodes in a specified cluster.)",
+            .examples{{"icebergAzureCluster", "SELECT * FROM icebergAzureCluster(cluster, connection_string|storage_account_url, container_name, blobpath, [account_name, account_key, format, compression])", ""}},
+            .categories{"DataLake"}},
+         .allow_readonly = false});
 #endif
 
 #if USE_HDFS
-template class TableFunctionObjectStorageCluster<HDFSClusterDefinition, StorageHDFSConfiguration>;
+    factory.registerFunction<TableFunctionIcebergHDFSCluster>(
+        {.documentation
+         = {.description = R"(The table function can be used to read the Iceberg table stored on HDFS virtual filesystem in parallel for many nodes in a specified cluster.)",
+            .examples{{"icebergHDFSCluster", "SELECT * FROM icebergHDFSCluster(cluster, uri, [format], [structure], [compression_method])", ""}},
+            .categories{"DataLake"}},
+         .allow_readonly = false});
 #endif
 }
+#endif
+
+#if USE_AWS_S3
+#if USE_PARQUET
+void registerTableFunctionDeltaLakeCluster(TableFunctionFactory & factory)
+{
+    factory.registerFunction<TableFunctionDeltaLakeCluster>(
+        {.documentation
+         = {.description = R"(The table function can be used to read the DeltaLake table stored on object store in parallel for many nodes in a specified cluster.)",
+            .examples{{"deltaLakeCluster", "SELECT * FROM deltaLakeCluster(cluster, url, access_key_id, secret_access_key)", ""}},
+            .categories{"DataLake"}},
+         .allow_readonly = false});
+}
+#endif
+
+void registerTableFunctionHudiCluster(TableFunctionFactory & factory)
+{
+    factory.registerFunction<TableFunctionHudiCluster>(
+        {.documentation
+         = {.description = R"(The table function can be used to read the Hudi table stored on object store in parallel for many nodes in a specified cluster.)",
+            .examples{{"hudiCluster", "SELECT * FROM hudiCluster(cluster, url, access_key_id, secret_access_key)", ""}},
+            .categories{"DataLake"}},
+         .allow_readonly = false});
+}
+#endif
+
+void registerDataLakeClusterTableFunctions(TableFunctionFactory & factory)
+{
+    UNUSED(factory);
+#if USE_AVRO
+    registerTableFunctionIcebergCluster(factory);
+#endif
+#if USE_AWS_S3
+#if USE_PARQUET
+    registerTableFunctionDeltaLakeCluster(factory);
+#endif
+    registerTableFunctionHudiCluster(factory);
+#endif
+}
+
+}

src/TableFunctions/TableFunctionObjectStorageCluster.h

@@ -33,6 +33,36 @@ struct HDFSClusterDefinition
     static constexpr auto storage_type_name = "HDFSCluster";
 };
 
+struct IcebergS3ClusterDefinition
+{
+    static constexpr auto name = "icebergS3Cluster";
+    static constexpr auto storage_type_name = "IcebergS3Cluster";
+};
+
+struct IcebergAzureClusterDefinition
+{
+    static constexpr auto name = "icebergAzureCluster";
+    static constexpr auto storage_type_name = "IcebergAzureCluster";
+};
+
+struct IcebergHDFSClusterDefinition
+{
+    static constexpr auto name = "icebergHDFSCluster";
+    static constexpr auto storage_type_name = "IcebergHDFSCluster";
+};
+
+struct DeltaLakeClusterDefinition
+{
+    static constexpr auto name = "deltaLakeCluster";
+    static constexpr auto storage_type_name = "DeltaLakeS3Cluster";
+};
+
+struct HudiClusterDefinition
+{
+    static constexpr auto name = "hudiCluster";
+    static constexpr auto storage_type_name = "HudiS3Cluster";
+};
+
 /**
 * Class implementing s3/hdfs/azureBlobStorageCluster(...) table functions,
 * which allow to process many files from S3/HDFS/Azure blob storage on a specific cluster.
@@ -79,4 +109,25 @@ using TableFunctionAzureBlobCluster = TableFunctionObjectStorageCluster<AzureClu
 #if USE_HDFS
 using TableFunctionHDFSCluster = TableFunctionObjectStorageCluster<HDFSClusterDefinition, StorageHDFSConfiguration>;
 #endif
+
+#if USE_AVRO && USE_AWS_S3
+using TableFunctionIcebergS3Cluster = TableFunctionObjectStorageCluster<IcebergS3ClusterDefinition, StorageS3IcebergConfiguration>;
+#endif
+
+#if USE_AVRO && USE_AZURE_BLOB_STORAGE
+using TableFunctionIcebergAzureCluster = TableFunctionObjectStorageCluster<IcebergAzureClusterDefinition, StorageAzureIcebergConfiguration>;
+#endif
+
+#if USE_AVRO && USE_HDFS
+using TableFunctionIcebergHDFSCluster = TableFunctionObjectStorageCluster<IcebergHDFSClusterDefinition, StorageHDFSIcebergConfiguration>;
+#endif
+
+#if USE_AWS_S3 && USE_PARQUET
+using TableFunctionDeltaLakeCluster = TableFunctionObjectStorageCluster<DeltaLakeClusterDefinition, StorageS3DeltaLakeConfiguration>;
+#endif
+
+#if USE_AWS_S3
+using TableFunctionHudiCluster = TableFunctionObjectStorageCluster<HudiClusterDefinition, StorageS3HudiConfiguration>;
+#endif
+
 }

src/TableFunctions/registerTableFunctions.cpp

@@ -66,6 +66,7 @@ void registerTableFunctions(bool use_legacy_mongodb_integration [[maybe_unused]]
     registerTableFunctionObjectStorage(factory);
     registerTableFunctionObjectStorageCluster(factory);
     registerDataLakeTableFunctions(factory);
+    registerDataLakeClusterTableFunctions(factory);
 }
 
 }

src/TableFunctions/registerTableFunctions.h

@@ -70,6 +70,7 @@ void registerTableFunctionExplain(TableFunctionFactory & factory);
 void registerTableFunctionObjectStorage(TableFunctionFactory & factory);
 void registerTableFunctionObjectStorageCluster(TableFunctionFactory & factory);
 void registerDataLakeTableFunctions(TableFunctionFactory & factory);
+void registerDataLakeClusterTableFunctions(TableFunctionFactory & factory);
 
 void registerTableFunctionTimeSeries(TableFunctionFactory & factory);
 

tests/integration/test_storage_iceberg/configs/config.d/cluster.xml

@@ -0,0 +1,20 @@
+<clickhouse>
+    <remote_servers>
+        <cluster_simple>
+            <shard>
+                <replica>
+                    <host>node1</host>
+                    <port>9000</port>
+                </replica>
+                <replica>
+                    <host>node2</host>
+                    <port>9000</port>
+                </replica>
+                <replica>
+                    <host>node3</host>
+                    <port>9000</port>
+                </replica>
+            </shard>
+        </cluster_simple>
+    </remote_servers>
+</clickhouse>

tests/integration/test_storage_iceberg/configs/config.d/query_log.xml

@@ -0,0 +1,6 @@
+<clickhouse>
+    <query_log>
+      <database>system</database>
+      <table>query_log</table>
+    </query_log>
+</clickhouse>

tests/integration/test_storage_iceberg/test.py

@@ -73,14 +73,38 @@ def started_cluster():
         cluster.add_instance(
             "node1",
             main_configs=[
+                "configs/config.d/query_log.xml",
+                "configs/config.d/cluster.xml",
                 "configs/config.d/named_collections.xml",
                 "configs/config.d/filesystem_caches.xml",
             ],
             user_configs=["configs/users.d/users.xml"],
             with_minio=True,
             with_azurite=True,
-            stay_alive=True,
             with_hdfs=with_hdfs,
+            stay_alive=True,
+        )
+        cluster.add_instance(
+            "node2",
+            main_configs=[
+                "configs/config.d/query_log.xml",
+                "configs/config.d/cluster.xml",
+                "configs/config.d/named_collections.xml",
+                "configs/config.d/filesystem_caches.xml",
+            ],
+            user_configs=["configs/users.d/users.xml"],
+            stay_alive=True,
+        )
+        cluster.add_instance(
+            "node3",
+            main_configs=[
+                "configs/config.d/query_log.xml",
+                "configs/config.d/cluster.xml",
+                "configs/config.d/named_collections.xml",
+                "configs/config.d/filesystem_caches.xml",
+            ],
+            user_configs=["configs/users.d/users.xml"],
+            stay_alive=True,
         )
 
         logging.info("Starting cluster...")
@@ -182,6 +206,7 @@ def get_creation_expression(
     cluster,
     format="Parquet",
     table_function=False,
+    run_on_cluster=False,
     **kwargs,
 ):
     if storage_type == "s3":
@@ -189,35 +214,56 @@ def get_creation_expression(
             bucket = kwargs["bucket"]
         else:
             bucket = cluster.minio_bucket
-        print(bucket)
-        if table_function:
-            return f"icebergS3(s3, filename = 'iceberg_data/default/{table_name}/', format={format}, url = 'http://minio1:9001/{bucket}/')"
+
+        if run_on_cluster:
+            assert table_function
+            return f"icebergS3Cluster('cluster_simple', s3, filename = 'iceberg_data/default/{table_name}/', format={format}, url = 'http://minio1:9001/{bucket}/')"
         else:
-            return f"""
-                DROP TABLE IF EXISTS {table_name};
-                CREATE TABLE {table_name}
-                ENGINE=IcebergS3(s3, filename = 'iceberg_data/default/{table_name}/', format={format}, url = 'http://minio1:9001/{bucket}/')"""
+            if table_function:
+                return f"icebergS3(s3, filename = 'iceberg_data/default/{table_name}/', format={format}, url = 'http://minio1:9001/{bucket}/')"
+            else:
+                return f"""
+                    DROP TABLE IF EXISTS {table_name};
+                    CREATE TABLE {table_name}
+                    ENGINE=IcebergS3(s3, filename = 'iceberg_data/default/{table_name}/', format={format}, url = 'http://minio1:9001/{bucket}/')"""
+
     elif storage_type == "azure":
-        if table_function:
+        if run_on_cluster:
+            assert table_function
             return f"""
-                icebergAzure(azure, container = '{cluster.azure_container_name}', storage_account_url = '{cluster.env_variables["AZURITE_STORAGE_ACCOUNT_URL"]}', blob_path = '/iceberg_data/default/{table_name}/', format={format})
+                icebergAzureCluster('cluster_simple', azure, container = '{cluster.azure_container_name}', storage_account_url = '{cluster.env_variables["AZURITE_STORAGE_ACCOUNT_URL"]}', blob_path = '/iceberg_data/default/{table_name}/', format={format})
             """
         else:
-            return f"""
-                DROP TABLE IF EXISTS {table_name};
-                CREATE TABLE {table_name}
-                ENGINE=IcebergAzure(azure, container = {cluster.azure_container_name}, storage_account_url = '{cluster.env_variables["AZURITE_STORAGE_ACCOUNT_URL"]}', blob_path = '/iceberg_data/default/{table_name}/', format={format})"""
+            if table_function:
+                return f"""
+                    icebergAzure(azure, container = '{cluster.azure_container_name}', storage_account_url = '{cluster.env_variables["AZURITE_STORAGE_ACCOUNT_URL"]}', blob_path = '/iceberg_data/default/{table_name}/', format={format})
+                """
+            else:
+                return f"""
+                    DROP TABLE IF EXISTS {table_name};
+                    CREATE TABLE {table_name}
+                    ENGINE=IcebergAzure(azure, container = {cluster.azure_container_name}, storage_account_url = '{cluster.env_variables["AZURITE_STORAGE_ACCOUNT_URL"]}', blob_path = '/iceberg_data/default/{table_name}/', format={format})"""
+
     elif storage_type == "hdfs":
-        if table_function:
+        if run_on_cluster:
+            assert table_function
             return f"""
-                icebergHDFS(hdfs, filename= 'iceberg_data/default/{table_name}/', format={format}, url = 'hdfs://hdfs1:9000/')
+                icebergHDFSCluster('cluster_simple', hdfs, filename= 'iceberg_data/default/{table_name}/', format={format}, url = 'hdfs://hdfs1:9000/')
             """
         else:
-            return f"""
-                DROP TABLE IF EXISTS {table_name};
-                CREATE TABLE {table_name}
-                ENGINE=IcebergHDFS(hdfs, filename = 'iceberg_data/default/{table_name}/', format={format}, url = 'hdfs://hdfs1:9000/');"""
+            if table_function:
+                return f"""
+                    icebergHDFS(hdfs, filename= 'iceberg_data/default/{table_name}/', format={format}, url = 'hdfs://hdfs1:9000/')
+                """
+            else:
+                return f"""
+                    DROP TABLE IF EXISTS {table_name};
+                    CREATE TABLE {table_name}
+                    ENGINE=IcebergHDFS(hdfs, filename = 'iceberg_data/default/{table_name}/', format={format}, url = 'hdfs://hdfs1:9000/');"""
+
     elif storage_type == "local":
+        assert not run_on_cluster
+
         if table_function:
             return f"""
                 icebergLocal(local, path = '/iceberg_data/default/{table_name}/', format={format})
@@ -227,6 +273,7 @@ def get_creation_expression(
                 DROP TABLE IF EXISTS {table_name};
                 CREATE TABLE {table_name}
                 ENGINE=IcebergLocal(local, path = '/iceberg_data/default/{table_name}/', format={format});"""
+
     else:
         raise Exception(f"Unknown iceberg storage type: {storage_type}")
 
@@ -492,6 +539,108 @@ def test_types(started_cluster, format_version, storage_type):
     )
 
 
+@pytest.mark.parametrize("format_version", ["1", "2"])
+@pytest.mark.parametrize("storage_type", ["s3", "azure", "hdfs"])
+def test_cluster_table_function(started_cluster, format_version, storage_type):
+    if is_arm() and storage_type == "hdfs":
+        pytest.skip("Disabled test IcebergHDFS for aarch64")
+
+    instance = started_cluster.instances["node1"]
+    spark = started_cluster.spark_session
+
+    TABLE_NAME = (
+        "test_iceberg_cluster_"
+        + format_version
+        + "_"
+        + storage_type
+        + "_"
+        + get_uuid_str()
+    )
+
+    def add_df(mode):
+        write_iceberg_from_df(
+            spark,
+            generate_data(spark, 0, 100),
+            TABLE_NAME,
+            mode=mode,
+            format_version=format_version,
+        )
+
+        files = default_upload_directory(
+            started_cluster,
+            storage_type,
+            f"/iceberg_data/default/{TABLE_NAME}/",
+            f"/iceberg_data/default/{TABLE_NAME}/",
+        )
+
+        logging.info(f"Adding another dataframe. result files: {files}")
+
+        return files
+
+    files = add_df(mode="overwrite")
+    for i in range(1, len(started_cluster.instances)):
+        files = add_df(mode="append")
+
+    logging.info(f"Setup complete. files: {files}")
+    assert len(files) == 5 + 4 * (len(started_cluster.instances) - 1)
+
+    clusters = instance.query(f"SELECT * FROM system.clusters")
+    logging.info(f"Clusters setup: {clusters}")
+
+    # Regular Query only node1
+    table_function_expr = get_creation_expression(
+        storage_type, TABLE_NAME, started_cluster, table_function=True
+    )
+    select_regular = (
+        instance.query(f"SELECT * FROM {table_function_expr}").strip().split()
+    )
+
+    # Cluster Query with node1 as coordinator
+    table_function_expr_cluster = get_creation_expression(
+        storage_type,
+        TABLE_NAME,
+        started_cluster,
+        table_function=True,
+        run_on_cluster=True,
+    )
+    select_cluster = (
+        instance.query(f"SELECT * FROM {table_function_expr_cluster}").strip().split()
+    )
+
+    # Simple size check
+    assert len(select_regular) == 600
+    assert len(select_cluster) == 600
+
+    # Actual check
+    assert select_cluster == select_regular
+
+    # Check query_log
+    for replica in started_cluster.instances.values():
+        replica.query("SYSTEM FLUSH LOGS")
+
+    for node_name, replica in started_cluster.instances.items():
+        cluster_secondary_queries = (
+            replica.query(
+                f"""
+                SELECT query, type, is_initial_query, read_rows, read_bytes FROM system.query_log
+                WHERE
+                    type = 'QueryStart' AND
+                    positionCaseInsensitive(query, '{storage_type}Cluster') != 0 AND
+                    position(query, '{TABLE_NAME}') != 0 AND
+                    position(query, 'system.query_log') = 0 AND
+                    NOT is_initial_query
+            """
+            )
+            .strip()
+            .split("\n")
+        )
+
+        logging.info(
+            f"[{node_name}] cluster_secondary_queries: {cluster_secondary_queries}"
+        )
+        assert len(cluster_secondary_queries) == 1
+
+
 @pytest.mark.parametrize("format_version", ["1", "2"])
 @pytest.mark.parametrize("storage_type", ["s3", "azure", "hdfs", "local"])
 def test_delete_files(started_cluster, format_version, storage_type):

tests/queries/0_stateless/03216_arrayWithConstant_limits.sql

@@ -1,3 +1,6 @@
 SELECT arrayWithConstant(96142475, ['qMUF']); -- { serverError TOO_LARGE_ARRAY_SIZE }
 SELECT arrayWithConstant(100000000, materialize([[[[[[[[[['Hello, world!']]]]]]]]]])); -- { serverError TOO_LARGE_ARRAY_SIZE }
 SELECT length(arrayWithConstant(10000000, materialize([[[[[[[[[['Hello world']]]]]]]]]])));
+
+CREATE TEMPORARY TABLE args (value Array(Int)) ENGINE=Memory AS SELECT [1, 1, 1, 1] as value FROM numbers(1, 100);
+SELECT length(arrayWithConstant(1000000, value)) FROM args FORMAT NULL;

tests/queries/0_stateless/03261_projection_mask.sql

@@ -0,0 +1,29 @@
+-- compact
+DROP TABLE IF EXISTS users;
+
+CREATE TABLE users (
+    uid Int16,
+    name String,
+    age Int16,
+    projection p1 (select age, count() group by age),
+) ENGINE = MergeTree order by uid
+SETTINGS lightweight_mutation_projection_mode = 'rebuild', min_bytes_for_wide_part = 10485760;
+
+INSERT INTO users VALUES (1231, 'John', 33), (1232, 'Mary', 34);
+
+DELETE FROM users WHERE age = 34;
+
+-- wide
+DROP TABLE IF EXISTS users;
+
+CREATE TABLE users (
+    uid Int16,
+    name String,
+    age Int16,
+    projection p1 (select age, count() group by age),
+) ENGINE = MergeTree order by uid
+SETTINGS lightweight_mutation_projection_mode = 'rebuild', min_bytes_for_wide_part = 0;
+
+INSERT INTO users VALUES (1231, 'John', 33), (1232, 'Mary', 34);
+
+DELETE FROM users WHERE age = 34;

utils/check-style/aspell-ignore/en/aspell-dict.txt

@@ -244,7 +244,10 @@ Deduplication
 DefaultTableEngine
 DelayedInserts
 DeliveryTag
+Deltalake
 DeltaLake
+deltalakeCluster
+deltaLakeCluster
 Denormalize
 DestroyAggregatesThreads
 DestroyAggregatesThreadsActive
@@ -377,10 +380,15 @@ Homebrew's
 HorizontalDivide
 Hostname
 HouseOps
+hudi
 Hudi
+hudiCluster
+HudiCluster
 HyperLogLog
 Hypot
 IANA
+icebergCluster
+IcebergCluster
 IDE
 IDEs
 IDNA