Merge a2f9329e18 into 733c57dae7

A better fix for #67476

docker/server/entrypoint.sh

@@ -109,7 +109,7 @@ if [ -n "$CLICKHOUSE_USER" ] && [ "$CLICKHOUSE_USER" != "default" ] || [ -n "$CL
           <networks>
             <ip>::/0</ip>
           </networks>
-          <password>${CLICKHOUSE_PASSWORD}</password>
+          <password><![CDATA[${CLICKHOUSE_PASSWORD//]]>/]]]]><![CDATA[>}]]></password>
           <quota>default</quota>
           <access_management>${CLICKHOUSE_ACCESS_MANAGEMENT}</access_management>
         </${CLICKHOUSE_USER}>

docs/en/interfaces/formats.md

@@ -1396,6 +1396,7 @@ SELECT * FROM json_each_row_nested
 - [input_format_json_ignore_unknown_keys_in_named_tuple](/docs/en/operations/settings/settings-formats.md/#input_format_json_ignore_unknown_keys_in_named_tuple) - ignore unknown keys in json object for named tuples. Default value - `false`.
 - [input_format_json_compact_allow_variable_number_of_columns](/docs/en/operations/settings/settings-formats.md/#input_format_json_compact_allow_variable_number_of_columns) - allow variable number of columns in JSONCompact/JSONCompactEachRow format, ignore extra columns and use default values on missing columns. Default value - `false`.
 - [input_format_json_throw_on_bad_escape_sequence](/docs/en/operations/settings/settings-formats.md/#input_format_json_throw_on_bad_escape_sequence) - throw an exception if JSON string contains bad escape sequence. If disabled, bad escape sequences will remain as is in the data. Default value - `true`.
+- [input_format_json_empty_as_default](/docs/en/operations/settings/settings-formats.md/#input_format_json_empty_as_default) - treat empty fields in JSON input as default values. Default value - `false`. For complex default expressions [input_format_defaults_for_omitted_fields](/docs/en/operations/settings/settings-formats.md/#input_format_defaults_for_omitted_fields) must be enabled too.
 - [output_format_json_quote_64bit_integers](/docs/en/operations/settings/settings-formats.md/#output_format_json_quote_64bit_integers) - controls quoting of 64-bit integers in JSON output format. Default value - `true`.
 - [output_format_json_quote_64bit_floats](/docs/en/operations/settings/settings-formats.md/#output_format_json_quote_64bit_floats) - controls quoting of 64-bit floats in JSON output format. Default value - `false`.
 - [output_format_json_quote_denormals](/docs/en/operations/settings/settings-formats.md/#output_format_json_quote_denormals) - enables '+nan', '-nan', '+inf', '-inf' outputs in JSON output format. Default value - `false`.

docs/en/operations/settings/settings-formats.md

@@ -752,6 +752,17 @@ Possible values:
 
 Default value: 0.
 
+### input_format_json_empty_as_default {#input_format_json_empty_as_default}
+
+When enabled, replace empty input fields in JSON with default values. For complex default expressions `input_format_defaults_for_omitted_fields` must be enabled too.
+
+Possible values:
+
++ 0 — Disable.
++ 1 — Enable.
+
+Default value: 0.
+
 ## TSV format settings {#tsv-format-settings}
 
 ### input_format_tsv_empty_as_default {#input_format_tsv_empty_as_default}

src/Common/ErrorCodes.cpp

@@ -609,6 +609,7 @@
     M(728, UNEXPECTED_DATA_TYPE) \
     M(729, ILLEGAL_TIME_SERIES_TAGS) \
     M(730, REFRESH_FAILED) \
+    M(731, QUERY_CACHE_USED_WITH_NON_THROW_OVERFLOW_MODE) \
     \
     M(900, DISTRIBUTED_CACHE_ERROR) \
     M(901, CANNOT_USE_DISTRIBUTED_CACHE) \

src/Common/atomicRename.cpp

@@ -57,11 +57,13 @@ namespace ErrorCodes
 namespace DB
 {
 
-static bool supportsAtomicRenameImpl()
+static std::optional<std::string> supportsAtomicRenameImpl()
 {
     VersionNumber renameat2_minimal_version(3, 15, 0);
     VersionNumber linux_version(Poco::Environment::osVersion());
-    return linux_version >= renameat2_minimal_version;
+    if (linux_version >= renameat2_minimal_version)
+        return std::nullopt;
+    return fmt::format("Linux kernel 3.15+ is required, got {}", linux_version.toString());
 }
 
 static bool renameat2(const std::string & old_path, const std::string & new_path, int flags)
@@ -97,10 +99,14 @@ static bool renameat2(const std::string & old_path, const std::string & new_path
     ErrnoException::throwFromPath(ErrorCodes::SYSTEM_ERROR, new_path, "Cannot rename {} to {}", old_path, new_path);
 }
 
-bool supportsAtomicRename()
+bool supportsAtomicRename(std::string * out_message)
 {
-    static bool supports = supportsAtomicRenameImpl();
+    static auto error = supportsAtomicRenameImpl();
-    return supports;
+    if (!error.has_value())
+        return true;
+    if (out_message)
+        *out_message = error.value();
+    return false;
 }
 
 }
@@ -152,16 +158,22 @@ static bool renameat2(const std::string & old_path, const std::string & new_path
 }
 
 
-static bool supportsAtomicRenameImpl()
+static std::optional<std::string> supportsAtomicRenameImpl()
 {
     auto fun = dlsym(RTLD_DEFAULT, "renamex_np");
-    return fun != nullptr;
+    if (fun != nullptr)
+        return std::nullopt;
+    return "macOS 10.12 or later is required";
 }
 
-bool supportsAtomicRename()
+bool supportsAtomicRename(std::string * out_message)
 {
-    static bool supports = supportsAtomicRenameImpl();
+    static auto error = supportsAtomicRenameImpl();
-    return supports;
+    if (!error.has_value())
+        return true;
+    if (out_message)
+        *out_message = error.value();
+    return false;
 }
 
 }
@@ -179,8 +191,10 @@ static bool renameat2(const std::string &, const std::string &, int)
     return false;
 }
 
-bool supportsAtomicRename()
+bool supportsAtomicRename(std::string * out_message)
 {
+    if (out_message)
+        *out_message = "only Linux and macOS are supported";
     return false;
 }
 

src/Common/atomicRename.h

@@ -6,7 +6,7 @@ namespace DB
 {
 
 /// Returns true, if the following functions supported by the system
-bool supportsAtomicRename();
+bool supportsAtomicRename(std::string * out_message = nullptr);
 
 /// Atomically rename old_path to new_path. If new_path exists, do not overwrite it and throw exception
 void renameNoReplace(const std::string & old_path, const std::string & new_path);

src/Core/Settings.h

@@ -1144,6 +1144,7 @@ class IColumn;
     M(Bool, input_format_try_infer_variants, false, "Try to infer the Variant type in text formats when there is more than one possible type for column/array elements", 0) \
     M(Bool, type_json_skip_duplicated_paths, false, "When enabled, during parsing JSON object into JSON type duplicated paths will be ignored and only the first one will be inserted instead of an exception", 0) \
     M(UInt64, input_format_json_max_depth, 1000, "Maximum depth of a field in JSON. This is not a strict limit, it does not have to be applied precisely.", 0) \
+    M(Bool, input_format_json_empty_as_default, false, "Treat empty fields in JSON input as default values.", 0) \
     M(Bool, input_format_try_infer_integers, true, "Try to infer integers instead of floats while schema inference in text formats", 0) \
     M(Bool, input_format_try_infer_dates, true, "Try to infer dates from string fields while schema inference in text formats", 0) \
     M(Bool, input_format_try_infer_datetimes, true, "Try to infer datetimes from string fields while schema inference in text formats", 0) \

src/Core/SettingsChangesHistory.cpp

@@ -71,6 +71,7 @@ static std::initializer_list<std::pair<ClickHouseVersion, SettingsChangesHistory
     },
     {"24.9",
         {
+            {"input_format_json_empty_as_default", false, false, "Added new setting to allow to treat empty fields in JSON input as default values."},
             {"input_format_try_infer_variants", false, false, "Try to infer Variant type in text formats when there is more than one possible type for column/array elements"},
             {"join_output_by_rowlist_perkey_rows_threshold", 0, 5, "The lower limit of per-key average rows in the right table to determine whether to output by row list in hash join."},
             {"create_if_not_exists", false, false, "New setting."},

src/DataTypes/Serializations/SerializationArray.cpp

@@ -11,6 +11,7 @@
 #include <IO/WriteBufferFromString.h>
 
 #include <Formats/FormatSettings.h>
+#include <Formats/JSONUtils.h>
 
 namespace DB
 {
@@ -615,28 +616,49 @@ void SerializationArray::serializeTextJSONPretty(const IColumn & column, size_t
 }
 
 
-void SerializationArray::deserializeTextJSON(IColumn & column, ReadBuffer & istr, const FormatSettings & settings) const
+template <typename ReturnType>
+ReturnType SerializationArray::deserializeTextJSONImpl(IColumn & column, ReadBuffer & istr, const FormatSettings & settings) const
 {
-    deserializeTextImpl(column, istr,
+    auto deserialize_nested = [&settings, this](IColumn & nested_column, ReadBuffer & buf) -> ReturnType
-        [&](IColumn & nested_column)
+    {
+        if constexpr (std::is_same_v<ReturnType, void>)
         {
             if (settings.null_as_default && !isColumnNullableOrLowCardinalityNullable(nested_column))
-                SerializationNullable::deserializeNullAsDefaultOrNestedTextJSON(nested_column, istr, settings, nested);
+                SerializationNullable::deserializeNullAsDefaultOrNestedTextJSON(nested_column, buf, settings, nested);
             else
-                nested->deserializeTextJSON(nested_column, istr, settings);
+                nested->deserializeTextJSON(nested_column, buf, settings);
-        }, false);
+        }
+        else
+        {
+            if (settings.null_as_default && !isColumnNullableOrLowCardinalityNullable(nested_column))
+                return SerializationNullable::tryDeserializeNullAsDefaultOrNestedTextJSON(nested_column, buf, settings, nested);
+            return nested->tryDeserializeTextJSON(nested_column, buf, settings);
+        }
+    };
+
+    if (settings.json.empty_as_default)
+        return deserializeTextImpl<ReturnType>(column, istr,
+            [&deserialize_nested, &istr](IColumn & nested_column) -> ReturnType
+            {
+                return JSONUtils::deserializeEmpyStringAsDefaultOrNested<ReturnType>(nested_column, istr, deserialize_nested);
+            }, false);
+    else
+        return deserializeTextImpl<ReturnType>(column, istr,
+            [&deserialize_nested, &istr](IColumn & nested_column) -> ReturnType
+            {
+                return deserialize_nested(nested_column, istr);
+            }, false);
+}
+
+
+void SerializationArray::deserializeTextJSON(IColumn & column, ReadBuffer & istr, const FormatSettings & settings) const
+{
+    deserializeTextJSONImpl<void>(column, istr, settings);
 }
 
 bool SerializationArray::tryDeserializeTextJSON(IColumn & column, ReadBuffer & istr, const FormatSettings & settings) const
 {
-    auto read_nested = [&](IColumn & nested_column)
+    return deserializeTextJSONImpl<bool>(column, istr, settings);
-    {
-        if (settings.null_as_default && !isColumnNullableOrLowCardinalityNullable(nested_column))
-            return SerializationNullable::tryDeserializeNullAsDefaultOrNestedTextJSON(nested_column, istr, settings, nested);
-        return nested->tryDeserializeTextJSON(nested_column, istr, settings);
-    };
-
-    return deserializeTextImpl<bool>(column, istr, std::move(read_nested), false);
 }
 
 

src/DataTypes/Serializations/SerializationArray.h

@@ -82,6 +82,10 @@ public:
         SerializationPtr create(const SerializationPtr & prev) const override;
         ColumnPtr create(const ColumnPtr & prev) const override;
     };
+
+private:
+    template <typename ReturnType>
+    ReturnType deserializeTextJSONImpl(IColumn & column, ReadBuffer & istr, const FormatSettings & settings) const;
 };
 
 }

src/DataTypes/Serializations/SerializationMap.cpp

@@ -6,6 +6,7 @@
 #include <Columns/ColumnMap.h>
 #include <Core/Field.h>
 #include <Formats/FormatSettings.h>
+#include <Formats/JSONUtils.h>
 #include <Common/assert_cast.h>
 #include <Common/quoteString.h>
 #include <IO/WriteHelpers.h>
@@ -316,28 +317,52 @@ void SerializationMap::serializeTextJSONPretty(const IColumn & column, size_t ro
 }
 
 
-void SerializationMap::deserializeTextJSON(IColumn & column, ReadBuffer & istr, const FormatSettings & settings) const
+template <typename ReturnType>
+ReturnType SerializationMap::deserializeTextJSONImpl(IColumn & column, ReadBuffer & istr, const FormatSettings & settings) const
 {
-    deserializeTextImpl(column, istr,
+    auto deserialize_nested = [&settings](IColumn & subcolumn, ReadBuffer & buf, const SerializationPtr & subcolumn_serialization) -> ReturnType
-        [&settings](ReadBuffer & buf, const SerializationPtr & subcolumn_serialization, IColumn & subcolumn)
+    {
+        if constexpr (std::is_same_v<ReturnType, void>)
         {
             if (settings.null_as_default && !isColumnNullableOrLowCardinalityNullable(subcolumn))
                 SerializationNullable::deserializeNullAsDefaultOrNestedTextJSON(subcolumn, buf, settings, subcolumn_serialization);
             else
                 subcolumn_serialization->deserializeTextJSON(subcolumn, buf, settings);
-        });
+        }
+        else
+        {
+            if (settings.null_as_default && !isColumnNullableOrLowCardinalityNullable(subcolumn))
+                return SerializationNullable::tryDeserializeNullAsDefaultOrNestedTextJSON(subcolumn, buf, settings, subcolumn_serialization);
+            return subcolumn_serialization->tryDeserializeTextJSON(subcolumn, buf, settings);
+        }
+    };
+
+    if (settings.json.empty_as_default)
+        return deserializeTextImpl<ReturnType>(column, istr,
+            [&deserialize_nested](ReadBuffer & buf, const SerializationPtr & subcolumn_serialization, IColumn & subcolumn) -> ReturnType
+            {
+                return JSONUtils::deserializeEmpyStringAsDefaultOrNested<ReturnType>(subcolumn, buf,
+                    [&deserialize_nested, &subcolumn_serialization](IColumn & subcolumn_, ReadBuffer & buf_) -> ReturnType
+                    {
+                        return deserialize_nested(subcolumn_, buf_, subcolumn_serialization);
+                    });
+            });
+    else
+        return deserializeTextImpl<ReturnType>(column, istr,
+            [&deserialize_nested](ReadBuffer & buf, const SerializationPtr & subcolumn_serialization, IColumn & subcolumn) -> ReturnType
+            {
+                return deserialize_nested(subcolumn, buf, subcolumn_serialization);
+            });
+}
+
+void SerializationMap::deserializeTextJSON(IColumn & column, ReadBuffer & istr, const FormatSettings & settings) const
+{
+    deserializeTextJSONImpl<void>(column, istr, settings);
 }
 
 bool SerializationMap::tryDeserializeTextJSON(IColumn & column, ReadBuffer & istr, const FormatSettings & settings) const
 {
-    auto reader = [&settings](ReadBuffer & buf, const SerializationPtr & subcolumn_serialization, IColumn & subcolumn)
+    return deserializeTextJSONImpl<bool>(column, istr, settings);
-    {
-        if (settings.null_as_default && !isColumnNullableOrLowCardinalityNullable(subcolumn))
-            return SerializationNullable::tryDeserializeNullAsDefaultOrNestedTextJSON(subcolumn, buf, settings, subcolumn_serialization);
-        return subcolumn_serialization->tryDeserializeTextJSON(subcolumn, buf, settings);
-    };
-
-    return deserializeTextImpl<bool>(column, istr, reader);
 }
 
 void SerializationMap::serializeTextXML(const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const

src/DataTypes/Serializations/SerializationMap.h

@@ -74,6 +74,9 @@ private:
 
     template <typename ReturnType = void, typename Reader>
     ReturnType deserializeTextImpl(IColumn & column, ReadBuffer & istr, Reader && reader) const;
+
+    template <typename ReturnType>
+    ReturnType deserializeTextJSONImpl(IColumn & column, ReadBuffer & istr, const FormatSettings & settings) const;
 };
 
 }

src/DataTypes/Serializations/SerializationTuple.cpp

@@ -5,6 +5,7 @@
 #include <Core/Field.h>
 #include <Columns/ColumnTuple.h>
 #include <Common/assert_cast.h>
+#include <Formats/JSONUtils.h>
 #include <IO/WriteHelpers.h>
 #include <IO/ReadHelpers.h>
 #include <IO/ReadBufferFromString.h>
@@ -313,27 +314,9 @@ void SerializationTuple::serializeTextJSONPretty(const IColumn & column, size_t
 }
 
 template <typename ReturnType>
-ReturnType SerializationTuple::deserializeTextJSONImpl(IColumn & column, ReadBuffer & istr, const FormatSettings & settings) const
+ReturnType SerializationTuple::deserializeTupleJSONImpl(IColumn & column, ReadBuffer & istr, const FormatSettings & settings, auto && deserialize_element) const
 {
-    static constexpr bool throw_exception = std::is_same_v<ReturnType, void>;
+    static constexpr auto throw_exception = std::is_same_v<ReturnType, void>;
-
-    auto deserialize_element = [&](IColumn & element_column, size_t element_pos)
-    {
-        if constexpr (throw_exception)
-        {
-            if (settings.null_as_default && !isColumnNullableOrLowCardinalityNullable(element_column))
-                SerializationNullable::deserializeNullAsDefaultOrNestedTextJSON(element_column, istr, settings, elems[element_pos]);
-            else
-                elems[element_pos]->deserializeTextJSON(element_column, istr, settings);
-            return true;
-        }
-        else
-        {
-            if (settings.null_as_default && !isColumnNullableOrLowCardinalityNullable(element_column))
-                return SerializationNullable::tryDeserializeNullAsDefaultOrNestedTextJSON(element_column, istr, settings, elems[element_pos]);
-            return elems[element_pos]->tryDeserializeTextJSON(element_column, istr, settings);
-        }
-    };
 
     if (settings.json.read_named_tuples_as_objects
         && have_explicit_names)
@@ -506,12 +489,51 @@ ReturnType SerializationTuple::deserializeTextJSONImpl(IColumn & column, ReadBuf
     }
 }
 
-void SerializationTuple::deserializeTextJSON(DB::IColumn & column, DB::ReadBuffer & istr, const DB::FormatSettings & settings) const
+template <typename ReturnType>
+ReturnType SerializationTuple::deserializeTextJSONImpl(IColumn & column, ReadBuffer & istr, const FormatSettings & settings) const
 {
-    deserializeTextJSONImpl(column, istr, settings);
+    auto deserialize_nested = [&settings](IColumn & nested_column, ReadBuffer & buf, const SerializationPtr & nested_column_serialization) -> ReturnType
+    {
+        if constexpr (std::is_same_v<ReturnType, void>)
+        {
+            if (settings.null_as_default && !isColumnNullableOrLowCardinalityNullable(nested_column))
+                SerializationNullable::deserializeNullAsDefaultOrNestedTextJSON(nested_column, buf, settings, nested_column_serialization);
+            else
+                nested_column_serialization->deserializeTextJSON(nested_column, buf, settings);
+        }
+        else
+        {
+            if (settings.null_as_default && !isColumnNullableOrLowCardinalityNullable(nested_column))
+                return SerializationNullable::tryDeserializeNullAsDefaultOrNestedTextJSON(nested_column, buf, settings, nested_column_serialization);
+            else
+                return nested_column_serialization->tryDeserializeTextJSON(nested_column, buf, settings);
+        }
+    };
+
+    if (settings.json.empty_as_default)
+        return deserializeTupleJSONImpl<ReturnType>(column, istr, settings,
+            [&deserialize_nested, &istr, this](IColumn & nested_column, size_t element_pos) -> ReturnType
+            {
+                return JSONUtils::deserializeEmpyStringAsDefaultOrNested<ReturnType>(nested_column, istr,
+                    [&deserialize_nested, element_pos, this](IColumn & nested_column_, ReadBuffer & buf) -> ReturnType
+                    {
+                        return deserialize_nested(nested_column_, buf, elems[element_pos]);
+                    });
+            });
+    else
+        return deserializeTupleJSONImpl<ReturnType>(column, istr, settings,
+            [&deserialize_nested, &istr, this](IColumn & nested_column, size_t element_pos) -> ReturnType
+            {
+                return deserialize_nested(nested_column, istr, elems[element_pos]);
+            });
 }
 
-bool SerializationTuple::tryDeserializeTextJSON(DB::IColumn & column, DB::ReadBuffer & istr, const DB::FormatSettings & settings) const
+void SerializationTuple::deserializeTextJSON(IColumn & column, ReadBuffer & istr, const FormatSettings & settings) const
+{
+    deserializeTextJSONImpl<void>(column, istr, settings);
+}
+
+bool SerializationTuple::tryDeserializeTextJSON(IColumn & column, ReadBuffer & istr, const FormatSettings & settings) const
 {
     return deserializeTextJSONImpl<bool>(column, istr, settings);
 }

src/DataTypes/Serializations/SerializationTuple.h

@@ -81,7 +81,10 @@ private:
     template <typename ReturnType = void>
     ReturnType deserializeTextImpl(IColumn & column, ReadBuffer & istr, const FormatSettings & settings, bool whole) const;
 
-    template <typename ReturnType = void>
+    template <typename ReturnType>
+    ReturnType deserializeTupleJSONImpl(IColumn & column, ReadBuffer & istr, const FormatSettings & settings, auto && deserialize_element) const;
+
+    template <typename ReturnType>
     ReturnType deserializeTextJSONImpl(IColumn & column, ReadBuffer & istr, const FormatSettings & settings) const;
 
     template <typename ReturnType = void>

src/Databases/DatabaseAtomic.cpp

@@ -197,8 +197,9 @@ void DatabaseAtomic::renameTable(ContextPtr local_context, const String & table_
             throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Moving tables between databases of different engines is not supported");
     }
 
-    if (exchange && !supportsAtomicRename())
+    std::string message;
-        throw Exception(ErrorCodes::NOT_IMPLEMENTED, "RENAME EXCHANGE is not supported");
+    if (exchange && !supportsAtomicRename(&message))
+        throw Exception(ErrorCodes::NOT_IMPLEMENTED, "RENAME EXCHANGE is not supported because exchanging files is not supported by the OS ({})", message);
 
     waitDatabaseStarted();
 

src/Formats/FormatFactory.cpp

@@ -152,6 +152,7 @@ FormatSettings getFormatSettings(const ContextPtr & context, const Settings & se
     format_settings.json.try_infer_objects_as_tuples = settings.input_format_json_try_infer_named_tuples_from_objects;
     format_settings.json.throw_on_bad_escape_sequence = settings.input_format_json_throw_on_bad_escape_sequence;
     format_settings.json.ignore_unnecessary_fields = settings.input_format_json_ignore_unnecessary_fields;
+    format_settings.json.empty_as_default = settings.input_format_json_empty_as_default;
     format_settings.json.type_json_skip_duplicated_paths = settings.type_json_skip_duplicated_paths;
     format_settings.null_as_default = settings.input_format_null_as_default;
     format_settings.force_null_for_omitted_fields = settings.input_format_force_null_for_omitted_fields;

src/Formats/FormatSettings.h

@@ -237,6 +237,7 @@ struct FormatSettings
         bool infer_incomplete_types_as_strings = true;
         bool throw_on_bad_escape_sequence = true;
         bool ignore_unnecessary_fields = true;
+        bool empty_as_default = false;
         bool type_json_skip_duplicated_paths = false;
     } json{};
 

src/Formats/JSONUtils.cpp

@@ -2,12 +2,14 @@
 #include <Formats/JSONUtils.h>
 #include <Formats/ReadSchemaUtils.h>
 #include <Formats/EscapingRuleUtils.h>
+#include <IO/PeekableReadBuffer.h>
 #include <IO/ReadBufferFromString.h>
 #include <IO/WriteBufferValidUTF8.h>
 #include <DataTypes/Serializations/SerializationNullable.h>
 #include <DataTypes/DataTypeNullable.h>
 #include <DataTypes/DataTypeObjectDeprecated.h>
 #include <DataTypes/DataTypeFactory.h>
+#include <Common/assert_cast.h>
 
 #include <base/find_symbols.h>
 
@@ -286,11 +288,19 @@ namespace JSONUtils
                 return true;
             }
 
-            if (as_nullable)
+            auto deserialize = [as_nullable, &format_settings, &serialization](IColumn & column_, ReadBuffer & buf) -> bool
-                return SerializationNullable::deserializeNullAsDefaultOrNestedTextJSON(column, in, format_settings, serialization);
+            {
+                if (as_nullable)
+                    return SerializationNullable::deserializeNullAsDefaultOrNestedTextJSON(column_, buf, format_settings, serialization);
 
-            serialization->deserializeTextJSON(column, in, format_settings);
+                serialization->deserializeTextJSON(column_, buf, format_settings);
-            return true;
+                return true;
+            };
+
+            if (format_settings.json.empty_as_default)
+                return JSONUtils::deserializeEmpyStringAsDefaultOrNested<bool, false>(column, in, deserialize);
+            else
+                return deserialize(column, in);
         }
         catch (Exception & e)
         {
@@ -920,6 +930,78 @@ namespace JSONUtils
         }
     }
 
+    template <typename ReturnType, bool default_column_return_value>
+    ReturnType deserializeEmpyStringAsDefaultOrNested(IColumn & column, ReadBuffer & istr, const NestedDeserialize<ReturnType> & deserialize_nested)
+    {
+        static constexpr auto throw_exception = std::is_same_v<ReturnType, void>;
+
+        static constexpr auto EMPTY_STRING = "\"\"";
+        static constexpr auto EMPTY_STRING_LENGTH = std::string_view(EMPTY_STRING).length();
+
+        if (istr.eof() || *istr.position() != EMPTY_STRING[0])
+            return deserialize_nested(column, istr);
+
+        auto do_deserialize = [](IColumn & column_, ReadBuffer & buf, auto && check_for_empty_string, auto && deserialize) -> ReturnType
+        {
+            if (check_for_empty_string(buf))
+            {
+                column_.insertDefault();
+                return ReturnType(default_column_return_value);
+            }
+            return deserialize(column_, buf);
+        };
+
+        if (istr.available() >= EMPTY_STRING_LENGTH)
+        {
+            /// We have enough data in buffer to check if we have an empty string.
+            auto check_for_empty_string = [](ReadBuffer & buf) -> bool
+            {
+                auto * pos = buf.position();
+                if (checkString(EMPTY_STRING, buf))
+                    return true;
+                buf.position() = pos;
+                return false;
+            };
+
+            return do_deserialize(column, istr, check_for_empty_string, deserialize_nested);
+        }
+
+        /// We don't have enough data in buffer to check if we have an empty string.
+        /// Use PeekableReadBuffer to make a checkpoint before checking for an
+        /// empty string and rollback if check was failed.
+
+        auto check_for_empty_string = [](ReadBuffer & buf) -> bool
+        {
+            auto & peekable_buf = assert_cast<PeekableReadBuffer &>(buf);
+            peekable_buf.setCheckpoint();
+            SCOPE_EXIT(peekable_buf.dropCheckpoint());
+            if (checkString(EMPTY_STRING, peekable_buf))
+                return true;
+            peekable_buf.rollbackToCheckpoint();
+            return false;
+        };
+
+        auto deserialize_nested_with_check = [&deserialize_nested](IColumn & column_, ReadBuffer & buf) -> ReturnType
+        {
+            auto & peekable_buf = assert_cast<PeekableReadBuffer &>(buf);
+            if constexpr (throw_exception)
+                deserialize_nested(column_, peekable_buf);
+            else if (!deserialize_nested(column_, peekable_buf))
+                return ReturnType(false);
+
+            if (unlikely(peekable_buf.hasUnreadData()))
+                throw Exception(ErrorCodes::LOGICAL_ERROR, "Incorrect state while parsing JSON: PeekableReadBuffer has unread data in own memory: {}", String(peekable_buf.position(), peekable_buf.available()));
+
+            return ReturnType(true);
+        };
+
+        PeekableReadBuffer peekable_buf(istr, true);
+        return do_deserialize(column, peekable_buf, check_for_empty_string, deserialize_nested_with_check);
+    }
+
+    template void deserializeEmpyStringAsDefaultOrNested<void, true>(IColumn & column, ReadBuffer & istr, const NestedDeserialize<void> & deserialize_nested);
+    template bool deserializeEmpyStringAsDefaultOrNested<bool, true>(IColumn & column, ReadBuffer & istr, const NestedDeserialize<bool> & deserialize_nested);
+    template bool deserializeEmpyStringAsDefaultOrNested<bool, false>(IColumn & column, ReadBuffer & istr, const NestedDeserialize<bool> & deserialize_nested);
 }
 
 }

src/Formats/JSONUtils.h

@@ -8,6 +8,7 @@
 #include <IO/Progress.h>
 #include <Core/NamesAndTypes.h>
 #include <Common/Stopwatch.h>
+#include <functional>
 #include <utility>
 
 namespace DB
@@ -146,6 +147,16 @@ namespace JSONUtils
 
     bool skipUntilFieldInObject(ReadBuffer & in, const String & desired_field_name, const FormatSettings::JSON & settings);
     void skipTheRestOfObject(ReadBuffer & in, const FormatSettings::JSON & settings);
+
+    template <typename ReturnType>
+    using NestedDeserialize = std::function<ReturnType(IColumn &, ReadBuffer &)>;
+
+    template <typename ReturnType, bool default_column_return_value = true>
+    ReturnType deserializeEmpyStringAsDefaultOrNested(IColumn & column, ReadBuffer & istr, const NestedDeserialize<ReturnType> & deserialize_nested);
+
+    extern template void deserializeEmpyStringAsDefaultOrNested<void, true>(IColumn & column, ReadBuffer & istr, const NestedDeserialize<void> & deserialize_nested);
+    extern template bool deserializeEmpyStringAsDefaultOrNested<bool, true>(IColumn & column, ReadBuffer & istr, const NestedDeserialize<bool> & deserialize_nested);
+    extern template bool deserializeEmpyStringAsDefaultOrNested<bool, false>(IColumn & column, ReadBuffer & istr, const NestedDeserialize<bool> & deserialize_nested);
 }
 
 }

src/Interpreters/ConcurrentHashJoin.cpp

@@ -85,7 +85,9 @@ ConcurrentHashJoin::ConcurrentHashJoin(
           CurrentMetrics::ConcurrentHashJoinPoolThreads,
           CurrentMetrics::ConcurrentHashJoinPoolThreadsActive,
           CurrentMetrics::ConcurrentHashJoinPoolThreadsScheduled,
-          slots))
+          /*max_threads_*/ slots,
+          /*max_free_threads_*/ 0,
+          /*queue_size_*/ slots))
     , stats_collecting_params(stats_collecting_params_)
 {
     hash_joins.resize(slots);

src/Interpreters/HashJoin/HashJoin.cpp

@@ -1236,6 +1236,7 @@ IBlocksStreamPtr HashJoin::getNonJoinedBlocks(const Block & left_sample_block,
 
 void HashJoin::reuseJoinedData(const HashJoin & join)
 {
+    have_compressed = join.have_compressed;
     data = join.data;
     from_storage_join = true;
 

src/Interpreters/InterpreterCreateQuery.cpp

@@ -968,6 +968,11 @@ void InterpreterCreateQuery::validateMaterializedViewColumnsAndEngine(const ASTC
         if (database && database->getEngineName() != "Atomic")
             throw Exception(ErrorCodes::INCORRECT_QUERY,
                 "Refreshable materialized views (except with APPEND) only support Atomic database engine, but database {} has engine {}", create.getDatabase(), database->getEngineName());
+
+        std::string message;
+        if (!supportsAtomicRename(&message))
+            throw Exception(ErrorCodes::NOT_IMPLEMENTED,
+                "Can't create refreshable materialized view because exchanging files is not supported by the OS ({})", message);
     }
 
     Block input_block;

src/Interpreters/executeQuery.cpp

@@ -99,6 +99,7 @@ namespace DB
 namespace ErrorCodes
 {
     extern const int QUERY_CACHE_USED_WITH_NONDETERMINISTIC_FUNCTIONS;
+    extern const int QUERY_CACHE_USED_WITH_NON_THROW_OVERFLOW_MODE;
     extern const int QUERY_CACHE_USED_WITH_SYSTEM_TABLE;
     extern const int INTO_OUTFILE_NOT_ALLOWED;
     extern const int INVALID_TRANSACTION;
@@ -1118,22 +1119,24 @@ static std::tuple<ASTPtr, BlockIO> executeQueryImpl(
             && settings.use_query_cache
             && !internal
             && client_info.query_kind == ClientInfo::QueryKind::INITIAL_QUERY
-            /// Bug 67476: Avoid that the query cache stores truncated results if the query ran with a non-THROW overflow mode and hit a limit.
-            /// This is more workaround than a fix ... unfortunately it is hard to detect from the perspective of the query cache that the
-            /// query result is truncated.
-            && (settings.read_overflow_mode == OverflowMode::THROW
-                && settings.read_overflow_mode_leaf == OverflowMode::THROW
-                && settings.group_by_overflow_mode == OverflowMode::THROW
-                && settings.sort_overflow_mode == OverflowMode::THROW
-                && settings.result_overflow_mode == OverflowMode::THROW
-                && settings.timeout_overflow_mode == OverflowMode::THROW
-                && settings.set_overflow_mode == OverflowMode::THROW
-                && settings.join_overflow_mode == OverflowMode::THROW
-                && settings.transfer_overflow_mode == OverflowMode::THROW
-                && settings.distinct_overflow_mode == OverflowMode::THROW)
             && (ast->as<ASTSelectQuery>() || ast->as<ASTSelectWithUnionQuery>());
         QueryCache::Usage query_cache_usage = QueryCache::Usage::None;
 
+        /// Bug 67476: If the query runs with a non-THROW overflow mode and hits a limit, the query cache will store a truncated result (if
+        /// enabled). This is incorrect. Unfortunately it is hard to detect from the perspective of the query cache that the query result
+        /// is truncated. Therefore throw an exception, to notify the user to disable either the query cache or use another overflow mode.
+        if (settings.use_query_cache && (settings.read_overflow_mode != OverflowMode::THROW
+            || settings.read_overflow_mode_leaf != OverflowMode::THROW
+            || settings.group_by_overflow_mode != OverflowMode::THROW
+            || settings.sort_overflow_mode != OverflowMode::THROW
+            || settings.result_overflow_mode != OverflowMode::THROW
+            || settings.timeout_overflow_mode != OverflowMode::THROW
+            || settings.set_overflow_mode != OverflowMode::THROW
+            || settings.join_overflow_mode != OverflowMode::THROW
+            || settings.transfer_overflow_mode != OverflowMode::THROW
+            || settings.distinct_overflow_mode != OverflowMode::THROW))
+            throw Exception(ErrorCodes::QUERY_CACHE_USED_WITH_NON_THROW_OVERFLOW_MODE, "use_query_cache and overflow_mode != 'throw' cannot be used together");
+
         /// If the query runs with "use_query_cache = 1", we first probe if the query cache already contains the query result (if yes:
         /// return result from cache). If doesn't, we execute the query normally and write the result into the query cache. Both steps use a
         /// hash of the AST, the current database and the settings as cache key. Unfortunately, the settings are in some places internally

src/Parsers/ParserShowColumnsQuery.cpp

@@ -1,6 +1,6 @@
 #include <Parsers/ParserShowColumnsQuery.h>
 
-#include <Parsers/ASTIdentifier_fwd.h>
+#include <Parsers/ASTIdentifier.h>
 #include <Parsers/ASTLiteral.h>
 #include <Parsers/ASTShowColumnsQuery.h>
 #include <Parsers/CommonParsers.h>
@@ -18,7 +18,6 @@ bool ParserShowColumnsQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expe
     ASTPtr from1;
     ASTPtr from2;
 
-    String from1_str;
     String from2_str;
 
     auto query = std::make_shared<ASTShowColumnsQuery>();
@@ -43,25 +42,18 @@ bool ParserShowColumnsQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expe
     else
         return false;
 
-    tryGetIdentifierNameInto(from1, from1_str);
+    const auto * table_id = from1->as<ASTIdentifier>();
-
+    if (!table_id)
-    bool abbreviated_form = from1_str.contains("."); // FROM database.table
+        return false;
-    if (abbreviated_form)
+    query->table = table_id->shortName();
-    {
+    if (table_id->compound())
-        std::vector<String> split;
+        query->database = table_id->name_parts[0];
-        boost::split(split, from1_str, boost::is_any_of("."));
-        query->database = split[0];
-        query->table = split[1];
-    }
     else
     {
         if (ParserKeyword(Keyword::FROM).ignore(pos, expected) || ParserKeyword(Keyword::IN).ignore(pos, expected))
             if (!ParserIdentifier().parse(pos, from2, expected))
                 return false;
-
         tryGetIdentifierNameInto(from2, from2_str);
-
-        query->table = from1_str;
         query->database = from2_str;
     }
 

src/Parsers/ParserShowIndexesQuery.cpp

@@ -1,6 +1,6 @@
 #include <Parsers/ParserShowIndexesQuery.h>
 
-#include <Parsers/ASTIdentifier_fwd.h>
+#include <Parsers/ASTIdentifier.h>
 #include <Parsers/ASTLiteral.h>
 #include <Parsers/ASTShowIndexesQuery.h>
 #include <Parsers/CommonParsers.h>
@@ -17,7 +17,6 @@ bool ParserShowIndexesQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expe
     ASTPtr from1;
     ASTPtr from2;
 
-    String from1_str;
     String from2_str;
 
     auto query = std::make_shared<ASTShowIndexesQuery>();
@@ -39,25 +38,18 @@ bool ParserShowIndexesQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expe
     else
         return false;
 
-    tryGetIdentifierNameInto(from1, from1_str);
+    const auto * table_id = from1->as<ASTIdentifier>();
-
+    if (!table_id)
-    bool abbreviated_form = from1_str.contains("."); // FROM database.table
+        return false;
-    if (abbreviated_form)
+    query->table = table_id->shortName();
-    {
+    if (table_id->compound())
-        std::vector<String> split;
+        query->database = table_id->name_parts[0];
-        boost::split(split, from1_str, boost::is_any_of("."));
-        query->database = split[0];
-        query->table = split[1];
-    }
     else
     {
         if (ParserKeyword(Keyword::FROM).ignore(pos, expected) || ParserKeyword(Keyword::IN).ignore(pos, expected))
             if (!ParserIdentifier().parse(pos, from2, expected))
                 return false;
-
         tryGetIdentifierNameInto(from2, from2_str);
-
-        query->table = from1_str;
         query->database = from2_str;
     }
 

src/Processors/QueryPlan/ReadFromMergeTree.cpp

@@ -2009,33 +2009,6 @@ void ReadFromMergeTree::initializePipeline(QueryPipelineBuilder & pipeline, cons
 {
     auto result = getAnalysisResult();
 
-    if (is_parallel_reading_from_replicas && context->canUseParallelReplicasOnInitiator()
-        && context->getSettingsRef().parallel_replicas_local_plan)
-    {
-        CoordinationMode mode = CoordinationMode::Default;
-        switch (result.read_type)
-        {
-            case ReadFromMergeTree::ReadType::Default:
-                mode = CoordinationMode::Default;
-                break;
-            case ReadFromMergeTree::ReadType::InOrder:
-                mode = CoordinationMode::WithOrder;
-                break;
-            case ReadFromMergeTree::ReadType::InReverseOrder:
-                mode = CoordinationMode::ReverseOrder;
-                break;
-            case ReadFromMergeTree::ReadType::ParallelReplicas:
-                throw Exception(ErrorCodes::LOGICAL_ERROR, "Read type can't be ParallelReplicas on initiator");
-        }
-
-        chassert(number_of_current_replica.has_value());
-        chassert(all_ranges_callback.has_value());
-
-        /// initialize working set from local replica
-        all_ranges_callback.value()(
-            InitialAllRangesAnnouncement(mode, result.parts_with_ranges.getDescriptions(), number_of_current_replica.value()));
-    }
-
     if (enable_remove_parts_from_snapshot_optimization)
     {
         /// Do not keep data parts in snapshot.

src/Storages/MergeTree/ReplicatedMergeTreeAttachThread.cpp

@@ -15,6 +15,7 @@ namespace ErrorCodes
 {
     extern const int SUPPORT_IS_DISABLED;
     extern const int REPLICA_STATUS_CHANGED;
+    extern const int LOGICAL_ERROR;
 }
 
 ReplicatedMergeTreeAttachThread::ReplicatedMergeTreeAttachThread(StorageReplicatedMergeTree & storage_)
@@ -117,6 +118,67 @@ void ReplicatedMergeTreeAttachThread::checkHasReplicaMetadataInZooKeeper(const z
     }
 }
 
+Int32 ReplicatedMergeTreeAttachThread::fixReplicaMetadataVersionIfNeeded(zkutil::ZooKeeperPtr zookeeper)
+{
+    const String & zookeeper_path = storage.zookeeper_path;
+    const String & replica_path = storage.replica_path;
+    const bool replica_readonly = storage.is_readonly;
+
+    for (size_t i = 0; i != 2; ++i)
+    {
+        String replica_metadata_version_str;
+        const bool replica_metadata_version_exists = zookeeper->tryGet(replica_path + "/metadata_version", replica_metadata_version_str);
+        if (!replica_metadata_version_exists)
+            return -1;
+
+        const Int32 metadata_version = parse<Int32>(replica_metadata_version_str);
+
+        if (metadata_version != 0 || replica_readonly)
+        {
+            /// No need to fix anything
+            return metadata_version;
+        }
+
+        Coordination::Stat stat;
+        zookeeper->get(fs::path(zookeeper_path) / "metadata", &stat);
+        if (stat.version == 0)
+        {
+            /// No need to fix anything
+            return metadata_version;
+        }
+
+        ReplicatedMergeTreeQueue & queue = storage.queue;
+        queue.pullLogsToQueue(zookeeper);
+        if (queue.getStatus().metadata_alters_in_queue != 0)
+        {
+            LOG_DEBUG(log, "No need to update metadata_version as there are ALTER_METADATA entries in the queue");
+            return metadata_version;
+        }
+
+        const Coordination::Requests ops = {
+            zkutil::makeSetRequest(fs::path(replica_path) / "metadata_version", std::to_string(stat.version), 0),
+            zkutil::makeCheckRequest(fs::path(zookeeper_path) / "metadata", stat.version),
+        };
+        Coordination::Responses ops_responses;
+        const auto code = zookeeper->tryMulti(ops, ops_responses);
+        if (code == Coordination::Error::ZOK)
+        {
+            LOG_DEBUG(log, "Successfully set metadata_version to {}", stat.version);
+            return stat.version;
+        }
+        if (code != Coordination::Error::ZBADVERSION)
+        {
+            throw zkutil::KeeperException(code);
+        }
+    }
+
+    /// Second attempt is only possible if metadata_version != 0 or metadata.version changed during the first attempt.
+    /// If metadata_version != 0, on second attempt we will return the new metadata_version.
+    /// If metadata.version changed, on second attempt we will either get metadata_version != 0 and return the new metadata_version or we will get metadata_alters_in_queue != 0 and return 0.
+    /// Either way, on second attempt this method should return.
+    throw Exception(ErrorCodes::LOGICAL_ERROR, "Failed to fix replica metadata_version in ZooKeeper after two attempts");
+}
+
 void ReplicatedMergeTreeAttachThread::runImpl()
 {
     storage.setZooKeeper();
@@ -160,11 +222,11 @@ void ReplicatedMergeTreeAttachThread::runImpl()
     /// Just in case it was not removed earlier due to connection loss
     zookeeper->tryRemove(replica_path + "/flags/force_restore_data");
 
-    String replica_metadata_version;
+    const Int32 replica_metadata_version = fixReplicaMetadataVersionIfNeeded(zookeeper);
-    const bool replica_metadata_version_exists = zookeeper->tryGet(replica_path + "/metadata_version", replica_metadata_version);
+    const bool replica_metadata_version_exists = replica_metadata_version != -1;
     if (replica_metadata_version_exists)
     {
-        storage.setInMemoryMetadata(metadata_snapshot->withMetadataVersion(parse<int>(replica_metadata_version)));
+        storage.setInMemoryMetadata(metadata_snapshot->withMetadataVersion(replica_metadata_version));
     }
     else
     {

src/Storages/MergeTree/ReplicatedMergeTreeAttachThread.h

@@ -48,6 +48,8 @@ private:
     void runImpl();
 
     void finalizeInitialization();
+
+    Int32 fixReplicaMetadataVersionIfNeeded(zkutil::ZooKeeperPtr zookeeper);
 };
 
 }

src/Storages/MergeTree/ReplicatedMergeTreeQueue.cpp

@@ -2222,6 +2222,7 @@ ReplicatedMergeTreeQueue::Status ReplicatedMergeTreeQueue::getStatus() const
     res.inserts_in_queue = 0;
     res.merges_in_queue = 0;
     res.part_mutations_in_queue = 0;
+    res.metadata_alters_in_queue = 0;
     res.queue_oldest_time = 0;
     res.inserts_oldest_time = 0;
     res.merges_oldest_time = 0;
@@ -2264,6 +2265,11 @@ ReplicatedMergeTreeQueue::Status ReplicatedMergeTreeQueue::getStatus() const
                 res.oldest_part_to_mutate_to = entry->new_part_name;
             }
         }
+
+        if (entry->type == LogEntry::ALTER_METADATA)
+        {
+            ++res.metadata_alters_in_queue;
+        }
     }
 
     return res;

src/Storages/MergeTree/ReplicatedMergeTreeQueue.h

@@ -473,6 +473,7 @@ public:
         UInt32 inserts_in_queue;
         UInt32 merges_in_queue;
         UInt32 part_mutations_in_queue;
+        UInt32 metadata_alters_in_queue;
         UInt32 queue_oldest_time;
         UInt32 inserts_oldest_time;
         UInt32 merges_oldest_time;

src/Storages/MergeTree/registerStorageMergeTree.cpp

@@ -277,7 +277,7 @@ static void extractZooKeeperPathAndReplicaNameFromEngineArgs(
 
     if (has_valid_arguments)
     {
-        if (is_replicated_database && local_context->getSettingsRef().database_replicated_allow_replicated_engine_arguments == 0)
+        if (!query.attach && is_replicated_database && local_context->getSettingsRef().database_replicated_allow_replicated_engine_arguments == 0)
         {
             throw Exception(ErrorCodes::BAD_ARGUMENTS,
                             "It's not allowed to specify explicit zookeeper_path and replica_name "
@@ -285,7 +285,7 @@ static void extractZooKeeperPathAndReplicaNameFromEngineArgs(
                             "specify them explicitly, enable setting "
                             "database_replicated_allow_replicated_engine_arguments.");
         }
-        else if (is_replicated_database && local_context->getSettingsRef().database_replicated_allow_replicated_engine_arguments == 1)
+        else if (!query.attach && is_replicated_database && local_context->getSettingsRef().database_replicated_allow_replicated_engine_arguments == 1)
         {
             LOG_WARNING(&Poco::Logger::get("registerStorageMergeTree"), "It's not recommended to explicitly specify "
                                                             "zookeeper_path and replica_name in ReplicatedMergeTree arguments");
@@ -305,7 +305,7 @@ static void extractZooKeeperPathAndReplicaNameFromEngineArgs(
             throw Exception(ErrorCodes::BAD_ARGUMENTS, "Replica name must be a string literal{}", verbose_help_message);
 
 
-        if (is_replicated_database && local_context->getSettingsRef().database_replicated_allow_replicated_engine_arguments == 2)
+        if (!query.attach && is_replicated_database && local_context->getSettingsRef().database_replicated_allow_replicated_engine_arguments == 2)
         {
             LOG_WARNING(&Poco::Logger::get("registerStorageMergeTree"), "Replacing user-provided ZooKeeper path and replica name ({}, {}) "
                                                                      "with default arguments", zookeeper_path, replica_name);

tests/integration/test_parallel_replicas_snapshot_from_initiator/configs/remote_servers.xml

@@ -0,0 +1,33 @@
+<clickhouse>
+    <remote_servers>
+        <parallel_replicas>
+            <shard>
+                <internal_replication>false</internal_replication>
+                <replica>
+                    <host>node0</host>
+                    <port>9000</port>
+                </replica>
+                <replica>
+                    <host>node1</host>
+                    <port>9000</port>
+                </replica>
+                <replica>
+                    <host>node2</host>
+                    <port>9000</port>
+                </replica>
+                <replica>
+                    <host>node3</host>
+                    <port>9000</port>
+                </replica>
+                <replica>
+                    <host>node4</host>
+                    <port>9000</port>
+                </replica>
+                <replica>
+                    <host>node5</host>
+                    <port>9000</port>
+                </replica>
+            </shard>
+        </parallel_replicas>
+    </remote_servers>
+</clickhouse>

tests/integration/test_parallel_replicas_snapshot_from_initiator/test.py

@@ -0,0 +1,73 @@
+import pytest
+
+from helpers.cluster import ClickHouseCluster
+
+cluster = ClickHouseCluster(__file__)
+
+nodes = [
+    cluster.add_instance(
+        f"node{num}", main_configs=["configs/remote_servers.xml"], with_zookeeper=True
+    )
+    for num in range(6)
+]
+
+
+@pytest.fixture(scope="module", autouse=True)
+def start_cluster():
+    try:
+        cluster.start()
+        yield cluster
+    finally:
+        cluster.shutdown()
+
+
+def _create_tables(table_name):
+    for idx, node in enumerate(nodes):
+        node.query(
+            f"DROP TABLE IF EXISTS {table_name}",
+            settings={"database_atomic_wait_for_drop_and_detach_synchronously": True},
+        )
+
+        node.query(
+            f"""
+            CREATE TABLE {table_name} (value Int64)
+            Engine=ReplicatedMergeTree('/test_parallel_replicas/shard/{table_name}', '{idx}')
+            ORDER BY ()
+            """
+        )
+
+    nodes[0].query(
+        f"INSERT INTO {table_name} SELECT * FROM numbers(1000)",
+        settings={"insert_deduplicate": 0},
+    )
+    nodes[0].query(f"SYSTEM SYNC REPLICA ON CLUSTER 'parallel_replicas' {table_name}")
+
+    for idx, node in enumerate(nodes):
+        node.query("SYSTEM STOP REPLICATED SENDS")
+        # the same data on all nodes except for a single value
+        node.query(
+            f"INSERT INTO {table_name} VALUES ({idx})",
+            settings={"insert_deduplicate": 0},
+        )
+
+
+# check that we use the state of data parts from the initiator node (for some sort of determinism of what is been read).
+# currently it is implemented only when we build local plan for the initiator node (we aim to make this behavior default)
+def test_initiator_snapshot_is_used_for_reading(start_cluster):
+    table_name = "t"
+    _create_tables(table_name)
+
+    for idx, node in enumerate(nodes):
+        expected = 499500 + idx  # sum of all integers 0..999 + idx
+        assert (
+            node.query(
+                f"SELECT sum(value) FROM {table_name}",
+                settings={
+                    "allow_experimental_parallel_reading_from_replicas": 2,
+                    "max_parallel_replicas": 100,
+                    "cluster_for_parallel_replicas": "parallel_replicas",
+                    "parallel_replicas_local_plan": True,
+                },
+            )
+            == f"{expected}\n"
+        )

tests/integration/test_replicated_database/test.py

@@ -1549,3 +1549,19 @@ def test_all_groups_cluster(started_cluster):
     assert "bad_settings_node\ndummy_node\n" == bad_settings_node.query(
         "select host_name from system.clusters where name='all_groups.db_cluster' order by host_name"
     )
+
+
+def test_detach_attach_table(started_cluster):
+    main_node.query("DROP DATABASE IF EXISTS detach_attach_db SYNC")
+    main_node.query(
+        "CREATE DATABASE detach_attach_db ENGINE = Replicated('/clickhouse/databases/detach_attach_db');"
+    )
+    main_node.query(
+        "CREATE TABLE detach_attach_db.detach_attach_table (k UInt64) ENGINE=ReplicatedMergeTree ORDER BY k;"
+    )
+    main_node.query("INSERT INTO detach_attach_db.detach_attach_table VALUES (1);")
+    main_node.query("DETACH TABLE detach_attach_db.detach_attach_table PERMANENTLY;")
+    main_node.query("ATTACH TABLE detach_attach_db.detach_attach_table;")
+    assert (
+        main_node.query("SELECT * FROM detach_attach_db.detach_attach_table;") == "1\n"
+    )

tests/queries/0_stateless/02494_query_cache_bugs.reference

@@ -23,23 +23,3 @@ Row 1:
 x: 1
 2
 -- Bug 67476: Queries with overflow mode != throw must not be cached by the query cache
-0
-0
-0
-0
-0
-0
-0
-0
-0
-0
-0
-0
-0
-0
-0
-0
-0
-0
-0
-0

tests/queries/0_stateless/02494_query_cache_bugs.sql

@@ -43,25 +43,15 @@ DROP TABLE IF EXISTS tab;
 CREATE TABLE tab(c UInt64) ENGINE = Memory;
 
 SYSTEM DROP QUERY CACHE;
-SELECT sum(c) FROM tab SETTINGS read_overflow_mode = 'break', use_query_cache = 1;
+SELECT sum(c) FROM tab SETTINGS read_overflow_mode = 'break', use_query_cache = 1; -- { serverError QUERY_CACHE_USED_WITH_NON_THROW_OVERFLOW_MODE }
-SELECT count(*) from system.query_cache;
+SELECT sum(c) FROM tab SETTINGS read_overflow_mode_leaf = 'break', use_query_cache = 1; -- { serverError QUERY_CACHE_USED_WITH_NON_THROW_OVERFLOW_MODE }
-SELECT sum(c) FROM tab SETTINGS read_overflow_mode_leaf = 'break', use_query_cache = 1;
+SELECT sum(c) FROM tab SETTINGS group_by_overflow_mode = 'break', use_query_cache = 1; -- { serverError QUERY_CACHE_USED_WITH_NON_THROW_OVERFLOW_MODE }
-SELECT count(*) from system.query_cache;
+SELECT sum(c) FROM tab SETTINGS sort_overflow_mode = 'break', use_query_cache = 1; -- { serverError QUERY_CACHE_USED_WITH_NON_THROW_OVERFLOW_MODE }
-SELECT sum(c) FROM tab SETTINGS group_by_overflow_mode = 'break', use_query_cache = 1;
+SELECT sum(c) FROM tab SETTINGS result_overflow_mode = 'break', use_query_cache = 1; -- { serverError QUERY_CACHE_USED_WITH_NON_THROW_OVERFLOW_MODE }
-SELECT count(*) from system.query_cache;
+SELECT sum(c) FROM tab SETTINGS timeout_overflow_mode = 'break', use_query_cache = 1; -- { serverError QUERY_CACHE_USED_WITH_NON_THROW_OVERFLOW_MODE }
-SELECT sum(c) FROM tab SETTINGS sort_overflow_mode = 'break', use_query_cache = 1;
+SELECT sum(c) FROM tab SETTINGS set_overflow_mode = 'break', use_query_cache = 1; -- { serverError QUERY_CACHE_USED_WITH_NON_THROW_OVERFLOW_MODE }
-SELECT count(*) from system.query_cache;
+SELECT sum(c) FROM tab SETTINGS join_overflow_mode = 'break', use_query_cache = 1; -- { serverError QUERY_CACHE_USED_WITH_NON_THROW_OVERFLOW_MODE }
-SELECT sum(c) FROM tab SETTINGS result_overflow_mode = 'break', use_query_cache = 1;
+SELECT sum(c) FROM tab SETTINGS transfer_overflow_mode = 'break', use_query_cache = 1; -- { serverError QUERY_CACHE_USED_WITH_NON_THROW_OVERFLOW_MODE }
-SELECT count(*) from system.query_cache;
+SELECT sum(c) FROM tab SETTINGS distinct_overflow_mode = 'break', use_query_cache = 1; -- { serverError QUERY_CACHE_USED_WITH_NON_THROW_OVERFLOW_MODE }
-SELECT sum(c) FROM tab SETTINGS timeout_overflow_mode = 'break', use_query_cache = 1;
-SELECT count(*) from system.query_cache;
-SELECT sum(c) FROM tab SETTINGS set_overflow_mode = 'break', use_query_cache = 1;
-SELECT count(*) from system.query_cache;
-SELECT sum(c) FROM tab SETTINGS join_overflow_mode = 'break', use_query_cache = 1;
-SELECT count(*) from system.query_cache;
-SELECT sum(c) FROM tab SETTINGS transfer_overflow_mode = 'break', use_query_cache = 1;
-SELECT count(*) from system.query_cache;
-SELECT sum(c) FROM tab SETTINGS distinct_overflow_mode = 'break', use_query_cache = 1;
-SELECT count(*) from system.query_cache;
 
 SYSTEM DROP QUERY CACHE;

tests/queries/0_stateless/02706_show_columns.reference

@@ -45,3 +45,7 @@ uint64	UInt64	NO	PRI SOR	\N
 int32	Int32	NO		\N	
 str	String	NO		\N	
 uint64	UInt64	NO	PRI SOR	\N	
+--- SHOW COLUMNS FROM table with dots
+int32	Nullable(Int32)	YES		\N	
+str	String	NO	SOR	\N	
+uint64	UInt64	NO	PRI SOR	\N	

tests/queries/0_stateless/02706_show_columns.sql

@@ -90,3 +90,18 @@ SHOW COLUMNS FROM database_123456789abcde.tab;
 DROP DATABASE database_123456789abcde;
 
 DROP TABLE tab;
+
+DROP TABLE IF EXISTS `tab.with.dots`;
+CREATE TABLE `tab.with.dots`
+(
+    `uint64` UInt64,
+    `int32` Nullable(Int32) COMMENT 'example comment',
+    `str` String,
+    INDEX idx str TYPE set(1000)
+)
+ENGINE = MergeTree
+PRIMARY KEY (uint64)
+ORDER BY (uint64, str);
+SELECT '--- SHOW COLUMNS FROM table with dots';
+SHOW COLUMNS FROM `tab.with.dots`;
+DROP TABLE `tab.with.dots`;

tests/queries/0_stateless/02724_show_indexes.reference

@@ -49,3 +49,10 @@ tbl	1	PRIMARY	1	a	A	0	\N	\N	\N	PRIMARY			YES
 --- Short form
 tbl	1	mmi_idx	1		\N	0	\N	\N	\N	MINMAX			YES	b
 tbl	1	PRIMARY	1	a	A	0	\N	\N	\N	PRIMARY			YES	
+--- SHOW INDEX FROM table with dots
+tab.with.dots	1	blf_idx	1		\N	0	\N	\N	\N	BLOOM_FILTER			YES	d, b
+tab.with.dots	1	mm1_idx	1		\N	0	\N	\N	\N	MINMAX			YES	a, c, d
+tab.with.dots	1	mm2_idx	1		\N	0	\N	\N	\N	MINMAX			YES	c, d, e
+tab.with.dots	1	PRIMARY	1	c	A	0	\N	\N	\N	PRIMARY			YES	
+tab.with.dots	1	PRIMARY	2	a	A	0	\N	\N	\N	PRIMARY			YES	
+tab.with.dots	1	set_idx	1		\N	0	\N	\N	\N	SET			YES	e

tests/queries/0_stateless/02724_show_indexes.sql

@@ -78,3 +78,22 @@ SHOW INDEX FROM database_123456789abcde.tbl;
 DROP DATABASE database_123456789abcde;
 
 DROP TABLE tbl;
+
+DROP TABLE IF EXISTS `tab.with.dots`;
+CREATE TABLE `tab.with.dots`
+(
+    a UInt64,
+    b UInt64,
+    c UInt64,
+    d UInt64,
+    e UInt64,
+    INDEX mm1_idx (a, c, d) TYPE minmax,
+    INDEX mm2_idx (c, d, e) TYPE minmax,
+    INDEX set_idx (e)       TYPE set(100),
+    INDEX blf_idx (d, b)    TYPE bloom_filter(0.8)
+)
+ENGINE = MergeTree
+PRIMARY KEY (c, a);
+SELECT '--- SHOW INDEX FROM table with dots';
+SHOW INDEX FROM `tab.with.dots`;
+DROP TABLE `tab.with.dots`;

tests/queries/0_stateless/03222_json_empty_as_default.reference

@@ -0,0 +1,47 @@
+-- Simple types
+-- { echoOn }
+SELECT x FROM format(JSONEachRow, 'x Date', '{"x":""}');
+1970-01-01
+SELECT x FROM format(JSONEachRow, 'x Date32', '{"x":""}');
+1970-01-01
+SELECT toTimeZone(x, 'UTC') FROM format(JSONEachRow, 'x DateTime', '{"x":""}');
+1970-01-01 00:00:00
+SELECT toTimeZone(x, 'UTC') FROM format(JSONEachRow, 'x DateTime64', '{"x":""}');
+1970-01-01 00:00:00.000
+SELECT x FROM format(JSONEachRow, 'x IPv4', '{"x":""}');
+0.0.0.0
+SELECT x FROM format(JSONEachRow, 'x IPv6', '{"x":""}');
+::
+SELECT x FROM format(JSONEachRow, 'x UUID', '{"x":""}');
+00000000-0000-0000-0000-000000000000
+-- { echoOn }
+SELECT COUNT(DISTINCT col) FROM table1;
+1
+-- { echoOn }
+SELECT * FROM table1 ORDER BY address ASC;
+::
+2001:db8:3333:4444:5555:6666:7777:8888
+-- Nullable
+-- { echoOn }
+SELECT x FROM format(JSONEachRow, 'x Nullable(IPv6)', '{"x":""}');
+\N
+-- Compound types
+SELECT x FROM format(JSONEachRow, 'x Array(UUID)', '{"x":["00000000-0000-0000-0000-000000000000","b15f852c-c41a-4fd6-9247-1929c841715e",""]}');
+['00000000-0000-0000-0000-000000000000','b15f852c-c41a-4fd6-9247-1929c841715e','00000000-0000-0000-0000-000000000000']
+SELECT x FROM format(JSONEachRow, 'x Array(Nullable(IPv6))', '{"x":["",""]}');
+[NULL,NULL]
+SELECT x FROM format(JSONEachRow, 'x Tuple(Date, IPv4, String)', '{"x":["", "", "abc"]}');
+('1970-01-01','0.0.0.0','abc')
+SELECT x FROM format(JSONEachRow, 'x Map(String, IPv6)', '{"x":{"abc": ""}}');
+{'abc':'::'}
+SELECT x FROM format(JSONEachRow, 'x Variant(Date, UUID)', '{"x":""}');
+\N
+-- Deep composition
+SELECT x FROM format(JSONEachRow, 'x Array(Array(IPv6))', '{"x":[["2001:db8:3333:4444:CCCC:DDDD:EEEE:FFFF", ""], ["", "2001:db8:3333:4444:5555:6666:7777:8888"]]}');
+[['2001:db8:3333:4444:cccc:dddd:eeee:ffff','::'],['::','2001:db8:3333:4444:5555:6666:7777:8888']]
+SELECT x FROM format(JSONEachRow, 'x Variant(Date, Array(UUID))', '{"x":["", "b15f852c-c41a-4fd6-9247-1929c841715e"]}');
+['00000000-0000-0000-0000-000000000000','b15f852c-c41a-4fd6-9247-1929c841715e']
+SELECT x FROM format(JSONEachRow, 'x Tuple(Array(UUID), Tuple(UUID, Map(String, IPv6)))', '{"x":[[""], ["",{"abc":""}]]}');
+(['00000000-0000-0000-0000-000000000000'],('00000000-0000-0000-0000-000000000000',{'abc':'::'}))
+SELECT x FROM format(JSONEachRow, 'x Map(Tuple(Date,IPv4), Variant(UUID,IPv6))', '{"x":{["",""]:""}}');
+{('1970-01-01','0.0.0.0'):NULL}

tests/queries/0_stateless/03222_json_empty_as_default.sql

@@ -0,0 +1,60 @@
+SET input_format_json_empty_as_default = 1, allow_experimental_variant_type = 1;
+
+-- Simple types
+-- { echoOn }
+SELECT x FROM format(JSONEachRow, 'x Date', '{"x":""}');
+SELECT x FROM format(JSONEachRow, 'x Date32', '{"x":""}');
+SELECT toTimeZone(x, 'UTC') FROM format(JSONEachRow, 'x DateTime', '{"x":""}');
+SELECT toTimeZone(x, 'UTC') FROM format(JSONEachRow, 'x DateTime64', '{"x":""}');
+SELECT x FROM format(JSONEachRow, 'x IPv4', '{"x":""}');
+SELECT x FROM format(JSONEachRow, 'x IPv6', '{"x":""}');
+SELECT x FROM format(JSONEachRow, 'x UUID', '{"x":""}');
+-- { echoOff }
+
+-- Simple type AggregateFunction
+DROP TABLE IF EXISTS table1;
+CREATE TABLE table1(col AggregateFunction(uniq, UInt64)) ENGINE=Memory();
+DROP TABLE IF EXISTS table2;
+CREATE TABLE table2(UserID UInt64) ENGINE=Memory();
+
+INSERT INTO table1 SELECT uniqState(UserID) FROM table2;
+INSERT INTO table1 SELECT x FROM format(JSONEachRow, 'x AggregateFunction(uniq, UInt64)' AS T, '{"x":""}');
+
+-- { echoOn }
+SELECT COUNT(DISTINCT col) FROM table1;
+-- { echoOff }
+
+DROP TABLE table1;
+DROP TABLE table2;
+
+-- The setting input_format_defaults_for_omitted_fields determines the default value if enabled.
+CREATE TABLE table1(address IPv6 DEFAULT toIPv6('2001:db8:3333:4444:5555:6666:7777:8888')) ENGINE=Memory();
+
+SET input_format_defaults_for_omitted_fields = 0;
+INSERT INTO table1 FORMAT JSONEachRow {"address":""};
+
+SET input_format_defaults_for_omitted_fields = 1;
+INSERT INTO table1 FORMAT JSONEachRow {"address":""};
+
+-- { echoOn }
+SELECT * FROM table1 ORDER BY address ASC;
+-- { echoOff }
+
+DROP TABLE table1;
+
+-- Nullable
+-- { echoOn }
+SELECT x FROM format(JSONEachRow, 'x Nullable(IPv6)', '{"x":""}');
+
+-- Compound types
+SELECT x FROM format(JSONEachRow, 'x Array(UUID)', '{"x":["00000000-0000-0000-0000-000000000000","b15f852c-c41a-4fd6-9247-1929c841715e",""]}');
+SELECT x FROM format(JSONEachRow, 'x Array(Nullable(IPv6))', '{"x":["",""]}');
+SELECT x FROM format(JSONEachRow, 'x Tuple(Date, IPv4, String)', '{"x":["", "", "abc"]}');
+SELECT x FROM format(JSONEachRow, 'x Map(String, IPv6)', '{"x":{"abc": ""}}');
+SELECT x FROM format(JSONEachRow, 'x Variant(Date, UUID)', '{"x":""}');
+
+-- Deep composition
+SELECT x FROM format(JSONEachRow, 'x Array(Array(IPv6))', '{"x":[["2001:db8:3333:4444:CCCC:DDDD:EEEE:FFFF", ""], ["", "2001:db8:3333:4444:5555:6666:7777:8888"]]}');
+SELECT x FROM format(JSONEachRow, 'x Variant(Date, Array(UUID))', '{"x":["", "b15f852c-c41a-4fd6-9247-1929c841715e"]}');
+SELECT x FROM format(JSONEachRow, 'x Tuple(Array(UUID), Tuple(UUID, Map(String, IPv6)))', '{"x":[[""], ["",{"abc":""}]]}');
+SELECT x FROM format(JSONEachRow, 'x Map(Tuple(Date,IPv4), Variant(UUID,IPv6))', '{"x":{["",""]:""}}');

tests/queries/0_stateless/03222_json_empty_as_default_small_read_buffer.reference

@@ -0,0 +1,8 @@
+Array(UUID)
+{"x":["00000000-0000-0000-0000-000000000000","b15f852c-c41a-4fd6-9247-1929c841715e","00000000-0000-0000-0000-000000000000"]}
+{"x":["00000000-0000-0000-0000-000000000000","b15f852c-c41a-4fd6-9247-1929c841715e","00000000-0000-0000-0000-000000000000"]}
+{"x":["00000000-0000-0000-0000-000000000000","b15f852c-c41a-4fd6-9247-1929c841715e","00000000-0000-0000-0000-000000000000"]}
+Tuple(Array(UUID), Tuple(UUID, Map(String, IPv6)))
+{"x":[["00000000-0000-0000-0000-000000000000"],["00000000-0000-0000-0000-000000000000",{"abc":"::"}]]}
+{"x":[["00000000-0000-0000-0000-000000000000"],["00000000-0000-0000-0000-000000000000",{"abc":"::"}]]}
+{"x":[["00000000-0000-0000-0000-000000000000"],["00000000-0000-0000-0000-000000000000",{"abc":"::"}]]}

tests/queries/0_stateless/03222_json_empty_as_default_small_read_buffer.sh

@@ -0,0 +1,31 @@
+#!/usr/bin/env bash
+# Tags: no-parallel
+
+CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
+# shellcheck source=../shell_config.sh
+. "$CURDIR"/../shell_config.sh
+
+DATA_FILE=$CLICKHOUSE_TEST_UNIQUE_NAME.json
+
+# Wrapper for clickhouse-client to always output in JSONEachRow format, that
+# way format settings will not affect output.
+function clickhouse_local()
+{
+    $CLICKHOUSE_LOCAL --output-format JSONEachRow "$@"
+}
+
+echo 'Array(UUID)'
+echo '{"x":["00000000-0000-0000-0000-000000000000","b15f852c-c41a-4fd6-9247-1929c841715e",""]}' > $DATA_FILE
+# Use increasingly smaller read buffers.
+clickhouse_local -q "SELECT x FROM file('$DATA_FILE', 'JSONEachRow', 'x Array(UUID)') SETTINGS input_format_json_empty_as_default=1, input_format_parallel_parsing=0, storage_file_read_method='read', max_read_buffer_size=4"
+clickhouse_local -q "SELECT x FROM file('$DATA_FILE', 'JSONEachRow', 'x Array(UUID)') SETTINGS input_format_json_empty_as_default=1, input_format_parallel_parsing=0, storage_file_read_method='read', max_read_buffer_size=2"
+clickhouse_local -q "SELECT x FROM file('$DATA_FILE', 'JSONEachRow', 'x Array(UUID)') SETTINGS input_format_json_empty_as_default=1, input_format_parallel_parsing=0, storage_file_read_method='read', max_read_buffer_size=1"
+
+echo 'Tuple(Array(UUID), Tuple(UUID, Map(String, IPv6)))'
+echo '{"x":[[""], ["",{"abc":""}]]}' > $DATA_FILE
+# Use increasingly smaller read buffers.
+clickhouse_local -q "SELECT x FROM file('$DATA_FILE', 'JSONEachRow', 'x Tuple(Array(UUID), Tuple(UUID, Map(String, IPv6)))') SETTINGS input_format_json_empty_as_default=1, input_format_parallel_parsing=0, storage_file_read_method='read', max_read_buffer_size=16"
+clickhouse_local -q "SELECT x FROM file('$DATA_FILE', 'JSONEachRow', 'x Tuple(Array(UUID), Tuple(UUID, Map(String, IPv6)))') SETTINGS input_format_json_empty_as_default=1, input_format_parallel_parsing=0, storage_file_read_method='read', max_read_buffer_size=8"
+clickhouse_local -q "SELECT x FROM file('$DATA_FILE', 'JSONEachRow', 'x Tuple(Array(UUID), Tuple(UUID, Map(String, IPv6)))') SETTINGS input_format_json_empty_as_default=1, input_format_parallel_parsing=0, storage_file_read_method='read', max_read_buffer_size=1"
+
+rm $DATA_FILE