From 50df893dc63b6af266d809e2d82fd92a332c5c5d Mon Sep 17 00:00:00 2001
From: Alexander Kuzmenkov <akuzm@yandex-team.ru>
Date: Fri, 15 May 2020 19:23:31 +0300
Subject: [PATCH 001/122] Assert that allocator & container are consistent wrt.
 inline memory

There are lots of places where HashTable is used with
AllocatorWithStackMemory, but the size of allocator inline memory is set
incorrectly, and it's less than the initial HashTable buffer size.
Because of this, HashTable is always allocated on the heap, and the
inline memory becomes a useless dead weight.

For PODArray, we have previously added a helper template that makes sure
these values are in sync, so there was only one such discrepancy left,
in the unit test.
---
 .../AggregateFunctionEntropy.h                | 14 +++------
 .../AggregateFunctionGroupUniqArray.h         | 15 ++++------
 .../AggregateFunctionTopK.h                   | 17 ++---------
 .../QuantileExactWeighted.h                   |  8 ++---
 src/Common/Allocator.h                        | 28 ++++++++++++-----
 src/Common/HashTable/HashTable.h              | 30 +++++++++++++++++++
 src/Common/HashTable/HashTableAllocator.h     |  4 +--
 src/Common/PODArray.h                         |  5 ++++
 src/Common/SpaceSaving.h                      |  7 ++---
 src/Common/tests/pod_array.cpp                | 18 +++++------
 src/Functions/array/arrayDistinct.cpp         | 16 ++++------
 src/Functions/array/arrayEnumerateExtended.h  | 30 ++++++++++++-------
 src/Functions/array/arrayEnumerateRanked.h    | 10 +++----
 src/Functions/array/arrayIntersect.cpp        | 26 ++++++++--------
 src/Functions/array/arrayUniq.cpp             | 25 +++++++++-------
 src/Interpreters/tests/hash_map3.cpp          |  5 ++--
 16 files changed, 145 insertions(+), 113 deletions(-)

diff --git a/src/AggregateFunctions/AggregateFunctionEntropy.h b/src/AggregateFunctions/AggregateFunctionEntropy.h
index 7586cebd8ec..59cacd0e9ca 100644
--- a/src/AggregateFunctions/AggregateFunctionEntropy.h
+++ b/src/AggregateFunctions/AggregateFunctionEntropy.h
@@ -23,18 +23,12 @@ struct EntropyData
 {
     using Weight = UInt64;
 
-    using HashingMap = HashMap<
-        Value, Weight,
-        HashCRC32<Value>,
-        HashTableGrower<4>,
-        HashTableAllocatorWithStackMemory<sizeof(std::pair<Value, Weight>) * (1 << 3)>>;
+    using HashingMap = HASH_TABLE_WITH_STACK_MEMORY(HashMap, Value, Weight,
+        HashCRC32<Value>, HashTableGrower<4>);
 
     /// For the case of pre-hashed values.
-    using TrivialMap = HashMap<
-        Value, Weight,
-        UInt128TrivialHash,
-        HashTableGrower<4>,
-        HashTableAllocatorWithStackMemory<sizeof(std::pair<Value, Weight>) * (1 << 3)>>;
+    using TrivialMap = HASH_TABLE_WITH_STACK_MEMORY(HashMap, Value, Weight,
+        UInt128TrivialHash, HashTableGrower<4>);
 
     using Map = std::conditional_t<std::is_same_v<UInt128, Value>, TrivialMap, HashingMap>;
 
diff --git a/src/AggregateFunctions/AggregateFunctionGroupUniqArray.h b/src/AggregateFunctions/AggregateFunctionGroupUniqArray.h
index 9dbf2c921c2..af97df209f6 100644
--- a/src/AggregateFunctions/AggregateFunctionGroupUniqArray.h
+++ b/src/AggregateFunctions/AggregateFunctionGroupUniqArray.h
@@ -28,12 +28,8 @@ template <typename T>
 struct AggregateFunctionGroupUniqArrayData
 {
     /// When creating, the hash table must be small.
-    using Set = HashSet<
-        T,
-        DefaultHash<T>,
-        HashTableGrower<4>,
-        HashTableAllocatorWithStackMemory<sizeof(T) * (1 << 4)>
-    >;
+    using Set = HASH_TABLE_WITH_STACK_MEMORY(HashSet, T, DefaultHash<T>,
+        HashTableGrower<4>);
 
     Set value;
 };
@@ -126,9 +122,10 @@ public:
 /// Generic implementation, it uses serialized representation as object descriptor.
 struct AggregateFunctionGroupUniqArrayGenericData
 {
-    static constexpr size_t INIT_ELEMS = 2; /// adjustable
-    static constexpr size_t ELEM_SIZE = sizeof(HashSetCellWithSavedHash<StringRef, StringRefHash>);
-    using Set = HashSetWithSavedHash<StringRef, StringRefHash, HashTableGrower<INIT_ELEMS>, HashTableAllocatorWithStackMemory<INIT_ELEMS * ELEM_SIZE>>;
+    static constexpr size_t INITIAL_SIZE_DEGREE = 1; /// adjustable
+
+    using Set = HASH_TABLE_WITH_STACK_MEMORY(HashSetWithSavedHash,
+        StringRef, StringRefHash, HashTableGrower<INITIAL_SIZE_DEGREE>);
 
     Set value;
 };
diff --git a/src/AggregateFunctions/AggregateFunctionTopK.h b/src/AggregateFunctions/AggregateFunctionTopK.h
index 9c5e62bb6d7..23eb0e7ff09 100644
--- a/src/AggregateFunctions/AggregateFunctionTopK.h
+++ b/src/AggregateFunctions/AggregateFunctionTopK.h
@@ -23,13 +23,8 @@ namespace DB
 template <typename T>
 struct AggregateFunctionTopKData
 {
-    using Set = SpaceSaving
-    <
-        T,
-        HashCRC32<T>,
-        HashTableGrower<4>,
-        HashTableAllocatorWithStackMemory<sizeof(T) * (1 << 4)>
-    >;
+    using Set = SpaceSaving<T, HashCRC32<T>>;
+
     Set value;
 };
 
@@ -109,13 +104,7 @@ public:
 /// Generic implementation, it uses serialized representation as object descriptor.
 struct AggregateFunctionTopKGenericData
 {
-    using Set = SpaceSaving
-    <
-        StringRef,
-        StringRefHash,
-        HashTableGrower<4>,
-        HashTableAllocatorWithStackMemory<sizeof(StringRef) * (1 << 4)>
-    >;
+    using Set = SpaceSaving<StringRef, StringRefHash>;
 
     Set value;
 };
diff --git a/src/AggregateFunctions/QuantileExactWeighted.h b/src/AggregateFunctions/QuantileExactWeighted.h
index 6053bddc947..7731ff980fe 100644
--- a/src/AggregateFunctions/QuantileExactWeighted.h
+++ b/src/AggregateFunctions/QuantileExactWeighted.h
@@ -33,12 +33,8 @@ struct QuantileExactWeighted
     using Hasher = std::conditional_t<std::is_same_v<Value, Decimal128>, Int128Hash, HashCRC32<UnderlyingType>>;
 
     /// When creating, the hash table must be small.
-    using Map = HashMap<
-        UnderlyingType, Weight,
-        Hasher,
-        HashTableGrower<4>,
-        HashTableAllocatorWithStackMemory<sizeof(std::pair<Value, Weight>) * (1 << 3)>
-    >;
+    using Map = HASH_TABLE_WITH_STACK_MEMORY(HashMap, UnderlyingType, Weight,
+        Hasher, HashTableGrower<4>);
 
     Map map;
 
diff --git a/src/Common/Allocator.h b/src/Common/Allocator.h
index 9add9299430..43d7e67c4bb 100644
--- a/src/Common/Allocator.h
+++ b/src/Common/Allocator.h
@@ -278,13 +278,15 @@ private:
 
 /** Allocator with optimization to place small memory ranges in automatic memory.
   */
-template <typename Base, size_t N, size_t Alignment>
+template <typename Base, size_t _initial_bytes, size_t Alignment>
 class AllocatorWithStackMemory : private Base
 {
 private:
-    alignas(Alignment) char stack_memory[N];
+    alignas(Alignment) char stack_memory[_initial_bytes];
 
 public:
+    static constexpr size_t initial_bytes = _initial_bytes;
+
     /// Do not use boost::noncopyable to avoid the warning about direct base
     /// being inaccessible due to ambiguity, when derived classes are also
     /// noncopiable (-Winaccessible-base).
@@ -295,10 +297,10 @@ public:
 
     void * alloc(size_t size)
     {
-        if (size <= N)
+        if (size <= initial_bytes)
         {
             if constexpr (Base::clear_memory)
-                memset(stack_memory, 0, N);
+                memset(stack_memory, 0, initial_bytes);
             return stack_memory;
         }
 
@@ -307,18 +309,18 @@ public:
 
     void free(void * buf, size_t size)
     {
-        if (size > N)
+        if (size > initial_bytes)
             Base::free(buf, size);
     }
 
     void * realloc(void * buf, size_t old_size, size_t new_size)
     {
         /// Was in stack_memory, will remain there.
-        if (new_size <= N)
+        if (new_size <= initial_bytes)
             return buf;
 
         /// Already was big enough to not fit in stack_memory.
-        if (old_size > N)
+        if (old_size > initial_bytes)
             return Base::realloc(buf, old_size, new_size, Alignment);
 
         /// Was in stack memory, but now will not fit there.
@@ -330,10 +332,20 @@ public:
 protected:
     static constexpr size_t getStackThreshold()
     {
-        return N;
+        return initial_bytes;
     }
 };
 
+// A constant that gives the number of initially available bytes in
+// the allocator. Used to check that this number is in sync with the
+// initial size of array or hash table that uses the allocator.
+template<typename TAllocator>
+constexpr size_t allocatorInitialBytes = 0;
+
+template<typename Base, size_t initial_bytes, size_t Alignment>
+constexpr size_t allocatorInitialBytes<AllocatorWithStackMemory<
+    Base, initial_bytes, Alignment>> = initial_bytes;
+
 
 #if !__clang__
 #pragma GCC diagnostic pop
diff --git a/src/Common/HashTable/HashTable.h b/src/Common/HashTable/HashTable.h
index 58b7cd81901..2f7272ca8ec 100644
--- a/src/Common/HashTable/HashTable.h
+++ b/src/Common/HashTable/HashTable.h
@@ -308,7 +308,16 @@ struct ZeroValueStorage<false, Cell>
     const Cell * zeroValue() const { return nullptr; }
 };
 
+// These templates give the initial hash table size, so that we can check
+// that it is in sync with initial allocator size.
+template <typename GenericGrower>
+constexpr size_t growerInitialCount = 0;
 
+template <size_t initial_size_degree>
+constexpr size_t growerInitialCount<HashTableGrower<initial_size_degree>>
+    = 1ULL << initial_size_degree;
+
+// The HashTable
 template
 <
     typename Key,
@@ -324,6 +333,17 @@ class HashTable :
     protected Cell::State,
     protected ZeroValueStorage<Cell::need_zero_value_storage, Cell>     /// empty base optimization
 {
+public:
+    // Export the initial buffer sizes for the ease of using allocators with
+    // inline memory.
+    static constexpr size_t initial_buffer_bytes
+        = growerInitialCount<Grower> * sizeof(Cell);
+
+    // If we use an allocator with inline memory, check that the initial
+    // size of the hash table is in sync with the amount of this memory.
+    static_assert(allocatorInitialBytes<Allocator> == 0
+        || allocatorInitialBytes<Allocator> == initial_buffer_bytes);
+
 protected:
     friend class const_iterator;
     friend class iterator;
@@ -1075,3 +1095,13 @@ public:
     }
 #endif
 };
+
+// A helper macro that declares hash table with allocator with stack memory,
+// and the initial size of the allocator is in sync with initial size of the
+// hash table.
+#define HASH_TABLE_WITH_STACK_MEMORY(HASH_TABLE_VARIANT, ...) \
+    HASH_TABLE_VARIANT<__VA_ARGS__, \
+        HashTableAllocatorWithStackMemory< \
+            HASH_TABLE_VARIANT<__VA_ARGS__>::initial_buffer_bytes \
+        > \
+    >
diff --git a/src/Common/HashTable/HashTableAllocator.h b/src/Common/HashTable/HashTableAllocator.h
index 99f9c979685..47e3fdfc4b6 100644
--- a/src/Common/HashTable/HashTableAllocator.h
+++ b/src/Common/HashTable/HashTableAllocator.h
@@ -10,5 +10,5 @@
   */
 using HashTableAllocator = Allocator<true /* clear_memory */, true /* mmap_populate */>;
 
-template <size_t N = 64>
-using HashTableAllocatorWithStackMemory = AllocatorWithStackMemory<HashTableAllocator, N>;
+template <size_t initial_bytes = 64>
+using HashTableAllocatorWithStackMemory = AllocatorWithStackMemory<HashTableAllocator, initial_bytes>;
diff --git a/src/Common/PODArray.h b/src/Common/PODArray.h
index 8fe1f74484e..8b30b314e91 100644
--- a/src/Common/PODArray.h
+++ b/src/Common/PODArray.h
@@ -85,6 +85,11 @@ protected:
 
     static_assert(pad_left <= EmptyPODArraySize && "Left Padding exceeds EmptyPODArraySize. Is the element size too large?");
 
+    // If we are using allocator with inline memory, the minimal size of
+    // array must be in sync with the size of this memory.
+    static_assert(allocatorInitialBytes<TAllocator> == 0
+                  || allocatorInitialBytes<TAllocator> == initial_bytes);
+
     char * c_start          = null;    /// Does not include pad_left.
     char * c_end            = null;
     char * c_end_of_storage = null;    /// Does not include pad_right.
diff --git a/src/Common/SpaceSaving.h b/src/Common/SpaceSaving.h
index 9ad7f6275d6..47b49add060 100644
--- a/src/Common/SpaceSaving.h
+++ b/src/Common/SpaceSaving.h
@@ -67,9 +67,7 @@ private:
 template
 <
     typename TKey,
-    typename Hash = DefaultHash<TKey>,
-    typename Grower = HashTableGrower<>,
-    typename Allocator = HashTableAllocator
+    typename Hash = DefaultHash<TKey>
 >
 class SpaceSaving
 {
@@ -380,7 +378,8 @@ private:
             counter_map[counter->key] = counter;
     }
 
-    using CounterMap = HashMap<TKey, Counter *, Hash, Grower, Allocator>;
+    using CounterMap = HASH_TABLE_WITH_STACK_MEMORY(HashMap,
+        TKey, Counter *, Hash, HashTableGrower<4>);
 
     CounterMap counter_map;
     std::vector<Counter *> counter_list;
diff --git a/src/Common/tests/pod_array.cpp b/src/Common/tests/pod_array.cpp
index de15b485411..6e9634ba3cf 100644
--- a/src/Common/tests/pod_array.cpp
+++ b/src/Common/tests/pod_array.cpp
@@ -18,9 +18,9 @@ static void test1()
 {
     using namespace DB;
 
-    static constexpr size_t initial_size = 8;
-    static constexpr size_t stack_threshold = 32;
-    using Array = PODArray<UInt64, initial_size, AllocatorWithStackMemory<Allocator<false>, stack_threshold>>;
+    static constexpr size_t initial_bytes = 32;
+    using Array = PODArray<UInt64, initial_bytes,
+        AllocatorWithStackMemory<Allocator<false>, initial_bytes>>;
 
     bool res = true;
 
@@ -139,9 +139,9 @@ static void test2()
 {
     using namespace DB;
 
-    static constexpr size_t initial_size = 8;
-    static constexpr size_t stack_threshold = 32;
-    using Array = PODArray<UInt64, initial_size, AllocatorWithStackMemory<Allocator<false>, stack_threshold>>;
+    static constexpr size_t initial_bytes = 32;
+    using Array = PODArray<UInt64, initial_bytes,
+        AllocatorWithStackMemory<Allocator<false>, initial_bytes>>;
 
     bool res = true;
 
@@ -389,9 +389,9 @@ static void test3()
 {
     using namespace DB;
 
-    static constexpr size_t initial_size = 8;
-    static constexpr size_t stack_threshold = 32;
-    using Array = PODArray<UInt64, initial_size, AllocatorWithStackMemory<Allocator<false>, stack_threshold>>;
+    static constexpr size_t initial_bytes = 32;
+    using Array = PODArray<UInt64, initial_bytes,
+        AllocatorWithStackMemory<Allocator<false>, initial_bytes>>;
 
     bool res = true;
 
diff --git a/src/Functions/array/arrayDistinct.cpp b/src/Functions/array/arrayDistinct.cpp
index d24de638865..d3da77e43d2 100644
--- a/src/Functions/array/arrayDistinct.cpp
+++ b/src/Functions/array/arrayDistinct.cpp
@@ -153,10 +153,8 @@ bool FunctionArrayDistinct::executeNumber(
     if (nullable_col)
         src_null_map = &nullable_col->getNullMapData();
 
-    using Set = ClearableHashSet<T,
-        DefaultHash<T>,
-        HashTableGrower<INITIAL_SIZE_DEGREE>,
-        HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(T)>>;
+    using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashSet,
+        T, DefaultHash<T>, HashTableGrower<INITIAL_SIZE_DEGREE>);
 
     Set set;
 
@@ -201,10 +199,8 @@ bool FunctionArrayDistinct::executeString(
 
     ColumnString & res_data_column_string = typeid_cast<ColumnString &>(res_data_col);
 
-    using Set = ClearableHashSet<StringRef,
-        StringRefHash,
-        HashTableGrower<INITIAL_SIZE_DEGREE>,
-        HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(StringRef)>>;
+    using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashSet,
+        StringRef, StringRefHash, HashTableGrower<INITIAL_SIZE_DEGREE>);
 
     const PaddedPODArray<UInt8> * src_null_map = nullptr;
 
@@ -249,8 +245,8 @@ void FunctionArrayDistinct::executeHashed(
     ColumnArray::Offsets & res_offsets,
     const ColumnNullable * nullable_col)
 {
-    using Set = ClearableHashSet<UInt128, UInt128TrivialHash, HashTableGrower<INITIAL_SIZE_DEGREE>,
-        HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(UInt128)>>;
+    using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashSet,
+        UInt128, UInt128TrivialHash, HashTableGrower<INITIAL_SIZE_DEGREE>);
 
     const PaddedPODArray<UInt8> * src_null_map = nullptr;
 
diff --git a/src/Functions/array/arrayEnumerateExtended.h b/src/Functions/array/arrayEnumerateExtended.h
index 69db59e698e..e97766c5827 100644
--- a/src/Functions/array/arrayEnumerateExtended.h
+++ b/src/Functions/array/arrayEnumerateExtended.h
@@ -64,36 +64,46 @@ private:
     template <typename T>
     struct MethodOneNumber
     {
-        using Set = ClearableHashMap<T, UInt32, DefaultHash<T>, HashTableGrower<INITIAL_SIZE_DEGREE>,
-                HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(T)>>;
+        using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
+            T, UInt32, DefaultHash<T>,
+            HashTableGrower<INITIAL_SIZE_DEGREE>);
+
         using Method = ColumnsHashing::HashMethodOneNumber<typename Set::value_type, UInt32, T, false>;
     };
 
     struct MethodString
     {
-        using Set = ClearableHashMap<StringRef, UInt32, StringRefHash, HashTableGrower<INITIAL_SIZE_DEGREE>,
-                HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(StringRef)>>;
+        using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
+            StringRef, UInt32, StringRefHash,
+            HashTableGrower<INITIAL_SIZE_DEGREE>);
+
         using Method = ColumnsHashing::HashMethodString<typename Set::value_type, UInt32, false, false>;
     };
 
     struct MethodFixedString
     {
-        using Set = ClearableHashMap<StringRef, UInt32, StringRefHash, HashTableGrower<INITIAL_SIZE_DEGREE>,
-                HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(StringRef)>>;
+        using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
+            StringRef, UInt32, StringRefHash,
+            HashTableGrower<INITIAL_SIZE_DEGREE>);
+
         using Method = ColumnsHashing::HashMethodFixedString<typename Set::value_type, UInt32, false, false>;
     };
 
     struct MethodFixed
     {
-        using Set = ClearableHashMap<UInt128, UInt32, UInt128HashCRC32, HashTableGrower<INITIAL_SIZE_DEGREE>,
-                HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(UInt128)>>;
+        using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
+            UInt128, UInt32, UInt128HashCRC32,
+            HashTableGrower<INITIAL_SIZE_DEGREE>);
+
         using Method = ColumnsHashing::HashMethodKeysFixed<typename Set::value_type, UInt128, UInt32, false, false, false>;
     };
 
     struct MethodHashed
     {
-        using Set =  ClearableHashMap<UInt128, UInt32, UInt128TrivialHash, HashTableGrower<INITIAL_SIZE_DEGREE>,
-                HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(UInt128)>>;
+        using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
+            UInt128, UInt32, UInt128TrivialHash,
+            HashTableGrower<INITIAL_SIZE_DEGREE>);
+
         using Method = ColumnsHashing::HashMethodHashed<typename Set::value_type, UInt32, false>;
     };
 
diff --git a/src/Functions/array/arrayEnumerateRanked.h b/src/Functions/array/arrayEnumerateRanked.h
index 133f5d7cb81..1700260dd9f 100644
--- a/src/Functions/array/arrayEnumerateRanked.h
+++ b/src/Functions/array/arrayEnumerateRanked.h
@@ -308,12 +308,10 @@ void FunctionArrayEnumerateRankedExtended<Derived>::executeMethodImpl(
     const size_t depth_to_look = arrays_depths.max_array_depth;
     const auto & offsets = *offsets_by_depth[depth_to_look - 1];
 
-    using Map = ClearableHashMap<
-        UInt128,
-        UInt32,
-        UInt128TrivialHash,
-        HashTableGrower<INITIAL_SIZE_DEGREE>,
-        HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(UInt128)>>;
+    using Map = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
+        UInt128, UInt32, UInt128TrivialHash,
+        HashTableGrower<INITIAL_SIZE_DEGREE>);
+
     Map indices;
 
     std::vector<size_t> indices_by_depth(depth_to_look);
diff --git a/src/Functions/array/arrayIntersect.cpp b/src/Functions/array/arrayIntersect.cpp
index cccad7c7a03..acd162974c9 100644
--- a/src/Functions/array/arrayIntersect.cpp
+++ b/src/Functions/array/arrayIntersect.cpp
@@ -418,16 +418,18 @@ void FunctionArrayIntersect::executeImpl(Block & block, const ColumnNumbers & ar
     TypeListNativeNumbers::forEach(NumberExecutor(arrays, not_nullable_nested_return_type, result_column));
     TypeListDecimalNumbers::forEach(DecimalExecutor(arrays, not_nullable_nested_return_type, result_column));
 
-    using DateMap = ClearableHashMap<DataTypeDate::FieldType, size_t, DefaultHash<DataTypeDate::FieldType>,
-            HashTableGrower<INITIAL_SIZE_DEGREE>,
-            HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(DataTypeDate::FieldType)>>;
+    using DateMap = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
+        DataTypeDate::FieldType, size_t, DefaultHash<DataTypeDate::FieldType>,
+        HashTableGrower<INITIAL_SIZE_DEGREE>);
 
-    using DateTimeMap = ClearableHashMap<DataTypeDateTime::FieldType, size_t, DefaultHash<DataTypeDateTime::FieldType>,
-            HashTableGrower<INITIAL_SIZE_DEGREE>,
-            HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(DataTypeDateTime::FieldType)>>;
+    using DateTimeMap = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
+        DataTypeDateTime::FieldType, size_t,
+        DefaultHash<DataTypeDateTime::FieldType>,
+        HashTableGrower<INITIAL_SIZE_DEGREE>);
 
-    using StringMap = ClearableHashMap<StringRef, size_t, StringRefHash, HashTableGrower<INITIAL_SIZE_DEGREE>,
-            HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(StringRef)>>;
+    using StringMap = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
+        StringRef, size_t, StringRefHash,
+        HashTableGrower<INITIAL_SIZE_DEGREE>);
 
     if (!result_column)
     {
@@ -455,8 +457,8 @@ void FunctionArrayIntersect::executeImpl(Block & block, const ColumnNumbers & ar
 template <typename T, size_t>
 void FunctionArrayIntersect::NumberExecutor::operator()()
 {
-    using Map = ClearableHashMap<T, size_t, DefaultHash<T>, HashTableGrower<INITIAL_SIZE_DEGREE>,
-            HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(T)>>;
+    using Map = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
+        T, size_t, DefaultHash<T>, HashTableGrower<INITIAL_SIZE_DEGREE>);
 
     if (!result && typeid_cast<const DataTypeNumber<T> *>(data_type.get()))
         result = execute<Map, ColumnVector<T>, true>(arrays, ColumnVector<T>::create());
@@ -465,8 +467,8 @@ void FunctionArrayIntersect::NumberExecutor::operator()()
 template <typename T, size_t>
 void FunctionArrayIntersect::DecimalExecutor::operator()()
 {
-    using Map = ClearableHashMap<T, size_t, DefaultHash<T>, HashTableGrower<INITIAL_SIZE_DEGREE>,
-            HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(T)>>;
+    using Map = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
+        T, size_t, DefaultHash<T>, HashTableGrower<INITIAL_SIZE_DEGREE>);
 
     if (!result)
         if (auto * decimal = typeid_cast<const DataTypeDecimal<T> *>(data_type.get()))
diff --git a/src/Functions/array/arrayUniq.cpp b/src/Functions/array/arrayUniq.cpp
index 02129781c13..a83c535d88d 100644
--- a/src/Functions/array/arrayUniq.cpp
+++ b/src/Functions/array/arrayUniq.cpp
@@ -66,36 +66,41 @@ private:
     template <typename T>
     struct MethodOneNumber
     {
-        using Set = ClearableHashSet<T, DefaultHash<T>, HashTableGrower<INITIAL_SIZE_DEGREE>,
-                HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(T)>>;
+        using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashSet,
+            T, DefaultHash<T>, HashTableGrower<INITIAL_SIZE_DEGREE>);
+
         using Method = ColumnsHashing::HashMethodOneNumber<typename Set::value_type, void, T, false>;
     };
 
     struct MethodString
     {
-        using Set = ClearableHashSet<StringRef, StringRefHash, HashTableGrower<INITIAL_SIZE_DEGREE>,
-                HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(StringRef)>>;
+        using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashSet,
+            StringRef, StringRefHash, HashTableGrower<INITIAL_SIZE_DEGREE>);
+
         using Method = ColumnsHashing::HashMethodString<typename Set::value_type, void, false, false>;
     };
 
     struct MethodFixedString
     {
-        using Set = ClearableHashSet<StringRef, StringRefHash, HashTableGrower<INITIAL_SIZE_DEGREE>,
-                HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(StringRef)>>;
+        using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashSet,
+            StringRef, StringRefHash, HashTableGrower<INITIAL_SIZE_DEGREE>);
+
         using Method = ColumnsHashing::HashMethodFixedString<typename Set::value_type, void, false, false>;
     };
 
     struct MethodFixed
     {
-        using Set = ClearableHashSet<UInt128, UInt128HashCRC32, HashTableGrower<INITIAL_SIZE_DEGREE>,
-                HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(UInt128)>>;
+        using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashSet,
+            UInt128, UInt128HashCRC32, HashTableGrower<INITIAL_SIZE_DEGREE>);
+
         using Method = ColumnsHashing::HashMethodKeysFixed<typename Set::value_type, UInt128, void, false, false, false>;
     };
 
     struct MethodHashed
     {
-        using Set = ClearableHashSet<UInt128, UInt128TrivialHash, HashTableGrower<INITIAL_SIZE_DEGREE>,
-                HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(UInt128)>>;
+        using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashSet,
+            UInt128, UInt128TrivialHash, HashTableGrower<INITIAL_SIZE_DEGREE>);
+
         using Method = ColumnsHashing::HashMethodHashed<typename Set::value_type, void, false>;
     };
 
diff --git a/src/Interpreters/tests/hash_map3.cpp b/src/Interpreters/tests/hash_map3.cpp
index 2207edc6cc1..804b084b2a1 100644
--- a/src/Interpreters/tests/hash_map3.cpp
+++ b/src/Interpreters/tests/hash_map3.cpp
@@ -61,12 +61,11 @@ struct Grower : public HashTableGrower<2>
 
 int main(int, char **)
 {
-    using Map = HashMapWithDump<
+    using Map = HASH_TABLE_WITH_STACK_MEMORY(HashMapWithDump,
         StringRef,
         UInt64,
         SimpleHash,
-        Grower,
-        HashTableAllocatorWithStackMemory<4 * 24>>;
+        Grower);
 
     Map map;
 

From 09ea7253cd0466fa29fe4f02414472aaa9cc72b5 Mon Sep 17 00:00:00 2001
From: Alexander Kuzmenkov <akuzm@yandex-team.ru>
Date: Wed, 20 May 2020 05:19:02 +0300
Subject: [PATCH 002/122] performance test for some functions that use hash
 tables

---
 tests/performance/functions_with_hash_tables.xml | 8 ++++++++
 1 file changed, 8 insertions(+)
 create mode 100644 tests/performance/functions_with_hash_tables.xml

diff --git a/tests/performance/functions_with_hash_tables.xml b/tests/performance/functions_with_hash_tables.xml
new file mode 100644
index 00000000000..3ff2633c4e7
--- /dev/null
+++ b/tests/performance/functions_with_hash_tables.xml
@@ -0,0 +1,8 @@
+<test>
+    <query>select arrayUniq(range(1 + (number % 10) * 10)) from numbers(1000000) format Null</query>
+    <query>select arrayDistinct(range(1 + (number % 10) * 10)) from numbers(1000000) format Null</query>
+    <query>select arrayEnumerateUniq(range(1 + (number % 10) * 10)) from numbers(1000000) format Null</query>
+    <query>select arrayIntersect(range((1 + number % 10)), range(1, (1 + number % 10) + 1)) from numbers(1000000) format Null</query>
+    <query>select groupUniqArray(number / 10) from numbers(1000 * 1000) group by number / 1000 format Null</query>
+    <query>select entropy(number / 10) from numbers(1000 * 1000) group by number / 1000 format Null</query>
+</test>

From 7ad8228854edecce8e20bd8a8db00ea97580d786 Mon Sep 17 00:00:00 2001
From: Alexander Kuzmenkov <akuzm@yandex-team.ru>
Date: Wed, 20 May 2020 05:19:19 +0300
Subject: [PATCH 003/122] [wip] Add memory stats to performance test

---
 docker/test/performance-comparison/compare.sh | 311 +++++++++++-------
 .../test/performance-comparison/entrypoint.sh |   2 +-
 docker/test/performance-comparison/eqmed.sql  |  41 ++-
 docker/test/performance-comparison/perf.py    |  23 +-
 docker/test/performance-comparison/report.py  |  45 +--
 5 files changed, 256 insertions(+), 166 deletions(-)

diff --git a/docker/test/performance-comparison/compare.sh b/docker/test/performance-comparison/compare.sh
index af88a3954d7..c748a09497d 100755
--- a/docker/test/performance-comparison/compare.sh
+++ b/docker/test/performance-comparison/compare.sh
@@ -133,7 +133,7 @@ function run_tests
     fi
 
     # Delete old report files.
-    for x in {test-times,skipped-tests,wall-clock-times,report-thresholds,client-times}.tsv
+    for x in {test-times,wall-clock-times}.tsv
     do
         rm -v "$x" ||:
         touch "$x"
@@ -220,66 +220,117 @@ function get_profiles
 # Build and analyze randomization distribution for all queries.
 function analyze_queries
 {
-rm -v analyze-commands.txt analyze-errors.log all-queries.tsv unstable-queries.tsv ./*-report.tsv raw-queries.tsv client-times.tsv report-thresholds.tsv ||:
+rm -v analyze-commands.txt analyze-errors.log all-queries.tsv unstable-queries.tsv ./*-report.tsv raw-queries.tsv ||:
+rm -rfv analyze ||:
+mkdir analyze ||:
+
+# FIXME This loop builds column definitons from TSVWithNamesAndTypes in an
+# absolutely atrocious way. This should be done by the file() function itself.
+for x in {right,left}-{addresses,{query,query-thread,trace,metric}-log}.tsv
+do
+    paste -d' ' \
+        <(sed -n '1{s/\t/\n/g;p;q}' "$x" | sed 's/\(^.*$\)/"\1"/') \
+        <(sed -n '2{s/\t/\n/g;p;q}' "$x" ) \
+        | tr '\n' ', ' | sed 's/,$//' > "$x.columns"
+done
 
 # Split the raw test output into files suitable for analysis.
 IFS=$'\n'
 for test_file in $(find . -maxdepth 1 -name "*-raw.tsv" -print)
 do
     test_name=$(basename "$test_file" "-raw.tsv")
-    sed -n "s/^query\t//p" < "$test_file" > "$test_name-queries.tsv"
-    sed -n "s/^client-time/$test_name/p" < "$test_file" >> "client-times.tsv"
-    sed -n "s/^report-threshold/$test_name/p" < "$test_file" >> "report-thresholds.tsv"
-    sed -n "s/^skipped/$test_name/p" < "$test_file" >> "skipped-tests.tsv"
+    sed -n "s/^query\t/$test_name\t/p" < "$test_file" >> "analyze/query-runs.tsv"
+    sed -n "s/^client-time/$test_name/p" < "$test_file" >> "analyze/client-times.tsv"
+    sed -n "s/^report-threshold/$test_name/p" < "$test_file" >> "analyze/report-thresholds.tsv"
+    sed -n "s/^skipped/$test_name/p" < "$test_file" >> "analyze/skipped-tests.tsv"
+    sed -n "s/^display-name/$test_name/p" < "$test_file" >> "analyze/query-display-names.tsv"
 done
 unset IFS
 
+# for each query run, prepare array of metrics from query log
+clickhouse-local --query "
+create view query_runs as select * from file('analyze/query-runs.tsv', TSV,
+    'test text, query_index int, query_id text, version UInt8, time float');
+
+create view left_query_log as select *
+    from file('left-query-log.tsv', TSVWithNamesAndTypes,
+        '$(cat "left-query-log.tsv.columns")');
+
+create view right_query_log as select *
+    from file('right-query-log.tsv', TSVWithNamesAndTypes,
+        '$(cat "right-query-log.tsv.columns")');
+
+create table query_metrics engine File(TSV, -- do not add header -- will parse with grep
+        'analyze/query-run-metrics.tsv')
+    as select
+        test, query_index, 0 run, version,
+        [
+            -- FIXME server-reported duration might be significantly less than
+            -- the client-reported one, when there is some slow cleanup that
+            -- happens after we reported the completion to the client (this did
+            -- happen with some queries).
+            -- To fix this, generate a unique query id for each run, that would
+            -- let us match the client and server times.
+            query_duration_ms / toFloat64(1000)
+            , toFloat64(memory_usage)
+            , query_runs.time
+        ] metrics
+    from (
+        select *, 0 version from left_query_log
+        union all
+        select *, 1 version from right_query_log
+    ) query_logs
+    right join query_runs
+    using (query_id, version)
+    ;
+"
+
 # This is a lateral join in bash... please forgive me.
-# We don't have arrayPermute(), so I have to make random permutations with 
+# We don't have arrayPermute(), so I have to make random permutations with
 # `order by rand`, and it becomes really slow if I do it for more than one
 # query. We also don't have lateral joins. So I just put all runs of each
 # query into a separate file, and then compute randomization distribution
 # for each file. I do this in parallel using GNU parallel.
+query_index=1
 IFS=$'\n'
-for test_file in $(find . -maxdepth 1 -name "*-queries.tsv" -print)
+for prefix in $(cut -f1,2 "analyze/query-run-metrics.tsv" | sort | uniq)
 do
-    test_name=$(basename "$test_file" "-queries.tsv")
-    query_index=1
-    for query in $(cut -d'	' -f1 "$test_file" | sort | uniq)
-    do
-        query_prefix="$test_name.q$query_index"
-        query_index=$((query_index + 1))
-        grep -F "$query	" "$test_file" > "$query_prefix.tmp"
-        printf "%s\0\n" \
-            "clickhouse-local \
-                --file \"$query_prefix.tmp\" \
-                --structure 'query text, run int, version UInt32, time float' \
-                --query \"$(cat "$script_dir/eqmed.sql")\" \
-                >> \"$test_name-report.tsv\"" \
-                2>> analyze-errors.log \
-            >> analyze-commands.txt
-    done
+    file="analyze/q$query_index.tmp"
+    grep -F "$prefix	" "analyze/query-run-metrics.tsv" > "$file" &
+    printf "%s\0\n" \
+        "clickhouse-local \
+            --file \"$file\" \
+            --structure 'test text, query text, run int, version UInt8, metrics Array(float)' \
+            --query \"$(cat "$script_dir/eqmed.sql")\" \
+            >> \"analyze/query-reports.tsv\"" \
+            2>> analyze/errors.log \
+        >> analyze/commands.txt
+
+    query_index=$((query_index + 1))
 done
 wait
 unset IFS
 
-parallel --null < analyze-commands.txt
+parallel --joblog analyze/parallel-log.txt --null < analyze/commands.txt
 }
 
 # Analyze results
 function report
 {
-
 rm -r report ||:
 mkdir report ||:
 
-
 rm ./*.{rep,svg} test-times.tsv test-dump.tsv unstable.tsv unstable-query-ids.tsv unstable-query-metrics.tsv changed-perf.tsv unstable-tests.tsv unstable-queries.tsv bad-tests.tsv slow-on-client.tsv all-queries.tsv ||:
 
-cat analyze-errors.log >> report/errors.log ||:
+cat analyze/errors.log >> report/errors.log ||:
 cat profile-errors.log >> report/errors.log ||:
 
 clickhouse-local --query "
+create view query_display_names as select * from
+    file('analyze/query-display-names.tsv', TSV,
+        'test text, query_index int, query_display_name text')
+    ;
+
 create table queries engine File(TSVWithNamesAndTypes, 'report/queries.tsv')
     as select
         -- FIXME Comparison mode doesn't make sense for queries that complete
@@ -296,21 +347,26 @@ create table queries engine File(TSVWithNamesAndTypes, 'report/queries.tsv')
         
         left, right, diff, stat_threshold,
         if(report_threshold > 0, report_threshold, 0.10) as report_threshold,
-        reports.test,
-        query
+        reports.test test, query_index, query_display_name
     from
         (
-            select *,
-                replaceAll(_file, '-report.tsv', '') test
-            from file('*-report.tsv', TSV, 'left float, right float, diff float, stat_threshold float, query text')
+            select left[1] as left, right[1] as right, diff[1] diff,
+                stat_threshold[1] stat_threshold, test, query_index
+            from file ('analyze/query-reports.tsv', TSV,
+                'left Array(float), right Array(float), diff Array(float),
+                    stat_threshold Array(float), test text, query_index int')
         ) reports
-        left join file('report-thresholds.tsv', TSV, 'test text, report_threshold float') thresholds
-        using test
+        left join file('analyze/report-thresholds.tsv', TSV, 'test text, report_threshold float') thresholds
+        on reports.test = thresholds.test
+        left join query_display_names
+        on reports.test = query_display_names.test
+            and reports.query_index = query_display_names.query_index
         ;
 
 -- keep the table in old format so that we can analyze new and old data together
 create table queries_old_format engine File(TSVWithNamesAndTypes, 'queries.rep')
-    as select short, changed_fail, unstable_fail, left, right, diff, stat_threshold, test, query
+    as select short, changed_fail, unstable_fail, left, right, diff,
+        stat_threshold, test, query_display_name query
     from queries
     ;
 
@@ -334,15 +390,16 @@ create table all_query_funs_json engine File(JSON, 'report/all-query-runs.json')
     ;
 
 create table changed_perf_tsv engine File(TSV, 'report/changed-perf.tsv') as
-    select left, right, diff, stat_threshold, changed_fail, test, query from queries where changed_show
-    order by abs(diff) desc;
+    select left, right, diff, stat_threshold, changed_fail, test, query_display_name
+    from queries where changed_show order by abs(diff) desc;
 
 create table unstable_queries_tsv engine File(TSV, 'report/unstable-queries.tsv') as
-    select left, right, diff, stat_threshold, unstable_fail, test, query from queries where unstable_show
-    order by stat_threshold desc;
+    select left, right, diff, stat_threshold, unstable_fail, test, query_display_name
+    from queries where unstable_show order by stat_threshold desc;
 
-create table queries_for_flamegraph engine File(TSVWithNamesAndTypes, 'report/queries-for-flamegraph.tsv') as
-    select query, test from queries where unstable_show or changed_show
+create table queries_for_flamegraph engine File(TSVWithNamesAndTypes,
+        'report/queries-for-flamegraph.tsv') as
+    select test, query_index from queries where unstable_show or changed_show
     ;
 
 create table unstable_tests_tsv engine File(TSV, 'report/bad-tests.tsv') as
@@ -350,23 +407,23 @@ create table unstable_tests_tsv engine File(TSV, 'report/bad-tests.tsv') as
     group by test having s > 0 order by s desc;
 
 create table query_time engine Memory as select *
-    from file('client-times.tsv', TSV, 'test text, query text, client float, server float');
+    from file('analyze/client-times.tsv', TSV,
+        'test text, query_index int, client float, server float');
 
 create table wall_clock engine Memory as select *
     from file('wall-clock-times.tsv', TSV, 'test text, real float, user float, system float');
 
 create table slow_on_client_tsv engine File(TSV, 'report/slow-on-client.tsv') as
-    select client, server, floor(client/server, 3) p, query
-    from query_time where p > 1.02 order by p desc;
+    select client, server, floor(client/server, 3) p, query_display_name
+    from query_time left join query_display_names using (test, query_index)
+    where p > 1.02 order by p desc;
 
 create table test_time engine Memory as
     select test, sum(client) total_client_time,
         maxIf(client, not short) query_max,
         minIf(client, not short) query_min,
-        count(*) queries,
-        sum(short) short_queries
-    from query_time full join queries
-    using test, query
+        count(*) queries, sum(short) short_queries
+    from query_time full join queries using (test, query_index)
     group by test;
 
 create table test_times_tsv engine File(TSV, 'report/test-times.tsv') as
@@ -378,40 +435,80 @@ create table test_times_tsv engine File(TSV, 'report/test-times.tsv') as
         floor(real / queries, 3) avg_real_per_query,
         floor(query_min, 3)
     from test_time
-        -- wall clock times are also measured for skipped tests, so don't
-        -- do full join
-        left join wall_clock using test
+    -- wall clock times are also measured for skipped tests, so don't
+    -- do full join
+    left join wall_clock using test
     order by avg_real_per_query desc;
 
 create table all_tests_tsv engine File(TSV, 'report/all-queries.tsv') as
     select changed_fail, unstable_fail,
         left, right, diff,
         floor(left > right ? left / right : right / left, 3),
-        stat_threshold, test, query
-    from queries order by test, query;
+        stat_threshold, test, query_display_name
+    from queries order by test, query_display_name;
 " 2> >(tee -a report/errors.log 1>&2)
 
-for x in {right,left}-{addresses,{query,query-thread,trace,metric}-log}.tsv
-do
-    # FIXME This loop builds column definitons from TSVWithNamesAndTypes in an
-    # absolutely atrocious way. This should be done by the file() function itself.
-    paste -d' ' \
-        <(sed -n '1{s/\t/\n/g;p;q}' "$x" | sed 's/\(^.*$\)/"\1"/') \
-        <(sed -n '2{s/\t/\n/g;p;q}' "$x" ) \
-        | tr '\n' ', ' | sed 's/,$//' > "$x.columns"
-done
-
+# Prepare source data for metrics and flamegraphs for unstable queries.
 for version in {right,left}
-do
-clickhouse-local --query "
+    do
+    rm -rf data
+    clickhouse-local --query "
 create view queries_for_flamegraph as
     select * from file('report/queries-for-flamegraph.tsv', TSVWithNamesAndTypes,
-        'query text, test text');
+        'test text, query_index int');
+
+create view query_runs as
+    with 0 as left, 1 as right
+    select * from file('analyze/query-runs.tsv', TSV,
+        'test text, query_index int, query_id text, version UInt8, time float')
+    where version = $version
+    ;
+
+create view query_display_names as select * from
+    file('analyze/query-display-names.tsv', TSV,
+        'test text, query_index int, query_display_name text')
+    ;
+
+create table unstable_query_runs engine File(TSVWithNamesAndTypes,
+        'unstable-query-runs.$version.rep') as
+    select test, query_index, query_display_name, query_id
+    from query_runs
+    join queries_for_flamegraph on
+        query_runs.test = queries_for_flamegraph.test
+        and query_runs.query_index = queries_for_flamegraph.query_index
+    left join query_display_names on
+        query_runs.test = query_display_names.test
+        and query_runs.query_index = query_display_names.query_index
+    ;
 
 create view query_log as select *
     from file('$version-query-log.tsv', TSVWithNamesAndTypes,
         '$(cat "$version-query-log.tsv.columns")');
 
+create table unstable_run_metrics engine File(TSVWithNamesAndTypes,
+        'unstable-run-metrics.$version.rep') as
+    select
+        test, query_index, query_id,
+        ProfileEvents.Values value, ProfileEvents.Names metric
+    from query_log array join ProfileEvents
+    join unstable_query_runs using (query_id)
+    ;
+
+create table unstable_run_metrics_2 engine File(TSVWithNamesAndTypes,
+        'unstable-run-metrics-2.$version.rep') as
+    select
+        test, query_index, query_id,
+        v, n
+    from (
+        select
+            test, query_index, query_id,
+            ['memory_usage', 'read_bytes', 'written_bytes', 'query_duration_ms'] n,
+            [memory_usage, read_bytes, written_bytes, query_duration_ms] v
+        from query_log
+        join unstable_query_runs using (query_id)
+    )
+    array join v, n;
+
 create view trace_log as select *
     from file('$version-trace-log.tsv', TSVWithNamesAndTypes,
         '$(cat "$version-trace-log.tsv.columns")');
@@ -423,88 +520,64 @@ create view addresses_src as select *
 create table addresses_join_$version engine Join(any, left, address) as
     select addr address, name from addresses_src;
 
-create table unstable_query_runs engine File(TSVWithNamesAndTypes,
-        'unstable-query-runs.$version.rep') as
-    select query, query_id from query_log
-    where query in (select query from queries_for_flamegraph)
-        and query_id not like 'prewarm %'
-    ;
-
-create table unstable_query_log engine File(Vertical,
-        'unstable-query-log.$version.rep') as
-    select * from query_log
-    where query_id in (select query_id from unstable_query_runs);
-
-create table unstable_run_metrics engine File(TSVWithNamesAndTypes,
-        'unstable-run-metrics.$version.rep') as
-    select ProfileEvents.Values value, ProfileEvents.Names metric, query_id, query
-    from query_log array join ProfileEvents
-    where query_id in (select query_id from unstable_query_runs)
-    ;
-
-create table unstable_run_metrics_2 engine File(TSVWithNamesAndTypes,
-        'unstable-run-metrics-2.$version.rep') as
-    select v, n, query_id, query
-    from
-        (select
-            ['memory_usage', 'read_bytes', 'written_bytes', 'query_duration_ms'] n,
-            [memory_usage, read_bytes, written_bytes, query_duration_ms] v,
-            query,
-            query_id
-        from query_log
-        where query_id in (select query_id from unstable_query_runs))
-    array join n, v;
-
 create table unstable_run_traces engine File(TSVWithNamesAndTypes,
         'unstable-run-traces.$version.rep') as
     select
+        test, query_index, query_id,
         count() value,
-        joinGet(addresses_join_$version, 'name', arrayJoin(trace)) metric,
-        unstable_query_runs.query_id,
-        any(unstable_query_runs.query) query
-    from unstable_query_runs
-    join trace_log on trace_log.query_id = unstable_query_runs.query_id
-    group by unstable_query_runs.query_id, metric
+        joinGet(addresses_join_$version, 'name', arrayJoin(trace)) metric
+    from trace_log
+    join unstable_query_runs using query_id
+    group by test, query_index, query_id, metric
     order by count() desc
     ;
 
 create table metric_devation engine File(TSVWithNamesAndTypes,
         'metric-deviation.$version.rep') as
-    select query, floor((q[3] - q[1])/q[2], 3) d,
-        quantilesExact(0, 0.5, 1)(value) q, metric
-    from (select * from unstable_run_metrics
-        union all select * from unstable_run_traces
-        union all select * from unstable_run_metrics_2) mm
-    join queries_for_flamegraph using query
-    group by query, metric
-    having d > 0.5
-    order by query desc, d desc
+    -- first goes the key used to split the file with grep
+    select test, query_index, query_display_name,
+        d, q, metric
+    from (
+        select
+            test, query_index,
+            floor((q[3] - q[1])/q[2], 3) d,
+            quantilesExact(0, 0.5, 1)(value) q, metric
+        from (select * from unstable_run_metrics
+            union all select * from unstable_run_traces
+            union all select * from unstable_run_metrics_2) mm
+        group by test, query_index, metric
+        having d > 0.5
+    ) metrics
+    left join unstable_query_runs using (test, query_index)
+    order by test, query_index, d desc
     ;
 
 create table stacks engine File(TSV, 'stacks.$version.rep') as
     select
-        query,
+        -- first goes the key used to split the file with grep
+        test, query_index, any(query_display_name),
         arrayStringConcat(
             arrayMap(x -> joinGet(addresses_join_$version, 'name', x),
                 arrayReverse(trace)
             ),
             ';'
         ) readable_trace,
-        count()
+        count() c
     from trace_log
     join unstable_query_runs using query_id
-    group by query, trace
+    group by test, query_index, trace
     ;
 " 2> >(tee -a report/errors.log 1>&2) # do not run in parallel because they use the same data dir for StorageJoins which leads to weird errors.
 done
 wait
 
+# Create per-query flamegraphs and files with metrics
 IFS=$'\n'
 for version in {right,left}
 do
-    for query in $(cut -d'	' -f1 "stacks.$version.rep" | sort | uniq)
+    for query in $(cut -d'	' -f1,2,3 "stacks.$version.rep" | sort | uniq)
     do
-        query_file=$(echo "$query" | cut -c-120 | sed 's/[/]/_/g')
+        query_file=$(echo "$query" | cut -c-120 | sed 's/[/	]/_/g')
 
         # Build separate .svg flamegraph for each query.
         grep -F "$query	" "stacks.$version.rep" \
diff --git a/docker/test/performance-comparison/entrypoint.sh b/docker/test/performance-comparison/entrypoint.sh
index 500dd872c8d..3b8a24ad5da 100755
--- a/docker/test/performance-comparison/entrypoint.sh
+++ b/docker/test/performance-comparison/entrypoint.sh
@@ -119,5 +119,5 @@ done
 
 dmesg -T > dmesg.log
 
-7z a /output/output.7z ./*.{log,tsv,html,txt,rep,svg} {right,left}/{performance,db/preprocessed_configs,scripts} ./report
+7z a /output/output.7z ./*.{log,tsv,html,txt,rep,svg} {right,left}/{performance,db/preprocessed_configs,scripts} report analyze
 cp compare.log /output
diff --git a/docker/test/performance-comparison/eqmed.sql b/docker/test/performance-comparison/eqmed.sql
index 4f8482ea74b..f7f8d6ac40d 100644
--- a/docker/test/performance-comparison/eqmed.sql
+++ b/docker/test/performance-comparison/eqmed.sql
@@ -1,32 +1,37 @@
--- input is table(query text, run UInt32, version int, time float)
+-- input is table(test text, query text, run UInt32, version int, metrics Array(float))
 select
-   floor(original_medians_array.time_by_version[1], 4) l,
-   floor(original_medians_array.time_by_version[2], 4) r,
-   floor((r - l) / l, 3) diff_percent,
-   floor(threshold / l, 3) threshold_percent,
-   query
+   arrayMap(x -> floor(x, 4), original_medians_array.medians_by_version[1] as l) l_rounded,
+   arrayMap(x -> floor(x, 4), original_medians_array.medians_by_version[2] as r) r_rounded,
+   arrayMap(x, y -> floor((y - x) / x, 3), l, r) diff_percent,
+   arrayMap(x, y -> floor(x / y, 3), threshold, l) threshold_percent,
+   test, query
 from
    (
       -- quantiles of randomization distributions
-      select quantileExact(0.999)(abs(time_by_label[1] - time_by_label[2]) as d) threshold
+      select quantileExactForEach(0.999)(
+        arrayMap(x, y -> abs(x - y), metrics_by_label[1], metrics_by_label[2]) as d
+      ) threshold
       ---- uncomment to see what the distribution is really like
-      --, uniqExact(d) u
+      --, uniqExact(d.1) u
       --, arraySort(x->x.1,
       --      arrayZip(
-      --          (sumMap([d], [1]) as f).1,
+      --          (sumMap([d.1], [1]) as f).1,
       --          f.2)) full_histogram
       from
          (
-            select virtual_run, groupArrayInsertAt(median_time, random_label) time_by_label -- make array 'random label' -> 'median time'
+            -- make array 'random label' -> '[median metric]'
+            select virtual_run, groupArrayInsertAt(median_metrics, random_label) metrics_by_label
             from (
-                  select medianExact(time) median_time, virtual_run, random_label -- get median times, grouping by random label
+                  -- get [median metric] arrays among virtual runs, grouping by random label
+                  select medianExactForEach(metrics) median_metrics, virtual_run, random_label
                   from (
-                        select *, toUInt32(rowNumberInAllBlocks() % 2) random_label -- randomly relabel measurements
+                        -- randomly relabel measurements
+                        select *, toUInt32(rowNumberInAllBlocks() % 2) random_label
                         from (
-                              select time, number virtual_run 
+                              select metrics, number virtual_run
                               from
                                 -- strip the query away before the join -- it might be several kB long;
-                                (select time, run, version from table) no_query,
+                                (select metrics, run, version from table) no_query,
                                 -- duplicate input measurements into many virtual runs
                                 numbers(1, 100000) nn
                               -- for each virtual run, randomly reorder measurements
@@ -40,19 +45,19 @@ from
       -- this select aggregates by virtual_run
    ) rd,
    (
-        select groupArrayInsertAt(median_time, version) time_by_version
+        select groupArrayInsertAt(median_metrics, version) medians_by_version
         from
         (
-            select medianExact(time) median_time, version
+            select medianExactForEach(metrics) median_metrics, version
             from table
             group by version
         ) original_medians
    ) original_medians_array,
    (
-        select any(query) query from table
+        select any(test) test, any(query) query from table
    ) any_query,
    (
-       select throwIf(uniq(query) != 1) from table
+       select throwIf(uniq((test, query)) != 1) from table
    ) check_single_query -- this subselect checks that there is only one query in the input table;
                         -- written this way so that it is not optimized away (#10523)
 ;
diff --git a/docker/test/performance-comparison/perf.py b/docker/test/performance-comparison/perf.py
index a749b158e83..be8155cdddb 100755
--- a/docker/test/performance-comparison/perf.py
+++ b/docker/test/performance-comparison/perf.py
@@ -29,6 +29,8 @@ parser.add_argument('--runs', type=int, default=int(os.environ.get('CHPC_RUNS',
 parser.add_argument('--no-long', type=bool, default=True, help='Skip the tests tagged as long.')
 args = parser.parse_args()
 
+test_name = os.path.splitext(os.path.basename(args.file[0].name))[0]
+
 tree = et.parse(args.file[0])
 root = tree.getroot()
 
@@ -141,19 +143,25 @@ test_queries = substitute_parameters(test_query_templates)
 
 report_stage_end('substitute2')
 
-for i, q in enumerate(test_queries):
+for query_index, q in enumerate(test_queries):
+    query_prefix = f'{test_name}.query{query_index}'
+
     # We have some crazy long queries (about 100kB), so trim them to a sane
-    # length.
+    # length. This means we can't use query text as an identifier and have to
+    # use the test name + the test-wide query index.
     query_display_name = q
     if len(query_display_name) > 1000:
         query_display_name = f'{query_display_name[:1000]}...({i})'
 
+    print(f'display-name\t{query_index}\t{tsv_escape(query_display_name)}')
+
     # Prewarm: run once on both servers. Helps to bring the data into memory,
     # precompile the queries, etc.
     try:
         for conn_index, c in enumerate(connections):
-            res = c.execute(q, query_id = f'prewarm {0} {query_display_name}')
-            print(f'prewarm\t{tsv_escape(query_display_name)}\t{conn_index}\t{c.last_query.elapsed}')
+            prewarm_id = f'{query_prefix}.prewarm0'
+            res = c.execute(q, query_id = prewarm_id)
+            print(f'prewarm\t{query_index}\t{prewarm_id}\t{conn_index}\t{c.last_query.elapsed}')
     except:
         # If prewarm fails for some query -- skip it, and try to test the others.
         # This might happen if the new test introduces some function that the
@@ -170,13 +178,14 @@ for i, q in enumerate(test_queries):
     start_seconds = time.perf_counter()
     server_seconds = 0
     for run in range(0, args.runs):
+        run_id = f'{query_prefix}.run{run}'
         for conn_index, c in enumerate(connections):
-            res = c.execute(q)
-            print(f'query\t{tsv_escape(query_display_name)}\t{run}\t{conn_index}\t{c.last_query.elapsed}')
+            res = c.execute(q, query_id = run_id)
+            print(f'query\t{query_index}\t{run_id}\t{conn_index}\t{c.last_query.elapsed}')
             server_seconds += c.last_query.elapsed
 
     client_seconds = time.perf_counter() - start_seconds
-    print(f'client-time\t{tsv_escape(query_display_name)}\t{client_seconds}\t{server_seconds}')
+    print(f'client-time\t{query_index}\t{client_seconds}\t{server_seconds}')
 
 report_stage_end('benchmark')
 
diff --git a/docker/test/performance-comparison/report.py b/docker/test/performance-comparison/report.py
index a6203ec40e6..c6b743a606e 100755
--- a/docker/test/performance-comparison/report.py
+++ b/docker/test/performance-comparison/report.py
@@ -25,6 +25,9 @@ very_unstable_queries = 0
 # max seconds to run one query by itself, not counting preparation
 allowed_single_run_time = 2
 
+color_bad='#ffb0c0'
+color_good='#b0d050'
+
 header_template = """
 <!DOCTYPE html>
 <html>
@@ -188,8 +191,8 @@ if args.report == 'main':
 
         print(tableStart('Changes in performance'))
         columns = [
-            'Old, s.',                                         # 0
-            'New, s.',                                         # 1
+            'Old, s',                                          # 0
+            'New, s',                                          # 1
             'Relative difference (new&nbsp;&minus;&nbsp;old) / old',   # 2
             'p&nbsp;<&nbsp;0.001 threshold',                   # 3
             # Failed                                           # 4
@@ -221,7 +224,7 @@ if args.report == 'main':
     slow_on_client_rows = tsvRows('report/slow-on-client.tsv')
     error_tests += len(slow_on_client_rows)
     printSimpleTable('Slow on client',
-        ['Client time, s.', 'Server time, s.', 'Ratio', 'Query'],
+        ['Client time, s', 'Server time, s', 'Ratio', 'Query'],
         slow_on_client_rows)
 
     def print_unstable_queries():
@@ -252,7 +255,7 @@ if args.report == 'main':
         for r in unstable_rows:
             if int(r[4]):
                 very_unstable_queries += 1
-                attrs[3] = 'style="background: #ffb0a0"'
+                attrs[3] = f'style="background: {color_bad}"'
             else:
                 attrs[3] = ''
 
@@ -266,7 +269,7 @@ if args.report == 'main':
     error_tests += len(run_error_rows)
     printSimpleTable('Run errors', ['Test', 'Error'], run_error_rows)
 
-    skipped_tests_rows = tsvRows('skipped-tests.tsv')
+    skipped_tests_rows = tsvRows('analyze/skipped-tests.tsv')
     printSimpleTable('Skipped tests', ['Test', 'Reason'], skipped_tests_rows)
 
     printSimpleTable('Tests with most unstable queries',
@@ -281,13 +284,13 @@ if args.report == 'main':
 
         columns = [
             'Test',                                          #0
-            'Wall clock time, s.',                           #1
-            'Total client time, s.',                         #2
+            'Wall clock time, s',                            #1
+            'Total client time, s',                          #2
             'Total queries',                                 #3
             'Ignored short queries',                         #4
-            'Longest query<br>(sum for all runs), s.',       #5
-            'Avg wall clock time<br>(sum for all runs), s.', #6
-            'Shortest query<br>(sum for all runs), s.',      #7
+            'Longest query<br>(sum for all runs), s',        #5
+            'Avg wall clock time<br>(sum for all runs), s',  #6
+            'Shortest query<br>(sum for all runs), s',       #7
             ]
 
         print(tableStart('Test times'))
@@ -300,13 +303,13 @@ if args.report == 'main':
             if float(r[6]) > 1.5 * total_runs:
                 # FIXME should be 15s max -- investigate parallel_insert
                 slow_average_tests += 1
-                attrs[6] = 'style="background: #ffb0a0"'
+                attrs[6] = f'style="background: {color_bad}"'
             else:
                 attrs[6] = ''
 
             if float(r[5]) > allowed_single_run_time * total_runs:
                 slow_average_tests += 1
-                attrs[5] = 'style="background: #ffb0a0"'
+                attrs[5] = f'style="background: {color_bad}"'
             else:
                 attrs[5] = ''
 
@@ -320,9 +323,9 @@ if args.report == 'main':
 
     print("""
     <p class="links">
-    <a href="output.7z">Test output</a>
     <a href="all-queries.html">All queries</a>
     <a href="compare.log">Log</a>
+    <a href="output.7z">Test output</a>
     </p>
     </body>
     </html>
@@ -382,8 +385,8 @@ elif args.report == 'all-queries':
         columns = [
             # Changed #0
             # Unstable #1
-            'Old, s.', #2
-            'New, s.', #3
+            'Old, s', #2
+            'New, s', #3
             'Relative difference (new&nbsp;&minus;&nbsp;old) / old', #4
             'Times speedup / slowdown',                 #5
             'p&nbsp;<&nbsp;0.001 threshold',          #6
@@ -399,21 +402,21 @@ elif args.report == 'all-queries':
         attrs[1] = None
         for r in rows:
             if int(r[1]):
-                attrs[6] = 'style="background: #ffb0a0"'
+                attrs[6] = f'style="background: {color_bad}"'
             else:
                 attrs[6] = ''
 
             if int(r[0]):
                 if float(r[4]) > 0.:
-                    attrs[4] = 'style="background: #ffb0a0"'
+                    attrs[4] = f'style="background: {color_bad}"'
                 else:
-                    attrs[4] = 'style="background: #adbdff"'
+                    attrs[4] = f'style="background: {color_good}"'
             else:
                 attrs[4] = ''
 
             if (float(r[2]) + float(r[3])) / 2 > allowed_single_run_time:
-                attrs[2] = 'style="background: #ffb0a0"'
-                attrs[3] = 'style="background: #ffb0a0"'
+                attrs[2] = f'style="background: {color_bad}"'
+                attrs[3] = f'style="background: {color_bad}"'
             else:
                 attrs[2] = ''
                 attrs[3] = ''
@@ -428,9 +431,9 @@ elif args.report == 'all-queries':
 
     print("""
     <p class="links">
-    <a href="output.7z">Test output</a>
     <a href="report.html">Main report</a>
     <a href="compare.log">Log</a>
+    <a href="output.7z">Test output</a>
     </p>
     </body>
     </html>

From e11fa03bddeb637255d19da1a2fbe5bf6f4a6a0e Mon Sep 17 00:00:00 2001
From: hcz <h.cz@qq.com>
Date: Thu, 21 May 2020 12:07:47 +0800
Subject: [PATCH 004/122] Add ArrowStream input and output format

---
 .../Formats/Impl/ArrowBlockInputFormat.cpp    | 70 ++++++++++++++-----
 .../Formats/Impl/ArrowBlockInputFormat.h      | 10 +--
 .../Formats/Impl/ArrowBlockOutputFormat.cpp   | 37 +++++++---
 .../Formats/Impl/ArrowBlockOutputFormat.h     |  6 +-
 4 files changed, 91 insertions(+), 32 deletions(-)

diff --git a/src/Processors/Formats/Impl/ArrowBlockInputFormat.cpp b/src/Processors/Formats/Impl/ArrowBlockInputFormat.cpp
index df73c93fd55..c8a098418b7 100644
--- a/src/Processors/Formats/Impl/ArrowBlockInputFormat.cpp
+++ b/src/Processors/Formats/Impl/ArrowBlockInputFormat.cpp
@@ -16,12 +16,12 @@ namespace DB
 
 namespace ErrorCodes
 {
-    extern const int BAD_ARGUMENTS;
+    extern const int UNKNOWN_EXCEPTION;
     extern const int CANNOT_READ_ALL_DATA;
 }
 
-ArrowBlockInputFormat::ArrowBlockInputFormat(ReadBuffer & in_, const Block & header_)
-    : IInputFormat(header_, in_)
+ArrowBlockInputFormat::ArrowBlockInputFormat(ReadBuffer & in_, const Block & header_, bool stream_)
+    : IInputFormat(header_, in_), stream{stream_}
 {
     prepareReader();
 }
@@ -31,11 +31,21 @@ Chunk ArrowBlockInputFormat::generate()
     Chunk res;
     const Block & header = getPort().getHeader();
 
-    if (record_batch_current >= record_batch_total)
+    if (!stream && record_batch_current >= record_batch_total)
         return res;
 
     std::vector<std::shared_ptr<arrow::RecordBatch>> single_batch(1);
-    arrow::Status read_status = file_reader->ReadRecordBatch(record_batch_current, &single_batch[0]);
+    arrow::Status read_status;
+    if (stream)
+    {
+        read_status = stream_reader->ReadNext(&single_batch[0]);
+    }
+    else
+    {
+        read_status = file_reader->ReadRecordBatch(record_batch_current, &single_batch[0]);
+        if (!single_batch[0])
+            return res;
+    }
     if (!read_status.ok())
         throw Exception{"Error while reading batch of Arrow data: " + read_status.ToString(),
                         ErrorCodes::CANNOT_READ_ALL_DATA};
@@ -57,30 +67,54 @@ void ArrowBlockInputFormat::resetParser()
 {
     IInputFormat::resetParser();
 
-    file_reader.reset();
+    if (stream)
+        stream_reader.reset();
+    else
+        file_reader.reset();
     prepareReader();
 }
 
 void ArrowBlockInputFormat::prepareReader()
 {
-    arrow::Status open_status = arrow::ipc::RecordBatchFileReader::Open(asArrowFile(in), &file_reader);
-    if (!open_status.ok())
-        throw Exception(open_status.ToString(), ErrorCodes::BAD_ARGUMENTS);
-    record_batch_total = file_reader->num_record_batches();
+    arrow::Status status;
+
+    if (stream)
+        status = arrow::ipc::RecordBatchStreamReader::Open(asArrowFile(in), &stream_reader);
+    else
+        status = arrow::ipc::RecordBatchFileReader::Open(asArrowFile(in), &file_reader);
+
+    if (!status.ok())
+        throw Exception{"Error while opening a table: " + status.ToString(), ErrorCodes::UNKNOWN_EXCEPTION};
+
+    if (stream)
+        record_batch_total = -1;
+    else
+        record_batch_total = file_reader->num_record_batches();
+
     record_batch_current = 0;
 }
 
 void registerInputFormatProcessorArrow(FormatFactory &factory)
 {
     factory.registerInputFormatProcessor(
-            "Arrow",
-            [](ReadBuffer & buf,
-               const Block & sample,
-               const RowInputFormatParams & /* params */,
-               const FormatSettings & /* format_settings */)
-            {
-                return std::make_shared<ArrowBlockInputFormat>(buf, sample);
-            });
+        "Arrow",
+        [](ReadBuffer & buf,
+           const Block & sample,
+           const RowInputFormatParams & /* params */,
+           const FormatSettings & /* format_settings */)
+        {
+            return std::make_shared<ArrowBlockInputFormat>(buf, sample, false);
+        });
+
+    factory.registerInputFormatProcessor(
+        "ArrowStream",
+        [](ReadBuffer & buf,
+           const Block & sample,
+           const RowInputFormatParams & /* params */,
+           const FormatSettings & /* format_settings */)
+        {
+            return std::make_shared<ArrowBlockInputFormat>(buf, sample, true);
+        });
 }
 
 }
diff --git a/src/Processors/Formats/Impl/ArrowBlockInputFormat.h b/src/Processors/Formats/Impl/ArrowBlockInputFormat.h
index 28ca033dd13..2e7f647b4f7 100644
--- a/src/Processors/Formats/Impl/ArrowBlockInputFormat.h
+++ b/src/Processors/Formats/Impl/ArrowBlockInputFormat.h
@@ -4,6 +4,7 @@
 
 #include <Processors/Formats/IInputFormat.h>
 
+namespace arrow { class RecordBatchReader; }
 namespace arrow::ipc { class RecordBatchFileReader; }
 
 namespace DB
@@ -14,7 +15,7 @@ class ReadBuffer;
 class ArrowBlockInputFormat : public IInputFormat
 {
 public:
-    ArrowBlockInputFormat(ReadBuffer & in_, const Block & header_);
+    ArrowBlockInputFormat(ReadBuffer & in_, const Block & header_, bool stream_);
 
     void resetParser() override;
 
@@ -24,12 +25,13 @@ protected:
     Chunk generate() override;
 
 private:
-    void prepareReader();
-
-private:
+    bool stream;
+    std::shared_ptr<arrow::RecordBatchReader> stream_reader;
     std::shared_ptr<arrow::ipc::RecordBatchFileReader> file_reader;
     int record_batch_total = 0;
     int record_batch_current = 0;
+
+    void prepareReader();
 };
 
 }
diff --git a/src/Processors/Formats/Impl/ArrowBlockOutputFormat.cpp b/src/Processors/Formats/Impl/ArrowBlockOutputFormat.cpp
index ed33eb08728..ef6bd315319 100644
--- a/src/Processors/Formats/Impl/ArrowBlockOutputFormat.cpp
+++ b/src/Processors/Formats/Impl/ArrowBlockOutputFormat.cpp
@@ -15,8 +15,8 @@ namespace ErrorCodes
     extern const int UNKNOWN_EXCEPTION;
 }
 
-ArrowBlockOutputFormat::ArrowBlockOutputFormat(WriteBuffer & out_, const Block & header_, const FormatSettings & format_settings_)
-    : IOutputFormat(header_, out_), format_settings{format_settings_}, arrow_ostream{std::make_shared<ArrowBufferedOutputStream>(out_)}
+ArrowBlockOutputFormat::ArrowBlockOutputFormat(WriteBuffer & out_, const Block & header_, bool stream_, const FormatSettings & format_settings_)
+    : IOutputFormat(header_, out_), stream{stream_}, format_settings{format_settings_}, arrow_ostream{std::make_shared<ArrowBufferedOutputStream>(out_)}
 {
 }
 
@@ -29,12 +29,7 @@ void ArrowBlockOutputFormat::consume(Chunk chunk)
     CHColumnToArrowColumn::chChunkToArrowTable(arrow_table, header, chunk, columns_num, "Arrow");
 
     if (!writer)
-    {
-        // TODO: should we use arrow::ipc::IpcOptions::alignment?
-        auto status = arrow::ipc::RecordBatchFileWriter::Open(arrow_ostream.get(), arrow_table->schema(), &writer);
-        if (!status.ok())
-            throw Exception{"Error while opening a table: " + status.ToString(), ErrorCodes::UNKNOWN_EXCEPTION};
-    }
+        prepareWriter(arrow_table->schema());
 
     // TODO: calculate row_group_size depending on a number of rows and table size
     auto status = writer->WriteTable(*arrow_table, format_settings.arrow.row_group_size);
@@ -53,6 +48,20 @@ void ArrowBlockOutputFormat::finalize()
     }
 }
 
+void ArrowBlockOutputFormat::prepareWriter(const std::shared_ptr<arrow::Schema> & schema)
+{
+    arrow::Status status;
+
+    // TODO: should we use arrow::ipc::IpcOptions::alignment?
+    if (stream)
+        status = arrow::ipc::RecordBatchStreamWriter::Open(arrow_ostream.get(), schema, &writer);
+    else
+        status = arrow::ipc::RecordBatchFileWriter::Open(arrow_ostream.get(), schema, &writer);
+
+    if (!status.ok())
+        throw Exception{"Error while opening a table writer: " + status.ToString(), ErrorCodes::UNKNOWN_EXCEPTION};
+}
+
 void registerOutputFormatProcessorArrow(FormatFactory & factory)
 {
     factory.registerOutputFormatProcessor(
@@ -62,7 +71,17 @@ void registerOutputFormatProcessorArrow(FormatFactory & factory)
            FormatFactory::WriteCallback,
            const FormatSettings & format_settings)
         {
-            return std::make_shared<ArrowBlockOutputFormat>(buf, sample, format_settings);
+            return std::make_shared<ArrowBlockOutputFormat>(buf, sample, false, format_settings);
+        });
+
+    factory.registerOutputFormatProcessor(
+        "ArrowStream",
+        [](WriteBuffer & buf,
+           const Block & sample,
+           FormatFactory::WriteCallback,
+           const FormatSettings & format_settings)
+        {
+            return std::make_shared<ArrowBlockOutputFormat>(buf, sample, true, format_settings);
         });
 }
 
diff --git a/src/Processors/Formats/Impl/ArrowBlockOutputFormat.h b/src/Processors/Formats/Impl/ArrowBlockOutputFormat.h
index 9ea84f83a0d..0cc6804705b 100644
--- a/src/Processors/Formats/Impl/ArrowBlockOutputFormat.h
+++ b/src/Processors/Formats/Impl/ArrowBlockOutputFormat.h
@@ -6,6 +6,7 @@
 #include <Processors/Formats/IOutputFormat.h>
 #include "ArrowBufferedStreams.h"
 
+namespace arrow { class Schema; }
 namespace arrow::ipc { class RecordBatchWriter; }
 
 namespace DB
@@ -14,7 +15,7 @@ namespace DB
 class ArrowBlockOutputFormat : public IOutputFormat
 {
 public:
-    ArrowBlockOutputFormat(WriteBuffer & out_, const Block & header_, const FormatSettings & format_settings_);
+    ArrowBlockOutputFormat(WriteBuffer & out_, const Block & header_, bool stream_, const FormatSettings & format_settings_);
 
     String getName() const override { return "ArrowBlockOutputFormat"; }
     void consume(Chunk) override;
@@ -23,9 +24,12 @@ public:
     String getContentType() const override { return "application/octet-stream"; }
 
 private:
+    bool stream;
     const FormatSettings format_settings;
     std::shared_ptr<ArrowBufferedOutputStream> arrow_ostream;
     std::shared_ptr<arrow::ipc::RecordBatchWriter> writer;
+
+    void prepareWriter(const std::shared_ptr<arrow::Schema> & schema);
 };
 
 }

From 4effdeaa0e2a07902a286d74a3d13e95d8a40689 Mon Sep 17 00:00:00 2001
From: hcz <h.cz@qq.com>
Date: Thu, 21 May 2020 12:13:49 +0800
Subject: [PATCH 005/122] Add tests

---
 .../0_stateless/01273_arrow_stream.reference  |  62 ++++++++++
 .../queries/0_stateless/01273_arrow_stream.sh | 115 ++++++++++++++++++
 2 files changed, 177 insertions(+)
 create mode 100644 tests/queries/0_stateless/01273_arrow_stream.reference
 create mode 100755 tests/queries/0_stateless/01273_arrow_stream.sh

diff --git a/tests/queries/0_stateless/01273_arrow_stream.reference b/tests/queries/0_stateless/01273_arrow_stream.reference
new file mode 100644
index 00000000000..0f4be2c74a0
--- /dev/null
+++ b/tests/queries/0_stateless/01273_arrow_stream.reference
@@ -0,0 +1,62 @@
+9999
+9998
+9997
+9996
+9995
+9994
+9993
+9992
+9991
+9990
+99999
+99998
+99997
+99996
+99995
+99994
+99993
+99992
+99991
+99990
+2
+1
+0
+999
+998
+997
+996
+995
+994
+993
+992
+991
+990
+ContextLock	Number of times the lock of Context was acquired or tried to acquire. This is global lock.
+Query	Number of queries to be interpreted and potentially executed. Does not include queries that failed to parse or were rejected due to AST size limits, quota limits or limits on the number of simultaneously running queries. May include internal queries initiated by ClickHouse itself. Does not count subqueries.
+original:
+-128	0	-32768	0	-2147483648	0	-9223372036854775808	0	-1.032	-1.064	string-1	fixedstring-1\0\0	2003-04-05	2003-02-03 04:05:06
+-108	108	-1016	1116	-1032	1132	-1064	1164	-1.032	-1.064	string-0	fixedstring\0\0\0\0	2001-02-03	2002-02-03 04:05:06
+127	255	32767	65535	2147483647	4294967295	9223372036854775807	9223372036854775807	-1.032	-1.064	string-2	fixedstring-2\0\0	2004-06-07	2004-02-03 04:05:06
+converted:
+-128	0	-32768	0	-2147483648	0	-9223372036854775808	0	-1.032	-1.064	string-1	fixedstring-1\0\0	2003-04-05	2003-02-03 04:05:06
+-108	108	-1016	1116	-1032	1132	-1064	1164	-1.032	-1.064	string-0	fixedstring\0\0\0\0	2001-02-03	2002-02-03 04:05:06
+127	255	32767	65535	2147483647	4294967295	9223372036854775807	9223372036854775807	-1.032	-1.064	string-2	fixedstring-2\0\0	2004-06-07	2004-02-03 04:05:06
+diff:
+dest:
+79	81	82	83	84	85	86	87	88	89	str01\0\0\0\0\0\0\0\0\0\0	fstr1\0\0\0\0\0\0\0\0\0\0	2003-03-04	1970-01-01 06:29:04
+80	81	82	83	84	85	86	87	88	89	str02	fstr2\0\0\0\0\0\0\0\0\0\0	2005-03-04	2006-08-09 10:11:12
+min:
+-128	0	0	0	0	0	0	0	-1	-1	string-1\0\0\0\0\0\0\0	fixedstring-1\0\0	2003-04-05	2003-02-03
+-108	108	8	92	-8	108	-40	-116	-1	-1	string-0\0\0\0\0\0\0\0	fixedstring\0\0\0\0	2001-02-03	2002-02-03
+79	81	82	83	84	85	86	87	88	89	str01\0\0\0\0\0\0\0\0\0\0	fstr1\0\0\0\0\0\0\0\0\0\0	2003-03-04	2004-05-06
+127	-1	-1	-1	-1	-1	-1	-1	-1	-1	string-2\0\0\0\0\0\0\0	fixedstring-2\0\0	2004-06-07	2004-02-03
+max:
+-128	0	-32768	0	-2147483648	0	-9223372036854775808	0	-1	-1	string-1	fixedstring-1\0\0	1970-01-01 06:22:27	2003-02-03 04:05:06
+-108	108	-1016	1116	-1032	1132	-1064	1164	-1	-1	string-0	fixedstring\0\0\0\0	1970-01-01 06:09:16	2002-02-03 04:05:06
+80	81	82	83	84	85	86	87	88	89	str02	fstr2	2005-03-04 05:06:07	2006-08-09 10:11:12
+127	255	32767	65535	2147483647	4294967295	9223372036854775807	9223372036854775807	-1	-1	string-2	fixedstring-2\0\0	1970-01-01 06:29:36	2004-02-03 04:05:06
+dest from null:
+-128	0	-32768	0	-2147483648	0	-9223372036854775808	0	-1.032	-1.064	string-1	fixedstring-1\0\0	2003-04-05	2003-02-03 04:05:06
+-108	108	-1016	1116	-1032	1132	-1064	1164	-1.032	-1.064	string-0	fixedstring\0\0\0\0	2001-02-03	2002-02-03 04:05:06
+127	255	32767	65535	2147483647	4294967295	9223372036854775807	9223372036854775807	-1.032	-1.064	string-2	fixedstring-2\0\0	2004-06-07	2004-02-03 04:05:06
+\N	\N	\N	\N	\N	\N	\N	\N	\N	\N	\N	\N	\N	\N
diff --git a/tests/queries/0_stateless/01273_arrow_stream.sh b/tests/queries/0_stateless/01273_arrow_stream.sh
new file mode 100755
index 00000000000..2fc630a3a8b
--- /dev/null
+++ b/tests/queries/0_stateless/01273_arrow_stream.sh
@@ -0,0 +1,115 @@
+#!/usr/bin/env bash
+
+set -e
+
+CUR_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
+. $CUR_DIR/../shell_config.sh
+
+
+${CLICKHOUSE_CLIENT} --query="DROP TABLE IF EXISTS contributors"
+${CLICKHOUSE_CLIENT} --query="CREATE TABLE contributors (name String) ENGINE = Memory"
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM system.contributors ORDER BY name DESC FORMAT ArrowStream" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO contributors FORMAT ArrowStream"
+# random results
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM contributors LIMIT 10" > /dev/null
+${CLICKHOUSE_CLIENT} --query="DROP TABLE contributors"
+
+${CLICKHOUSE_CLIENT} --query="DROP TABLE IF EXISTS arrow_numbers"
+${CLICKHOUSE_CLIENT} --query="CREATE TABLE arrow_numbers (number UInt64) ENGINE = Memory"
+# less than default block size (65k)
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM system.numbers LIMIT 10000 FORMAT ArrowStream" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_numbers FORMAT ArrowStream"
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_numbers ORDER BY number DESC LIMIT 10"
+${CLICKHOUSE_CLIENT} --query="TRUNCATE TABLE arrow_numbers"
+
+# More than default block size
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM system.numbers LIMIT 100000 FORMAT ArrowStream" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_numbers FORMAT ArrowStream"
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_numbers ORDER BY number DESC LIMIT 10"
+${CLICKHOUSE_CLIENT} --query="TRUNCATE TABLE arrow_numbers"
+
+${CLICKHOUSE_CLIENT} --max_block_size=2 --query="SELECT * FROM system.numbers LIMIT 3 FORMAT ArrowStream" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_numbers FORMAT ArrowStream"
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_numbers ORDER BY number DESC LIMIT 10"
+
+${CLICKHOUSE_CLIENT} --query="TRUNCATE TABLE arrow_numbers"
+${CLICKHOUSE_CLIENT} --max_block_size=1 --query="SELECT * FROM system.numbers LIMIT 1000 FORMAT ArrowStream" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_numbers FORMAT ArrowStream"
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_numbers ORDER BY number DESC LIMIT 10"
+
+${CLICKHOUSE_CLIENT} --query="DROP TABLE arrow_numbers"
+
+${CLICKHOUSE_CLIENT} --query="DROP TABLE IF EXISTS arrow_events"
+${CLICKHOUSE_CLIENT} --query="CREATE TABLE arrow_events (event String, value UInt64, description String) ENGINE = Memory"
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM system.events FORMAT ArrowStream" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_events FORMAT ArrowStream"
+${CLICKHOUSE_CLIENT} --query="SELECT event, description FROM arrow_events WHERE event IN ('ContextLock', 'Query') ORDER BY event"
+${CLICKHOUSE_CLIENT} --query="DROP TABLE arrow_events"
+
+
+
+${CLICKHOUSE_CLIENT} --query="DROP TABLE IF EXISTS arrow_types1"
+${CLICKHOUSE_CLIENT} --query="DROP TABLE IF EXISTS arrow_types2"
+${CLICKHOUSE_CLIENT} --query="DROP TABLE IF EXISTS arrow_types3"
+${CLICKHOUSE_CLIENT} --query="DROP TABLE IF EXISTS arrow_types4"
+${CLICKHOUSE_CLIENT} --query="CREATE TABLE arrow_types1       (int8 Int8, uint8 UInt8, int16 Int16, uint16 UInt16, int32 Int32, uint32 UInt32, int64 Int64, uint64 UInt64, float32 Float32, float64 Float64, string String, fixedstring FixedString(15), date Date, datetime DateTime) ENGINE = Memory"
+${CLICKHOUSE_CLIENT} --query="CREATE TABLE arrow_types2       (int8 Int8, uint8 UInt8, int16 Int16, uint16 UInt16, int32 Int32, uint32 UInt32, int64 Int64, uint64 UInt64, float32 Float32, float64 Float64, string String, fixedstring FixedString(15), date Date, datetime DateTime) ENGINE = Memory"
+# convert min type
+${CLICKHOUSE_CLIENT} --query="CREATE TABLE arrow_types3       (int8 Int8,  uint8 Int8,  int16 Int8,   uint16 Int8,  int32 Int8,   uint32 Int8,  int64 Int8,   uint64 Int8,    float32 Int8,    float64 Int8, string FixedString(15), fixedstring FixedString(15), date Date,    datetime Date) ENGINE = Memory"
+# convert max type
+${CLICKHOUSE_CLIENT} --query="CREATE TABLE arrow_types4       (int8 Int64, uint8 Int64, int16 Int64, uint16 Int64, int32 Int64,  uint32 Int64, int64 Int64,  uint64 Int64,   float32 Int64,   float64 Int64, string String,          fixedstring String, date DateTime, datetime DateTime) ENGINE = Memory"
+
+${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types1 values (     -108,         108,       -1016,          1116,       -1032,          1132,       -1064,          1164,          -1.032,          -1.064,    'string-0',               'fixedstring', '2001-02-03', '2002-02-03 04:05:06')"
+
+# min
+${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types1 values (     -128,           0,      -32768,             0, -2147483648,             0, -9223372036854775808, 0,             -1.032,          -1.064,    'string-1',             'fixedstring-1', '2003-04-05', '2003-02-03 04:05:06')"
+
+# max
+${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types1 values (      127,         255,       32767,         65535,  2147483647,    4294967295, 9223372036854775807, 9223372036854775807, -1.032,     -1.064,    'string-2',             'fixedstring-2', '2004-06-07', '2004-02-03 04:05:06')"
+
+# 'SELECT -127,-128,-129,126,127,128,255,256,257,-32767,-32768,-32769,32766,32767,32768,65535,65536,65537,  -2147483647,-2147483648,-2147483649,2147483646,2147483647,2147483648,4294967295,4294967296,4294967297,   -9223372036854775807,-9223372036854775808,9223372036854775806,9223372036854775807,9223372036854775808,18446744073709551615';
+
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types1 FORMAT ArrowStream" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types2 FORMAT ArrowStream"
+
+echo original:
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types1 ORDER BY int8" | tee ${CLICKHOUSE_TMP}/arrow_all_types_1.dump
+echo converted:
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types2 ORDER BY int8" | tee ${CLICKHOUSE_TMP}/arrow_all_types_2.dump
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types1 ORDER BY int8 FORMAT ArrowStream" > ${CLICKHOUSE_TMP}/arrow_all_types_1.arrow
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types2 ORDER BY int8 FORMAT ArrowStream" > ${CLICKHOUSE_TMP}/arrow_all_types_2.arrow
+echo diff:
+diff ${CLICKHOUSE_TMP}/arrow_all_types_1.dump ${CLICKHOUSE_TMP}/arrow_all_types_2.dump
+
+${CLICKHOUSE_CLIENT} --query="TRUNCATE TABLE arrow_types2"
+${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types3 values (       79,          81,          82,            83,          84,            85,          86,            87,              88,              89,         'str01',                  'fstr1', '2003-03-04', '2004-05-06')"
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types3 ORDER BY int8 FORMAT ArrowStream" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types2 FORMAT ArrowStream"
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types1 ORDER BY int8 FORMAT ArrowStream" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types3 FORMAT ArrowStream"
+
+${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types4 values (       80,          81,          82,            83,          84,            85,          86,            87,              88,              89,         'str02',                  'fstr2', '2005-03-04 05:06:07', '2006-08-09 10:11:12')"
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types4 ORDER BY int8 FORMAT ArrowStream" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types2 FORMAT ArrowStream"
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types1 ORDER BY int8 FORMAT ArrowStream" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types4 FORMAT ArrowStream"
+
+echo dest:
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types2 ORDER BY int8"
+echo min:
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types3 ORDER BY int8"
+echo max:
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types4 ORDER BY int8"
+
+
+${CLICKHOUSE_CLIENT} --query="DROP TABLE IF EXISTS arrow_types5"
+${CLICKHOUSE_CLIENT} --query="DROP TABLE IF EXISTS arrow_types6"
+${CLICKHOUSE_CLIENT} --query="TRUNCATE TABLE arrow_types2"
+${CLICKHOUSE_CLIENT} --query="CREATE TABLE arrow_types5       (int8 Nullable(Int8), uint8 Nullable(UInt8), int16 Nullable(Int16), uint16 Nullable(UInt16), int32 Nullable(Int32), uint32 Nullable(UInt32), int64 Nullable(Int64), uint64 Nullable(UInt64), float32 Nullable(Float32), float64 Nullable(Float64), string Nullable(String), fixedstring Nullable(FixedString(15)), date Nullable(Date), datetime Nullable(DateTime)) ENGINE = Memory"
+${CLICKHOUSE_CLIENT} --query="CREATE TABLE arrow_types6       (int8 Nullable(Int8), uint8 Nullable(UInt8), int16 Nullable(Int16), uint16 Nullable(UInt16), int32 Nullable(Int32), uint32 Nullable(UInt32), int64 Nullable(Int64), uint64 Nullable(UInt64), float32 Nullable(Float32), float64 Nullable(Float64), string Nullable(String), fixedstring Nullable(FixedString(15)), date Nullable(Date), datetime Nullable(DateTime)) ENGINE = Memory"
+${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types5 values (               NULL,                  NULL,                  NULL,                    NULL,                  NULL,                    NULL,                  NULL,                    NULL,                      NULL,                      NULL,                    NULL,                                  NULL,                NULL,                        NULL)"
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types5 ORDER BY int8 FORMAT ArrowStream" > ${CLICKHOUSE_TMP}/arrow_all_types_5.arrow
+#${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types5 FORMAT ArrowStream" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types6 FORMAT ArrowStream"
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types5 ORDER BY int8 FORMAT ArrowStream" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types6 FORMAT ArrowStream"
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types1 ORDER BY int8 FORMAT ArrowStream" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types6 FORMAT ArrowStream"
+echo dest from null:
+${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types6 ORDER BY int8"
+
+${CLICKHOUSE_CLIENT} --query="DROP TABLE arrow_types5"
+${CLICKHOUSE_CLIENT} --query="DROP TABLE arrow_types6"
+
+
+${CLICKHOUSE_CLIENT} --query="DROP TABLE arrow_types1"
+${CLICKHOUSE_CLIENT} --query="DROP TABLE arrow_types2"
+${CLICKHOUSE_CLIENT} --query="DROP TABLE arrow_types3"
+${CLICKHOUSE_CLIENT} --query="DROP TABLE arrow_types4"
+

From ab37791134b7317c99679ef5e4cb738c2ff9d9a4 Mon Sep 17 00:00:00 2001
From: hcz <h.cz@qq.com>
Date: Thu, 21 May 2020 15:28:10 +0800
Subject: [PATCH 006/122] Fix arrow stream termination

---
 src/Processors/Formats/Impl/ArrowBlockInputFormat.cpp | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/src/Processors/Formats/Impl/ArrowBlockInputFormat.cpp b/src/Processors/Formats/Impl/ArrowBlockInputFormat.cpp
index c8a098418b7..5468fbcac16 100644
--- a/src/Processors/Formats/Impl/ArrowBlockInputFormat.cpp
+++ b/src/Processors/Formats/Impl/ArrowBlockInputFormat.cpp
@@ -39,12 +39,12 @@ Chunk ArrowBlockInputFormat::generate()
     if (stream)
     {
         read_status = stream_reader->ReadNext(&single_batch[0]);
+        if (!single_batch[0])
+            return res;
     }
     else
     {
         read_status = file_reader->ReadRecordBatch(record_batch_current, &single_batch[0]);
-        if (!single_batch[0])
-            return res;
     }
     if (!read_status.ok())
         throw Exception{"Error while reading batch of Arrow data: " + read_status.ToString(),

From 95e2f0fa0769e87cbf77cb8d264c75a6fb198a8d Mon Sep 17 00:00:00 2001
From: hcz <h.cz@qq.com>
Date: Thu, 21 May 2020 18:20:41 +0800
Subject: [PATCH 007/122] Add brief docs about apache arrow

---
 docs/en/interfaces/formats.md | 20 ++++++++++++++++----
 1 file changed, 16 insertions(+), 4 deletions(-)

diff --git a/docs/en/interfaces/formats.md b/docs/en/interfaces/formats.md
index 4697fbca7e1..ed411a1b0d4 100644
--- a/docs/en/interfaces/formats.md
+++ b/docs/en/interfaces/formats.md
@@ -37,6 +37,8 @@ The supported formats are:
 | [Avro](#data-format-avro)                                       | ✔     | ✔      |
 | [AvroConfluent](#data-format-avro-confluent)                    | ✔     | ✗      |
 | [Parquet](#data-format-parquet)                                 | ✔     | ✔      |
+| [Arrow](#data-format-arrow)                                     | ✔     | ✔      |
+| [ArrowStream](#data-format-arrow-stream)                        | ✔     | ✔      |
 | [ORC](#data-format-orc)                                         | ✔     | ✗      |
 | [RowBinary](#rowbinary)                                         | ✔     | ✔      |
 | [RowBinaryWithNamesAndTypes](#rowbinarywithnamesandtypes)       | ✔     | ✔      |
@@ -985,9 +987,9 @@ See also [how to read/write length-delimited protobuf messages in popular langua
 
 ## Avro {#data-format-avro}
 
-[Apache Avro](http://avro.apache.org/) is a row-oriented data serialization framework developed within Apache’s Hadoop project.
+[Apache Avro](https://avro.apache.org/) is a row-oriented data serialization framework developed within Apache’s Hadoop project.
 
-ClickHouse Avro format supports reading and writing [Avro data files](http://avro.apache.org/docs/current/spec.html#Object+Container+Files).
+ClickHouse Avro format supports reading and writing [Avro data files](https://avro.apache.org/docs/current/spec.html#Object+Container+Files).
 
 ### Data Types Matching {#data_types-matching}
 
@@ -1009,7 +1011,7 @@ The table below shows supported data types and how they match ClickHouse [data t
 | `long (timestamp-millis)` \*                | [DateTime64(3)](../sql-reference/data-types/datetime.md)                                                              | `long (timestamp-millis)` \* |
 | `long (timestamp-micros)` \*                | [DateTime64(6)](../sql-reference/data-types/datetime.md)                                                              | `long (timestamp-micros)` \* |
 
-\* [Avro logical types](http://avro.apache.org/docs/current/spec.html#Logical+Types)
+\* [Avro logical types](https://avro.apache.org/docs/current/spec.html#Logical+Types)
 
 Unsupported Avro data types: `record` (non-root), `map`
 
@@ -1095,7 +1097,7 @@ SELECT * FROM topic1_stream;
 
 ## Parquet {#data-format-parquet}
 
-[Apache Parquet](http://parquet.apache.org/) is a columnar storage format widespread in the Hadoop ecosystem. ClickHouse supports read and write operations for this format.
+[Apache Parquet](https://parquet.apache.org/) is a columnar storage format widespread in the Hadoop ecosystem. ClickHouse supports read and write operations for this format.
 
 ### Data Types Matching {#data_types-matching-2}
 
@@ -1141,6 +1143,16 @@ $ clickhouse-client --query="SELECT * FROM {some_table} FORMAT Parquet" > {some_
 
 To exchange data with Hadoop, you can use [HDFS table engine](../engines/table-engines/integrations/hdfs.md).
 
+## Arrow {data-format-arrow}
+
+[Apache Arrow](https://arrow.apache.org/) comes with two built-in columnar storage formats. ClickHouse supports read and write operations for these formats.
+
+`Arrow` is Apache Arrow's "file mode" format. It is designed for in-memory random access.
+
+## ArrowStream {data-format-arrow-stream}
+
+`ArrowStream` is Apache Arrow's "stream mode" format. It is designed for in-memory stream processing.
+
 ## ORC {#data-format-orc}
 
 [Apache ORC](https://orc.apache.org/) is a columnar storage format widespread in the Hadoop ecosystem. You can only insert data in this format to ClickHouse.

From 409cff11df96b810f3dbe8adb75c62d71894d10f Mon Sep 17 00:00:00 2001
From: Azat Khuzhin <a3at.mail@gmail.com>
Date: Fri, 22 May 2020 00:37:47 +0300
Subject: [PATCH 008/122] Add port() function

It uses domain() (getURLHost()) internally so it has the same
limitations:
- no ipv6 support
- no host-without-dots support (i.e. "foobar")
---
 .../sql-reference/functions/url-functions.md  |  4 +
 src/Functions/URL/port.cpp                    | 95 +++++++++++++++++++
 src/Functions/URL/registerFunctionsURL.cpp    |  2 +
 .../queries/0_stateless/01284_port.reference  | 21 ++++
 tests/queries/0_stateless/01284_port.sql      | 25 +++++
 5 files changed, 147 insertions(+)
 create mode 100644 src/Functions/URL/port.cpp
 create mode 100644 tests/queries/0_stateless/01284_port.reference
 create mode 100644 tests/queries/0_stateless/01284_port.sql

diff --git a/docs/en/sql-reference/functions/url-functions.md b/docs/en/sql-reference/functions/url-functions.md
index 143bd42b08e..f35c621a6e2 100644
--- a/docs/en/sql-reference/functions/url-functions.md
+++ b/docs/en/sql-reference/functions/url-functions.md
@@ -117,6 +117,10 @@ Returns the part of the domain that includes top-level subdomains up to the “f
 
 For example, `cutToFirstSignificantSubdomain('https://news.yandex.com.tr/') = 'yandex.com.tr'`.
 
+### port(URL) {#port}
+
+Returns the port or zero if there is no port in the URL (or in case of validation error).
+
 ### path {#path}
 
 Returns the path. Example: `/top/news.html` The path does not include the query string.
diff --git a/src/Functions/URL/port.cpp b/src/Functions/URL/port.cpp
new file mode 100644
index 00000000000..d4876e9f647
--- /dev/null
+++ b/src/Functions/URL/port.cpp
@@ -0,0 +1,95 @@
+#include <Functions/FunctionFactory.h>
+#include <Functions/FunctionStringOrArrayToT.h>
+#include <Common/StringUtils/StringUtils.h>
+#include "domain.h"
+
+
+namespace DB
+{
+
+namespace ErrorCodes
+{
+    extern const int ILLEGAL_TYPE_OF_ARGUMENT;
+}
+
+struct ExtractPort
+{
+    static constexpr auto name = "port";
+    static constexpr auto is_fixed_to_constant = true;
+
+    static void vector(const ColumnString::Chars & data, const ColumnString::Offsets & offsets, PaddedPODArray<UInt16> & res)
+    {
+        size_t size = offsets.size();
+
+        ColumnString::Offset prev_offset = 0;
+        for (size_t i = 0; i < size; ++i)
+        {
+            res[i] = parse(data, prev_offset, offsets[i] - prev_offset - 1);
+            prev_offset = offsets[i];
+        }
+    }
+
+    static void vectorFixedToConstant(const ColumnString::Chars & data, size_t n, UInt16 & res) { res = parse(data, 0, n); }
+
+    static void vectorFixedToVector(const ColumnString::Chars & data, size_t n, PaddedPODArray<UInt16> & res)
+    {
+        size_t size = data.size() / n;
+
+        for (size_t i = 0; i < size; ++i)
+        {
+            res[i] = parse(data, i * n, n);
+        }
+    }
+
+    [[noreturn]] static void array(const ColumnString::Offsets &, PaddedPODArray<UInt16> &)
+    {
+        throw Exception("Cannot apply function port to Array argument", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT);
+    }
+
+private:
+    static UInt16 parse(const ColumnString::Chars & buf, size_t offset, size_t size)
+    {
+        const char * p = reinterpret_cast<const char *>(&buf[0]) + offset;
+        const char * end = p + size;
+
+        StringRef host = getURLHost(p, size);
+        if (!host.size)
+            return 0;
+        if (host.size == size)
+            return 0;
+
+        p = host.data + host.size;
+        if (*p++ != ':')
+            return 0;
+
+        Int64 port = 0;
+        while (p < end)
+        {
+            if (*p == '/')
+                break;
+            if (!isNumericASCII(*p))
+                return 0;
+
+            port = (port * 10) + (*p - '0');
+            if (port < 0 || port > UInt16(-1))
+                return 0;
+            ++p;
+        }
+        return port;
+    }
+};
+
+struct NamePort
+{
+    static constexpr auto name = "port";
+};
+
+using FunctionPort = FunctionStringOrArrayToT<ExtractPort, NamePort, UInt16>;
+
+void registerFunctionPort(FunctionFactory & factory)
+{
+    factory.registerFunction<FunctionPort>();
+}
+
+}
+
diff --git a/src/Functions/URL/registerFunctionsURL.cpp b/src/Functions/URL/registerFunctionsURL.cpp
index 66a847185f3..9ba5261f728 100644
--- a/src/Functions/URL/registerFunctionsURL.cpp
+++ b/src/Functions/URL/registerFunctionsURL.cpp
@@ -8,6 +8,7 @@ void registerFunctionDomain(FunctionFactory & factory);
 void registerFunctionDomainWithoutWWW(FunctionFactory & factory);
 void registerFunctionFirstSignificantSubdomain(FunctionFactory & factory);
 void registerFunctionTopLevelDomain(FunctionFactory & factory);
+void registerFunctionPort(FunctionFactory & factory);
 void registerFunctionPath(FunctionFactory & factory);
 void registerFunctionPathFull(FunctionFactory & factory);
 void registerFunctionQueryString(FunctionFactory & factory);
@@ -33,6 +34,7 @@ void registerFunctionsURL(FunctionFactory & factory)
     registerFunctionDomainWithoutWWW(factory);
     registerFunctionFirstSignificantSubdomain(factory);
     registerFunctionTopLevelDomain(factory);
+    registerFunctionPort(factory);
     registerFunctionPath(factory);
     registerFunctionPathFull(factory);
     registerFunctionQueryString(factory);
diff --git a/tests/queries/0_stateless/01284_port.reference b/tests/queries/0_stateless/01284_port.reference
new file mode 100644
index 00000000000..5bf1c035ca8
--- /dev/null
+++ b/tests/queries/0_stateless/01284_port.reference
@@ -0,0 +1,21 @@
+ipv4
+0
+80
+80
+80
+80
+hostname
+0
+80
+80
+80
+80
+ipv6
+0
+0
+0
+0
+0
+0
+host-no-dot
+0
diff --git a/tests/queries/0_stateless/01284_port.sql b/tests/queries/0_stateless/01284_port.sql
new file mode 100644
index 00000000000..3ba5aecd691
--- /dev/null
+++ b/tests/queries/0_stateless/01284_port.sql
@@ -0,0 +1,25 @@
+select 'ipv4';
+select port('http://127.0.0.1/');
+select port('http://127.0.0.1:80');
+select port('http://127.0.0.1:80/');
+select port('//127.0.0.1:80/');
+select port('127.0.0.1:80');
+select 'hostname';
+select port('http://foobar.com/');
+select port('http://foobar.com:80');
+select port('http://foobar.com:80/');
+select port('//foobar.com:80/');
+select port('foobar.com:80');
+
+--
+-- Known limitations of domain() (getURLHost())
+--
+select 'ipv6';
+select port('http://[2001:db8::8a2e:370:7334]/');
+select port('http://[2001:db8::8a2e:370:7334]:80');
+select port('http://[2001:db8::8a2e:370:7334]:80/');
+select port('//[2001:db8::8a2e:370:7334]:80/');
+select port('[2001:db8::8a2e:370:7334]:80');
+select port('2001:db8::8a2e:370:7334:80');
+select 'host-no-dot';
+select port('//foobar:80/');

From aa5247f1fed85e64479985dc0de65621118a249d Mon Sep 17 00:00:00 2001
From: Azat Khuzhin <a3at.mail@gmail.com>
Date: Thu, 21 May 2020 23:58:55 +0300
Subject: [PATCH 009/122] Get back 01281_group_by_limit_memory_tracking
 (without flackiness)

---
 ...1_group_by_limit_memory_tracking.reference |  0
 .../01281_group_by_limit_memory_tracking.sh   | 34 +++++++++++++++++++
 2 files changed, 34 insertions(+)
 create mode 100644 tests/queries/0_stateless/01281_group_by_limit_memory_tracking.reference
 create mode 100755 tests/queries/0_stateless/01281_group_by_limit_memory_tracking.sh

diff --git a/tests/queries/0_stateless/01281_group_by_limit_memory_tracking.reference b/tests/queries/0_stateless/01281_group_by_limit_memory_tracking.reference
new file mode 100644
index 00000000000..e69de29bb2d
diff --git a/tests/queries/0_stateless/01281_group_by_limit_memory_tracking.sh b/tests/queries/0_stateless/01281_group_by_limit_memory_tracking.sh
new file mode 100755
index 00000000000..79a731b9273
--- /dev/null
+++ b/tests/queries/0_stateless/01281_group_by_limit_memory_tracking.sh
@@ -0,0 +1,34 @@
+#!/usr/bin/env bash
+
+# Regression for MemoryTracker that had been incorrectly accounted
+# (it was reseted before deallocation)
+#
+# For this will be used:
+# - two-level group by
+# - max_memory_usage_for_user
+# - one users' query in background (to avoid reseting max_memory_usage_for_user)
+
+CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
+. $CURDIR/../shell_config.sh
+
+set -o pipefail
+
+function execute_null()
+{
+    ${CLICKHOUSE_CLIENT} --format Null -n "$@"
+}
+
+function execute_group_by()
+{
+    local opts=(
+        --max_memory_usage_for_user=$((150<<20))
+        --max_threads=2
+    )
+    execute_null "${opts[@]}" <<<'SELECT uniq(number) FROM numbers_mt(toUInt64(1e6)) GROUP BY number % 5e5'
+}
+
+execute_null <<<'SELECT sleep(3)' &
+execute_group_by
+# if memory accounting will be incorrect, the second query will be failed with MEMORY_LIMIT_EXCEEDED
+execute_group_by
+wait

From 9de1a231c9b003c306570cedbf380b4bff922294 Mon Sep 17 00:00:00 2001
From: Azat Khuzhin <a3at.mail@gmail.com>
Date: Sat, 23 May 2020 21:14:52 +0300
Subject: [PATCH 010/122] Add optional default_port argument for the port()
 function

---
 .../sql-reference/functions/url-functions.md  |   4 +-
 src/Functions/URL/port.cpp                    | 110 ++++++++++++------
 .../queries/0_stateless/01284_port.reference  |   3 +
 tests/queries/0_stateless/01284_port.sql      |   9 ++
 4 files changed, 86 insertions(+), 40 deletions(-)

diff --git a/docs/en/sql-reference/functions/url-functions.md b/docs/en/sql-reference/functions/url-functions.md
index f35c621a6e2..34477529649 100644
--- a/docs/en/sql-reference/functions/url-functions.md
+++ b/docs/en/sql-reference/functions/url-functions.md
@@ -117,9 +117,9 @@ Returns the part of the domain that includes top-level subdomains up to the “f
 
 For example, `cutToFirstSignificantSubdomain('https://news.yandex.com.tr/') = 'yandex.com.tr'`.
 
-### port(URL) {#port}
+### port(URL[, default_port = 0]) {#port}
 
-Returns the port or zero if there is no port in the URL (or in case of validation error).
+Returns the port or `default_port` if there is no port in the URL (or in case of validation error).
 
 ### path {#path}
 
diff --git a/src/Functions/URL/port.cpp b/src/Functions/URL/port.cpp
index d4876e9f647..e703da2602b 100644
--- a/src/Functions/URL/port.cpp
+++ b/src/Functions/URL/port.cpp
@@ -1,6 +1,10 @@
 #include <Functions/FunctionFactory.h>
-#include <Functions/FunctionStringOrArrayToT.h>
+#include <Functions/IFunctionImpl.h>
 #include <Common/StringUtils/StringUtils.h>
+#include <DataTypes/DataTypesNumber.h>
+#include <Columns/ColumnsNumber.h>
+#include <Columns/ColumnArray.h>
+#include <Columns/ColumnConst.h>
 #include "domain.h"
 
 
@@ -9,83 +13,113 @@ namespace DB
 
 namespace ErrorCodes
 {
+    extern const int ILLEGAL_COLUMN;
     extern const int ILLEGAL_TYPE_OF_ARGUMENT;
+    extern const int NUMBER_OF_ARGUMENTS_DOESNT_MATCH;
 }
 
-struct ExtractPort
+struct FunctionPort : public IFunction
 {
     static constexpr auto name = "port";
-    static constexpr auto is_fixed_to_constant = true;
+    static FunctionPtr create(const Context &) { return std::make_shared<FunctionPort>(); }
 
-    static void vector(const ColumnString::Chars & data, const ColumnString::Offsets & offsets, PaddedPODArray<UInt16> & res)
+    String getName() const override { return name; }
+    bool isVariadic() const override { return true; }
+    size_t getNumberOfArguments() const override { return 0; }
+    bool useDefaultImplementationForConstants() const override { return true; }
+    ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {1}; }
+
+    DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override
+    {
+        if (arguments.size() != 1 && arguments.size() != 2)
+            throw Exception("Number of arguments for function " + getName() + " doesn't match: passed "
+                            + std::to_string(arguments.size()) + ", should be 1 or 2",
+                            ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH);
+
+        if (!WhichDataType(arguments[0].type).isString())
+            throw Exception("Illegal type " + arguments[0].type->getName() + " of first argument of function " + getName() + ". Must be String.",
+                ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT);
+
+        if (arguments.size() == 2 && !WhichDataType(arguments[1].type).isUInt16())
+            throw Exception("Illegal type " + arguments[1].type->getName() + " of second argument of function " + getName() + ". Must be UInt16.",
+                ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT);
+
+        return std::make_shared<DataTypeUInt16>();
+    }
+
+
+    void executeImpl(Block & block, const ColumnNumbers & arguments, size_t result, size_t) override
+    {
+        UInt16 default_port = 0;
+        if (arguments.size() == 2)
+        {
+            const auto * port_column = checkAndGetColumn<ColumnConst>(block.getByPosition(arguments[1]).column.get());
+            if (!port_column)
+                throw Exception("Second argument for function " + getName() + " must be constant UInt16", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT);
+            default_port = port_column->getValue<UInt16>();
+        }
+
+        const ColumnPtr url_column = block.getByPosition(arguments[0]).column;
+        if (const ColumnString * url_strs = checkAndGetColumn<ColumnString>(url_column.get()))
+        {
+            auto col_res = ColumnVector<UInt16>::create();
+            typename ColumnVector<UInt16>::Container & vec_res = col_res->getData();
+            vec_res.resize(url_column->size());
+
+            vector(default_port, url_strs->getChars(), url_strs->getOffsets(), vec_res);
+            block.getByPosition(result).column = std::move(col_res);
+        }
+        else
+            throw Exception(
+                "Illegal column " + block.getByPosition(arguments[0]).column->getName() + " of argument of function " + getName(),
+                ErrorCodes::ILLEGAL_COLUMN);
+}
+
+private:
+    static void vector(UInt16 default_port, const ColumnString::Chars & data, const ColumnString::Offsets & offsets, PaddedPODArray<UInt16> & res)
     {
         size_t size = offsets.size();
 
         ColumnString::Offset prev_offset = 0;
         for (size_t i = 0; i < size; ++i)
         {
-            res[i] = parse(data, prev_offset, offsets[i] - prev_offset - 1);
+            res[i] = extractPort(default_port, data, prev_offset, offsets[i] - prev_offset - 1);
             prev_offset = offsets[i];
         }
-    }
+}
 
-    static void vectorFixedToConstant(const ColumnString::Chars & data, size_t n, UInt16 & res) { res = parse(data, 0, n); }
-
-    static void vectorFixedToVector(const ColumnString::Chars & data, size_t n, PaddedPODArray<UInt16> & res)
-    {
-        size_t size = data.size() / n;
-
-        for (size_t i = 0; i < size; ++i)
-        {
-            res[i] = parse(data, i * n, n);
-        }
-    }
-
-    [[noreturn]] static void array(const ColumnString::Offsets &, PaddedPODArray<UInt16> &)
-    {
-        throw Exception("Cannot apply function port to Array argument", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT);
-    }
-
-private:
-    static UInt16 parse(const ColumnString::Chars & buf, size_t offset, size_t size)
+    static UInt16 extractPort(UInt16 default_port, const ColumnString::Chars & buf, size_t offset, size_t size)
     {
         const char * p = reinterpret_cast<const char *>(&buf[0]) + offset;
         const char * end = p + size;
 
         StringRef host = getURLHost(p, size);
         if (!host.size)
-            return 0;
+            return default_port;
         if (host.size == size)
-            return 0;
+            return default_port;
 
         p = host.data + host.size;
         if (*p++ != ':')
-            return 0;
+            return default_port;
 
-        Int64 port = 0;
+        Int64 port = default_port;
         while (p < end)
         {
             if (*p == '/')
                 break;
             if (!isNumericASCII(*p))
-                return 0;
+                return default_port;
 
             port = (port * 10) + (*p - '0');
             if (port < 0 || port > UInt16(-1))
-                return 0;
+                return default_port;
             ++p;
         }
         return port;
     }
 };
 
-struct NamePort
-{
-    static constexpr auto name = "port";
-};
-
-using FunctionPort = FunctionStringOrArrayToT<ExtractPort, NamePort, UInt16>;
-
 void registerFunctionPort(FunctionFactory & factory)
 {
     factory.registerFunction<FunctionPort>();
diff --git a/tests/queries/0_stateless/01284_port.reference b/tests/queries/0_stateless/01284_port.reference
index 5bf1c035ca8..7e776595065 100644
--- a/tests/queries/0_stateless/01284_port.reference
+++ b/tests/queries/0_stateless/01284_port.reference
@@ -10,6 +10,9 @@ hostname
 80
 80
 80
+default-port
+80
+80
 ipv6
 0
 0
diff --git a/tests/queries/0_stateless/01284_port.sql b/tests/queries/0_stateless/01284_port.sql
index 3ba5aecd691..9c31a5d42ad 100644
--- a/tests/queries/0_stateless/01284_port.sql
+++ b/tests/queries/0_stateless/01284_port.sql
@@ -11,6 +11,15 @@ select port('http://foobar.com:80/');
 select port('//foobar.com:80/');
 select port('foobar.com:80');
 
+select 'default-port';
+select port('http://127.0.0.1/', toUInt16(80));
+select port('http://foobar.com/', toUInt16(80));
+
+-- unsupported
+/* ILLEGAL_TYPE_OF_ARGUMENT */ select port(toFixedString('', 1)); -- { serverError 43; }
+/* ILLEGAL_TYPE_OF_ARGUMENT */ select port('', 1); -- { serverError 43; }
+/* NUMBER_OF_ARGUMENTS_DOESNT_MATCH */ select port('', 1, 1); -- { serverError 42; }
+
 --
 -- Known limitations of domain() (getURLHost())
 --

From 96260cf33a12e906e8db3303c057713db954f1b9 Mon Sep 17 00:00:00 2001
From: Albert Kidrachev <kidrachev@sandbox-410109003>
Date: Mon, 25 May 2020 02:55:00 +0300
Subject: [PATCH 011/122] init

---
 src/Common/SharedBlockRowRef.h                | 30 +++++++
 src/Interpreters/InterpreterSelectQuery.cpp   |  1 +
 .../OptimizedPartialSortingTransform.cpp      | 83 +++++++++++++++++++
 .../OptimizedPartialSortingTransform.h        | 34 ++++++++
 src/Processors/ya.make                        |  1 +
 5 files changed, 149 insertions(+)
 create mode 100644 src/Processors/Transforms/OptimizedPartialSortingTransform.cpp
 create mode 100644 src/Processors/Transforms/OptimizedPartialSortingTransform.h

diff --git a/src/Common/SharedBlockRowRef.h b/src/Common/SharedBlockRowRef.h
index 193f7e4dd05..e9fd076da07 100644
--- a/src/Common/SharedBlockRowRef.h
+++ b/src/Common/SharedBlockRowRef.h
@@ -2,6 +2,7 @@
 
 #include <algorithm>
 #include <Core/Block.h>
+#include <Core/SortDescription.h>
 #include <Columns/IColumn.h>
 #include <boost/smart_ptr/intrusive_ptr.hpp>
 
@@ -86,4 +87,33 @@ struct SharedBlockRowRef
     }
 };
 
+struct SharedBlockRowWithSortDescriptionRef : SharedBlockRowRef
+{
+    SortDescription * description = nullptr;
+
+    void set(SharedBlockPtr & shared_block_, ColumnRawPtrs * columns_, size_t row_num_) = delete;
+
+    bool operator< (const SharedBlockRowRef & other) const
+    {
+        size_t size = columns->size();
+        for (size_t i = 0; i < size; ++i)
+        {
+            int res = (*description)[i].direction * (*columns)[i]->compareAt(row_num, other.row_num, *(*other.columns)[i], 1);
+            if (res < 0)
+                return true;
+            else if (res > 0)
+                return false;
+        }
+        return false;
+    }
+
+    void set(SharedBlockPtr & shared_block_, ColumnRawPtrs * columns_, size_t row_num_, SortDescription * description_)
+    {
+        shared_block = shared_block_;
+        columns = columns_;
+        row_num = row_num_;
+        description = description_;
+    }
+};
+
 }
diff --git a/src/Interpreters/InterpreterSelectQuery.cpp b/src/Interpreters/InterpreterSelectQuery.cpp
index c8d842dfa53..3d95265cb27 100644
--- a/src/Interpreters/InterpreterSelectQuery.cpp
+++ b/src/Interpreters/InterpreterSelectQuery.cpp
@@ -84,6 +84,7 @@
 #include <Processors/Transforms/MergingAggregatedMemoryEfficientTransform.h>
 #include <Processors/Transforms/TotalsHavingTransform.h>
 #include <Processors/Transforms/PartialSortingTransform.h>
+#include <Processors/Transforms/OptimizedPartialSortingTransform.h>
 #include <Processors/Transforms/LimitsCheckingTransform.h>
 #include <Processors/Transforms/MergeSortingTransform.h>
 #include <Processors/Transforms/DistinctTransform.h>
diff --git a/src/Processors/Transforms/OptimizedPartialSortingTransform.cpp b/src/Processors/Transforms/OptimizedPartialSortingTransform.cpp
new file mode 100644
index 00000000000..e753052a2b3
--- /dev/null
+++ b/src/Processors/Transforms/OptimizedPartialSortingTransform.cpp
@@ -0,0 +1,83 @@
+#include <Processors/Transforms/OptimizedPartialSortingTransform.h>
+#include <Interpreters/sortBlock.h>
+#include <Common/PODArray.h>
+
+namespace DB
+{
+
+OptimizedPartialSortingTransform::OptimizedPartialSortingTransform(
+    const Block & header_, SortDescription & description_, UInt64 limit_)
+    : ISimpleTransform(header_, header_, false)
+    , description(description_), limit(limit_)
+    , threshold_shared_block(nullptr)
+{
+}
+
+static ColumnRawPtrs extractColumns(const Block & block, const SortDescription& description)
+{
+    size_t size = description.size();
+    ColumnRawPtrs res;
+    res.reserve(size);
+
+    for (size_t i = 0; i < size; ++i)
+    {
+        const IColumn * column = !description[i].column_name.empty()
+            ? block.getByName(description[i].column_name).column.get()
+            : block.safeGetByPosition(description[i].column_number).column.get();
+        res.emplace_back(column);
+    }
+
+    return res;
+}
+
+void OptimizedPartialSortingTransform::transform(Chunk & chunk)
+{
+    if (read_rows)
+        read_rows->add(chunk.getNumRows());
+
+    auto block = getInputPort().getHeader().cloneWithColumns(chunk.detachColumns());
+    chunk.clear();
+
+    SharedBlockPtr shared_block = new detail::SharedBlock(std::move(block));
+    UInt64 rows_num = shared_block->rows();
+
+
+    if (threshold_shared_block) {
+        SharedBlockRowWithSortDescriptionRef row;
+        IColumn::Filter filter(rows_num);
+        ColumnRawPtrs shared_block_columns = extractColumns(*shared_block, description);
+        size_t filtered_count = 0;
+
+        for (UInt64 i = 0; i < rows_num; ++i) {
+            row.set(shared_block, &shared_block_columns, i, &description);
+
+            if (threshold_row < row)
+            {
+                ++filtered_count;
+                filter[i] = 1;
+            }
+        }
+
+        if (filtered_count)
+        {
+            for (auto & column : shared_block->getColumns())
+            {
+                column = column->filter(filter, filtered_count);
+            }
+        }
+    }
+
+    sortBlock(*shared_block, description, limit);
+
+    if (!threshold_shared_block && limit && limit < rows_num)
+    {
+        Block threshold_block = shared_block->cloneWithColumns(shared_block->getColumns());
+        threshold_shared_block = new detail::SharedBlock(std::move(threshold_block));
+        threshold_block_columns = extractColumns(*threshold_shared_block, description);
+        threshold_row.set(threshold_shared_block, &threshold_block_columns, limit - 1, &description);
+    }
+
+    chunk.setColumns(shared_block->getColumns(), shared_block->rows());
+}
+
+}
diff --git a/src/Processors/Transforms/OptimizedPartialSortingTransform.h b/src/Processors/Transforms/OptimizedPartialSortingTransform.h
new file mode 100644
index 00000000000..20e72bd836f
--- /dev/null
+++ b/src/Processors/Transforms/OptimizedPartialSortingTransform.h
@@ -0,0 +1,34 @@
+#pragma once
+#include <Processors/ISimpleTransform.h>
+#include <Processors/RowsBeforeLimitCounter.h>
+#include <Common/SharedBlockRowRef.h>
+#include <Core/SortDescription.h>
+
+
+namespace DB
+{
+class OptimizedPartialSortingTransform : public ISimpleTransform
+{
+public:
+    OptimizedPartialSortingTransform(
+        const Block & header_,
+        SortDescription & description_,
+        UInt64 limit_ = 0);
+
+    String getName() const override { return "OptimizedPartialSortingTransform"; }
+
+    void setRowsBeforeLimitCounter(RowsBeforeLimitCounterPtr counter) { read_rows.swap(counter); }
+
+protected:
+    void transform(Chunk & chunk) override;
+
+private:
+    SortDescription description;
+    UInt64 limit;
+    RowsBeforeLimitCounterPtr read_rows;
+    SharedBlockRowWithSortDescriptionRef threshold_row;
+    SharedBlockPtr threshold_shared_block;
+    ColumnRawPtrs threshold_block_columns;
+};
+
+}
diff --git a/src/Processors/ya.make b/src/Processors/ya.make
index 62320f1c147..5952341527b 100644
--- a/src/Processors/ya.make
+++ b/src/Processors/ya.make
@@ -129,6 +129,7 @@ SRCS(
     Transforms/MergeSortingTransform.cpp
     Transforms/MergingAggregatedMemoryEfficientTransform.cpp
     Transforms/MergingAggregatedTransform.cpp
+    Transforms/OptimizedPartialSortingTransform.cpp
     Transforms/PartialSortingTransform.cpp
     Transforms/ReverseTransform.cpp
     Transforms/RollupTransform.cpp

From 3abf7b7dcdc26acda9aae9663e9e026f1b5cb631 Mon Sep 17 00:00:00 2001
From: Albert Kidrachev <kidrachev@sandbox-410109003>
Date: Mon, 25 May 2020 03:32:08 +0300
Subject: [PATCH 012/122] fix

---
 src/Interpreters/InterpreterSelectQuery.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/Interpreters/InterpreterSelectQuery.cpp b/src/Interpreters/InterpreterSelectQuery.cpp
index 3d95265cb27..c3fce45ac61 100644
--- a/src/Interpreters/InterpreterSelectQuery.cpp
+++ b/src/Interpreters/InterpreterSelectQuery.cpp
@@ -2109,7 +2109,7 @@ void InterpreterSelectQuery::executeOrder(QueryPipeline & pipeline, InputSorting
         if (stream_type != QueryPipeline::StreamType::Main)
             return nullptr;
 
-        return std::make_shared<PartialSortingTransform>(header, output_order_descr, limit);
+        return std::make_shared<OptimizedPartialSortingTransform>(header, output_order_descr, limit);
     });
 
     pipeline.addSimpleTransform([&](const Block & header, QueryPipeline::StreamType stream_type) -> ProcessorPtr

From f2e3bcdd15d5436ea2c8866457964f6c3572978f Mon Sep 17 00:00:00 2001
From: hcz <h.cz@qq.com>
Date: Mon, 25 May 2020 10:50:55 +0800
Subject: [PATCH 013/122] Improve code style in src and test files of Arrow
 formats

---
 src/Processors/Formats/Impl/ArrowBlockInputFormat.cpp |  9 +++++++--
 src/Processors/Formats/Impl/ArrowBlockInputFormat.h   |  4 ++++
 tests/queries/0_stateless/01273_arrow.reference       |  2 --
 tests/queries/0_stateless/01273_arrow.sh              | 11 -----------
 .../queries/0_stateless/01273_arrow_stream.reference  |  2 --
 tests/queries/0_stateless/01273_arrow_stream.sh       | 11 -----------
 6 files changed, 11 insertions(+), 28 deletions(-)

diff --git a/src/Processors/Formats/Impl/ArrowBlockInputFormat.cpp b/src/Processors/Formats/Impl/ArrowBlockInputFormat.cpp
index 5468fbcac16..e0eb0e21567 100644
--- a/src/Processors/Formats/Impl/ArrowBlockInputFormat.cpp
+++ b/src/Processors/Formats/Impl/ArrowBlockInputFormat.cpp
@@ -31,11 +31,15 @@ Chunk ArrowBlockInputFormat::generate()
     Chunk res;
     const Block & header = getPort().getHeader();
 
-    if (!stream && record_batch_current >= record_batch_total)
-        return res;
+    if (!stream)
+    {
+        if (record_batch_current >= record_batch_total)
+            return res;
+    }
 
     std::vector<std::shared_ptr<arrow::RecordBatch>> single_batch(1);
     arrow::Status read_status;
+
     if (stream)
     {
         read_status = stream_reader->ReadNext(&single_batch[0]);
@@ -46,6 +50,7 @@ Chunk ArrowBlockInputFormat::generate()
     {
         read_status = file_reader->ReadRecordBatch(record_batch_current, &single_batch[0]);
     }
+
     if (!read_status.ok())
         throw Exception{"Error while reading batch of Arrow data: " + read_status.ToString(),
                         ErrorCodes::CANNOT_READ_ALL_DATA};
diff --git a/src/Processors/Formats/Impl/ArrowBlockInputFormat.h b/src/Processors/Formats/Impl/ArrowBlockInputFormat.h
index 2e7f647b4f7..5ad112efde9 100644
--- a/src/Processors/Formats/Impl/ArrowBlockInputFormat.h
+++ b/src/Processors/Formats/Impl/ArrowBlockInputFormat.h
@@ -25,9 +25,13 @@ protected:
     Chunk generate() override;
 
 private:
+    // Whether to use ArrowStream format
     bool stream;
+    // This field is only used for ArrowStream format
     std::shared_ptr<arrow::RecordBatchReader> stream_reader;
+    // The following fields are used only for Arrow format
     std::shared_ptr<arrow::ipc::RecordBatchFileReader> file_reader;
+
     int record_batch_total = 0;
     int record_batch_current = 0;
 
diff --git a/tests/queries/0_stateless/01273_arrow.reference b/tests/queries/0_stateless/01273_arrow.reference
index 0f4be2c74a0..0dc503f65e4 100644
--- a/tests/queries/0_stateless/01273_arrow.reference
+++ b/tests/queries/0_stateless/01273_arrow.reference
@@ -31,8 +31,6 @@
 992
 991
 990
-ContextLock	Number of times the lock of Context was acquired or tried to acquire. This is global lock.
-Query	Number of queries to be interpreted and potentially executed. Does not include queries that failed to parse or were rejected due to AST size limits, quota limits or limits on the number of simultaneously running queries. May include internal queries initiated by ClickHouse itself. Does not count subqueries.
 original:
 -128	0	-32768	0	-2147483648	0	-9223372036854775808	0	-1.032	-1.064	string-1	fixedstring-1\0\0	2003-04-05	2003-02-03 04:05:06
 -108	108	-1016	1116	-1032	1132	-1064	1164	-1.032	-1.064	string-0	fixedstring\0\0\0\0	2001-02-03	2002-02-03 04:05:06
diff --git a/tests/queries/0_stateless/01273_arrow.sh b/tests/queries/0_stateless/01273_arrow.sh
index f659a81f118..deb0aa20ccc 100755
--- a/tests/queries/0_stateless/01273_arrow.sh
+++ b/tests/queries/0_stateless/01273_arrow.sh
@@ -34,14 +34,6 @@ ${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_numbers ORDER BY number DESC L
 
 ${CLICKHOUSE_CLIENT} --query="DROP TABLE arrow_numbers"
 
-${CLICKHOUSE_CLIENT} --query="DROP TABLE IF EXISTS arrow_events"
-${CLICKHOUSE_CLIENT} --query="CREATE TABLE arrow_events (event String, value UInt64, description String) ENGINE = Memory"
-${CLICKHOUSE_CLIENT} --query="SELECT * FROM system.events FORMAT Arrow" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_events FORMAT Arrow"
-${CLICKHOUSE_CLIENT} --query="SELECT event, description FROM arrow_events WHERE event IN ('ContextLock', 'Query') ORDER BY event"
-${CLICKHOUSE_CLIENT} --query="DROP TABLE arrow_events"
-
-
-
 ${CLICKHOUSE_CLIENT} --query="DROP TABLE IF EXISTS arrow_types1"
 ${CLICKHOUSE_CLIENT} --query="DROP TABLE IF EXISTS arrow_types2"
 ${CLICKHOUSE_CLIENT} --query="DROP TABLE IF EXISTS arrow_types3"
@@ -61,8 +53,6 @@ ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types1 values (     -128,
 # max
 ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types1 values (      127,         255,       32767,         65535,  2147483647,    4294967295, 9223372036854775807, 9223372036854775807, -1.032,     -1.064,    'string-2',             'fixedstring-2', '2004-06-07', '2004-02-03 04:05:06')"
 
-# 'SELECT -127,-128,-129,126,127,128,255,256,257,-32767,-32768,-32769,32766,32767,32768,65535,65536,65537,  -2147483647,-2147483648,-2147483649,2147483646,2147483647,2147483648,4294967295,4294967296,4294967297,   -9223372036854775807,-9223372036854775808,9223372036854775806,9223372036854775807,9223372036854775808,18446744073709551615';
-
 ${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types1 FORMAT Arrow" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types2 FORMAT Arrow"
 
 echo original:
@@ -98,7 +88,6 @@ ${CLICKHOUSE_CLIENT} --query="CREATE TABLE arrow_types5       (int8 Nullable(Int
 ${CLICKHOUSE_CLIENT} --query="CREATE TABLE arrow_types6       (int8 Nullable(Int8), uint8 Nullable(UInt8), int16 Nullable(Int16), uint16 Nullable(UInt16), int32 Nullable(Int32), uint32 Nullable(UInt32), int64 Nullable(Int64), uint64 Nullable(UInt64), float32 Nullable(Float32), float64 Nullable(Float64), string Nullable(String), fixedstring Nullable(FixedString(15)), date Nullable(Date), datetime Nullable(DateTime)) ENGINE = Memory"
 ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types5 values (               NULL,                  NULL,                  NULL,                    NULL,                  NULL,                    NULL,                  NULL,                    NULL,                      NULL,                      NULL,                    NULL,                                  NULL,                NULL,                        NULL)"
 ${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types5 ORDER BY int8 FORMAT Arrow" > ${CLICKHOUSE_TMP}/arrow_all_types_5.arrow
-#${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types5 FORMAT Arrow" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types6 FORMAT Arrow"
 ${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types5 ORDER BY int8 FORMAT Arrow" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types6 FORMAT Arrow"
 ${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types1 ORDER BY int8 FORMAT Arrow" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types6 FORMAT Arrow"
 echo dest from null:
diff --git a/tests/queries/0_stateless/01273_arrow_stream.reference b/tests/queries/0_stateless/01273_arrow_stream.reference
index 0f4be2c74a0..0dc503f65e4 100644
--- a/tests/queries/0_stateless/01273_arrow_stream.reference
+++ b/tests/queries/0_stateless/01273_arrow_stream.reference
@@ -31,8 +31,6 @@
 992
 991
 990
-ContextLock	Number of times the lock of Context was acquired or tried to acquire. This is global lock.
-Query	Number of queries to be interpreted and potentially executed. Does not include queries that failed to parse or were rejected due to AST size limits, quota limits or limits on the number of simultaneously running queries. May include internal queries initiated by ClickHouse itself. Does not count subqueries.
 original:
 -128	0	-32768	0	-2147483648	0	-9223372036854775808	0	-1.032	-1.064	string-1	fixedstring-1\0\0	2003-04-05	2003-02-03 04:05:06
 -108	108	-1016	1116	-1032	1132	-1064	1164	-1.032	-1.064	string-0	fixedstring\0\0\0\0	2001-02-03	2002-02-03 04:05:06
diff --git a/tests/queries/0_stateless/01273_arrow_stream.sh b/tests/queries/0_stateless/01273_arrow_stream.sh
index 2fc630a3a8b..a4bb403cb5a 100755
--- a/tests/queries/0_stateless/01273_arrow_stream.sh
+++ b/tests/queries/0_stateless/01273_arrow_stream.sh
@@ -34,14 +34,6 @@ ${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_numbers ORDER BY number DESC L
 
 ${CLICKHOUSE_CLIENT} --query="DROP TABLE arrow_numbers"
 
-${CLICKHOUSE_CLIENT} --query="DROP TABLE IF EXISTS arrow_events"
-${CLICKHOUSE_CLIENT} --query="CREATE TABLE arrow_events (event String, value UInt64, description String) ENGINE = Memory"
-${CLICKHOUSE_CLIENT} --query="SELECT * FROM system.events FORMAT ArrowStream" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_events FORMAT ArrowStream"
-${CLICKHOUSE_CLIENT} --query="SELECT event, description FROM arrow_events WHERE event IN ('ContextLock', 'Query') ORDER BY event"
-${CLICKHOUSE_CLIENT} --query="DROP TABLE arrow_events"
-
-
-
 ${CLICKHOUSE_CLIENT} --query="DROP TABLE IF EXISTS arrow_types1"
 ${CLICKHOUSE_CLIENT} --query="DROP TABLE IF EXISTS arrow_types2"
 ${CLICKHOUSE_CLIENT} --query="DROP TABLE IF EXISTS arrow_types3"
@@ -61,8 +53,6 @@ ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types1 values (     -128,
 # max
 ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types1 values (      127,         255,       32767,         65535,  2147483647,    4294967295, 9223372036854775807, 9223372036854775807, -1.032,     -1.064,    'string-2',             'fixedstring-2', '2004-06-07', '2004-02-03 04:05:06')"
 
-# 'SELECT -127,-128,-129,126,127,128,255,256,257,-32767,-32768,-32769,32766,32767,32768,65535,65536,65537,  -2147483647,-2147483648,-2147483649,2147483646,2147483647,2147483648,4294967295,4294967296,4294967297,   -9223372036854775807,-9223372036854775808,9223372036854775806,9223372036854775807,9223372036854775808,18446744073709551615';
-
 ${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types1 FORMAT ArrowStream" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types2 FORMAT ArrowStream"
 
 echo original:
@@ -98,7 +88,6 @@ ${CLICKHOUSE_CLIENT} --query="CREATE TABLE arrow_types5       (int8 Nullable(Int
 ${CLICKHOUSE_CLIENT} --query="CREATE TABLE arrow_types6       (int8 Nullable(Int8), uint8 Nullable(UInt8), int16 Nullable(Int16), uint16 Nullable(UInt16), int32 Nullable(Int32), uint32 Nullable(UInt32), int64 Nullable(Int64), uint64 Nullable(UInt64), float32 Nullable(Float32), float64 Nullable(Float64), string Nullable(String), fixedstring Nullable(FixedString(15)), date Nullable(Date), datetime Nullable(DateTime)) ENGINE = Memory"
 ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types5 values (               NULL,                  NULL,                  NULL,                    NULL,                  NULL,                    NULL,                  NULL,                    NULL,                      NULL,                      NULL,                    NULL,                                  NULL,                NULL,                        NULL)"
 ${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types5 ORDER BY int8 FORMAT ArrowStream" > ${CLICKHOUSE_TMP}/arrow_all_types_5.arrow
-#${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types5 FORMAT ArrowStream" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types6 FORMAT ArrowStream"
 ${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types5 ORDER BY int8 FORMAT ArrowStream" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types6 FORMAT ArrowStream"
 ${CLICKHOUSE_CLIENT} --query="SELECT * FROM arrow_types1 ORDER BY int8 FORMAT ArrowStream" | ${CLICKHOUSE_CLIENT} --query="INSERT INTO arrow_types6 FORMAT ArrowStream"
 echo dest from null:

From 548fc7141a47583ae33ab02e716cd338a0b7300e Mon Sep 17 00:00:00 2001
From: Ivan Blinkov <github@blinkov.ru>
Date: Mon, 25 May 2020 21:12:03 +0300
Subject: [PATCH 014/122] Enable log_queries by default

---
 src/Core/Settings.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/Core/Settings.h b/src/Core/Settings.h
index be7c09ec6c2..142e0872d72 100644
--- a/src/Core/Settings.h
+++ b/src/Core/Settings.h
@@ -156,7 +156,7 @@ struct Settings : public SettingsCollection<Settings>
     M(SettingUInt64, priority, 0, "Priority of the query. 1 - the highest, higher value - lower priority; 0 - do not use priorities.", 0) \
     M(SettingInt64, os_thread_priority, 0, "If non zero - set corresponding 'nice' value for query processing threads. Can be used to adjust query priority for OS scheduler.", 0) \
     \
-    M(SettingBool, log_queries, 0, "Log requests and write the log to the system table.", 0) \
+    M(SettingBool, log_queries, 1, "Log requests and write the log to the system table.", 0) \
     M(SettingLogQueriesType, log_queries_min_type, QueryLogElementType::QUERY_START, "query_log minimal type to log, possible values (from low to high): QUERY_START, QUERY_FINISH, EXCEPTION_BEFORE_START, EXCEPTION_WHILE_PROCESSING.", 0) \
     M(SettingUInt64, log_queries_cut_to_length, 100000, "If query length is greater than specified threshold (in bytes), then cut query when writing to query log. Also limit length of printed query in ordinary text log.", 0) \
     \

From 7142e68d7abab49f0d734e907f07ac139c6702ec Mon Sep 17 00:00:00 2001
From: Alexander Kuzmenkov <akuzm@yandex-team.ru>
Date: Tue, 26 May 2020 08:54:04 +0300
Subject: [PATCH 015/122] fixup

---
 .../AggregateFunctionEntropy.h                |  6 ++---
 .../AggregateFunctionGroupUniqArray.h         |  9 +++----
 .../QuantileExactWeighted.h                   |  3 +--
 src/Common/HashTable/ClearableHashMap.h       | 11 ++++++++
 src/Common/HashTable/ClearableHashSet.h       | 12 +++++++++
 src/Common/HashTable/HashMap.h                | 11 ++++++++
 src/Common/HashTable/HashSet.h                | 17 +++++++++++++
 src/Common/HashTable/HashTable.h              | 21 +++-------------
 src/Common/SpaceSaving.h                      |  3 +--
 src/Functions/array/arrayDistinct.cpp         | 12 ++++-----
 src/Functions/array/arrayEnumerateExtended.h  | 25 ++++++++-----------
 src/Functions/array/arrayEnumerateRanked.h    |  5 ++--
 src/Functions/array/arrayIntersect.cpp        | 23 ++++++++---------
 src/Functions/array/arrayUniq.cpp             | 20 +++++++--------
 src/Interpreters/tests/hash_map3.cpp          |  6 +++--
 .../functions_with_hash_tables.xml            |  8 +++---
 16 files changed, 108 insertions(+), 84 deletions(-)

diff --git a/src/AggregateFunctions/AggregateFunctionEntropy.h b/src/AggregateFunctions/AggregateFunctionEntropy.h
index 59cacd0e9ca..ff233a5ac93 100644
--- a/src/AggregateFunctions/AggregateFunctionEntropy.h
+++ b/src/AggregateFunctions/AggregateFunctionEntropy.h
@@ -23,12 +23,10 @@ struct EntropyData
 {
     using Weight = UInt64;
 
-    using HashingMap = HASH_TABLE_WITH_STACK_MEMORY(HashMap, Value, Weight,
-        HashCRC32<Value>, HashTableGrower<4>);
+    using HashingMap = HashMapWithStackMemory<Value, Weight, HashCRC32<Value>, 4>;
 
     /// For the case of pre-hashed values.
-    using TrivialMap = HASH_TABLE_WITH_STACK_MEMORY(HashMap, Value, Weight,
-        UInt128TrivialHash, HashTableGrower<4>);
+    using TrivialMap = HashMapWithStackMemory<Value, Weight, UInt128TrivialHash, 4>;
 
     using Map = std::conditional_t<std::is_same_v<UInt128, Value>, TrivialMap, HashingMap>;
 
diff --git a/src/AggregateFunctions/AggregateFunctionGroupUniqArray.h b/src/AggregateFunctions/AggregateFunctionGroupUniqArray.h
index af97df209f6..88b1c87f526 100644
--- a/src/AggregateFunctions/AggregateFunctionGroupUniqArray.h
+++ b/src/AggregateFunctions/AggregateFunctionGroupUniqArray.h
@@ -28,8 +28,7 @@ template <typename T>
 struct AggregateFunctionGroupUniqArrayData
 {
     /// When creating, the hash table must be small.
-    using Set = HASH_TABLE_WITH_STACK_MEMORY(HashSet, T, DefaultHash<T>,
-        HashTableGrower<4>);
+    using Set = HashSetWithStackMemory<T, DefaultHash<T>, 4>;
 
     Set value;
 };
@@ -122,10 +121,10 @@ public:
 /// Generic implementation, it uses serialized representation as object descriptor.
 struct AggregateFunctionGroupUniqArrayGenericData
 {
-    static constexpr size_t INITIAL_SIZE_DEGREE = 1; /// adjustable
+    static constexpr size_t INITIAL_SIZE_DEGREE = 3; /// adjustable
 
-    using Set = HASH_TABLE_WITH_STACK_MEMORY(HashSetWithSavedHash,
-        StringRef, StringRefHash, HashTableGrower<INITIAL_SIZE_DEGREE>);
+    using Set = HashSetWithSavedHashWithStackMemory<StringRef, StringRefHash,
+        INITIAL_SIZE_DEGREE>;
 
     Set value;
 };
diff --git a/src/AggregateFunctions/QuantileExactWeighted.h b/src/AggregateFunctions/QuantileExactWeighted.h
index 7731ff980fe..666dbd6b622 100644
--- a/src/AggregateFunctions/QuantileExactWeighted.h
+++ b/src/AggregateFunctions/QuantileExactWeighted.h
@@ -33,8 +33,7 @@ struct QuantileExactWeighted
     using Hasher = std::conditional_t<std::is_same_v<Value, Decimal128>, Int128Hash, HashCRC32<UnderlyingType>>;
 
     /// When creating, the hash table must be small.
-    using Map = HASH_TABLE_WITH_STACK_MEMORY(HashMap, UnderlyingType, Weight,
-        Hasher, HashTableGrower<4>);
+    using Map = HashMapWithStackMemory<UnderlyingType, Weight, Hasher, 4>;
 
     Map map;
 
diff --git a/src/Common/HashTable/ClearableHashMap.h b/src/Common/HashTable/ClearableHashMap.h
index 4370f6b6dc7..fda01dcf4bc 100644
--- a/src/Common/HashTable/ClearableHashMap.h
+++ b/src/Common/HashTable/ClearableHashMap.h
@@ -43,3 +43,14 @@ public:
         this->m_size = 0;
     }
 };
+
+template <typename Key, typename Mapped, typename Hash,
+    size_t initial_size_degree>
+using ClearableHashMapWithStackMemory = ClearableHashMap<
+    Key,
+    Mapped,
+    Hash,
+    HashTableGrower<initial_size_degree>,
+    HashTableAllocatorWithStackMemory<
+        (1ULL << initial_size_degree)
+        * sizeof(ClearableHashMapCell<Key, Mapped, Hash>)>>;
diff --git a/src/Common/HashTable/ClearableHashSet.h b/src/Common/HashTable/ClearableHashSet.h
index 824ec9d8e5f..dc057afacd8 100644
--- a/src/Common/HashTable/ClearableHashSet.h
+++ b/src/Common/HashTable/ClearableHashSet.h
@@ -84,3 +84,15 @@ public:
         this->m_size = 0;
     }
 };
+
+template <typename Key, typename Hash, size_t initial_size_degree>
+using ClearableHashSetWithStackMemory = ClearableHashSet<
+    Key,
+    Hash,
+    HashTableGrower<initial_size_degree>,
+    HashTableAllocatorWithStackMemory<
+        (1ULL << initial_size_degree)
+        * sizeof(
+            ClearableHashTableCell<
+                Key,
+                HashTableCell<Key, Hash, ClearableHashSetState>>)>>;
diff --git a/src/Common/HashTable/HashMap.h b/src/Common/HashTable/HashMap.h
index cdc4a003af8..3ecbd9f263c 100644
--- a/src/Common/HashTable/HashMap.h
+++ b/src/Common/HashTable/HashMap.h
@@ -239,3 +239,14 @@ template <
     typename Grower = HashTableGrower<>,
     typename Allocator = HashTableAllocator>
 using HashMapWithSavedHash = HashMapTable<Key, HashMapCellWithSavedHash<Key, Mapped, Hash>, Hash, Grower, Allocator>;
+
+template <typename Key, typename Mapped, typename Hash,
+    size_t initial_size_degree>
+using HashMapWithStackMemory = HashMapTable<
+    Key,
+    HashMapCellWithSavedHash<Key, Mapped, Hash>,
+    Hash,
+    HashTableGrower<initial_size_degree>,
+    HashTableAllocatorWithStackMemory<
+        (1ULL << initial_size_degree)
+        * sizeof(HashMapCellWithSavedHash<Key, Mapped, Hash>)>>;
diff --git a/src/Common/HashTable/HashSet.h b/src/Common/HashTable/HashSet.h
index c1970e898a1..c79e05073fc 100644
--- a/src/Common/HashTable/HashSet.h
+++ b/src/Common/HashTable/HashSet.h
@@ -93,6 +93,14 @@ template
 >
 using HashSet = HashSetTable<Key, HashTableCell<Key, Hash>, Hash, Grower, Allocator>;
 
+template <typename Key, typename Hash, size_t initial_size_degree>
+using HashSetWithStackMemory = HashSet<
+    Key,
+    Hash,
+    HashTableGrower<initial_size_degree>,
+    HashTableAllocatorWithStackMemory<
+        (1ULL << initial_size_degree)
+        * sizeof(HashTableCell<Key, Hash>)>>;
 
 template
 <
@@ -102,3 +110,12 @@ template
     typename Allocator = HashTableAllocator
 >
 using HashSetWithSavedHash = HashSetTable<Key, HashSetCellWithSavedHash<Key, Hash>, Hash, Grower, Allocator>;
+
+template <typename Key, typename Hash, size_t initial_size_degree>
+using HashSetWithSavedHashWithStackMemory = HashSetWithSavedHash<
+    Key,
+    Hash,
+    HashTableGrower<initial_size_degree>,
+    HashTableAllocatorWithStackMemory<
+        (1ULL << initial_size_degree)
+        * sizeof(HashSetCellWithSavedHash<Key, Hash>)>>;
diff --git a/src/Common/HashTable/HashTable.h b/src/Common/HashTable/HashTable.h
index 2f7272ca8ec..9d66c7b7a89 100644
--- a/src/Common/HashTable/HashTable.h
+++ b/src/Common/HashTable/HashTable.h
@@ -208,6 +208,7 @@ struct HashTableGrower
     /// The state of this structure is enough to get the buffer size of the hash table.
 
     UInt8 size_degree = initial_size_degree;
+    static constexpr auto initial_count = 1ULL << initial_size_degree;
 
     /// The size of the hash table in the cells.
     size_t bufSize() const               { return 1ULL << size_degree; }
@@ -255,6 +256,7 @@ struct HashTableGrower
 template <size_t key_bits>
 struct HashTableFixedGrower
 {
+    static constexpr auto initial_count = 1ULL << key_bits;
     size_t bufSize() const               { return 1ULL << key_bits; }
     size_t place(size_t x) const         { return x; }
     /// You could write __builtin_unreachable(), but the compiler does not optimize everything, and it turns out less efficiently.
@@ -308,14 +310,6 @@ struct ZeroValueStorage<false, Cell>
     const Cell * zeroValue() const { return nullptr; }
 };
 
-// These templates give the initial hash table size, so that we can check
-// that it is in sync with initial allocator size.
-template <typename GenericGrower>
-constexpr size_t growerInitialCount = 0;
-
-template <size_t initial_size_degree>
-constexpr size_t growerInitialCount<HashTableGrower<initial_size_degree>>
-    = 1ULL << initial_size_degree;
 
 // The HashTable
 template
@@ -337,7 +331,7 @@ public:
     // Export the initial buffer sizes for the ease of using allocators with
     // inline memory.
     static constexpr size_t initial_buffer_bytes
-        = growerInitialCount<Grower> * sizeof(Cell);
+        = Grower::initial_count * sizeof(Cell);
 
     // If we use an allocator with inline memory, check that the initial
     // size of the hash table is in sync with the amount of this memory.
@@ -1096,12 +1090,3 @@ public:
 #endif
 };
 
-// A helper macro that declares hash table with allocator with stack memory,
-// and the initial size of the allocator is in sync with initial size of the
-// hash table.
-#define HASH_TABLE_WITH_STACK_MEMORY(HASH_TABLE_VARIANT, ...) \
-    HASH_TABLE_VARIANT<__VA_ARGS__, \
-        HashTableAllocatorWithStackMemory< \
-            HASH_TABLE_VARIANT<__VA_ARGS__>::initial_buffer_bytes \
-        > \
-    >
diff --git a/src/Common/SpaceSaving.h b/src/Common/SpaceSaving.h
index 47b49add060..56063340240 100644
--- a/src/Common/SpaceSaving.h
+++ b/src/Common/SpaceSaving.h
@@ -378,8 +378,7 @@ private:
             counter_map[counter->key] = counter;
     }
 
-    using CounterMap = HASH_TABLE_WITH_STACK_MEMORY(HashMap,
-        TKey, Counter *, Hash, HashTableGrower<4>);
+    using CounterMap = HashMapWithStackMemory<TKey, Counter *, Hash, 4>;
 
     CounterMap counter_map;
     std::vector<Counter *> counter_list;
diff --git a/src/Functions/array/arrayDistinct.cpp b/src/Functions/array/arrayDistinct.cpp
index d3da77e43d2..bb26567e8a5 100644
--- a/src/Functions/array/arrayDistinct.cpp
+++ b/src/Functions/array/arrayDistinct.cpp
@@ -153,8 +153,8 @@ bool FunctionArrayDistinct::executeNumber(
     if (nullable_col)
         src_null_map = &nullable_col->getNullMapData();
 
-    using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashSet,
-        T, DefaultHash<T>, HashTableGrower<INITIAL_SIZE_DEGREE>);
+    using Set = ClearableHashSetWithStackMemory<T, DefaultHash<T>,
+        INITIAL_SIZE_DEGREE>;
 
     Set set;
 
@@ -199,8 +199,8 @@ bool FunctionArrayDistinct::executeString(
 
     ColumnString & res_data_column_string = typeid_cast<ColumnString &>(res_data_col);
 
-    using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashSet,
-        StringRef, StringRefHash, HashTableGrower<INITIAL_SIZE_DEGREE>);
+    using Set = ClearableHashSetWithStackMemory<StringRef, StringRefHash,
+        INITIAL_SIZE_DEGREE>;
 
     const PaddedPODArray<UInt8> * src_null_map = nullptr;
 
@@ -245,8 +245,8 @@ void FunctionArrayDistinct::executeHashed(
     ColumnArray::Offsets & res_offsets,
     const ColumnNullable * nullable_col)
 {
-    using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashSet,
-        UInt128, UInt128TrivialHash, HashTableGrower<INITIAL_SIZE_DEGREE>);
+    using Set = ClearableHashSetWithStackMemory<UInt128, UInt128TrivialHash,
+        INITIAL_SIZE_DEGREE>;
 
     const PaddedPODArray<UInt8> * src_null_map = nullptr;
 
diff --git a/src/Functions/array/arrayEnumerateExtended.h b/src/Functions/array/arrayEnumerateExtended.h
index e97766c5827..e2f83ae8826 100644
--- a/src/Functions/array/arrayEnumerateExtended.h
+++ b/src/Functions/array/arrayEnumerateExtended.h
@@ -64,45 +64,40 @@ private:
     template <typename T>
     struct MethodOneNumber
     {
-        using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
-            T, UInt32, DefaultHash<T>,
-            HashTableGrower<INITIAL_SIZE_DEGREE>);
+        using Set = ClearableHashMapWithStackMemory<T, UInt32, DefaultHash<T>,
+            INITIAL_SIZE_DEGREE>;
 
         using Method = ColumnsHashing::HashMethodOneNumber<typename Set::value_type, UInt32, T, false>;
     };
 
     struct MethodString
     {
-        using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
-            StringRef, UInt32, StringRefHash,
-            HashTableGrower<INITIAL_SIZE_DEGREE>);
+        using Set = ClearableHashMapWithStackMemory<StringRef, UInt32, StringRefHash,
+            INITIAL_SIZE_DEGREE>;
 
         using Method = ColumnsHashing::HashMethodString<typename Set::value_type, UInt32, false, false>;
     };
 
     struct MethodFixedString
     {
-        using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
-            StringRef, UInt32, StringRefHash,
-            HashTableGrower<INITIAL_SIZE_DEGREE>);
+        using Set = ClearableHashMapWithStackMemory<StringRef, UInt32, StringRefHash,
+            INITIAL_SIZE_DEGREE>;
 
         using Method = ColumnsHashing::HashMethodFixedString<typename Set::value_type, UInt32, false, false>;
     };
 
     struct MethodFixed
     {
-        using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
-            UInt128, UInt32, UInt128HashCRC32,
-            HashTableGrower<INITIAL_SIZE_DEGREE>);
+        using Set = ClearableHashMapWithStackMemory<UInt128, UInt32, UInt128HashCRC32,
+            INITIAL_SIZE_DEGREE>;
 
         using Method = ColumnsHashing::HashMethodKeysFixed<typename Set::value_type, UInt128, UInt32, false, false, false>;
     };
 
     struct MethodHashed
     {
-        using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
-            UInt128, UInt32, UInt128TrivialHash,
-            HashTableGrower<INITIAL_SIZE_DEGREE>);
+        using Set = ClearableHashMapWithStackMemory<UInt128, UInt32, UInt128TrivialHash,
+            INITIAL_SIZE_DEGREE>;
 
         using Method = ColumnsHashing::HashMethodHashed<typename Set::value_type, UInt32, false>;
     };
diff --git a/src/Functions/array/arrayEnumerateRanked.h b/src/Functions/array/arrayEnumerateRanked.h
index 1700260dd9f..c021b5fbfd1 100644
--- a/src/Functions/array/arrayEnumerateRanked.h
+++ b/src/Functions/array/arrayEnumerateRanked.h
@@ -308,9 +308,8 @@ void FunctionArrayEnumerateRankedExtended<Derived>::executeMethodImpl(
     const size_t depth_to_look = arrays_depths.max_array_depth;
     const auto & offsets = *offsets_by_depth[depth_to_look - 1];
 
-    using Map = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
-        UInt128, UInt32, UInt128TrivialHash,
-        HashTableGrower<INITIAL_SIZE_DEGREE>);
+    using Map = ClearableHashMapWithStackMemory<UInt128, UInt32,
+        UInt128TrivialHash, INITIAL_SIZE_DEGREE>;
 
     Map indices;
 
diff --git a/src/Functions/array/arrayIntersect.cpp b/src/Functions/array/arrayIntersect.cpp
index acd162974c9..3525914b0d3 100644
--- a/src/Functions/array/arrayIntersect.cpp
+++ b/src/Functions/array/arrayIntersect.cpp
@@ -418,18 +418,15 @@ void FunctionArrayIntersect::executeImpl(Block & block, const ColumnNumbers & ar
     TypeListNativeNumbers::forEach(NumberExecutor(arrays, not_nullable_nested_return_type, result_column));
     TypeListDecimalNumbers::forEach(DecimalExecutor(arrays, not_nullable_nested_return_type, result_column));
 
-    using DateMap = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
-        DataTypeDate::FieldType, size_t, DefaultHash<DataTypeDate::FieldType>,
-        HashTableGrower<INITIAL_SIZE_DEGREE>);
+    using DateMap = ClearableHashMapWithStackMemory<DataTypeDate::FieldType,
+        size_t, DefaultHash<DataTypeDate::FieldType>, INITIAL_SIZE_DEGREE>;
 
-    using DateTimeMap = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
+    using DateTimeMap = ClearableHashMapWithStackMemory<
         DataTypeDateTime::FieldType, size_t,
-        DefaultHash<DataTypeDateTime::FieldType>,
-        HashTableGrower<INITIAL_SIZE_DEGREE>);
+        DefaultHash<DataTypeDateTime::FieldType>, INITIAL_SIZE_DEGREE>;
 
-    using StringMap = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
-        StringRef, size_t, StringRefHash,
-        HashTableGrower<INITIAL_SIZE_DEGREE>);
+    using StringMap = ClearableHashMapWithStackMemory<StringRef, size_t,
+        StringRefHash, INITIAL_SIZE_DEGREE>;
 
     if (!result_column)
     {
@@ -457,8 +454,8 @@ void FunctionArrayIntersect::executeImpl(Block & block, const ColumnNumbers & ar
 template <typename T, size_t>
 void FunctionArrayIntersect::NumberExecutor::operator()()
 {
-    using Map = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
-        T, size_t, DefaultHash<T>, HashTableGrower<INITIAL_SIZE_DEGREE>);
+    using Map = ClearableHashMapWithStackMemory<T, size_t, DefaultHash<T>,
+        INITIAL_SIZE_DEGREE>;
 
     if (!result && typeid_cast<const DataTypeNumber<T> *>(data_type.get()))
         result = execute<Map, ColumnVector<T>, true>(arrays, ColumnVector<T>::create());
@@ -467,8 +464,8 @@ void FunctionArrayIntersect::NumberExecutor::operator()()
 template <typename T, size_t>
 void FunctionArrayIntersect::DecimalExecutor::operator()()
 {
-    using Map = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashMap,
-        T, size_t, DefaultHash<T>, HashTableGrower<INITIAL_SIZE_DEGREE>);
+    using Map = ClearableHashMapWithStackMemory<T, size_t, DefaultHash<T>,
+        INITIAL_SIZE_DEGREE);
 
     if (!result)
         if (auto * decimal = typeid_cast<const DataTypeDecimal<T> *>(data_type.get()))
diff --git a/src/Functions/array/arrayUniq.cpp b/src/Functions/array/arrayUniq.cpp
index a83c535d88d..d9a44f6d8d3 100644
--- a/src/Functions/array/arrayUniq.cpp
+++ b/src/Functions/array/arrayUniq.cpp
@@ -66,40 +66,40 @@ private:
     template <typename T>
     struct MethodOneNumber
     {
-        using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashSet,
-            T, DefaultHash<T>, HashTableGrower<INITIAL_SIZE_DEGREE>);
+        using Set = ClearableHashSetWithStackMemory<T, DefaultHash<T>,
+            INITIAL_SIZE_DEGREE>;
 
         using Method = ColumnsHashing::HashMethodOneNumber<typename Set::value_type, void, T, false>;
     };
 
     struct MethodString
     {
-        using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashSet,
-            StringRef, StringRefHash, HashTableGrower<INITIAL_SIZE_DEGREE>);
+        using Set = ClearableHashSetWithStackMemory<StringRef, StringRefHash,
+            INITIAL_SIZE_DEGREE>;
 
         using Method = ColumnsHashing::HashMethodString<typename Set::value_type, void, false, false>;
     };
 
     struct MethodFixedString
     {
-        using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashSet,
-            StringRef, StringRefHash, HashTableGrower<INITIAL_SIZE_DEGREE>);
+        using Set = ClearableHashSetWithStackMemory<StringRef, StringRefHash,
+            INITIAL_SIZE_DEGREE>;
 
         using Method = ColumnsHashing::HashMethodFixedString<typename Set::value_type, void, false, false>;
     };
 
     struct MethodFixed
     {
-        using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashSet,
-            UInt128, UInt128HashCRC32, HashTableGrower<INITIAL_SIZE_DEGREE>);
+        using Set = ClearableHashSetWithStackMemory<UInt128, UInt128HashCRC32,
+            INITIAL_SIZE_DEGREE>;
 
         using Method = ColumnsHashing::HashMethodKeysFixed<typename Set::value_type, UInt128, void, false, false, false>;
     };
 
     struct MethodHashed
     {
-        using Set = HASH_TABLE_WITH_STACK_MEMORY(ClearableHashSet,
-            UInt128, UInt128TrivialHash, HashTableGrower<INITIAL_SIZE_DEGREE>);
+        using Set = ClearableHashSetWithStackMemory<UInt128, UInt128TrivialHash,
+            INITIAL_SIZE_DEGREE>;
 
         using Method = ColumnsHashing::HashMethodHashed<typename Set::value_type, void, false>;
     };
diff --git a/src/Interpreters/tests/hash_map3.cpp b/src/Interpreters/tests/hash_map3.cpp
index 804b084b2a1..1b4ce8eac77 100644
--- a/src/Interpreters/tests/hash_map3.cpp
+++ b/src/Interpreters/tests/hash_map3.cpp
@@ -61,11 +61,13 @@ struct Grower : public HashTableGrower<2>
 
 int main(int, char **)
 {
-    using Map = HASH_TABLE_WITH_STACK_MEMORY(HashMapWithDump,
+    using Map = HashMapWithDump<
         StringRef,
         UInt64,
         SimpleHash,
-        Grower);
+        Grower,
+        HashTableAllocatorWithStackMemory<
+            4 * sizeof(HashMapCell<StringRef, UInt64, SimpleHash>)>>;
 
     Map map;
 
diff --git a/tests/performance/functions_with_hash_tables.xml b/tests/performance/functions_with_hash_tables.xml
index 3ff2633c4e7..b247f0e4f6f 100644
--- a/tests/performance/functions_with_hash_tables.xml
+++ b/tests/performance/functions_with_hash_tables.xml
@@ -1,8 +1,8 @@
 <test>
-    <query>select arrayUniq(range(1 + (number % 10) * 10)) from numbers(1000000) format Null</query>
-    <query>select arrayDistinct(range(1 + (number % 10) * 10)) from numbers(1000000) format Null</query>
-    <query>select arrayEnumerateUniq(range(1 + (number % 10) * 10)) from numbers(1000000) format Null</query>
+    <query>select arrayUniq(range(1 + (number % 100) * 10)) from numbers(1000000) format Null</query>
+    <query>select arrayDistinct(range(1 + (number % 100) * 10)) from numbers(1000000) format Null</query>
+    <query>select arrayEnumerateUniq(range(1 + (number % 100) * 10)) from numbers(1000000) format Null</query>
     <query>select arrayIntersect(range((1 + number % 10)), range(1, (1 + number % 10) + 1)) from numbers(1000000) format Null</query>
-    <query>select groupUniqArray(number / 10) from numbers(1000 * 1000) group by number / 1000 format Null</query>
+    <query>select groupUniqArray(rand() % 100) from numbers(1000 * 1000) group by number / 1000 format Null</query>
     <query>select entropy(number / 10) from numbers(1000 * 1000) group by number / 1000 format Null</query>
 </test>

From bf941755f2e776304d0c318d8d0aa6dc6dc068b3 Mon Sep 17 00:00:00 2001
From: Alexander Kuzmenkov <akuzm@yandex-team.ru>
Date: Tue, 26 May 2020 09:00:51 +0300
Subject: [PATCH 016/122] fixup

---
 src/Common/HashTable/HashTable.h       | 8 ++------
 src/Functions/array/arrayIntersect.cpp | 2 +-
 2 files changed, 3 insertions(+), 7 deletions(-)

diff --git a/src/Common/HashTable/HashTable.h b/src/Common/HashTable/HashTable.h
index 9d66c7b7a89..528e719c05b 100644
--- a/src/Common/HashTable/HashTable.h
+++ b/src/Common/HashTable/HashTable.h
@@ -328,13 +328,10 @@ class HashTable :
     protected ZeroValueStorage<Cell::need_zero_value_storage, Cell>     /// empty base optimization
 {
 public:
-    // Export the initial buffer sizes for the ease of using allocators with
-    // inline memory.
-    static constexpr size_t initial_buffer_bytes
-        = Grower::initial_count * sizeof(Cell);
-
     // If we use an allocator with inline memory, check that the initial
     // size of the hash table is in sync with the amount of this memory.
+    static constexpr size_t initial_buffer_bytes
+        = Grower::initial_count * sizeof(Cell);
     static_assert(allocatorInitialBytes<Allocator> == 0
         || allocatorInitialBytes<Allocator> == initial_buffer_bytes);
 
@@ -1089,4 +1086,3 @@ public:
     }
 #endif
 };
-
diff --git a/src/Functions/array/arrayIntersect.cpp b/src/Functions/array/arrayIntersect.cpp
index 3525914b0d3..24db3c0cd08 100644
--- a/src/Functions/array/arrayIntersect.cpp
+++ b/src/Functions/array/arrayIntersect.cpp
@@ -465,7 +465,7 @@ template <typename T, size_t>
 void FunctionArrayIntersect::DecimalExecutor::operator()()
 {
     using Map = ClearableHashMapWithStackMemory<T, size_t, DefaultHash<T>,
-        INITIAL_SIZE_DEGREE);
+        INITIAL_SIZE_DEGREE>;
 
     if (!result)
         if (auto * decimal = typeid_cast<const DataTypeDecimal<T> *>(data_type.get()))

From 27df8d5aca9ee12a82a6f83337fd65bfd256db4d Mon Sep 17 00:00:00 2001
From: Alexander Kuzmenkov <akuzm@yandex-team.ru>
Date: Tue, 26 May 2020 09:06:46 +0300
Subject: [PATCH 017/122] fixup

---
 tests/performance/functions_with_hash_tables.xml | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/tests/performance/functions_with_hash_tables.xml b/tests/performance/functions_with_hash_tables.xml
index b247f0e4f6f..72570a5925c 100644
--- a/tests/performance/functions_with_hash_tables.xml
+++ b/tests/performance/functions_with_hash_tables.xml
@@ -1,8 +1,8 @@
 <test>
-    <query>select arrayUniq(range(1 + (number % 100) * 10)) from numbers(1000000) format Null</query>
-    <query>select arrayDistinct(range(1 + (number % 100) * 10)) from numbers(1000000) format Null</query>
-    <query>select arrayEnumerateUniq(range(1 + (number % 100) * 10)) from numbers(1000000) format Null</query>
-    <query>select arrayIntersect(range((1 + number % 10)), range(1, (1 + number % 10) + 1)) from numbers(1000000) format Null</query>
+    <query>select arrayUniq(range(1 + (number % 100) * 10)) from numbers(100000) format Null</query>
+    <query>select arrayDistinct(range(1 + (number % 100) * 10)) from numbers(10000) format Null</query>
+    <query>select arrayEnumerateUniq(range(1 + (number % 100) * 10)) from numbers(100000) format Null</query>
+    <query>select arrayIntersect(range((1 + number % 100)), range(1, (1 + number % 100) + 1)) from numbers(100000) format Null</query>
     <query>select groupUniqArray(rand() % 100) from numbers(1000 * 1000) group by number / 1000 format Null</query>
     <query>select entropy(number / 10) from numbers(1000 * 1000) group by number / 1000 format Null</query>
 </test>

From 5780240e884c9c3915721f159b3c5fd33b40111d Mon Sep 17 00:00:00 2001
From: alesapin <alesapin@gmail.com>
Date: Tue, 26 May 2020 12:46:04 +0300
Subject: [PATCH 018/122] Fix primry key check

---
 src/Storages/MergeTree/IMergeTreeDataPart.cpp               | 4 ++--
 .../integration/test_version_update_after_mutation/test.py  | 6 +++---
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/src/Storages/MergeTree/IMergeTreeDataPart.cpp b/src/Storages/MergeTree/IMergeTreeDataPart.cpp
index 6e048a956da..78548cdddb2 100644
--- a/src/Storages/MergeTree/IMergeTreeDataPart.cpp
+++ b/src/Storages/MergeTree/IMergeTreeDataPart.cpp
@@ -841,7 +841,7 @@ void IMergeTreeDataPart::checkConsistencyBase() const
 
     if (!checksums.empty())
     {
-        if (storage.hasPrimaryKey() && !checksums.files.count("primary.idx"))
+        if (storage.isPrimaryKeyDefined() && !checksums.files.count("primary.idx"))
             throw Exception("No checksum for primary.idx", ErrorCodes::NO_FILE_IN_DATA_PART);
 
         if (storage.format_version >= MERGE_TREE_DATA_MIN_FORMAT_VERSION_WITH_CUSTOM_PARTITIONING)
@@ -875,7 +875,7 @@ void IMergeTreeDataPart::checkConsistencyBase() const
         };
 
         /// Check that the primary key index is not empty.
-        if (storage.hasPrimaryKey())
+        if (storage.isPrimaryKeyDefined())
             check_file_not_empty(volume->getDisk(), path + "primary.idx");
 
         if (storage.format_version >= MERGE_TREE_DATA_MIN_FORMAT_VERSION_WITH_CUSTOM_PARTITIONING)
diff --git a/tests/integration/test_version_update_after_mutation/test.py b/tests/integration/test_version_update_after_mutation/test.py
index 3f2063bdddb..9e3471175fa 100644
--- a/tests/integration/test_version_update_after_mutation/test.py
+++ b/tests/integration/test_version_update_after_mutation/test.py
@@ -6,9 +6,9 @@ from helpers.test_tools import assert_eq_with_retry
 
 cluster = ClickHouseCluster(__file__)
 
-node1 = cluster.add_instance('node1', with_zookeeper=True, image='yandex/clickhouse-server:20.1.6.3', with_installed_binary=True, stay_alive=True)
-node2 = cluster.add_instance('node2', with_zookeeper=True, image='yandex/clickhouse-server:20.1.6.3', with_installed_binary=True, stay_alive=True)
-node3 = cluster.add_instance('node3', with_zookeeper=True, image='yandex/clickhouse-server:20.1.6.3', with_installed_binary=True, stay_alive=True)
+node1 = cluster.add_instance('node1', with_zookeeper=True, image='yandex/clickhouse-server:20.1.10.70', with_installed_binary=True, stay_alive=True)
+node2 = cluster.add_instance('node2', with_zookeeper=True, image='yandex/clickhouse-server:20.1.10.70', with_installed_binary=True, stay_alive=True)
+node3 = cluster.add_instance('node3', with_zookeeper=True, image='yandex/clickhouse-server:20.1.10.70', with_installed_binary=True, stay_alive=True)
 
 @pytest.fixture(scope="module")
 def start_cluster():

From 12384551da5fd4d280eb95c356da449c31761d01 Mon Sep 17 00:00:00 2001
From: Nikolai Kochetov <nik-kochetov@yandex-team.ru>
Date: Tue, 26 May 2020 16:19:24 +0300
Subject: [PATCH 019/122] Fix totals and extremes header for views.

---
 src/Processors/Pipe.h            |  2 +-
 src/Processors/QueryPipeline.cpp | 13 ++++++++++-
 src/Processors/QueryPipeline.h   |  3 ++-
 src/Storages/StorageView.cpp     | 37 ++++++++++++++++++++++----------
 4 files changed, 41 insertions(+), 14 deletions(-)

diff --git a/src/Processors/Pipe.h b/src/Processors/Pipe.h
index 984fa7605c6..ec5514915a7 100644
--- a/src/Processors/Pipe.h
+++ b/src/Processors/Pipe.h
@@ -23,7 +23,7 @@ public:
     /// Will connect pipes outputs with transform inputs automatically.
     Pipe(Pipes && pipes, ProcessorPtr transform);
     /// Create pipe from output port. If pipe was created that way, it possibly will not have tree shape.
-    Pipe(OutputPort * port);
+    explicit Pipe(OutputPort * port);
 
     Pipe(const Pipe & other) = delete;
     Pipe(Pipe && other) = default;
diff --git a/src/Processors/QueryPipeline.cpp b/src/Processors/QueryPipeline.cpp
index 13787a3fd3b..26736672412 100644
--- a/src/Processors/QueryPipeline.cpp
+++ b/src/Processors/QueryPipeline.cpp
@@ -703,6 +703,11 @@ void QueryPipeline::initRowsBeforeLimit()
 Pipe QueryPipeline::getPipe() &&
 {
     resize(1);
+    return std::move(std::move(*this).getPipes()[0]);
+}
+
+Pipes QueryPipeline::getPipes() &&
+{
     Pipe pipe(std::move(processors), streams.at(0), totals_having_port, extremes_port);
     pipe.max_parallel_streams = streams.maxParallelStreams();
 
@@ -721,7 +726,13 @@ Pipe QueryPipeline::getPipe() &&
     if (extremes_port)
         pipe.setExtremesPort(extremes_port);
 
-    return pipe;
+    Pipes pipes;
+    pipes.emplace_back(std::move(pipe));
+
+    for (size_t i = 1; i < streams.size(); ++i)
+        pipes.emplace_back(Pipe(streams[i]));
+
+    return pipes;
 }
 
 PipelineExecutorPtr QueryPipeline::execute()
diff --git a/src/Processors/QueryPipeline.h b/src/Processors/QueryPipeline.h
index 45e38ffa715..d248853131c 100644
--- a/src/Processors/QueryPipeline.h
+++ b/src/Processors/QueryPipeline.h
@@ -155,8 +155,9 @@ public:
     /// Set upper limit for the recommend number of threads
     void setMaxThreads(size_t max_threads_) { max_threads = max_threads_; }
 
-    /// Convert query pipeline to single pipe.
+    /// Convert query pipeline to single or several pipes.
     Pipe getPipe() &&;
+    Pipes getPipes() &&;
 
 private:
     /// Destruction order: processors, header, locks, temporary storages, local contexts
diff --git a/src/Storages/StorageView.cpp b/src/Storages/StorageView.cpp
index 636c7f9d64d..01c85c5238a 100644
--- a/src/Storages/StorageView.cpp
+++ b/src/Storages/StorageView.cpp
@@ -19,6 +19,8 @@
 #include <Processors/Sources/SourceFromInputStream.h>
 #include <Processors/Transforms/MaterializingTransform.h>
 #include <Processors/Transforms/ConvertingTransform.h>
+#include <DataStreams/MaterializingBlockInputStream.h>
+#include <DataStreams/ConvertingBlockInputStream.h>
 
 
 namespace DB
@@ -62,29 +64,42 @@ Pipes StorageView::read(
     if (context.getSettings().enable_optimize_predicate_expression)
         current_inner_query = getRuntimeViewQuery(*query_info.query->as<const ASTSelectQuery>(), context);
 
-    QueryPipeline pipeline;
     InterpreterSelectWithUnionQuery interpreter(current_inner_query, context, {}, column_names);
     /// FIXME res may implicitly use some objects owned be pipeline, but them will be destructed after return
     if (query_info.force_tree_shaped_pipeline)
     {
+        QueryPipeline pipeline;
         BlockInputStreams streams = interpreter.executeWithMultipleStreams(pipeline);
+
+        for (auto & stream : streams)
+        {
+            stream = std::make_shared<MaterializingBlockInputStream>(stream);
+            stream = std::make_shared<ConvertingBlockInputStream>(stream, getSampleBlockForColumns(column_names),
+                                                                  ConvertingBlockInputStream::MatchColumnsMode::Name);
+        }
+
         for (auto & stream : streams)
             pipes.emplace_back(std::make_shared<SourceFromInputStream>(std::move(stream)));
     }
     else
-        /// TODO: support multiple streams here. Need more general interface than pipes.
-        pipes.emplace_back(interpreter.executeWithProcessors().getPipe());
-
-    /// It's expected that the columns read from storage are not constant.
-    /// Because method 'getSampleBlockForColumns' is used to obtain a structure of result in InterpreterSelectQuery.
-    for (auto & pipe : pipes)
     {
-        pipe.addSimpleTransform(std::make_shared<MaterializingTransform>(pipe.getHeader()));
+        auto pipeline = interpreter.executeWithProcessors();
+
+        /// It's expected that the columns read from storage are not constant.
+        /// Because method 'getSampleBlockForColumns' is used to obtain a structure of result in InterpreterSelectQuery.
+        pipeline.addSimpleTransform([](const Block & header)
+        {
+            return std::make_shared<MaterializingTransform>(header);
+        });
 
         /// And also convert to expected structure.
-        pipe.addSimpleTransform(std::make_shared<ConvertingTransform>(
-            pipe.getHeader(), getSampleBlockForColumns(column_names),
-            ConvertingTransform::MatchColumnsMode::Name));
+        pipeline.addSimpleTransform([&](const Block & header)
+        {
+            return std::make_shared<ConvertingTransform>(header, getSampleBlockForColumns(column_names),
+                                                         ConvertingTransform::MatchColumnsMode::Name);
+        });
+
+        pipes = std::move(pipeline).getPipes();
     }
 
     return pipes;

From 8074e2027fd76e03725598fbdf5e15565301441b Mon Sep 17 00:00:00 2001
From: Nikolai Kochetov <nik-kochetov@yandex-team.ru>
Date: Tue, 26 May 2020 16:32:44 +0300
Subject: [PATCH 020/122] Added test.

---
 .../queries/0_stateless/01284_view_and_extremes_bug.reference | 1 +
 tests/queries/0_stateless/01284_view_and_extremes_bug.sql     | 4 ++++
 2 files changed, 5 insertions(+)
 create mode 100644 tests/queries/0_stateless/01284_view_and_extremes_bug.reference
 create mode 100644 tests/queries/0_stateless/01284_view_and_extremes_bug.sql

diff --git a/tests/queries/0_stateless/01284_view_and_extremes_bug.reference b/tests/queries/0_stateless/01284_view_and_extremes_bug.reference
new file mode 100644
index 00000000000..216e97ce082
--- /dev/null
+++ b/tests/queries/0_stateless/01284_view_and_extremes_bug.reference
@@ -0,0 +1 @@
+World
diff --git a/tests/queries/0_stateless/01284_view_and_extremes_bug.sql b/tests/queries/0_stateless/01284_view_and_extremes_bug.sql
new file mode 100644
index 00000000000..c444441a258
--- /dev/null
+++ b/tests/queries/0_stateless/01284_view_and_extremes_bug.sql
@@ -0,0 +1,4 @@
+drop table if exists view_bug_const;
+CREATE VIEW view_bug_const AS SELECT 'World' AS hello FROM (SELECT number FROM system.numbers LIMIT 1) AS n1 JOIN (SELECT number FROM system.numbers LIMIT 1) AS n2 USING (number);
+select * from view_bug_const;
+drop table if exists view_bug_const;

From e09bcb4290b374b654d2a71ddd9fbde5beeeaa1f Mon Sep 17 00:00:00 2001
From: Maxim Akhmedov <max42@yandex-team.ru>
Date: Tue, 26 May 2020 17:05:06 +0300
Subject: [PATCH 021/122] Make possible adding nested storages to
 MultipleAccessStorage in run-time.

---
 src/Access/MultipleAccessStorage.cpp | 8 ++++++++
 src/Access/MultipleAccessStorage.h   | 2 ++
 2 files changed, 10 insertions(+)

diff --git a/src/Access/MultipleAccessStorage.cpp b/src/Access/MultipleAccessStorage.cpp
index 0dd1f142f31..9b80c16e487 100644
--- a/src/Access/MultipleAccessStorage.cpp
+++ b/src/Access/MultipleAccessStorage.cpp
@@ -143,6 +143,14 @@ const IAccessStorage & MultipleAccessStorage::getStorage(const UUID & id) const
     return const_cast<MultipleAccessStorage *>(this)->getStorage(id);
 }
 
+void MultipleAccessStorage::addStorage(std::unique_ptr<Storage> nested_storage)
+{
+    /// Note that IStorage::storage_name is not changed. It is ok as this method
+    /// is considered as a temporary solution allowing third-party Arcadia applications
+    /// using CH as a library to register their own access storages. Do not remove
+    /// this method without providing any alternative :)
+    nested_storages.emplace_back(std::move(nested_storage));
+}
 
 AccessEntityPtr MultipleAccessStorage::readImpl(const UUID & id) const
 {
diff --git a/src/Access/MultipleAccessStorage.h b/src/Access/MultipleAccessStorage.h
index ec8c8f2a101..06fb3d45c05 100644
--- a/src/Access/MultipleAccessStorage.h
+++ b/src/Access/MultipleAccessStorage.h
@@ -25,6 +25,8 @@ public:
     const Storage & getStorage(const UUID & id) const;
     Storage & getStorage(const UUID & id);
 
+    void addStorage(std::unique_ptr<Storage> nested_storage);
+
     Storage & getStorageByIndex(size_t i) { return *(nested_storages[i]); }
     const Storage & getStorageByIndex(size_t i) const { return *(nested_storages[i]); }
 

From 9d0c28d96c1987d4f4412a679354e22ad0f6304c Mon Sep 17 00:00:00 2001
From: alesapin <alesapin@gmail.com>
Date: Tue, 26 May 2020 17:14:08 +0300
Subject: [PATCH 022/122] Add missing methods

---
 src/Storages/IStorage.cpp                     | 24 +++++++++++++++----
 src/Storages/IStorage.h                       | 11 +++++++--
 src/Storages/MergeTree/IMergeTreeDataPart.cpp |  4 ++--
 src/Storages/MergeTree/MergeTreeData.cpp      |  6 ++---
 4 files changed, 34 insertions(+), 11 deletions(-)

diff --git a/src/Storages/IStorage.cpp b/src/Storages/IStorage.cpp
index 64e06a2534e..6564fee827e 100644
--- a/src/Storages/IStorage.cpp
+++ b/src/Storages/IStorage.cpp
@@ -444,9 +444,14 @@ void IStorage::setPartitionKey(const StorageMetadataKeyField & partition_key_)
     partition_key = partition_key_;
 }
 
+bool IStorage::isPartitionKeyDefined() const
+{
+    return partition_key.definition_ast != nullptr;
+}
+
 bool IStorage::hasPartitionKey() const
 {
-    return partition_key.expression != nullptr;
+    return !partition_key.column_names.empty();
 }
 
 Names IStorage::getColumnsRequiredForPartitionKey() const
@@ -466,9 +471,14 @@ void IStorage::setSortingKey(const StorageMetadataKeyField & sorting_key_)
     sorting_key = sorting_key_;
 }
 
+bool IStorage::isSortingKeyDefined() const
+{
+    return sorting_key.definition_ast != nullptr;
+}
+
 bool IStorage::hasSortingKey() const
 {
-    return sorting_key.expression != nullptr;
+    return !sorting_key.column_names.empty();
 }
 
 Names IStorage::getColumnsRequiredForSortingKey() const
@@ -502,7 +512,7 @@ bool IStorage::isPrimaryKeyDefined() const
 
 bool IStorage::hasPrimaryKey() const
 {
-    return primary_key.expression != nullptr;
+    return !primary_key.column_names.empty();
 }
 
 Names IStorage::getColumnsRequiredForPrimaryKey() const
@@ -529,9 +539,15 @@ void IStorage::setSamplingKey(const StorageMetadataKeyField & sampling_key_)
     sampling_key = sampling_key_;
 }
 
+
+bool IStorage::isSamplingKeyDefined() const
+{
+    return sampling_key.definition_ast != nullptr;
+}
+
 bool IStorage::hasSamplingKey() const
 {
-    return sampling_key.expression != nullptr;
+    return !sampling_key.column_names.empty();
 }
 
 Names IStorage::getColumnsRequiredForSampling() const
diff --git a/src/Storages/IStorage.h b/src/Storages/IStorage.h
index f707b3a4b5c..655ced0d8cf 100644
--- a/src/Storages/IStorage.h
+++ b/src/Storages/IStorage.h
@@ -453,6 +453,8 @@ public:
     void setPartitionKey(const StorageMetadataKeyField & partition_key_);
     /// Returns ASTExpressionList of partition key expression for storage or nullptr if there is none.
     ASTPtr getPartitionKeyAST() const { return partition_key.definition_ast; }
+    /// Storage has user-defined (in CREATE query) partition key.
+    bool isPartitionKeyDefined() const;
     /// Storage has partition key.
     bool hasPartitionKey() const;
     /// Returns column names that need to be read to calculate partition key.
@@ -466,7 +468,9 @@ public:
     void setSortingKey(const StorageMetadataKeyField & sorting_key_);
     /// Returns ASTExpressionList of sorting key expression for storage or nullptr if there is none.
     ASTPtr getSortingKeyAST() const { return sorting_key.definition_ast; }
-    /// Storage has sorting key.
+    /// Storage has user-defined (in CREATE query) sorting key.
+    bool isSortingKeyDefined() const;
+    /// Storage has sorting key. It means, that it contains at least one column.
     bool hasSortingKey() const;
     /// Returns column names that need to be read to calculate sorting key.
     Names getColumnsRequiredForSortingKey() const;
@@ -483,7 +487,8 @@ public:
     ASTPtr getPrimaryKeyAST() const { return primary_key.definition_ast; }
     /// Storage has user-defined (in CREATE query) sorting key.
     bool isPrimaryKeyDefined() const;
-    /// Storage has primary key (maybe part of some other key).
+    /// Storage has primary key (maybe part of some other key). It means, that
+    /// it contains at least one column.
     bool hasPrimaryKey() const;
     /// Returns column names that need to be read to calculate primary key.
     Names getColumnsRequiredForPrimaryKey() const;
@@ -498,6 +503,8 @@ public:
     void setSamplingKey(const StorageMetadataKeyField & sampling_key_);
     /// Returns sampling expression AST for storage or nullptr if there is none.
     ASTPtr getSamplingKeyAST() const { return sampling_key.definition_ast; }
+    /// Storage has user-defined (in CREATE query) sampling key.
+    bool isSamplingKeyDefined() const;
     /// Storage has sampling key.
     bool hasSamplingKey() const;
     /// Returns column names that need to be read to calculate sampling key.
diff --git a/src/Storages/MergeTree/IMergeTreeDataPart.cpp b/src/Storages/MergeTree/IMergeTreeDataPart.cpp
index 78548cdddb2..6e048a956da 100644
--- a/src/Storages/MergeTree/IMergeTreeDataPart.cpp
+++ b/src/Storages/MergeTree/IMergeTreeDataPart.cpp
@@ -841,7 +841,7 @@ void IMergeTreeDataPart::checkConsistencyBase() const
 
     if (!checksums.empty())
     {
-        if (storage.isPrimaryKeyDefined() && !checksums.files.count("primary.idx"))
+        if (storage.hasPrimaryKey() && !checksums.files.count("primary.idx"))
             throw Exception("No checksum for primary.idx", ErrorCodes::NO_FILE_IN_DATA_PART);
 
         if (storage.format_version >= MERGE_TREE_DATA_MIN_FORMAT_VERSION_WITH_CUSTOM_PARTITIONING)
@@ -875,7 +875,7 @@ void IMergeTreeDataPart::checkConsistencyBase() const
         };
 
         /// Check that the primary key index is not empty.
-        if (storage.isPrimaryKeyDefined())
+        if (storage.hasPrimaryKey())
             check_file_not_empty(volume->getDisk(), path + "primary.idx");
 
         if (storage.format_version >= MERGE_TREE_DATA_MIN_FORMAT_VERSION_WITH_CUSTOM_PARTITIONING)
diff --git a/src/Storages/MergeTree/MergeTreeData.cpp b/src/Storages/MergeTree/MergeTreeData.cpp
index d80dce167b4..48cf3934820 100644
--- a/src/Storages/MergeTree/MergeTreeData.cpp
+++ b/src/Storages/MergeTree/MergeTreeData.cpp
@@ -250,10 +250,10 @@ StorageInMemoryMetadata MergeTreeData::getInMemoryMetadata() const
 {
     StorageInMemoryMetadata metadata(getColumns(), getIndices(), getConstraints());
 
-    if (hasPartitionKey())
+    if (isPartitionKeyDefined())
         metadata.partition_by_ast = getPartitionKeyAST()->clone();
 
-    if (hasSortingKey())
+    if (isSortingKeyDefined())
         metadata.order_by_ast = getSortingKeyAST()->clone();
 
     if (isPrimaryKeyDefined())
@@ -262,7 +262,7 @@ StorageInMemoryMetadata MergeTreeData::getInMemoryMetadata() const
     if (ttl_table_ast)
         metadata.ttl_for_table_ast = ttl_table_ast->clone();
 
-    if (hasSamplingKey())
+    if (isSamplingKeyDefined())
         metadata.sample_by_ast = getSamplingKeyAST()->clone();
 
     if (settings_ast)

From b2057159c1d12fb016b7918b7a00b11ee73967fc Mon Sep 17 00:00:00 2001
From: Alexey Milovidov <milovidov@yandex-team.ru>
Date: Tue, 26 May 2020 17:24:20 +0300
Subject: [PATCH 023/122] Experiment on parsing backslash escapes in a more
 convenient way #10922

---
 src/IO/ReadHelpers.cpp                        | 16 +++-
 src/IO/tests/CMakeLists.txt                   |  3 -
 src/IO/tests/mempbrk.cpp                      | 90 -------------------
 ...escape_sequences_php_mysql_style.reference |  3 +
 ...01284_escape_sequences_php_mysql_style.sql |  6 ++
 5 files changed, 22 insertions(+), 96 deletions(-)
 delete mode 100644 src/IO/tests/mempbrk.cpp
 create mode 100644 tests/queries/0_stateless/01284_escape_sequences_php_mysql_style.reference
 create mode 100644 tests/queries/0_stateless/01284_escape_sequences_php_mysql_style.sql

diff --git a/src/IO/ReadHelpers.cpp b/src/IO/ReadHelpers.cpp
index dccb413af2c..65c3cfab55a 100644
--- a/src/IO/ReadHelpers.cpp
+++ b/src/IO/ReadHelpers.cpp
@@ -283,7 +283,9 @@ static void parseComplexEscapeSequence(Vector & s, ReadBuffer & buf)
     if (buf.eof())
         throw Exception("Cannot parse escape sequence", ErrorCodes::CANNOT_PARSE_ESCAPE_SEQUENCE);
 
-    if (*buf.position() == 'x')
+    char char_after_backslash = *buf.position();
+
+    if (char_after_backslash == 'x')
     {
         ++buf.position();
         /// escape sequence of the form \xAA
@@ -291,7 +293,7 @@ static void parseComplexEscapeSequence(Vector & s, ReadBuffer & buf)
         readPODBinary(hex_code, buf);
         s.push_back(unhex2(hex_code));
     }
-    else if (*buf.position() == 'N')
+    else if (char_after_backslash == 'N')
     {
         /// Support for NULLs: \N sequence must be parsed as empty string.
         ++buf.position();
@@ -299,7 +301,15 @@ static void parseComplexEscapeSequence(Vector & s, ReadBuffer & buf)
     else
     {
         /// The usual escape sequence of a single character.
-        s.push_back(parseEscapeSequence(*buf.position()));
+        char decoded_char = parseEscapeSequence(char_after_backslash);
+
+        /// For convenience using LIKE and regular expressions,
+        /// we leave backslash when user write something like 'Hello 100\%':
+        /// it is parsed like Hello 100\% instead of Hello 100%
+        if (decoded_char != '\\' && !isControlASCII(decoded_char))
+            s.push_back('\\');
+
+        s.push_back(decoded_char);
         ++buf.position();
     }
 }
diff --git a/src/IO/tests/CMakeLists.txt b/src/IO/tests/CMakeLists.txt
index 2767ce6e271..dfbbfa77853 100644
--- a/src/IO/tests/CMakeLists.txt
+++ b/src/IO/tests/CMakeLists.txt
@@ -37,9 +37,6 @@ target_link_libraries (parse_int_perf2 PRIVATE clickhouse_common_io)
 add_executable (read_write_int read_write_int.cpp)
 target_link_libraries (read_write_int PRIVATE clickhouse_common_io)
 
-add_executable (mempbrk mempbrk.cpp)
-target_link_libraries (mempbrk PRIVATE clickhouse_common_io)
-
 add_executable (o_direct_and_dirty_pages o_direct_and_dirty_pages.cpp)
 target_link_libraries (o_direct_and_dirty_pages PRIVATE clickhouse_common_io)
 
diff --git a/src/IO/tests/mempbrk.cpp b/src/IO/tests/mempbrk.cpp
deleted file mode 100644
index 55097d989af..00000000000
--- a/src/IO/tests/mempbrk.cpp
+++ /dev/null
@@ -1,90 +0,0 @@
-#include <string>
-#include <iostream>
-#include <iomanip>
-
-#include <Common/Stopwatch.h>
-
-#include <Core/Types.h>
-#include <IO/ReadHelpers.h>
-#include <IO/WriteHelpers.h>
-#include <IO/ReadBufferFromFileDescriptor.h>
-#include <IO/WriteBufferFromFileDescriptor.h>
-
-#include <common/find_symbols.h>
-
-namespace DB
-{
-namespace ErrorCodes
-{
-    extern const int CANNOT_PARSE_ESCAPE_SEQUENCE;
-}
-}
-
-
-namespace test
-{
-static void readEscapedString(DB::String & s, DB::ReadBuffer & buf)
-    {
-        s = "";
-        while (!buf.eof())
-        {
-            const char * next_pos = find_first_symbols<'\b', '\f', '\n', '\r', '\t', '\0', '\\'>(buf.position(), buf.buffer().end());
-
-            s.append(buf.position(), next_pos - buf.position());
-            buf.position() += next_pos - buf.position();
-
-            if (!buf.hasPendingData())
-                continue;
-
-            if (*buf.position() == '\t' || *buf.position() == '\n')
-                return;
-
-            if (*buf.position() == '\\')
-            {
-                ++buf.position();
-                if (buf.eof())
-                    throw DB::Exception("Cannot parse escape sequence", DB::ErrorCodes::CANNOT_PARSE_ESCAPE_SEQUENCE);
-                s += DB::parseEscapeSequence(*buf.position());
-                ++buf.position();
-            }
-        }
-    }
-}
-
-
-int main(int, char **)
-{
-    try
-    {
-        DB::ReadBufferFromFileDescriptor in(STDIN_FILENO);
-//        DB::WriteBufferFromFileDescriptor out(STDOUT_FILENO);
-        std::string s;
-        size_t rows = 0;
-
-        Stopwatch watch;
-
-        while (!in.eof())
-        {
-            test::readEscapedString(s, in);
-            in.ignore();
-
-            ++rows;
-
-/*            DB::writeEscapedString(s, out);
-            DB::writeChar('\n', out);*/
-        }
-
-        watch.stop();
-        std::cerr << std::fixed << std::setprecision(2)
-            << "Read " << rows << " rows (" << in.count() / 1000000.0 << " MB) in " << watch.elapsedSeconds() << " sec., "
-            << rows / watch.elapsedSeconds() << " rows/sec. (" << in.count() / watch.elapsedSeconds() / 1000000 << " MB/s.)"
-            << std::endl;
-    }
-    catch (const DB::Exception & e)
-    {
-        std::cerr << e.what() << ", " << e.displayText() << std::endl;
-        return 1;
-    }
-
-    return 0;
-}
diff --git a/tests/queries/0_stateless/01284_escape_sequences_php_mysql_style.reference b/tests/queries/0_stateless/01284_escape_sequences_php_mysql_style.reference
new file mode 100644
index 00000000000..19a5fc680ff
--- /dev/null
+++ b/tests/queries/0_stateless/01284_escape_sequences_php_mysql_style.reference
@@ -0,0 +1,3 @@
+a\\_\\c\\l\\i\\c\\k\\h\\o\\u\\s	a\\_\\c\\l\\i\\c\\k\\h\\o\\u\\s\\e
+1	0	1	1
+1	1	0	0	0
diff --git a/tests/queries/0_stateless/01284_escape_sequences_php_mysql_style.sql b/tests/queries/0_stateless/01284_escape_sequences_php_mysql_style.sql
new file mode 100644
index 00000000000..36ad7b4506c
--- /dev/null
+++ b/tests/queries/0_stateless/01284_escape_sequences_php_mysql_style.sql
@@ -0,0 +1,6 @@
+SELECT 'a\_\c\l\i\c\k\h\o\u\s\e', 'a\\_\\c\\l\\i\\c\\k\\h\\o\\u\\s\\e';
+select 'aXb' like 'a_b', 'aXb' like 'a\_b', 'a_b' like 'a\_b', 'a_b' like 'a\\_b';
+SELECT match('Hello', '\w+'), match('Hello', '\\w+'), match('Hello', '\\\w+'), match('Hello', '\w\+'), match('Hello', 'w+');
+
+SELECT match('Hello', '\He\l\l\o'); -- { serverError 427 }
+SELECT match('Hello', '\H\e\l\l\o'); -- { serverError 427 }

From 009f57fc2fc84e4cffe08eea118fa0543ca4c66c Mon Sep 17 00:00:00 2001
From: Alexey Milovidov <milovidov@yandex-team.ru>
Date: Tue, 26 May 2020 21:26:17 +0300
Subject: [PATCH 024/122] Fix flacky test #11190

---
 .../01247_dist_on_dist_group_by_sharding_key_optimization.sql | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/tests/queries/0_stateless/01247_dist_on_dist_group_by_sharding_key_optimization.sql b/tests/queries/0_stateless/01247_dist_on_dist_group_by_sharding_key_optimization.sql
index 7cf171c8c73..29e45ffdb80 100644
--- a/tests/queries/0_stateless/01247_dist_on_dist_group_by_sharding_key_optimization.sql
+++ b/tests/queries/0_stateless/01247_dist_on_dist_group_by_sharding_key_optimization.sql
@@ -1,5 +1,9 @@
 -- TODO: correct testing with real unique shards
 
+-- Avoid "Connection failed at try №1" messages.
+SET send_logs_level = 'none';
+SET connect_timeout_with_failover_ms = 5000;
+
 set optimize_distributed_group_by_sharding_key=1;
 
 drop table if exists dist_01247;

From 90f49c0c17ff5ce8182a5d45543c5c3aab9985f4 Mon Sep 17 00:00:00 2001
From: Alexey Milovidov <milovidov@yandex-team.ru>
Date: Tue, 26 May 2020 21:37:20 +0300
Subject: [PATCH 025/122] Fix error in performance test

---
 tests/performance/agg_functions_min_max_any.xml | 4 ----
 1 file changed, 4 deletions(-)

diff --git a/tests/performance/agg_functions_min_max_any.xml b/tests/performance/agg_functions_min_max_any.xml
index 4591182cdfc..9c16cb88970 100644
--- a/tests/performance/agg_functions_min_max_any.xml
+++ b/tests/performance/agg_functions_min_max_any.xml
@@ -35,10 +35,6 @@
 <query>select max(PageCharset) from test.hits where PageCharset != '' group by intHash32(UserID) % 1000000 FORMAT Null</query>
 <query>select any(PageCharset) from test.hits where PageCharset != '' group by intHash32(UserID) % 1000000 FORMAT Null</query>
 <query>select anyHeavy(PageCharset) from test.hits where PageCharset != '' group by intHash32(UserID) % 1000000 FORMAT Null</query>
-<query>select min(OriginalURL) from test.hits where OriginalURL != '' group by intHash32(UserID) % 1000000 FORMAT Null</query>
-<query>select max(OriginalURL) from test.hits where OriginalURL != '' group by intHash32(UserID) % 1000000 FORMAT Null</query>
-<query>select any(OriginalURL) from test.hits where OriginalURL != '' group by intHash32(UserID) % 1000000 FORMAT Null</query>
-<query>select anyHeavy(OriginalURL) from test.hits where OriginalURL != '' group by intHash32(UserID) % 1000000 FORMAT Null</query>
 <query>select min(SocialNetwork) from test.hits where SocialNetwork != '' group by intHash32(UserID) % 1000000 FORMAT Null</query>
 <query>select max(SocialNetwork) from test.hits where SocialNetwork != '' group by intHash32(UserID) % 1000000 FORMAT Null</query>
 <query>select any(SocialNetwork) from test.hits where SocialNetwork != '' group by intHash32(UserID) % 1000000 FORMAT Null</query>

From 15e7c618bb5023ed2d63465e2e3eed7e99503799 Mon Sep 17 00:00:00 2001
From: alesapin <alesapin@gmail.com>
Date: Tue, 26 May 2020 22:21:07 +0300
Subject: [PATCH 026/122] Fix cctz linking with dirty hack

---
 contrib/cctz-cmake/CMakeLists.txt | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/contrib/cctz-cmake/CMakeLists.txt b/contrib/cctz-cmake/CMakeLists.txt
index df9fd6aa61c..d353fc9d4de 100644
--- a/contrib/cctz-cmake/CMakeLists.txt
+++ b/contrib/cctz-cmake/CMakeLists.txt
@@ -618,7 +618,11 @@ if (USE_INTERNAL_CCTZ)
 
         add_library(tzdata STATIC ${TZ_OBJS})
         set_target_properties(tzdata PROPERTIES LINKER_LANGUAGE C)
-        target_link_libraries(cctz -Wl,--whole-archive tzdata -Wl,--no-whole-archive) # whole-archive prevents symbols from being discarded
+        # whole-archive prevents symbols from being discarded for unknown reason
+        # CMake can shuffle each of target_link_libraries arguments with other
+        # libraries in linker command. To avoid this we hardcode whole-archive
+        # library into single string.
+        target_link_libraries(cctz PUBLIC "-Wl,--whole-archive contrib/cctz-cmake/libtzdata.a -Wl,--no-whole-archive")
     endif ()
 
 else ()

From 4a92ad11ec02a4a94b7b3567cd02a62df4dd37ea Mon Sep 17 00:00:00 2001
From: Ivan Blinkov <github@blinkov.ru>
Date: Tue, 26 May 2020 23:27:31 +0300
Subject: [PATCH 027/122] Update test.py

---
 tests/integration/test_system_queries/test.py | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/tests/integration/test_system_queries/test.py b/tests/integration/test_system_queries/test.py
index 1761017362a..9397cd20ec7 100644
--- a/tests/integration/test_system_queries/test.py
+++ b/tests/integration/test_system_queries/test.py
@@ -95,6 +95,10 @@ def test_RELOAD_CONFIG_AND_MACROS(started_cluster):
 
 def test_SYSTEM_FLUSH_LOGS(started_cluster):
     instance = cluster.instances['ch1']
+    instance.query('''
+        SET log_queries = 0;
+        TRUNCATE TABLE system.query_log;
+    ''')
     for i in range(4):
         # Sleep to execute flushing from background thread at first query
         # by expiration of flush_interval_millisecond and test probable race condition.
@@ -105,7 +109,10 @@ def test_SYSTEM_FLUSH_LOGS(started_cluster):
             SET log_queries = 0;
             SYSTEM FLUSH LOGS;
             SELECT count() FROM system.query_log;''')
-        instance.query('TRUNCATE TABLE system.query_log')
+        instance.query('''
+            SET log_queries = 0;
+            TRUNCATE TABLE system.query_log;
+        ''')
         assert TSV(result) == TSV('4')
 
 

From a5bde57acfd00acf1af44f794daa4ebb23583201 Mon Sep 17 00:00:00 2001
From: Saulius Valatka <saulius.valatka@adform.com>
Date: Tue, 26 May 2020 23:43:20 +0300
Subject: [PATCH 028/122] add reference to stream-loader (#11206)

---
 docs/en/interfaces/third-party/integrations.md | 1 +
 1 file changed, 1 insertion(+)

diff --git a/docs/en/interfaces/third-party/integrations.md b/docs/en/interfaces/third-party/integrations.md
index 5e9e8f841bf..716e774871b 100644
--- a/docs/en/interfaces/third-party/integrations.md
+++ b/docs/en/interfaces/third-party/integrations.md
@@ -25,6 +25,7 @@ toc_title: Integrations
 -   Message queues
     -   [Kafka](https://kafka.apache.org)
         -   [clickhouse\_sinker](https://github.com/housepower/clickhouse_sinker) (uses [Go client](https://github.com/ClickHouse/clickhouse-go/))
+        -   [stream-loader-clickhouse](https://github.com/adform/stream-loader)
 -   Stream processing
     -   [Flink](https://flink.apache.org)
         -   [flink-clickhouse-sink](https://github.com/ivi-ru/flink-clickhouse-sink)

From fb4089ab8e28c2e66ada5556344a225ffc629579 Mon Sep 17 00:00:00 2001
From: Alexey Milovidov <milovidov@yandex-team.ru>
Date: Wed, 27 May 2020 00:48:29 +0300
Subject: [PATCH 029/122] Fix error

---
 src/IO/ReadHelpers.cpp | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/src/IO/ReadHelpers.cpp b/src/IO/ReadHelpers.cpp
index 65c3cfab55a..0797e63cdd9 100644
--- a/src/IO/ReadHelpers.cpp
+++ b/src/IO/ReadHelpers.cpp
@@ -306,8 +306,15 @@ static void parseComplexEscapeSequence(Vector & s, ReadBuffer & buf)
         /// For convenience using LIKE and regular expressions,
         /// we leave backslash when user write something like 'Hello 100\%':
         /// it is parsed like Hello 100\% instead of Hello 100%
-        if (decoded_char != '\\' && !isControlASCII(decoded_char))
+        if (decoded_char != '\\'
+            && decoded_char != '\''
+            && decoded_char != '"'
+            && decoded_char != '`'  /// MySQL style identifiers
+            && decoded_char != '/'  /// JavaScript in HTML
+            && !isControlASCII(decoded_char))
+        {
             s.push_back('\\');
+        }
 
         s.push_back(decoded_char);
         ++buf.position();

From 734cc78e5838bafc35a3abc544fd7609c09902ca Mon Sep 17 00:00:00 2001
From: Azat Khuzhin <a3at.mail@gmail.com>
Date: Sun, 24 May 2020 13:16:34 +0300
Subject: [PATCH 030/122] jemalloc: make include_linux_*/jemalloc/*.h common
 for aarch64 and x86_64

This patch can go upstream, regardless of the jemalloc update
eventually.

Also I noticed that header with jemalloc version differs (someone forgot
to update?)
---
 .../jemalloc/jemalloc_defs.h                  |   0
 .../jemalloc/jemalloc_macros.h                |   0
 .../jemalloc/jemalloc_protos.h                |   0
 .../jemalloc/jemalloc_typedefs.h              |  18 +--
 .../jemalloc/jemalloc_macros.h                | 123 ------------------
 .../jemalloc/jemalloc_defs.h                  |  43 ------
 .../jemalloc/jemalloc_protos.h                |  66 ----------
 .../jemalloc/jemalloc_typedefs.h              |  77 -----------
 8 files changed, 9 insertions(+), 318 deletions(-)
 rename contrib/jemalloc-cmake/{include_linux_aarch64 => include}/jemalloc/jemalloc_defs.h (100%)
 rename contrib/jemalloc-cmake/{include_linux_x86_64 => include}/jemalloc/jemalloc_macros.h (100%)
 rename contrib/jemalloc-cmake/{include_linux_aarch64 => include}/jemalloc/jemalloc_protos.h (100%)
 rename contrib/jemalloc-cmake/{include_linux_aarch64 => include}/jemalloc/jemalloc_typedefs.h (88%)
 delete mode 100644 contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/jemalloc_macros.h
 delete mode 100644 contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/jemalloc_defs.h
 delete mode 100644 contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/jemalloc_protos.h
 delete mode 100644 contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/jemalloc_typedefs.h

diff --git a/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/jemalloc_defs.h b/contrib/jemalloc-cmake/include/jemalloc/jemalloc_defs.h
similarity index 100%
rename from contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/jemalloc_defs.h
rename to contrib/jemalloc-cmake/include/jemalloc/jemalloc_defs.h
diff --git a/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/jemalloc_macros.h b/contrib/jemalloc-cmake/include/jemalloc/jemalloc_macros.h
similarity index 100%
rename from contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/jemalloc_macros.h
rename to contrib/jemalloc-cmake/include/jemalloc/jemalloc_macros.h
diff --git a/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/jemalloc_protos.h b/contrib/jemalloc-cmake/include/jemalloc/jemalloc_protos.h
similarity index 100%
rename from contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/jemalloc_protos.h
rename to contrib/jemalloc-cmake/include/jemalloc/jemalloc_protos.h
diff --git a/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/jemalloc_typedefs.h b/contrib/jemalloc-cmake/include/jemalloc/jemalloc_typedefs.h
similarity index 88%
rename from contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/jemalloc_typedefs.h
rename to contrib/jemalloc-cmake/include/jemalloc/jemalloc_typedefs.h
index 1a58874306e..5f94f16f937 100644
--- a/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/jemalloc_typedefs.h
+++ b/contrib/jemalloc-cmake/include/jemalloc/jemalloc_typedefs.h
@@ -65,13 +65,13 @@ typedef bool (extent_merge_t)(extent_hooks_t *, void *, size_t, void *, size_t,
     bool, unsigned);
 
 struct extent_hooks_s {
-	extent_alloc_t		*alloc;
-	extent_dalloc_t		*dalloc;
-	extent_destroy_t	*destroy;
-	extent_commit_t		*commit;
-	extent_decommit_t	*decommit;
-	extent_purge_t		*purge_lazy;
-	extent_purge_t		*purge_forced;
-	extent_split_t		*split;
-	extent_merge_t		*merge;
+    extent_alloc_t		*alloc;
+    extent_dalloc_t		*dalloc;
+    extent_destroy_t	*destroy;
+    extent_commit_t		*commit;
+    extent_decommit_t	*decommit;
+    extent_purge_t		*purge_lazy;
+    extent_purge_t		*purge_forced;
+    extent_split_t		*split;
+    extent_merge_t		*merge;
 };
diff --git a/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/jemalloc_macros.h b/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/jemalloc_macros.h
deleted file mode 100644
index 79b13337fbb..00000000000
--- a/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/jemalloc_macros.h
+++ /dev/null
@@ -1,123 +0,0 @@
-#include <stdlib.h>
-#include <stdbool.h>
-#include <stdint.h>
-#include <limits.h>
-#include <strings.h>
-
-#define JEMALLOC_VERSION "5.1.0-97-gcd2931ad9bbd78208565716ab102e86d858c2fff"
-#define JEMALLOC_VERSION_MAJOR 5
-#define JEMALLOC_VERSION_MINOR 1
-#define JEMALLOC_VERSION_BUGFIX 0
-#define JEMALLOC_VERSION_NREV 97
-#define JEMALLOC_VERSION_GID "cd2931ad9bbd78208565716ab102e86d858c2fff"
-#define JEMALLOC_VERSION_GID_IDENT cd2931ad9bbd78208565716ab102e86d858c2fff
-
-#define MALLOCX_LG_ALIGN(la)	((int)(la))
-#if LG_SIZEOF_PTR == 2
-#  define MALLOCX_ALIGN(a)	((int)(ffs((int)(a))-1))
-#else
-#  define MALLOCX_ALIGN(a)						\
-     ((int)(((size_t)(a) < (size_t)INT_MAX) ? ffs((int)(a))-1 :	\
-     ffs((int)(((size_t)(a))>>32))+31))
-#endif
-#define MALLOCX_ZERO	((int)0x40)
-/*
- * Bias tcache index bits so that 0 encodes "automatic tcache management", and 1
- * encodes MALLOCX_TCACHE_NONE.
- */
-#define MALLOCX_TCACHE(tc)	((int)(((tc)+2) << 8))
-#define MALLOCX_TCACHE_NONE	MALLOCX_TCACHE(-1)
-/*
- * Bias arena index bits so that 0 encodes "use an automatically chosen arena".
- */
-#define MALLOCX_ARENA(a)	((((int)(a))+1) << 20)
-
-/*
- * Use as arena index in "arena.<i>.{purge,decay,dss}" and
- * "stats.arenas.<i>.*" mallctl interfaces to select all arenas.  This
- * definition is intentionally specified in raw decimal format to support
- * cpp-based string concatenation, e.g.
- *
- *   #define STRINGIFY_HELPER(x) #x
- *   #define STRINGIFY(x) STRINGIFY_HELPER(x)
- *
- *   mallctl("arena." STRINGIFY(MALLCTL_ARENAS_ALL) ".purge", NULL, NULL, NULL,
- *       0);
- */
-#define MALLCTL_ARENAS_ALL	4096
-/*
- * Use as arena index in "stats.arenas.<i>.*" mallctl interfaces to select
- * destroyed arenas.
- */
-#define MALLCTL_ARENAS_DESTROYED	4097
-
-#if defined(__cplusplus) && defined(JEMALLOC_USE_CXX_THROW)
-#  define JEMALLOC_CXX_THROW throw()
-#else
-#  define JEMALLOC_CXX_THROW
-#endif
-
-#if defined(_MSC_VER)
-#  define JEMALLOC_ATTR(s)
-#  define JEMALLOC_ALIGNED(s) __declspec(align(s))
-#  define JEMALLOC_ALLOC_SIZE(s)
-#  define JEMALLOC_ALLOC_SIZE2(s1, s2)
-#  ifndef JEMALLOC_EXPORT
-#    ifdef DLLEXPORT
-#      define JEMALLOC_EXPORT __declspec(dllexport)
-#    else
-#      define JEMALLOC_EXPORT __declspec(dllimport)
-#    endif
-#  endif
-#  define JEMALLOC_FORMAT_PRINTF(s, i)
-#  define JEMALLOC_NOINLINE __declspec(noinline)
-#  ifdef __cplusplus
-#    define JEMALLOC_NOTHROW __declspec(nothrow)
-#  else
-#    define JEMALLOC_NOTHROW
-#  endif
-#  define JEMALLOC_SECTION(s) __declspec(allocate(s))
-#  define JEMALLOC_RESTRICT_RETURN __declspec(restrict)
-#  if _MSC_VER >= 1900 && !defined(__EDG__)
-#    define JEMALLOC_ALLOCATOR __declspec(allocator)
-#  else
-#    define JEMALLOC_ALLOCATOR
-#  endif
-#elif defined(JEMALLOC_HAVE_ATTR)
-#  define JEMALLOC_ATTR(s) __attribute__((s))
-#  define JEMALLOC_ALIGNED(s) JEMALLOC_ATTR(aligned(s))
-#  ifdef JEMALLOC_HAVE_ATTR_ALLOC_SIZE
-#    define JEMALLOC_ALLOC_SIZE(s) JEMALLOC_ATTR(alloc_size(s))
-#    define JEMALLOC_ALLOC_SIZE2(s1, s2) JEMALLOC_ATTR(alloc_size(s1, s2))
-#  else
-#    define JEMALLOC_ALLOC_SIZE(s)
-#    define JEMALLOC_ALLOC_SIZE2(s1, s2)
-#  endif
-#  ifndef JEMALLOC_EXPORT
-#    define JEMALLOC_EXPORT JEMALLOC_ATTR(visibility("default"))
-#  endif
-#  ifdef JEMALLOC_HAVE_ATTR_FORMAT_GNU_PRINTF
-#    define JEMALLOC_FORMAT_PRINTF(s, i) JEMALLOC_ATTR(format(gnu_printf, s, i))
-#  elif defined(JEMALLOC_HAVE_ATTR_FORMAT_PRINTF)
-#    define JEMALLOC_FORMAT_PRINTF(s, i) JEMALLOC_ATTR(format(printf, s, i))
-#  else
-#    define JEMALLOC_FORMAT_PRINTF(s, i)
-#  endif
-#  define JEMALLOC_NOINLINE JEMALLOC_ATTR(noinline)
-#  define JEMALLOC_NOTHROW JEMALLOC_ATTR(nothrow)
-#  define JEMALLOC_SECTION(s) JEMALLOC_ATTR(section(s))
-#  define JEMALLOC_RESTRICT_RETURN
-#  define JEMALLOC_ALLOCATOR
-#else
-#  define JEMALLOC_ATTR(s)
-#  define JEMALLOC_ALIGNED(s)
-#  define JEMALLOC_ALLOC_SIZE(s)
-#  define JEMALLOC_ALLOC_SIZE2(s1, s2)
-#  define JEMALLOC_EXPORT
-#  define JEMALLOC_FORMAT_PRINTF(s, i)
-#  define JEMALLOC_NOINLINE
-#  define JEMALLOC_NOTHROW
-#  define JEMALLOC_SECTION(s)
-#  define JEMALLOC_RESTRICT_RETURN
-#  define JEMALLOC_ALLOCATOR
-#endif
diff --git a/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/jemalloc_defs.h b/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/jemalloc_defs.h
deleted file mode 100644
index d1389237a77..00000000000
--- a/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/jemalloc_defs.h
+++ /dev/null
@@ -1,43 +0,0 @@
-/* include/jemalloc/jemalloc_defs.h.  Generated from jemalloc_defs.h.in by configure.  */
-/* Defined if __attribute__((...)) syntax is supported. */
-#define JEMALLOC_HAVE_ATTR
-
-/* Defined if alloc_size attribute is supported. */
-#define JEMALLOC_HAVE_ATTR_ALLOC_SIZE
-
-/* Defined if format(printf, ...) attribute is supported. */
-#define JEMALLOC_HAVE_ATTR_FORMAT_PRINTF
-
-/*
- * Define overrides for non-standard allocator-related functions if they are
- * present on the system.
- */
-#define JEMALLOC_OVERRIDE_MEMALIGN
-#define JEMALLOC_OVERRIDE_VALLOC
-
-/*
- * At least Linux omits the "const" in:
- *
- *   size_t malloc_usable_size(const void *ptr);
- *
- * Match the operating system's prototype.
- */
-#define JEMALLOC_USABLE_SIZE_CONST
-
-/*
- * If defined, specify throw() for the public function prototypes when compiling
- * with C++.  The only justification for this is to match the prototypes that
- * glibc defines.
- */
-#define JEMALLOC_USE_CXX_THROW
-
-#ifdef _MSC_VER
-#  ifdef _WIN64
-#    define LG_SIZEOF_PTR_WIN 3
-#  else
-#    define LG_SIZEOF_PTR_WIN 2
-#  endif
-#endif
-
-/* sizeof(void *) == 2^LG_SIZEOF_PTR. */
-#define LG_SIZEOF_PTR 3
diff --git a/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/jemalloc_protos.h b/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/jemalloc_protos.h
deleted file mode 100644
index ff025e30fa7..00000000000
--- a/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/jemalloc_protos.h
+++ /dev/null
@@ -1,66 +0,0 @@
-/*
- * The je_ prefix on the following public symbol declarations is an artifact
- * of namespace management, and should be omitted in application code unless
- * JEMALLOC_NO_DEMANGLE is defined (see jemalloc_mangle.h).
- */
-extern JEMALLOC_EXPORT const char	*je_malloc_conf;
-extern JEMALLOC_EXPORT void		(*je_malloc_message)(void *cbopaque,
-    const char *s);
-
-JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
-    void JEMALLOC_NOTHROW	*je_malloc(size_t size)
-    JEMALLOC_CXX_THROW JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE(1);
-JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
-    void JEMALLOC_NOTHROW	*je_calloc(size_t num, size_t size)
-    JEMALLOC_CXX_THROW JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE2(1, 2);
-JEMALLOC_EXPORT int JEMALLOC_NOTHROW	je_posix_memalign(void **memptr,
-    size_t alignment, size_t size) JEMALLOC_CXX_THROW JEMALLOC_ATTR(nonnull(1));
-JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
-    void JEMALLOC_NOTHROW	*je_aligned_alloc(size_t alignment,
-    size_t size) JEMALLOC_CXX_THROW JEMALLOC_ATTR(malloc)
-    JEMALLOC_ALLOC_SIZE(2);
-JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
-    void JEMALLOC_NOTHROW	*je_realloc(void *ptr, size_t size)
-    JEMALLOC_CXX_THROW JEMALLOC_ALLOC_SIZE(2);
-JEMALLOC_EXPORT void JEMALLOC_NOTHROW	je_free(void *ptr)
-    JEMALLOC_CXX_THROW;
-
-JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
-    void JEMALLOC_NOTHROW	*je_mallocx(size_t size, int flags)
-    JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE(1);
-JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
-    void JEMALLOC_NOTHROW	*je_rallocx(void *ptr, size_t size,
-    int flags) JEMALLOC_ALLOC_SIZE(2);
-JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW	je_xallocx(void *ptr, size_t size,
-    size_t extra, int flags);
-JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW	je_sallocx(const void *ptr,
-    int flags) JEMALLOC_ATTR(pure);
-JEMALLOC_EXPORT void JEMALLOC_NOTHROW	je_dallocx(void *ptr, int flags);
-JEMALLOC_EXPORT void JEMALLOC_NOTHROW	je_sdallocx(void *ptr, size_t size,
-    int flags);
-JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW	je_nallocx(size_t size, int flags)
-    JEMALLOC_ATTR(pure);
-
-JEMALLOC_EXPORT int JEMALLOC_NOTHROW	je_mallctl(const char *name,
-    void *oldp, size_t *oldlenp, void *newp, size_t newlen);
-JEMALLOC_EXPORT int JEMALLOC_NOTHROW	je_mallctlnametomib(const char *name,
-    size_t *mibp, size_t *miblenp);
-JEMALLOC_EXPORT int JEMALLOC_NOTHROW	je_mallctlbymib(const size_t *mib,
-    size_t miblen, void *oldp, size_t *oldlenp, void *newp, size_t newlen);
-JEMALLOC_EXPORT void JEMALLOC_NOTHROW	je_malloc_stats_print(
-    void (*write_cb)(void *, const char *), void *je_cbopaque,
-    const char *opts);
-JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW	je_malloc_usable_size(
-    JEMALLOC_USABLE_SIZE_CONST void *ptr) JEMALLOC_CXX_THROW;
-
-#ifdef JEMALLOC_OVERRIDE_MEMALIGN
-JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
-    void JEMALLOC_NOTHROW	*je_memalign(size_t alignment, size_t size)
-    JEMALLOC_CXX_THROW JEMALLOC_ATTR(malloc);
-#endif
-
-#ifdef JEMALLOC_OVERRIDE_VALLOC
-JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
-    void JEMALLOC_NOTHROW	*je_valloc(size_t size) JEMALLOC_CXX_THROW
-    JEMALLOC_ATTR(malloc);
-#endif
diff --git a/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/jemalloc_typedefs.h b/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/jemalloc_typedefs.h
deleted file mode 100644
index 1a58874306e..00000000000
--- a/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/jemalloc_typedefs.h
+++ /dev/null
@@ -1,77 +0,0 @@
-typedef struct extent_hooks_s extent_hooks_t;
-
-/*
- * void *
- * extent_alloc(extent_hooks_t *extent_hooks, void *new_addr, size_t size,
- *     size_t alignment, bool *zero, bool *commit, unsigned arena_ind);
- */
-typedef void *(extent_alloc_t)(extent_hooks_t *, void *, size_t, size_t, bool *,
-    bool *, unsigned);
-
-/*
- * bool
- * extent_dalloc(extent_hooks_t *extent_hooks, void *addr, size_t size,
- *     bool committed, unsigned arena_ind);
- */
-typedef bool (extent_dalloc_t)(extent_hooks_t *, void *, size_t, bool,
-    unsigned);
-
-/*
- * void
- * extent_destroy(extent_hooks_t *extent_hooks, void *addr, size_t size,
- *     bool committed, unsigned arena_ind);
- */
-typedef void (extent_destroy_t)(extent_hooks_t *, void *, size_t, bool,
-    unsigned);
-
-/*
- * bool
- * extent_commit(extent_hooks_t *extent_hooks, void *addr, size_t size,
- *     size_t offset, size_t length, unsigned arena_ind);
- */
-typedef bool (extent_commit_t)(extent_hooks_t *, void *, size_t, size_t, size_t,
-    unsigned);
-
-/*
- * bool
- * extent_decommit(extent_hooks_t *extent_hooks, void *addr, size_t size,
- *     size_t offset, size_t length, unsigned arena_ind);
- */
-typedef bool (extent_decommit_t)(extent_hooks_t *, void *, size_t, size_t,
-    size_t, unsigned);
-
-/*
- * bool
- * extent_purge(extent_hooks_t *extent_hooks, void *addr, size_t size,
- *     size_t offset, size_t length, unsigned arena_ind);
- */
-typedef bool (extent_purge_t)(extent_hooks_t *, void *, size_t, size_t, size_t,
-    unsigned);
-
-/*
- * bool
- * extent_split(extent_hooks_t *extent_hooks, void *addr, size_t size,
- *     size_t size_a, size_t size_b, bool committed, unsigned arena_ind);
- */
-typedef bool (extent_split_t)(extent_hooks_t *, void *, size_t, size_t, size_t,
-    bool, unsigned);
-
-/*
- * bool
- * extent_merge(extent_hooks_t *extent_hooks, void *addr_a, size_t size_a,
- *     void *addr_b, size_t size_b, bool committed, unsigned arena_ind);
- */
-typedef bool (extent_merge_t)(extent_hooks_t *, void *, size_t, void *, size_t,
-    bool, unsigned);
-
-struct extent_hooks_s {
-	extent_alloc_t		*alloc;
-	extent_dalloc_t		*dalloc;
-	extent_destroy_t	*destroy;
-	extent_commit_t		*commit;
-	extent_decommit_t	*decommit;
-	extent_purge_t		*purge_lazy;
-	extent_purge_t		*purge_forced;
-	extent_split_t		*split;
-	extent_merge_t		*merge;
-};

From b53c6ddfd9a9960fc94486b85db1d1d07f86fbc7 Mon Sep 17 00:00:00 2001
From: Azat Khuzhin <a3at.mail@gmail.com>
Date: Sun, 24 May 2020 02:37:11 +0300
Subject: [PATCH 031/122] jemalloc: bump to 5.2.1

---
 contrib/jemalloc | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/contrib/jemalloc b/contrib/jemalloc
index cd2931ad9bb..ea6b3e973b4 160000
--- a/contrib/jemalloc
+++ b/contrib/jemalloc
@@ -1 +1 @@
-Subproject commit cd2931ad9bbd78208565716ab102e86d858c2fff
+Subproject commit ea6b3e973b477b8061e0076bb257dbd7f3faa756

From bbb38664c10cbed66c82922595102ff28d8b3f71 Mon Sep 17 00:00:00 2001
From: Azat Khuzhin <a3at.mail@gmail.com>
Date: Sun, 24 May 2020 12:55:21 +0300
Subject: [PATCH 032/122] jemalloc: add missing safety_check.c

---
 contrib/jemalloc-cmake/CMakeLists.txt | 1 +
 1 file changed, 1 insertion(+)

diff --git a/contrib/jemalloc-cmake/CMakeLists.txt b/contrib/jemalloc-cmake/CMakeLists.txt
index 9081de593b8..e187ca2926d 100644
--- a/contrib/jemalloc-cmake/CMakeLists.txt
+++ b/contrib/jemalloc-cmake/CMakeLists.txt
@@ -55,6 +55,7 @@ if (ENABLE_JEMALLOC)
             ${LIBRARY_DIR}/src/ticker.c
             ${LIBRARY_DIR}/src/tsd.c
             ${LIBRARY_DIR}/src/witness.c
+            ${LIBRARY_DIR}/src/safety_check.c
         )
         if (OS_DARWIN)
             list(APPEND SRCS ${LIBRARY_DIR}/src/zone.c)

From b2ff1b3f20d15722e5097fd8c7622572eb809d01 Mon Sep 17 00:00:00 2001
From: Azat Khuzhin <a3at.mail@gmail.com>
Date: Sun, 24 May 2020 13:05:20 +0300
Subject: [PATCH 033/122] jemalloc: merge jemalloc_preamble.h for x86_64

---
 .../jemalloc/internal/jemalloc_preamble.h     | 23 +++++++++++++++++--
 1 file changed, 21 insertions(+), 2 deletions(-)

diff --git a/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/internal/jemalloc_preamble.h b/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/internal/jemalloc_preamble.h
index c150785fb4a..e5e34925b55 100644
--- a/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/internal/jemalloc_preamble.h
+++ b/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/internal/jemalloc_preamble.h
@@ -21,7 +21,7 @@
 #  include "jemalloc/jemalloc.h"
 #endif
 
-#if (defined(JEMALLOC_OSATOMIC) || defined(JEMALLOC_OSSPIN))
+#if defined(JEMALLOC_OSATOMIC)
 #include <libkern/OSAtomic.h>
 #endif
 
@@ -161,7 +161,26 @@ static const bool config_log =
     false
 #endif
     ;
-#ifdef JEMALLOC_HAVE_SCHED_GETCPU
+/*
+ * Are extra safety checks enabled; things like checking the size of sized
+ * deallocations, double-frees, etc.
+ */
+static const bool config_opt_safety_checks =
+#ifdef JEMALLOC_OPT_SAFETY_CHECKS
+    true
+#elif defined(JEMALLOC_DEBUG)
+    /*
+     * This lets us only guard safety checks by one flag instead of two; fast
+     * checks can guard solely by config_opt_safety_checks and run in debug mode
+     * too.
+     */
+    true
+#else
+    false
+#endif
+    ;
+
+#if defined(_WIN32) || defined(JEMALLOC_HAVE_SCHED_GETCPU)
 /* Currently percpu_arena depends on sched_getcpu. */
 #define JEMALLOC_PERCPU_ARENA
 #endif

From 2590ee64d902cf18be3ba5ad715803bfe94370b3 Mon Sep 17 00:00:00 2001
From: Azat Khuzhin <a3at.mail@gmail.com>
Date: Sun, 24 May 2020 12:55:55 +0300
Subject: [PATCH 034/122] jemalloc: merge jemalloc_preamble.h for aarch64

---
 .../jemalloc/internal/jemalloc_preamble.h     | 23 +++++++++++++++++--
 1 file changed, 21 insertions(+), 2 deletions(-)

diff --git a/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/internal/jemalloc_preamble.h b/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/internal/jemalloc_preamble.h
index c150785fb4a..e5e34925b55 100644
--- a/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/internal/jemalloc_preamble.h
+++ b/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/internal/jemalloc_preamble.h
@@ -21,7 +21,7 @@
 #  include "jemalloc/jemalloc.h"
 #endif
 
-#if (defined(JEMALLOC_OSATOMIC) || defined(JEMALLOC_OSSPIN))
+#if defined(JEMALLOC_OSATOMIC)
 #include <libkern/OSAtomic.h>
 #endif
 
@@ -161,7 +161,26 @@ static const bool config_log =
     false
 #endif
     ;
-#ifdef JEMALLOC_HAVE_SCHED_GETCPU
+/*
+ * Are extra safety checks enabled; things like checking the size of sized
+ * deallocations, double-frees, etc.
+ */
+static const bool config_opt_safety_checks =
+#ifdef JEMALLOC_OPT_SAFETY_CHECKS
+    true
+#elif defined(JEMALLOC_DEBUG)
+    /*
+     * This lets us only guard safety checks by one flag instead of two; fast
+     * checks can guard solely by config_opt_safety_checks and run in debug mode
+     * too.
+     */
+    true
+#else
+    false
+#endif
+    ;
+
+#if defined(_WIN32) || defined(JEMALLOC_HAVE_SCHED_GETCPU)
 /* Currently percpu_arena depends on sched_getcpu. */
 #define JEMALLOC_PERCPU_ARENA
 #endif

From 2f256b0d2de001d064efa5c755b5a10da4ddb013 Mon Sep 17 00:00:00 2001
From: Azat Khuzhin <a3at.mail@gmail.com>
Date: Sun, 24 May 2020 13:00:41 +0300
Subject: [PATCH 035/122] jemalloc: merge jemalloc_internal_defs.h.in for
 x86_64

- Use upstream JEMALLOC_PURGE_MADVISE_FREE
---
 .../internal/jemalloc_internal_defs.h.in      | 56 +++++++++----------
 1 file changed, 25 insertions(+), 31 deletions(-)

diff --git a/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/internal/jemalloc_internal_defs.h.in b/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/internal/jemalloc_internal_defs.h.in
index 63f7f765023..71a93253441 100644
--- a/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/internal/jemalloc_internal_defs.h.in
+++ b/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/internal/jemalloc_internal_defs.h.in
@@ -1,11 +1,6 @@
 /* include/jemalloc/internal/jemalloc_internal_defs.h.  Generated from jemalloc_internal_defs.h.in by configure.  */
 #ifndef JEMALLOC_INTERNAL_DEFS_H_
 #define JEMALLOC_INTERNAL_DEFS_H_
-
-#ifndef _GNU_SOURCE
-    #define _GNU_SOURCE
-#endif
-
 /*
  * If JEMALLOC_PREFIX is defined via --with-jemalloc-prefix, it will cause all
  * public APIs to be prefixed.  This makes it possible, with some care, to use
@@ -24,7 +19,7 @@
 #define JEMALLOC_OVERRIDE___LIBC_MEMALIGN
 #define JEMALLOC_OVERRIDE___LIBC_REALLOC
 #define JEMALLOC_OVERRIDE___LIBC_VALLOC
-#define JEMALLOC_OVERRIDE___POSIX_MEMALIGN
+/* #undef JEMALLOC_OVERRIDE___POSIX_MEMALIGN */
 
 /*
  * JEMALLOC_PRIVATE_NAMESPACE is used as a prefix for all library-private APIs.
@@ -54,25 +49,13 @@
 
 /* Defined if GCC __atomic atomics are available. */
 #define JEMALLOC_GCC_ATOMIC_ATOMICS 1
+/* and the 8-bit variant support. */
+#define JEMALLOC_GCC_U8_ATOMIC_ATOMICS 1
 
 /* Defined if GCC __sync atomics are available. */
 #define JEMALLOC_GCC_SYNC_ATOMICS 1
-
-/*
- * Defined if __sync_add_and_fetch(uint32_t *, uint32_t) and
- * __sync_sub_and_fetch(uint32_t *, uint32_t) are available, despite
- * __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4 not being defined (which means the
- * functions are defined in libgcc instead of being inlines).
- */
-/* #undef JE_FORCE_SYNC_COMPARE_AND_SWAP_4 */
-
-/*
- * Defined if __sync_add_and_fetch(uint64_t *, uint64_t) and
- * __sync_sub_and_fetch(uint64_t *, uint64_t) are available, despite
- * __GCC_HAVE_SYNC_COMPARE_AND_SWAP_8 not being defined (which means the
- * functions are defined in libgcc instead of being inlines).
- */
-/* #undef JE_FORCE_SYNC_COMPARE_AND_SWAP_8 */
+/* and the 8-bit variant support. */
+#define JEMALLOC_GCC_U8_SYNC_ATOMICS 1
 
 /*
  * Defined if __builtin_clz() and __builtin_clzl() are available.
@@ -84,20 +67,13 @@
  */
 /* #undef JEMALLOC_OS_UNFAIR_LOCK */
 
-/*
- * Defined if OSSpin*() functions are available, as provided by Darwin, and
- * documented in the spinlock(3) manual page.
- */
-/* #undef JEMALLOC_OSSPIN */
-
 /* Defined if syscall(2) is usable. */
 #define JEMALLOC_USE_SYSCALL
 
 /*
  * Defined if secure_getenv(3) is available.
  */
-// Don't want dependency on newer GLIBC
-//#define JEMALLOC_HAVE_SECURE_GETENV
+#define JEMALLOC_HAVE_SECURE_GETENV
 
 /*
  * Defined if issetugid(2) is available.
@@ -160,6 +136,9 @@
 /* JEMALLOC_STATS enables statistics calculation. */
 #define JEMALLOC_STATS
 
+/* JEMALLOC_EXPERIMENTAL_SMALLOCX_API enables experimental smallocx API. */
+/* #undef JEMALLOC_EXPERIMENTAL_SMALLOCX_API */
+
 /* JEMALLOC_PROF enables allocation profiling. */
 /* #undef JEMALLOC_PROF */
 
@@ -240,6 +219,12 @@
 #define JEMALLOC_INTERNAL_FFSL __builtin_ffsl
 #define JEMALLOC_INTERNAL_FFS __builtin_ffs
 
+/*
+ * popcount*() functions to use for bitmapping.
+ */
+#define JEMALLOC_INTERNAL_POPCOUNTL __builtin_popcountl
+#define JEMALLOC_INTERNAL_POPCOUNT __builtin_popcount
+
 /*
  * If defined, explicitly attempt to more uniformly distribute large allocation
  * pointer alignments across all cache indices.
@@ -252,6 +237,12 @@
  */
 /* #undef JEMALLOC_LOG */
 
+/*
+ * If defined, use readlinkat() (instead of readlink()) to follow
+ * /etc/malloc_conf.
+ */
+/* #undef JEMALLOC_READLINKAT */
+
 /*
  * Darwin (OS X) uses zones to work around Mach-O symbol override shortcomings.
  */
@@ -288,7 +279,7 @@
  *                                 MADV_FREE, though typically with higher
  *                                 system overhead.
  */
-//#define JEMALLOC_PURGE_MADVISE_FREE
+#define JEMALLOC_PURGE_MADVISE_FREE
 #define JEMALLOC_PURGE_MADVISE_DONTNEED
 #define JEMALLOC_PURGE_MADVISE_DONTNEED_ZEROS
 
@@ -370,4 +361,7 @@
  */
 #define JEMALLOC_STRERROR_R_RETURNS_CHAR_WITH_GNU_SOURCE
 
+/* Performs additional safety checks when defined. */
+/* #undef JEMALLOC_OPT_SAFETY_CHECKS */
+
 #endif /* JEMALLOC_INTERNAL_DEFS_H_ */

From 82f2b1a986965fc850188460f4873566101692a0 Mon Sep 17 00:00:00 2001
From: Azat Khuzhin <a3at.mail@gmail.com>
Date: Sun, 24 May 2020 13:02:23 +0300
Subject: [PATCH 036/122] jemalloc: merge jemalloc_internal_defs.h.in for
 aarch64

- keep LG_PAGE/LG_HUGEPAGE
- use upstream JEMALLOC_PURGE_MADVISE_FREE
---
 .../internal/jemalloc_internal_defs.h.in      | 49 +++++++------------
 1 file changed, 17 insertions(+), 32 deletions(-)

diff --git a/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/internal/jemalloc_internal_defs.h.in b/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/internal/jemalloc_internal_defs.h.in
index 089f1a773aa..231d80ceb5c 100644
--- a/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/internal/jemalloc_internal_defs.h.in
+++ b/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/internal/jemalloc_internal_defs.h.in
@@ -1,12 +1,6 @@
 /* include/jemalloc/internal/jemalloc_internal_defs.h.  Generated from jemalloc_internal_defs.h.in by configure.  */
 #ifndef JEMALLOC_INTERNAL_DEFS_H_
 #define JEMALLOC_INTERNAL_DEFS_H_
-
-
-#ifndef _GNU_SOURCE
-    #define _GNU_SOURCE
-#endif
-
 /*
  * If JEMALLOC_PREFIX is defined via --with-jemalloc-prefix, it will cause all
  * public APIs to be prefixed.  This makes it possible, with some care, to use
@@ -25,7 +19,7 @@
 #define JEMALLOC_OVERRIDE___LIBC_MEMALIGN
 #define JEMALLOC_OVERRIDE___LIBC_REALLOC
 #define JEMALLOC_OVERRIDE___LIBC_VALLOC
-#define JEMALLOC_OVERRIDE___POSIX_MEMALIGN
+/* #undef JEMALLOC_OVERRIDE___POSIX_MEMALIGN */
 
 /*
  * JEMALLOC_PRIVATE_NAMESPACE is used as a prefix for all library-private APIs.
@@ -39,9 +33,9 @@
  * Hyper-threaded CPUs may need a special instruction inside spin loops in
  * order to yield to another virtual CPU.
  */
-#define CPU_SPINWAIT
+#define CPU_SPINWAIT __asm__ volatile("pause")
 /* 1 if CPU_SPINWAIT is defined, 0 otherwise. */
-#define HAVE_CPU_SPINWAIT 0
+#define HAVE_CPU_SPINWAIT 1
 
 /*
  * Number of significant bits in virtual addresses.  This may be less than the
@@ -55,25 +49,13 @@
 
 /* Defined if GCC __atomic atomics are available. */
 #define JEMALLOC_GCC_ATOMIC_ATOMICS 1
+/* and the 8-bit variant support. */
+#define JEMALLOC_GCC_U8_ATOMIC_ATOMICS 1
 
 /* Defined if GCC __sync atomics are available. */
 #define JEMALLOC_GCC_SYNC_ATOMICS 1
-
-/*
- * Defined if __sync_add_and_fetch(uint32_t *, uint32_t) and
- * __sync_sub_and_fetch(uint32_t *, uint32_t) are available, despite
- * __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4 not being defined (which means the
- * functions are defined in libgcc instead of being inlines).
- */
-/* #undef JE_FORCE_SYNC_COMPARE_AND_SWAP_4 */
-
-/*
- * Defined if __sync_add_and_fetch(uint64_t *, uint64_t) and
- * __sync_sub_and_fetch(uint64_t *, uint64_t) are available, despite
- * __GCC_HAVE_SYNC_COMPARE_AND_SWAP_8 not being defined (which means the
- * functions are defined in libgcc instead of being inlines).
- */
-/* #undef JE_FORCE_SYNC_COMPARE_AND_SWAP_8 */
+/* and the 8-bit variant support. */
+#define JEMALLOC_GCC_U8_SYNC_ATOMICS 1
 
 /*
  * Defined if __builtin_clz() and __builtin_clzl() are available.
@@ -85,12 +67,6 @@
  */
 /* #undef JEMALLOC_OS_UNFAIR_LOCK */
 
-/*
- * Defined if OSSpin*() functions are available, as provided by Darwin, and
- * documented in the spinlock(3) manual page.
- */
-/* #undef JEMALLOC_OSSPIN */
-
 /* Defined if syscall(2) is usable. */
 #define JEMALLOC_USE_SYSCALL
 
@@ -243,6 +219,12 @@
 #define JEMALLOC_INTERNAL_FFSL __builtin_ffsl
 #define JEMALLOC_INTERNAL_FFS __builtin_ffs
 
+/*
+ * popcount*() functions to use for bitmapping.
+ */
+#define JEMALLOC_INTERNAL_POPCOUNTL __builtin_popcountl
+#define JEMALLOC_INTERNAL_POPCOUNT __builtin_popcount
+
 /*
  * If defined, explicitly attempt to more uniformly distribute large allocation
  * pointer alignments across all cache indices.
@@ -297,7 +279,7 @@
  *                                 MADV_FREE, though typically with higher
  *                                 system overhead.
  */
-// #define JEMALLOC_PURGE_MADVISE_FREE
+#define JEMALLOC_PURGE_MADVISE_FREE
 #define JEMALLOC_PURGE_MADVISE_DONTNEED
 #define JEMALLOC_PURGE_MADVISE_DONTNEED_ZEROS
 
@@ -379,4 +361,7 @@
  */
 #define JEMALLOC_STRERROR_R_RETURNS_CHAR_WITH_GNU_SOURCE
 
+/* Performs additional safety checks when defined. */
+/* #undef JEMALLOC_OPT_SAFETY_CHECKS */
+
 #endif /* JEMALLOC_INTERNAL_DEFS_H_ */

From 9f1cefd8dfea9e28b1d1b2a8ae7eb64dedb8ceaf Mon Sep 17 00:00:00 2001
From: Azat Khuzhin <a3at.mail@gmail.com>
Date: Sun, 24 May 2020 13:09:10 +0300
Subject: [PATCH 037/122] jemalloc: disable secure_getenv()

---
 .../jemalloc/internal/jemalloc_internal_defs.h.in               | 2 +-
 .../jemalloc/internal/jemalloc_internal_defs.h.in               | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/internal/jemalloc_internal_defs.h.in b/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/internal/jemalloc_internal_defs.h.in
index 231d80ceb5c..3852f8b6382 100644
--- a/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/internal/jemalloc_internal_defs.h.in
+++ b/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/internal/jemalloc_internal_defs.h.in
@@ -73,7 +73,7 @@
 /*
  * Defined if secure_getenv(3) is available.
  */
-#define JEMALLOC_HAVE_SECURE_GETENV
+// #define JEMALLOC_HAVE_SECURE_GETENV
 
 /*
  * Defined if issetugid(2) is available.
diff --git a/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/internal/jemalloc_internal_defs.h.in b/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/internal/jemalloc_internal_defs.h.in
index 71a93253441..d5cf0e719ef 100644
--- a/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/internal/jemalloc_internal_defs.h.in
+++ b/contrib/jemalloc-cmake/include_linux_x86_64/jemalloc/internal/jemalloc_internal_defs.h.in
@@ -73,7 +73,7 @@
 /*
  * Defined if secure_getenv(3) is available.
  */
-#define JEMALLOC_HAVE_SECURE_GETENV
+// #define JEMALLOC_HAVE_SECURE_GETENV
 
 /*
  * Defined if issetugid(2) is available.

From 5cd31dbf0deda64dadcf6a3f6b8f5ad33eb9726a Mon Sep 17 00:00:00 2001
From: Azat Khuzhin <a3at.mail@gmail.com>
Date: Sun, 24 May 2020 13:12:00 +0300
Subject: [PATCH 038/122] jemalloc: add _GNU_SOURCE for RTLD_NEXT

---
 contrib/jemalloc-cmake/CMakeLists.txt | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/contrib/jemalloc-cmake/CMakeLists.txt b/contrib/jemalloc-cmake/CMakeLists.txt
index e187ca2926d..e5f52f4dc8e 100644
--- a/contrib/jemalloc-cmake/CMakeLists.txt
+++ b/contrib/jemalloc-cmake/CMakeLists.txt
@@ -90,6 +90,8 @@ if (ENABLE_JEMALLOC)
         endif ()
 
         target_compile_options(jemalloc PRIVATE -Wno-redundant-decls)
+        # for RTLD_NEXT
+        target_compile_options(jemalloc PRIVATE -D_GNU_SOURCE)
     else ()
         find_library(LIBRARY_JEMALLOC jemalloc)
         find_path(INCLUDE_JEMALLOC jemalloc/jemalloc.h)

From bc572dc5f22b139a60a9a0dc188f2e38ed0ba9f9 Mon Sep 17 00:00:00 2001
From: Azat Khuzhin <a3at.mail@gmail.com>
Date: Sun, 24 May 2020 13:20:07 +0300
Subject: [PATCH 039/122] jemalloc: sync common (non per-arch) pre-generated
 headers

$ for f in include/jemalloc/jemalloc_*.h; do cp /src/oss/jemalloc/$f $f; done
---
 .../include/jemalloc/jemalloc_defs.h            |  6 ++++++
 .../include/jemalloc/jemalloc_macros.h          | 17 ++++++++++++-----
 .../include/jemalloc/jemalloc_rename.h          |  1 +
 3 files changed, 19 insertions(+), 5 deletions(-)

diff --git a/contrib/jemalloc-cmake/include/jemalloc/jemalloc_defs.h b/contrib/jemalloc-cmake/include/jemalloc/jemalloc_defs.h
index d1389237a77..6a03a231a0e 100644
--- a/contrib/jemalloc-cmake/include/jemalloc/jemalloc_defs.h
+++ b/contrib/jemalloc-cmake/include/jemalloc/jemalloc_defs.h
@@ -5,6 +5,12 @@
 /* Defined if alloc_size attribute is supported. */
 #define JEMALLOC_HAVE_ATTR_ALLOC_SIZE
 
+/* Defined if format_arg(...) attribute is supported. */
+#define JEMALLOC_HAVE_ATTR_FORMAT_ARG
+
+/* Defined if format(gnu_printf, ...) attribute is supported. */
+#define JEMALLOC_HAVE_ATTR_FORMAT_GNU_PRINTF
+
 /* Defined if format(printf, ...) attribute is supported. */
 #define JEMALLOC_HAVE_ATTR_FORMAT_PRINTF
 
diff --git a/contrib/jemalloc-cmake/include/jemalloc/jemalloc_macros.h b/contrib/jemalloc-cmake/include/jemalloc/jemalloc_macros.h
index 7432f1cda53..34235894285 100644
--- a/contrib/jemalloc-cmake/include/jemalloc/jemalloc_macros.h
+++ b/contrib/jemalloc-cmake/include/jemalloc/jemalloc_macros.h
@@ -4,12 +4,13 @@
 #include <limits.h>
 #include <strings.h>
 
-#define JEMALLOC_VERSION "5.1.0-56-g41b7372eadee941b9164751b8d4963f915d3ceae"
+#define JEMALLOC_VERSION "5.2.1-0-gea6b3e973b477b8061e0076bb257dbd7f3faa756"
 #define JEMALLOC_VERSION_MAJOR 5
-#define JEMALLOC_VERSION_MINOR 1
-#define JEMALLOC_VERSION_BUGFIX 0
-#define JEMALLOC_VERSION_NREV 56
-#define JEMALLOC_VERSION_GID "41b7372eadee941b9164751b8d4963f915d3ceae"
+#define JEMALLOC_VERSION_MINOR 2
+#define JEMALLOC_VERSION_BUGFIX 1
+#define JEMALLOC_VERSION_NREV 0
+#define JEMALLOC_VERSION_GID "ea6b3e973b477b8061e0076bb257dbd7f3faa756"
+#define JEMALLOC_VERSION_GID_IDENT ea6b3e973b477b8061e0076bb257dbd7f3faa756
 
 #define MALLOCX_LG_ALIGN(la)	((int)(la))
 #if LG_SIZEOF_PTR == 2
@@ -68,6 +69,7 @@
 #      define JEMALLOC_EXPORT __declspec(dllimport)
 #    endif
 #  endif
+#  define JEMALLOC_FORMAT_ARG(i)
 #  define JEMALLOC_FORMAT_PRINTF(s, i)
 #  define JEMALLOC_NOINLINE __declspec(noinline)
 #  ifdef __cplusplus
@@ -95,6 +97,11 @@
 #  ifndef JEMALLOC_EXPORT
 #    define JEMALLOC_EXPORT JEMALLOC_ATTR(visibility("default"))
 #  endif
+#  ifdef JEMALLOC_HAVE_ATTR_FORMAT_ARG
+#    define JEMALLOC_FORMAT_ARG(i) JEMALLOC_ATTR(__format_arg__(3))
+#  else
+#    define JEMALLOC_FORMAT_ARG(i)
+#  endif
 #  ifdef JEMALLOC_HAVE_ATTR_FORMAT_GNU_PRINTF
 #    define JEMALLOC_FORMAT_PRINTF(s, i) JEMALLOC_ATTR(format(gnu_printf, s, i))
 #  elif defined(JEMALLOC_HAVE_ATTR_FORMAT_PRINTF)
diff --git a/contrib/jemalloc-cmake/include/jemalloc/jemalloc_rename.h b/contrib/jemalloc-cmake/include/jemalloc/jemalloc_rename.h
index a2ea2dd3533..2e94f7a0cc3 100644
--- a/contrib/jemalloc-cmake/include/jemalloc/jemalloc_rename.h
+++ b/contrib/jemalloc-cmake/include/jemalloc/jemalloc_rename.h
@@ -17,6 +17,7 @@
 #  define je_malloc_stats_print malloc_stats_print
 #  define je_malloc_usable_size malloc_usable_size
 #  define je_mallocx mallocx
+#  define je_smallocx_ea6b3e973b477b8061e0076bb257dbd7f3faa756 smallocx_ea6b3e973b477b8061e0076bb257dbd7f3faa756
 #  define je_nallocx nallocx
 #  define je_posix_memalign posix_memalign
 #  define je_rallocx rallocx

From a8749c5d5c668a079239112f96a1fafe072a9ad5 Mon Sep 17 00:00:00 2001
From: Azat Khuzhin <a3at.mail@gmail.com>
Date: Sun, 24 May 2020 19:17:50 +0300
Subject: [PATCH 040/122] jemalloc: aarch64 does not have pause

---
 .../jemalloc/internal/jemalloc_internal_defs.h.in             | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/internal/jemalloc_internal_defs.h.in b/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/internal/jemalloc_internal_defs.h.in
index 3852f8b6382..cbd2740e1f1 100644
--- a/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/internal/jemalloc_internal_defs.h.in
+++ b/contrib/jemalloc-cmake/include_linux_aarch64/jemalloc/internal/jemalloc_internal_defs.h.in
@@ -33,9 +33,9 @@
  * Hyper-threaded CPUs may need a special instruction inside spin loops in
  * order to yield to another virtual CPU.
  */
-#define CPU_SPINWAIT __asm__ volatile("pause")
+#define CPU_SPINWAIT
 /* 1 if CPU_SPINWAIT is defined, 0 otherwise. */
-#define HAVE_CPU_SPINWAIT 1
+#define HAVE_CPU_SPINWAIT 9
 
 /*
  * Number of significant bits in virtual addresses.  This may be less than the

From b694d6ed8c78eacb248a0cd20255e81433f88d27 Mon Sep 17 00:00:00 2001
From: Azat Khuzhin <a3at.mail@gmail.com>
Date: Tue, 26 May 2020 21:45:51 +0300
Subject: [PATCH 041/122] jemalloc: disable oversize_threshold (slower)

---
 contrib/jemalloc-cmake/CMakeLists.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/contrib/jemalloc-cmake/CMakeLists.txt b/contrib/jemalloc-cmake/CMakeLists.txt
index e5f52f4dc8e..a02a63ba999 100644
--- a/contrib/jemalloc-cmake/CMakeLists.txt
+++ b/contrib/jemalloc-cmake/CMakeLists.txt
@@ -17,7 +17,7 @@ if (ENABLE_JEMALLOC)
         #
         # By enabling percpu_arena number of arenas limited to number of CPUs and hence
         # this problem should go away.
-        set (JEMALLOC_CONFIG_MALLOC_CONF "percpu_arena:percpu" CACHE STRING "Change default configuration string of JEMalloc" )
+        set (JEMALLOC_CONFIG_MALLOC_CONF "percpu_arena:percpu,oversize_threshold:0" CACHE STRING "Change default configuration string of JEMalloc" )
         message (STATUS "jemalloc malloc_conf: ${JEMALLOC_CONFIG_MALLOC_CONF}")
 
         set (LIBRARY_DIR "${ClickHouse_SOURCE_DIR}/contrib/jemalloc")

From 210e67d8317252252402aeba6ef6d05051f7fc7e Mon Sep 17 00:00:00 2001
From: Azat Khuzhin <a3at.mail@gmail.com>
Date: Tue, 26 May 2020 21:51:06 +0300
Subject: [PATCH 042/122] jemalloc: allow changing default malloc_conf without
 purging cmake cache

---
 contrib/jemalloc-cmake/CMakeLists.txt | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/contrib/jemalloc-cmake/CMakeLists.txt b/contrib/jemalloc-cmake/CMakeLists.txt
index a02a63ba999..63c5a5e66ea 100644
--- a/contrib/jemalloc-cmake/CMakeLists.txt
+++ b/contrib/jemalloc-cmake/CMakeLists.txt
@@ -17,7 +17,13 @@ if (ENABLE_JEMALLOC)
         #
         # By enabling percpu_arena number of arenas limited to number of CPUs and hence
         # this problem should go away.
-        set (JEMALLOC_CONFIG_MALLOC_CONF "percpu_arena:percpu,oversize_threshold:0" CACHE STRING "Change default configuration string of JEMalloc" )
+        set (JEMALLOC_CONFIG_MALLOC_CONF "percpu_arena:percpu,oversize_threshold:0")
+        # CACHE variable is empty, to allow changing defaults without necessity
+        # to purge cache
+        set (JEMALLOC_CONFIG_MALLOC_CONF_OVERRIDE "" CACHE STRING "Change default configuration string of JEMalloc" )
+        if (JEMALLOC_CONFIG_MALLOC_CONF_OVERRIDE)
+            set (JEMALLOC_CONFIG_MALLOC_CONF "${JEMALLOC_CONFIG_MALLOC_CONF_OVERRIDE}")
+        endif()
         message (STATUS "jemalloc malloc_conf: ${JEMALLOC_CONFIG_MALLOC_CONF}")
 
         set (LIBRARY_DIR "${ClickHouse_SOURCE_DIR}/contrib/jemalloc")

From ec894e7cb3e0e3f9f7c880b00d44ee3b83e028ad Mon Sep 17 00:00:00 2001
From: Azat Khuzhin <a3at.mail@gmail.com>
Date: Sun, 24 May 2020 19:39:31 +0300
Subject: [PATCH 043/122] Check for POPCNT instruction

---
 programs/main.cpp                             | 19 ++++++++++++++++---
 ...heck_cpu_instructions_at_startup.reference |  1 +
 ...01103_check_cpu_instructions_at_startup.sh | 13 +++++++++----
 3 files changed, 26 insertions(+), 7 deletions(-)

diff --git a/programs/main.cpp b/programs/main.cpp
index 2eb226d3c00..b1bff1de1b0 100644
--- a/programs/main.cpp
+++ b/programs/main.cpp
@@ -127,9 +127,10 @@ enum class InstructionFail
     SSSE3 = 2,
     SSE4_1 = 3,
     SSE4_2 = 4,
-    AVX = 5,
-    AVX2 = 6,
-    AVX512 = 7
+    POPCNT = 5,
+    AVX = 6,
+    AVX2 = 7,
+    AVX512 = 8
 };
 
 const char * instructionFailToString(InstructionFail fail)
@@ -146,6 +147,8 @@ const char * instructionFailToString(InstructionFail fail)
             return "SSE4.1";
         case InstructionFail::SSE4_2:
             return "SSE4.2";
+        case InstructionFail::POPCNT:
+            return "POPCNT";
         case InstructionFail::AVX:
             return "AVX";
         case InstructionFail::AVX2:
@@ -189,6 +192,16 @@ void checkRequiredInstructionsImpl(volatile InstructionFail & fail)
     __asm__ volatile ("pcmpgtq %%xmm0, %%xmm0" : : : "xmm0");
 #endif
 
+    /// Defined by -msse4.2
+#if defined(__POPCNT__)
+    fail = InstructionFail::POPCNT;
+    {
+        uint64_t a = 0;
+        uint64_t b = 0;
+        __asm__ volatile ("popcnt %1, %0" : "=r"(a) :"r"(b) : );
+    }
+#endif
+
 #if defined(__AVX__)
     fail = InstructionFail::AVX;
     __asm__ volatile ("vaddpd %%ymm0, %%ymm0, %%ymm0" : : : "ymm0");
diff --git a/tests/queries/0_stateless/01103_check_cpu_instructions_at_startup.reference b/tests/queries/0_stateless/01103_check_cpu_instructions_at_startup.reference
index 59c88f013dd..03ed07cf1a4 100644
--- a/tests/queries/0_stateless/01103_check_cpu_instructions_at_startup.reference
+++ b/tests/queries/0_stateless/01103_check_cpu_instructions_at_startup.reference
@@ -1,4 +1,5 @@
 Instruction check fail. The CPU does not support SSSE3 instruction set.
 Instruction check fail. The CPU does not support SSE4.1 instruction set.
 Instruction check fail. The CPU does not support SSE4.2 instruction set.
+Instruction check fail. The CPU does not support POPCNT instruction set.
 1
diff --git a/tests/queries/0_stateless/01103_check_cpu_instructions_at_startup.sh b/tests/queries/0_stateless/01103_check_cpu_instructions_at_startup.sh
index 15466e1889a..5ae4f8b3dd2 100755
--- a/tests/queries/0_stateless/01103_check_cpu_instructions_at_startup.sh
+++ b/tests/queries/0_stateless/01103_check_cpu_instructions_at_startup.sh
@@ -9,8 +9,13 @@ ${CLICKHOUSE_LOCAL} --query "SELECT max(value LIKE '%sanitize%') FROM system.bui
 
 command=$(command -v ${CLICKHOUSE_LOCAL})
 
-qemu-x86_64-static -cpu qemu64                        $command --query "SELECT 1" 2>&1 | grep -v -F "warning: TCG doesn't support requested feature" ||:
-qemu-x86_64-static -cpu qemu64,+ssse3                 $command --query "SELECT 1" 2>&1 | grep -v -F "warning: TCG doesn't support requested feature" ||:
-qemu-x86_64-static -cpu qemu64,+ssse3,+sse4.1         $command --query "SELECT 1" 2>&1 | grep -v -F "warning: TCG doesn't support requested feature" ||:
-qemu-x86_64-static -cpu qemu64,+ssse3,+sse4.1,+sse4.2 $command --query "SELECT 1" 2>&1 | grep -v -F "warning: TCG doesn't support requested feature" ||:
+function run_with_cpu()
+{
+    qemu-x86_64-static -cpu "$@" $command --query "SELECT 1" 2>&1 | grep -v -F "warning: TCG doesn't support requested feature" ||:
+}
 
+run_with_cpu qemu64
+run_with_cpu qemu64,+ssse3
+run_with_cpu qemu64,+ssse3,+sse4.1
+run_with_cpu qemu64,+ssse3,+sse4.1,+sse4.2
+run_with_cpu qemu64,+ssse3,+sse4.1,+sse4.2,+popcnt

From efc0b977078af8a80a2afd6a2849efeb78148201 Mon Sep 17 00:00:00 2001
From: Mikhail Filimonov <mfilimonov@altinity.com>
Date: Fri, 22 May 2020 18:58:08 +0200
Subject: [PATCH 044/122] Fix Kafka performance issue related to reschedules
 based on limits, which were always applied.

---
 src/Storages/Kafka/KafkaBlockInputStream.h    |  1 +
 .../Kafka/ReadBufferFromKafkaConsumer.h       |  1 +
 src/Storages/Kafka/StorageKafka.cpp           | 31 ++++++++++----
 tests/integration/test_storage_kafka/test.py  | 40 ++++++++++++++++++-
 4 files changed, 64 insertions(+), 9 deletions(-)

diff --git a/src/Storages/Kafka/KafkaBlockInputStream.h b/src/Storages/Kafka/KafkaBlockInputStream.h
index 1f94ee332d3..e3052122894 100644
--- a/src/Storages/Kafka/KafkaBlockInputStream.h
+++ b/src/Storages/Kafka/KafkaBlockInputStream.h
@@ -25,6 +25,7 @@ public:
     void readSuffixImpl() override;
 
     void commit();
+    bool isStalled() const { return buffer->isStalled(); }
 
 private:
     StorageKafka & storage;
diff --git a/src/Storages/Kafka/ReadBufferFromKafkaConsumer.h b/src/Storages/Kafka/ReadBufferFromKafkaConsumer.h
index 12da701a55d..e90e3b48881 100644
--- a/src/Storages/Kafka/ReadBufferFromKafkaConsumer.h
+++ b/src/Storages/Kafka/ReadBufferFromKafkaConsumer.h
@@ -38,6 +38,7 @@ public:
 
     bool hasMorePolledMessages() const;
     bool polledDataUnusable() const { return (was_stopped || rebalance_happened); }
+    bool isStalled() const { return stalled; }
 
     void storeLastReadMessageOffset();
     void resetToLastCommitted(const char * msg);
diff --git a/src/Storages/Kafka/StorageKafka.cpp b/src/Storages/Kafka/StorageKafka.cpp
index d1dfbe9d875..7731cf3c06a 100644
--- a/src/Storages/Kafka/StorageKafka.cpp
+++ b/src/Storages/Kafka/StorageKafka.cpp
@@ -51,6 +51,7 @@ namespace
 {
     const auto RESCHEDULE_MS = 500;
     const auto CLEANUP_TIMEOUT_MS = 3000;
+    const auto MAX_THREAD_WORK_DURATION_MS = 60000;  // once per minute leave do reschedule (we can't lock threads in pool forever)
 
     /// Configuration prefix
     const String CONFIG_PREFIX = "kafka";
@@ -386,6 +387,8 @@ void StorageKafka::threadFunc()
         size_t dependencies_count = DatabaseCatalog::instance().getDependencies(table_id).size();
         if (dependencies_count)
         {
+            auto start_time = std::chrono::steady_clock::now();
+
             // Keep streaming as long as there are attached views and streaming is not cancelled
             while (!stream_cancelled && num_created_consumers > 0)
             {
@@ -394,9 +397,21 @@ void StorageKafka::threadFunc()
 
                 LOG_DEBUG(log, "Started streaming to {} attached views", dependencies_count);
 
-                // Reschedule if not limited
-                if (!streamToViews())
+                // Exit the loop & reschedule if some stream stalled
+                auto some_stream_is_stalled = streamToViews();
+                if (some_stream_is_stalled)
+                {
+                    LOG_TRACE(log, "Stream(s) stalled. Reschedule.");
                     break;
+                }
+
+                auto ts = std::chrono::steady_clock::now();
+                auto duration = std::chrono::duration_cast<std::chrono::milliseconds>(ts-start_time);
+                if (duration.count() > MAX_THREAD_WORK_DURATION_MS)
+                {
+                    LOG_TRACE(log, "Thread work duration limit exceeded. Reschedule.");
+                    break;
+                }
             }
         }
     }
@@ -459,15 +474,15 @@ bool StorageKafka::streamToViews()
     // It will be cancelled on underlying layer (kafka buffer)
     std::atomic<bool> stub = {false};
     copyData(*in, *block_io.out, &stub);
+
+    bool some_stream_is_stalled = false;
     for (auto & stream : streams)
+    {
+        some_stream_is_stalled = some_stream_is_stalled || stream->as<KafkaBlockInputStream>()->isStalled();
         stream->as<KafkaBlockInputStream>()->commit();
+    }
 
-    // Check whether the limits were applied during query execution
-    bool limits_applied = false;
-    const BlockStreamProfileInfo & info = in->getProfileInfo();
-    limits_applied = info.hasAppliedLimit();
-
-    return limits_applied;
+    return some_stream_is_stalled;
 }
 
 void registerStorageKafka(StorageFactory & factory)
diff --git a/tests/integration/test_storage_kafka/test.py b/tests/integration/test_storage_kafka/test.py
index 8e329a1cb60..d71347e83f1 100644
--- a/tests/integration/test_storage_kafka/test.py
+++ b/tests/integration/test_storage_kafka/test.py
@@ -1017,7 +1017,10 @@ def test_kafka_flush_by_block_size(kafka_cluster):
 
     time.sleep(1)
 
-    result = instance.query('SELECT count() FROM test.view')
+    # TODO: due to https://github.com/ClickHouse/ClickHouse/pull/11149
+    # second flush happens earlier than expected, so we have 2 parts here instead of one
+    # flush by block size works correctly, so the feature checked by the test is working correctly
+    result = instance.query("SELECT count() FROM test.view WHERE _part='all_1_1_0'")
     # print(result)
 
     # kafka_cluster.open_bash_shell('instance')
@@ -1390,6 +1393,41 @@ def test_commits_of_unprocessed_messages_on_drop(kafka_cluster):
     assert TSV(result) == TSV('{0}\t{0}\t{0}'.format(i[0]-1)), 'Missing data!'
 
 
+
+@pytest.mark.timeout(120)
+def test_bad_reschedule(kafka_cluster):
+    messages = [json.dumps({'key': j+1, 'value': j+1}) for j in range(20000)]
+    kafka_produce('test_bad_reschedule', messages)
+
+    instance.query('''
+        CREATE TABLE test.kafka (key UInt64, value UInt64)
+            ENGINE = Kafka
+            SETTINGS kafka_broker_list = 'kafka1:19092',
+                    kafka_topic_list = 'test_bad_reschedule',
+                    kafka_group_name = 'test_bad_reschedule',
+                    kafka_format = 'JSONEachRow',
+                    kafka_max_block_size = 1000;
+
+        CREATE MATERIALIZED VIEW test.destination Engine=Log AS
+        SELECT
+            key,
+            now() as consume_ts,
+            value,
+            _topic,
+            _key,
+            _offset,
+            _partition,
+            _timestamp
+        FROM test.kafka;
+    ''')
+
+    while int(instance.query("SELECT count() FROM test.destination")) < 20000:
+        print("Waiting for consume")
+        time.sleep(1)
+
+    assert int(instance.query("SELECT max(consume_ts) - min(consume_ts) FROM test.destination")) < 8
+
+
 @pytest.mark.timeout(1200)
 def test_kafka_duplicates_when_commit_failed(kafka_cluster):
     messages = [json.dumps({'key': j+1, 'value': 'x' * 300}) for j in range(22)]

From c73b8372e786ba6de958e8be222bf20081c97afd Mon Sep 17 00:00:00 2001
From: filimonov <1549571+filimonov@users.noreply.github.com>
Date: Wed, 27 May 2020 07:44:13 +0200
Subject: [PATCH 045/122] Update test.py

---
 tests/integration/test_storage_kafka/test.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tests/integration/test_storage_kafka/test.py b/tests/integration/test_storage_kafka/test.py
index d71347e83f1..13577864870 100644
--- a/tests/integration/test_storage_kafka/test.py
+++ b/tests/integration/test_storage_kafka/test.py
@@ -1017,7 +1017,7 @@ def test_kafka_flush_by_block_size(kafka_cluster):
 
     time.sleep(1)
 
-    # TODO: due to https://github.com/ClickHouse/ClickHouse/pull/11149
+    # TODO: due to https://github.com/ClickHouse/ClickHouse/issues/11216
     # second flush happens earlier than expected, so we have 2 parts here instead of one
     # flush by block size works correctly, so the feature checked by the test is working correctly
     result = instance.query("SELECT count() FROM test.view WHERE _part='all_1_1_0'")

From cd008f9b7fb427895ab3bcaffd163aa9746de166 Mon Sep 17 00:00:00 2001
From: Alexander Kuzmenkov <36882414+akuzm@users.noreply.github.com>
Date: Wed, 27 May 2020 10:58:29 +0300
Subject: [PATCH 046/122] Update functions_with_hash_tables.xml

---
 tests/performance/functions_with_hash_tables.xml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tests/performance/functions_with_hash_tables.xml b/tests/performance/functions_with_hash_tables.xml
index 72570a5925c..2e9a88107bd 100644
--- a/tests/performance/functions_with_hash_tables.xml
+++ b/tests/performance/functions_with_hash_tables.xml
@@ -1,6 +1,6 @@
 <test>
     <query>select arrayUniq(range(1 + (number % 100) * 10)) from numbers(100000) format Null</query>
-    <query>select arrayDistinct(range(1 + (number % 100) * 10)) from numbers(10000) format Null</query>
+    <query>select arrayDistinct(range(1 + (number % 100) * 10)) from numbers(100000) format Null</query>
     <query>select arrayEnumerateUniq(range(1 + (number % 100) * 10)) from numbers(100000) format Null</query>
     <query>select arrayIntersect(range((1 + number % 100)), range(1, (1 + number % 100) + 1)) from numbers(100000) format Null</query>
     <query>select groupUniqArray(rand() % 100) from numbers(1000 * 1000) group by number / 1000 format Null</query>

From d941c483b2923c378536631fc479c06f557a7e44 Mon Sep 17 00:00:00 2001
From: Ivan Blinkov <github@blinkov.ru>
Date: Wed, 27 May 2020 11:25:36 +0300
Subject: [PATCH 047/122] Update test.py

---
 tests/integration/test_system_queries/test.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/tests/integration/test_system_queries/test.py b/tests/integration/test_system_queries/test.py
index 9397cd20ec7..6f36a13b184 100644
--- a/tests/integration/test_system_queries/test.py
+++ b/tests/integration/test_system_queries/test.py
@@ -97,6 +97,7 @@ def test_SYSTEM_FLUSH_LOGS(started_cluster):
     instance = cluster.instances['ch1']
     instance.query('''
         SET log_queries = 0;
+        SYSTEM FLUSH LOGS;
         TRUNCATE TABLE system.query_log;
     ''')
     for i in range(4):
@@ -104,13 +105,13 @@ def test_SYSTEM_FLUSH_LOGS(started_cluster):
         # by expiration of flush_interval_millisecond and test probable race condition.
         time.sleep(0.5)
         result = instance.query('''
-            SET log_queries = 1;
             SELECT 1 FORMAT Null;
             SET log_queries = 0;
             SYSTEM FLUSH LOGS;
             SELECT count() FROM system.query_log;''')
         instance.query('''
             SET log_queries = 0;
+            SYSTEM FLUSH LOGS;
             TRUNCATE TABLE system.query_log;
         ''')
         assert TSV(result) == TSV('4')

From f29d3b8cc916d9ff750186f89e4932b076481193 Mon Sep 17 00:00:00 2001
From: alesapin <alesapin@gmail.com>
Date: Wed, 27 May 2020 11:58:11 +0300
Subject: [PATCH 048/122] Fix cmake one more time

---
 contrib/cctz-cmake/CMakeLists.txt | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/contrib/cctz-cmake/CMakeLists.txt b/contrib/cctz-cmake/CMakeLists.txt
index d353fc9d4de..76a1fe6e8ac 100644
--- a/contrib/cctz-cmake/CMakeLists.txt
+++ b/contrib/cctz-cmake/CMakeLists.txt
@@ -622,7 +622,8 @@ if (USE_INTERNAL_CCTZ)
         # CMake can shuffle each of target_link_libraries arguments with other
         # libraries in linker command. To avoid this we hardcode whole-archive
         # library into single string.
-        target_link_libraries(cctz PUBLIC "-Wl,--whole-archive contrib/cctz-cmake/libtzdata.a -Wl,--no-whole-archive")
+        add_dependencies(cctz tzdata)
+        target_link_libraries(cctz PRIVATE "-Wl,--whole-archive contrib/cctz-cmake/libtzdata.a -Wl,--no-whole-archive")
     endif ()
 
 else ()

From f452c50d86f897e40bc167abf56674d41b77c7c5 Mon Sep 17 00:00:00 2001
From: alesapin <alesapin@gmail.com>
Date: Wed, 27 May 2020 12:00:33 +0300
Subject: [PATCH 049/122] Replace private with public

---
 contrib/cctz-cmake/CMakeLists.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/contrib/cctz-cmake/CMakeLists.txt b/contrib/cctz-cmake/CMakeLists.txt
index 76a1fe6e8ac..707ebe30a13 100644
--- a/contrib/cctz-cmake/CMakeLists.txt
+++ b/contrib/cctz-cmake/CMakeLists.txt
@@ -623,7 +623,7 @@ if (USE_INTERNAL_CCTZ)
         # libraries in linker command. To avoid this we hardcode whole-archive
         # library into single string.
         add_dependencies(cctz tzdata)
-        target_link_libraries(cctz PRIVATE "-Wl,--whole-archive contrib/cctz-cmake/libtzdata.a -Wl,--no-whole-archive")
+        target_link_libraries(cctz PUBLIC "-Wl,--whole-archive contrib/cctz-cmake/libtzdata.a -Wl,--no-whole-archive")
     endif ()
 
 else ()

From d04962808e1f0c4fe53db55aab765741f8b6b945 Mon Sep 17 00:00:00 2001
From: alesapin <alesapin@gmail.com>
Date: Wed, 27 May 2020 12:14:40 +0300
Subject: [PATCH 050/122] Actually it's INTERFACE linking

---
 contrib/cctz-cmake/CMakeLists.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/contrib/cctz-cmake/CMakeLists.txt b/contrib/cctz-cmake/CMakeLists.txt
index 707ebe30a13..2c44b25b17a 100644
--- a/contrib/cctz-cmake/CMakeLists.txt
+++ b/contrib/cctz-cmake/CMakeLists.txt
@@ -623,7 +623,7 @@ if (USE_INTERNAL_CCTZ)
         # libraries in linker command. To avoid this we hardcode whole-archive
         # library into single string.
         add_dependencies(cctz tzdata)
-        target_link_libraries(cctz PUBLIC "-Wl,--whole-archive contrib/cctz-cmake/libtzdata.a -Wl,--no-whole-archive")
+        target_link_libraries(cctz INTERFACE "-Wl,--whole-archive contrib/cctz-cmake/libtzdata.a -Wl,--no-whole-archive")
     endif ()
 
 else ()

From 61974e0047e9826ab84d5d4b07f93ec065b285dc Mon Sep 17 00:00:00 2001
From: Nikolai Sorokin <sonikint@gmail.com>
Date: Mon, 27 Apr 2020 17:47:59 +0300
Subject: [PATCH 051/122] TTL Expressions WHERE and GROUP BY draft

---
 src/DataStreams/TTLBlockInputStream.cpp  | 225 +++++++++++++++++++++--
 src/DataStreams/TTLBlockInputStream.h    |   8 +
 src/Interpreters/ExpressionAnalyzer.cpp  |   9 +-
 src/Interpreters/SyntaxAnalyzer.cpp      |  17 +-
 src/Interpreters/SyntaxAnalyzer.h        |   2 +-
 src/Parsers/ASTTTLElement.cpp            |  67 ++++++-
 src/Parsers/ASTTTLElement.h              |  38 +++-
 src/Parsers/ExpressionElementParsers.cpp |  84 ++++++++-
 src/Parsers/ExpressionListParsers.cpp    |   9 +
 src/Parsers/ExpressionListParsers.h      |  11 +-
 src/Storages/MergeTree/MergeTreeData.cpp |  44 ++++-
 src/Storages/MergeTree/MergeTreeData.h   |   9 +
 12 files changed, 479 insertions(+), 44 deletions(-)

diff --git a/src/DataStreams/TTLBlockInputStream.cpp b/src/DataStreams/TTLBlockInputStream.cpp
index d2d5d6a92f9..d816cba4b2f 100644
--- a/src/DataStreams/TTLBlockInputStream.cpp
+++ b/src/DataStreams/TTLBlockInputStream.cpp
@@ -67,6 +67,30 @@ TTLBlockInputStream::TTLBlockInputStream(
             default_expr_list, storage.getColumns().getAllPhysical());
         defaults_expression = ExpressionAnalyzer{default_expr_list, syntax_result, storage.global_context}.getActions(true);
     }
+
+    if (storage.hasRowsTTL() && !storage.rows_ttl_entry.group_by_keys.empty())
+    {
+        ColumnNumbers keys;
+        for (const auto & key : storage.rows_ttl_entry.group_by_keys)
+            keys.push_back(header.getPositionByName(key));
+        agg_key_columns.resize(storage.rows_ttl_entry.group_by_keys.size());
+
+        AggregateDescriptions aggregates = storage.rows_ttl_entry.aggregate_descriptions;
+        for (auto & descr : aggregates)
+            if (descr.arguments.empty())
+                for (const auto & name : descr.argument_names)
+                    descr.arguments.push_back(header.getPositionByName(name));
+        agg_aggregate_columns.resize(storage.rows_ttl_entry.aggregate_descriptions.size());
+
+        const Settings & settings = storage.global_context.getSettingsRef();
+
+        Aggregator::Params params(header, keys, aggregates,
+            false, settings.max_rows_to_group_by, settings.group_by_overflow_mode,
+            SettingUInt64(0), SettingUInt64(0),
+            settings.max_bytes_before_external_group_by, settings.empty_result_for_aggregation_by_empty_set,
+            storage.global_context.getTemporaryVolume(), settings.max_threads, settings.min_free_disk_space_for_temporary_data);
+        aggregator = std::make_unique<Aggregator>(params);
+    }
 }
 
 bool TTLBlockInputStream::isTTLExpired(time_t ttl) const
@@ -77,7 +101,8 @@ bool TTLBlockInputStream::isTTLExpired(time_t ttl) const
 Block TTLBlockInputStream::readImpl()
 {
     /// Skip all data if table ttl is expired for part
-    if (storage.hasRowsTTL() && isTTLExpired(old_ttl_infos.table_ttl.max))
+    if (storage.hasRowsTTL() && !storage.rows_ttl_entry.where_expression &&
+        storage.rows_ttl_entry.group_by_keys.empty() && isTTLExpired(old_ttl_infos.table_ttl.max))
     {
         rows_removed = data_part->rows_count;
         return {};
@@ -85,7 +110,43 @@ Block TTLBlockInputStream::readImpl()
 
     Block block = children.at(0)->read();
     if (!block)
+    {
+        if (aggregator && !agg_result.empty())
+        {
+            MutableColumns result_columns;
+            const auto & column_names = header.getNames();
+            for (const auto & column_name : column_names)
+            {
+                const IColumn * values_column = header.getByName(column_name).column.get();
+                MutableColumnPtr result_column = values_column->cloneEmpty();
+                result_columns.emplace_back(std::move(result_column));
+            }
+
+            auto aggregated_res = aggregator->convertToBlocks(agg_result, true, 1);
+            for (auto & agg_block : aggregated_res)
+            {
+                for (const auto & [name, actions] : storage.rows_ttl_entry.group_by_aggregations)
+                    actions->execute(agg_block);
+                for (const auto & name : storage.rows_ttl_entry.group_by_keys)
+                {
+                    const IColumn * values_column = agg_block.getByName(name).column.get();
+                    auto & result_column = result_columns[header.getPositionByName(name)];
+                    result_column->insertRangeFrom(*values_column, 0, agg_block.rows());
+                }
+                for (const auto & [name, res_column] : storage.rows_ttl_entry.group_by_aggregations_res_column)
+                {
+                    const IColumn * values_column = agg_block.getByName(res_column).column.get();
+                    auto & result_column = result_columns[header.getPositionByName(name)];
+                    result_column->insertRangeFrom(*values_column, 0, agg_block.rows());
+                }
+            }
+
+            block = header.cloneWithColumns(std::move(result_columns));
+            agg_result.invalidate();
+        }
+
         return block;
+    }
 
     if (storage.hasRowsTTL() && (force || isTTLExpired(old_ttl_infos.table_ttl.min)))
         removeRowsWithExpiredTableTTL(block);
@@ -114,35 +175,171 @@ void TTLBlockInputStream::readSuffixImpl()
 void TTLBlockInputStream::removeRowsWithExpiredTableTTL(Block & block)
 {
     storage.rows_ttl_entry.expression->execute(block);
+    if (storage.rows_ttl_entry.where_expression)
+        storage.rows_ttl_entry.where_expression->execute(block);
 
     const IColumn * ttl_column =
         block.getByName(storage.rows_ttl_entry.result_column).column.get();
 
+    const IColumn * where_filter_column = storage.rows_ttl_entry.where_expression ?
+        block.getByName(storage.rows_ttl_entry.where_filter_column).column.get() : nullptr;
+
     const auto & column_names = header.getNames();
-    MutableColumns result_columns;
-    result_columns.reserve(column_names.size());
 
-    for (auto it = column_names.begin(); it != column_names.end(); ++it)
+    if (!aggregator)
     {
-        const IColumn * values_column = block.getByName(*it).column.get();
-        MutableColumnPtr result_column = values_column->cloneEmpty();
-        result_column->reserve(block.rows());
+        MutableColumns result_columns;
+        result_columns.reserve(column_names.size());
+        for (auto it = column_names.begin(); it != column_names.end(); ++it)
+        {
+            const IColumn * values_column = block.getByName(*it).column.get();
+            MutableColumnPtr result_column = values_column->cloneEmpty();
+            result_column->reserve(block.rows());
 
+            for (size_t i = 0; i < block.rows(); ++i)
+            {
+                UInt32 cur_ttl = getTimestampByIndex(ttl_column, i);
+                bool where_filter_passed = !where_filter_column || where_filter_column->getBool(i);
+                if (!isTTLExpired(cur_ttl) || !where_filter_passed)
+                {
+                    new_ttl_infos.table_ttl.update(cur_ttl);
+                    result_column->insertFrom(*values_column, i);
+                }
+                else if (it == column_names.begin())
+                    ++rows_removed;
+            }
+            result_columns.emplace_back(std::move(result_column));
+        }
+        block = header.cloneWithColumns(std::move(result_columns));
+    }
+    else
+    {
+        MutableColumns result_columns;
+        MutableColumns aggregate_columns;
+
+        for (const auto & column_name : column_names)
+        {
+            const IColumn * values_column = block.getByName(column_name).column.get();
+            MutableColumnPtr result_column = values_column->cloneEmpty();
+            result_column->reserve(block.rows());
+            result_columns.emplace_back(std::move(result_column));
+
+            MutableColumnPtr aggregate_column = values_column->cloneEmpty();
+            aggregate_column->reserve(block.rows());
+            aggregate_columns.emplace_back(std::move(aggregate_column));
+        }
+
+        size_t rows_aggregated = 0;
+        size_t current_key_start = 0;
+        size_t rows_with_current_key = 0; 
         for (size_t i = 0; i < block.rows(); ++i)
         {
             UInt32 cur_ttl = getTimestampByIndex(ttl_column, i);
-            if (!isTTLExpired(cur_ttl))
+            bool where_filter_passed = !where_filter_column || where_filter_column->getBool(i);
+            bool ttl_expired = isTTLExpired(cur_ttl) && where_filter_passed;
+
+            bool same_as_current = true;
+            if (current_key_value.empty())
+            {
+                same_as_current = false;
+                current_key_value.resize(storage.rows_ttl_entry.group_by_keys.size());
+            }
+            for (size_t j = 0; j < storage.rows_ttl_entry.group_by_keys.size(); ++j)
+            {
+                const String & key_column = storage.rows_ttl_entry.group_by_keys[j];
+                const IColumn * values_column = block.getByName(key_column).column.get();
+                if (!same_as_current)
+                    values_column->get(i, current_key_value[j]);
+                else
+                {
+                    Field value;
+                    values_column->get(i, value);
+                    if (value != current_key_value[j])
+                    {
+                        current_key_value[j] = value;
+                        same_as_current = false;
+                    }
+                }
+            }
+            if (!same_as_current)
+            {
+                if (rows_with_current_key)
+                {
+                    Columns aggregate_chunk;
+                    aggregate_chunk.reserve(aggregate_columns.size());
+                    for (const auto & name : column_names)
+                    {
+                        const auto & column = aggregate_columns[header.getPositionByName(name)];
+                        ColumnPtr chunk_column = column->cut(current_key_start, rows_with_current_key);
+                        aggregate_chunk.emplace_back(std::move(chunk_column));
+                    }
+                    aggregator->executeOnBlock(aggregate_chunk, rows_with_current_key, agg_result, agg_key_columns,
+                                            agg_aggregate_columns, agg_no_more_keys);
+                }
+                if (!agg_result.empty())
+                {
+                    auto aggregated_res = aggregator->convertToBlocks(agg_result, true, 1);
+                    for (auto & agg_block : aggregated_res)
+                    {
+                        for (const auto & [name, actions] : storage.rows_ttl_entry.group_by_aggregations)
+                            actions->execute(agg_block);
+                        for (const auto & name : storage.rows_ttl_entry.group_by_keys)
+                        {
+                            const IColumn * values_column = agg_block.getByName(name).column.get();
+                            auto & result_column = result_columns[header.getPositionByName(name)];
+                            result_column->insertRangeFrom(*values_column, 0, agg_block.rows());
+                        }
+                        for (const auto & [name, res_column] : storage.rows_ttl_entry.group_by_aggregations_res_column)
+                        {
+                            const IColumn * values_column = agg_block.getByName(res_column).column.get();
+                            auto & result_column = result_columns[header.getPositionByName(name)];
+                            result_column->insertRangeFrom(*values_column, 0, agg_block.rows());
+                        }
+                    }
+                }
+                agg_result.invalidate();
+                current_key_start = rows_aggregated;
+                rows_with_current_key = 0;
+            }
+            if (ttl_expired)
+            {
+                ++rows_with_current_key;
+                ++rows_aggregated;
+                for (const auto & name : column_names)
+                {
+                    const IColumn * values_column = block.getByName(name).column.get();
+                    auto & column = aggregate_columns[header.getPositionByName(name)];
+                    column->insertFrom(*values_column, i);
+                }
+            }
+            else
             {
                 new_ttl_infos.table_ttl.update(cur_ttl);
-                result_column->insertFrom(*values_column, i);
+                for (const auto & name : column_names)
+                {
+                    const IColumn * values_column = block.getByName(name).column.get();
+                    auto & column = result_columns[header.getPositionByName(name)];
+                    column->insertFrom(*values_column, i);
+                }
             }
-            else if (it == column_names.begin())
-                ++rows_removed;
         }
-        result_columns.emplace_back(std::move(result_column));
-    }
 
-    block = header.cloneWithColumns(std::move(result_columns));
+        if (rows_with_current_key)
+        {
+            Columns aggregate_chunk;
+            aggregate_chunk.reserve(aggregate_columns.size());
+            for (const auto & name : column_names)
+            {
+                const auto & column = aggregate_columns[header.getPositionByName(name)];
+                ColumnPtr chunk_column = column->cut(current_key_start, rows_with_current_key);
+                aggregate_chunk.emplace_back(std::move(chunk_column));
+            }
+            aggregator->executeOnBlock(aggregate_chunk, rows_with_current_key, agg_result, agg_key_columns,
+                                       agg_aggregate_columns, agg_no_more_keys);
+        }
+
+        block = header.cloneWithColumns(std::move(result_columns));
+    }
 }
 
 void TTLBlockInputStream::removeValuesWithExpiredColumnTTL(Block & block)
diff --git a/src/DataStreams/TTLBlockInputStream.h b/src/DataStreams/TTLBlockInputStream.h
index 3896e5232f8..821f8dd9284 100644
--- a/src/DataStreams/TTLBlockInputStream.h
+++ b/src/DataStreams/TTLBlockInputStream.h
@@ -3,6 +3,7 @@
 #include <Storages/MergeTree/MergeTreeData.h>
 #include <Storages/MergeTree/IMergeTreeDataPart.h>
 #include <Core/Block.h>
+#include <Interpreters/Aggregator.h>
 
 #include <common/DateLUT.h>
 
@@ -39,6 +40,13 @@ private:
     time_t current_time;
     bool force;
 
+    std::unique_ptr<Aggregator> aggregator;
+    std::vector<Field> current_key_value;
+    AggregatedDataVariants agg_result;
+    ColumnRawPtrs agg_key_columns; 
+    Aggregator::AggregateColumns agg_aggregate_columns;
+    bool agg_no_more_keys;
+
     IMergeTreeDataPart::TTLInfos old_ttl_infos;
     IMergeTreeDataPart::TTLInfos new_ttl_infos;
     NameSet empty_columns;
diff --git a/src/Interpreters/ExpressionAnalyzer.cpp b/src/Interpreters/ExpressionAnalyzer.cpp
index 3341855b8c6..c432b920a9f 100644
--- a/src/Interpreters/ExpressionAnalyzer.cpp
+++ b/src/Interpreters/ExpressionAnalyzer.cpp
@@ -192,10 +192,10 @@ void ExpressionAnalyzer::analyzeAggregation()
 
     if (has_aggregation)
     {
-        getSelectQuery(); /// assertSelect()
+        // getSelectQuery(); /// assertSelect()
 
         /// Find out aggregation keys.
-        if (select_query->groupBy())
+        if (select_query && select_query->groupBy())
         {
             NameSet unique_keys;
             ASTs & group_asts = select_query->groupBy()->children;
@@ -926,7 +926,10 @@ void ExpressionAnalyzer::appendExpression(ExpressionActionsChain & chain, const
 
 ExpressionActionsPtr ExpressionAnalyzer::getActions(bool add_aliases, bool project_result)
 {
-    ExpressionActionsPtr actions = std::make_shared<ExpressionActions>(sourceColumns(), context);
+    NamesAndTypesList columns(sourceColumns());
+    for (const auto & col : aggregated_columns)
+        columns.push_back(col);
+    ExpressionActionsPtr actions = std::make_shared<ExpressionActions>(columns, context);
     NamesWithAliases result_columns;
     Names result_names;
 
diff --git a/src/Interpreters/SyntaxAnalyzer.cpp b/src/Interpreters/SyntaxAnalyzer.cpp
index b3d566dbdc8..ec7f2154dad 100644
--- a/src/Interpreters/SyntaxAnalyzer.cpp
+++ b/src/Interpreters/SyntaxAnalyzer.cpp
@@ -839,7 +839,7 @@ SyntaxAnalyzerResultPtr SyntaxAnalyzer::analyzeSelect(
     return std::make_shared<const SyntaxAnalyzerResult>(result);
 }
 
-SyntaxAnalyzerResultPtr SyntaxAnalyzer::analyze(ASTPtr & query, const NamesAndTypesList & source_columns, ConstStoragePtr storage) const
+SyntaxAnalyzerResultPtr SyntaxAnalyzer::analyze(ASTPtr & query, const NamesAndTypesList & source_columns, ConstStoragePtr storage, bool allow_aggregates) const
 {
     if (query->as<ASTSelectQuery>())
         throw Exception("Not select analyze for select asts.", ErrorCodes::LOGICAL_ERROR);
@@ -855,7 +855,20 @@ SyntaxAnalyzerResultPtr SyntaxAnalyzer::analyze(ASTPtr & query, const NamesAndTy
 
     optimizeIf(query, result.aliases, settings.optimize_if_chain_to_miltiif);
 
-    assertNoAggregates(query, "in wrong place");
+    if (allow_aggregates) 
+    {
+        GetAggregatesVisitor::Data data;
+        GetAggregatesVisitor(data).visit(query);
+
+        /// There can not be other aggregate functions within the aggregate functions.
+        for (const ASTFunction * node : data.aggregates)
+            for (auto & arg : node->arguments->children)
+                assertNoAggregates(arg, "inside another aggregate function");
+        result.aggregates = data.aggregates;
+    }
+    else 
+        assertNoAggregates(query, "in wrong place");
+
     result.collectUsedColumns(query);
     return std::make_shared<const SyntaxAnalyzerResult>(result);
 }
diff --git a/src/Interpreters/SyntaxAnalyzer.h b/src/Interpreters/SyntaxAnalyzer.h
index dda0add38db..abacb25ac4d 100644
--- a/src/Interpreters/SyntaxAnalyzer.h
+++ b/src/Interpreters/SyntaxAnalyzer.h
@@ -86,7 +86,7 @@ public:
     {}
 
     /// Analyze and rewrite not select query
-    SyntaxAnalyzerResultPtr analyze(ASTPtr & query, const NamesAndTypesList & source_columns_, ConstStoragePtr storage = {}) const;
+    SyntaxAnalyzerResultPtr analyze(ASTPtr & query, const NamesAndTypesList & source_columns_, ConstStoragePtr storage = {}, bool allow_aggregations = false) const;
 
     /// Analyze and rewrite select query
     SyntaxAnalyzerResultPtr analyzeSelect(
diff --git a/src/Parsers/ASTTTLElement.cpp b/src/Parsers/ASTTTLElement.cpp
index 7e03a73e36d..fb35266465b 100644
--- a/src/Parsers/ASTTTLElement.cpp
+++ b/src/Parsers/ASTTTLElement.cpp
@@ -7,21 +7,80 @@
 namespace DB
 {
 
+ASTPtr ASTTTLElement::clone() const 
+{
+    auto clone = std::make_shared<ASTTTLElement>(*this);
+    clone->children.clear();
+    clone->positions.clear();
+    
+    for (auto expr : {Expression::TTL, Expression::WHERE})
+        clone->setExpression(expr, getExpression(expr, true));
+
+    for (auto & [name, expr] : clone->group_by_aggregations)
+        expr = expr->clone();
+
+    return clone;
+}
+
 void ASTTTLElement::formatImpl(const FormatSettings & settings, FormatState & state, FormatStateStacked frame) const
 {
-    children.front()->formatImpl(settings, state, frame);
-    if (destination_type == PartDestinationType::DISK)
+    ttl()->formatImpl(settings, state, frame);
+    if (mode == Mode::MOVE && destination_type == PartDestinationType::DISK)
     {
         settings.ostr << " TO DISK " << quoteString(destination_name);
     }
-    else if (destination_type == PartDestinationType::VOLUME)
+    else if (mode == Mode::MOVE && destination_type == PartDestinationType::VOLUME)
     {
         settings.ostr << " TO VOLUME " << quoteString(destination_name);
     }
-    else if (destination_type == PartDestinationType::DELETE)
+    else if (mode == Mode::GROUP_BY)
+    {
+        settings.ostr << " GROUP BY ";
+        for (size_t i = 0; i < group_by_key_columns.size(); ++i)
+        {
+            settings.ostr << group_by_key_columns[i];
+            if (i + 1 != group_by_key_columns.size())
+                settings.ostr << ", ";
+        }
+        settings.ostr << " SET ";
+        for (size_t i =  0; i < group_by_aggregations.size(); ++i)
+        {
+            settings.ostr << group_by_aggregations[i].first << " = ";
+            group_by_aggregations[i].second->formatImpl(settings, state, frame);
+            if (i + 1 != group_by_aggregations.size())
+                settings.ostr << ", ";
+        }
+    }
+    else if (mode == Mode::DELETE)
     {
         /// It would be better to output "DELETE" here but that will break compatibility with earlier versions.
     }
+
+    if (where()) 
+    {
+        settings.ostr << " WHERE ";
+        where()->formatImpl(settings, state, frame);
+    }
+}
+
+void ASTTTLElement::setExpression(Expression expr, ASTPtr && ast) 
+{
+    auto it = positions.find(expr);
+    if (it == positions.end())
+    {
+        positions[expr] = children.size();
+        children.emplace_back(ast);
+    }
+    else
+        children[it->second] = ast;
+}
+
+ASTPtr ASTTTLElement::getExpression(Expression expr, bool clone) const
+{
+    auto it = positions.find(expr);
+    if (it != positions.end())
+        return clone ? children[it->second]->clone() : children[it->second];
+    return {};
 }
 
 }
diff --git a/src/Parsers/ASTTTLElement.h b/src/Parsers/ASTTTLElement.h
index 02f70094e04..1a2d7d3c723 100644
--- a/src/Parsers/ASTTTLElement.h
+++ b/src/Parsers/ASTTTLElement.h
@@ -6,31 +6,53 @@
 
 namespace DB
 {
+
 /** Element of TTL expression.
   */
 class ASTTTLElement : public IAST
 {
 public:
+    enum class Expression : uint8_t
+    {
+        TTL,
+        WHERE
+    };
+
+    enum class Mode : uint8_t
+    {
+        DELETE,
+        MOVE,
+        GROUP_BY
+    };
+
+    Mode mode;
     PartDestinationType destination_type;
     String destination_name;
+    std::vector<String> group_by_key_columns;
+    std::vector<std::pair<String, ASTPtr>> group_by_aggregations;
 
-    ASTTTLElement(PartDestinationType destination_type_, const String & destination_name_)
-        : destination_type(destination_type_)
+    ASTTTLElement(Mode mode_, PartDestinationType destination_type_, const String & destination_name_)
+        : mode(mode_) 
+        , destination_type(destination_type_)
         , destination_name(destination_name_)
     {
     }
 
     String getID(char) const override { return "TTLElement"; }
 
-    ASTPtr clone() const override
-    {
-        auto clone = std::make_shared<ASTTTLElement>(*this);
-        clone->cloneChildren();
-        return clone;
-    }
+    ASTPtr clone() const override;
+
+    const ASTPtr ttl() const { return getExpression(Expression::TTL); }
+    const ASTPtr where() const { return getExpression(Expression::WHERE); }
+
+    void setExpression(Expression expr, ASTPtr && ast);
+    ASTPtr getExpression(Expression expr, bool clone = false) const;
 
 protected:
     void formatImpl(const FormatSettings & settings, FormatState & state, FormatStateStacked frame) const override;
+
+private:
+    std::unordered_map<Expression, size_t> positions;
 };
 
 }
diff --git a/src/Parsers/ExpressionElementParsers.cpp b/src/Parsers/ExpressionElementParsers.cpp
index 70a8b282a72..2734aa67193 100644
--- a/src/Parsers/ExpressionElementParsers.cpp
+++ b/src/Parsers/ExpressionElementParsers.cpp
@@ -1455,23 +1455,50 @@ bool ParserTTLElement::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
     ParserKeyword s_to_disk("TO DISK");
     ParserKeyword s_to_volume("TO VOLUME");
     ParserKeyword s_delete("DELETE");
+    ParserKeyword s_where("WHERE");
+    ParserKeyword s_group_by("GROUP BY");
+    ParserKeyword s_set("SET");
+    ParserToken s_comma(TokenType::Comma);
+    ParserToken s_eq(TokenType::Equals);
+    ParserIdentifier parser_identifier;
     ParserStringLiteral parser_string_literal;
     ParserExpression parser_exp;
+    ParserIdentifierList parser_identifier_list;
 
-    ASTPtr expr_elem;
-    if (!parser_exp.parse(pos, expr_elem, expected))
+
+    ASTPtr ttl_expr;
+    if (!parser_exp.parse(pos, ttl_expr, expected))
         return false;
 
+    ASTPtr where_expr;
+
+    std::vector<String> group_by_key_columns;
+    std::vector<std::pair<String, ASTPtr>> group_by_aggregations;
+
+    ASTTTLElement::Mode mode;
     PartDestinationType destination_type = PartDestinationType::DELETE;
     String destination_name;
     if (s_to_disk.ignore(pos))
+    {
+        mode = ASTTTLElement::Mode::MOVE;
         destination_type = PartDestinationType::DISK;
+    }
     else if (s_to_volume.ignore(pos))
+    {
+        mode = ASTTTLElement::Mode::MOVE;
         destination_type = PartDestinationType::VOLUME;
+    }
+    else if (s_group_by.ignore(pos)) 
+    {
+        mode = ASTTTLElement::Mode::GROUP_BY;
+    }
     else
+    {
         s_delete.ignore(pos);
+        mode = ASTTTLElement::Mode::DELETE;
+    }
 
-    if (destination_type == PartDestinationType::DISK || destination_type == PartDestinationType::VOLUME)
+    if (mode == ASTTTLElement::Mode::MOVE)
     {
         ASTPtr ast_space_name;
         if (!parser_string_literal.parse(pos, ast_space_name, expected))
@@ -1479,10 +1506,57 @@ bool ParserTTLElement::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
 
         destination_name = ast_space_name->as<ASTLiteral &>().value.get<const String &>();
     }
+    else if (mode == ASTTTLElement::Mode::GROUP_BY)
+    {
+        ASTPtr ast_group_by_key_columns;
+        if (!parser_identifier_list.parse(pos, ast_group_by_key_columns, expected))
+            return false;
+        for (const auto identifier : ast_group_by_key_columns->children)
+        {
+            String identifier_str;
+            if (!tryGetIdentifierNameInto(identifier, identifier_str))
+                return false;
+            group_by_key_columns.emplace_back(std::move(identifier_str));
+        }        
 
-    node = std::make_shared<ASTTTLElement>(destination_type, destination_name);
-    node->children.push_back(expr_elem);
+        if (!s_set.ignore(pos))
+            return false;
+        while (true)
+        {
+            if (!group_by_aggregations.empty() && !s_comma.ignore(pos))
+                break;
+            
+            ASTPtr name;
+            ASTPtr value;
+            if (!parser_identifier.parse(pos, name, expected))
+                return false;
+            if (!s_eq.ignore(pos))
+                return false;
+            if (!parser_exp.parse(pos, value, expected))
+                return false;
 
+            String name_str;
+            if (!tryGetIdentifierNameInto(name, name_str))
+                return false;
+            group_by_aggregations.emplace_back(name_str, std::move(value));
+        }
+    }
+
+    if ((mode == ASTTTLElement::Mode::MOVE || mode == ASTTTLElement::Mode::DELETE) && s_where.ignore(pos))
+    {
+        if (!parser_exp.parse(pos, where_expr, expected))
+            return false;
+    }
+
+    auto ttl_element = std::make_shared<ASTTTLElement>(mode, destination_type, destination_name);
+    ttl_element->setExpression(ASTTTLElement::Expression::TTL, std::move(ttl_expr));
+    if (where_expr)
+        ttl_element->setExpression(ASTTTLElement::Expression::WHERE, std::move(where_expr));
+
+    ttl_element->group_by_key_columns = std::move(group_by_key_columns);
+    ttl_element->group_by_aggregations = std::move(group_by_aggregations);
+
+    node = ttl_element;
     return true;
 }
 
diff --git a/src/Parsers/ExpressionListParsers.cpp b/src/Parsers/ExpressionListParsers.cpp
index a967ae19691..062d1dd4483 100644
--- a/src/Parsers/ExpressionListParsers.cpp
+++ b/src/Parsers/ExpressionListParsers.cpp
@@ -742,4 +742,13 @@ bool ParserKeyValuePairsList::parseImpl(Pos & pos, ASTPtr & node, Expected & exp
     return parser.parse(pos, node, expected);
 }
 
+
+bool ParserIdentifierList::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
+{
+    return ParserList(
+        std::make_unique<ParserIdentifier>(),
+        std::make_unique<ParserToken>(TokenType::Comma))
+        .parse(pos, node, expected);
 }
+
+}
\ No newline at end of file
diff --git a/src/Parsers/ExpressionListParsers.h b/src/Parsers/ExpressionListParsers.h
index 0cef29b6d67..ba939d8d160 100644
--- a/src/Parsers/ExpressionListParsers.h
+++ b/src/Parsers/ExpressionListParsers.h
@@ -421,4 +421,13 @@ protected:
     bool parseImpl(Pos & pos, ASTPtr & node, Expected & expected) override;
 };
 
-}
+
+/** A comma-separated list of identifiers, probably empty. */
+class ParserIdentifierList : public IParserBase
+{
+protected:
+    const char * getName() const override { return "list of identifiers"; }
+    bool parseImpl(Pos & pos, ASTPtr & node, Expected & expected) override;
+};
+
+}
\ No newline at end of file
diff --git a/src/Storages/MergeTree/MergeTreeData.cpp b/src/Storages/MergeTree/MergeTreeData.cpp
index 48cf3934820..9eb14ed5475 100644
--- a/src/Storages/MergeTree/MergeTreeData.cpp
+++ b/src/Storages/MergeTree/MergeTreeData.cpp
@@ -615,19 +615,50 @@ void MergeTreeData::setTTLExpressions(const ColumnsDescription & new_columns,
 
     auto new_column_ttls = new_columns.getColumnTTLs();
 
-    auto create_ttl_entry = [this, &new_columns](ASTPtr ttl_ast)
+    auto create_ttl_entry = [this, &new_columns](ASTPtr ttl_expr_ast)
     {
         TTLEntry result;
 
-        auto syntax_result = SyntaxAnalyzer(global_context).analyze(ttl_ast, new_columns.getAllPhysical());
-        result.expression = ExpressionAnalyzer(ttl_ast, syntax_result, global_context).getActions(false);
+        auto ttl_syntax_result = SyntaxAnalyzer(global_context).analyze(ttl_expr_ast, new_columns.getAllPhysical());
+        result.expression = ExpressionAnalyzer(ttl_expr_ast, ttl_syntax_result, global_context).getActions(false);
+        result.result_column = ttl_expr_ast->getColumnName();
+
         result.destination_type = PartDestinationType::DELETE;
-        result.result_column = ttl_ast->getColumnName();
 
         checkTTLExpression(result.expression, result.result_column);
         return result;
     };
 
+    auto create_rows_ttl_entry = [this, &new_columns, &create_ttl_entry](const ASTTTLElement * ttl_element)
+    {
+        auto result = create_ttl_entry(ttl_element->ttl());
+        if (ttl_element->mode == ASTTTLElement::Mode::DELETE || ttl_element->mode ==ASTTTLElement::Mode::MOVE)         
+        {
+            if (ASTPtr where_expr_ast = ttl_element->where()) 
+            {
+                auto where_syntax_result = SyntaxAnalyzer(global_context).analyze(where_expr_ast, new_columns.getAllPhysical());
+                result.where_expression = ExpressionAnalyzer(where_expr_ast, where_syntax_result, global_context).getActions(false);
+                result.where_filter_column = where_expr_ast->getColumnName();
+            }
+        }
+        else if (ttl_element->mode == ASTTTLElement::Mode::GROUP_BY) 
+        {
+            result.group_by_keys = ttl_element->group_by_key_columns;
+            for (auto [name, value] : ttl_element->group_by_aggregations)
+            {
+                auto syntax_result = SyntaxAnalyzer(global_context).analyze(value, new_columns.getAllPhysical(), {}, true);
+                auto expr_analyzer = ExpressionAnalyzer(value, syntax_result, global_context);
+
+                result.group_by_aggregations.emplace_back(name, expr_analyzer.getActions(false));
+                result.group_by_aggregations_res_column.emplace_back(name, value->getColumnName());
+
+                for (const auto descr : expr_analyzer.getAnalyzedData().aggregate_descriptions)
+                    result.aggregate_descriptions.push_back(descr);
+            }
+        }
+        return result;
+    };
+
     if (!new_column_ttls.empty())
     {
         NameSet columns_ttl_forbidden;
@@ -672,7 +703,8 @@ void MergeTreeData::setTTLExpressions(const ColumnsDescription & new_columns,
                     throw Exception("More than one DELETE TTL expression is not allowed", ErrorCodes::BAD_TTL_EXPRESSION);
                 }
 
-                auto new_rows_ttl_entry = create_ttl_entry(ttl_element->children[0]);
+                LOG_DEBUG(log, "ttl_element->size is " << ttl_element->size());
+                auto new_rows_ttl_entry = create_rows_ttl_entry(ttl_element);
                 if (!only_check)
                     update_rows_ttl_entry = new_rows_ttl_entry;
 
@@ -680,7 +712,7 @@ void MergeTreeData::setTTLExpressions(const ColumnsDescription & new_columns,
             }
             else
             {
-                auto new_ttl_entry = create_ttl_entry(ttl_element->children[0]);
+                auto new_ttl_entry = create_rows_ttl_entry(ttl_element);
 
                 new_ttl_entry.entry_ast = ttl_element_ptr;
                 new_ttl_entry.destination_type = ttl_element->destination_type;
diff --git a/src/Storages/MergeTree/MergeTreeData.h b/src/Storages/MergeTree/MergeTreeData.h
index cd1d4912a29..dfcddb40f3c 100644
--- a/src/Storages/MergeTree/MergeTreeData.h
+++ b/src/Storages/MergeTree/MergeTreeData.h
@@ -20,6 +20,7 @@
 #include <Storages/MergeTree/MergeTreePartsMover.h>
 #include <Interpreters/PartLog.h>
 #include <Disks/StoragePolicy.h>
+#include <Interpreters/Aggregator.h>
 
 #include <boost/multi_index_container.hpp>
 #include <boost/multi_index/ordered_index.hpp>
@@ -652,6 +653,14 @@ public:
         ExpressionActionsPtr expression;
         String result_column;
 
+        ExpressionActionsPtr where_expression;
+        String where_filter_column;
+
+        Names group_by_keys;
+        std::vector<std::pair<String, ExpressionActionsPtr>> group_by_aggregations;
+        std::vector<std::pair<String, String>> group_by_aggregations_res_column;
+        AggregateDescriptions aggregate_descriptions;
+
         /// Name and type of a destination are only valid in table-level context.
         PartDestinationType destination_type;
         String destination_name;

From eeaf608aa912543540106ec5bc4363b88d3324f3 Mon Sep 17 00:00:00 2001
From: Nikolai Sorokin <sonikint@gmail.com>
Date: Tue, 12 May 2020 17:27:00 +0300
Subject: [PATCH 052/122] Refactor code; support prefix of PK as GROUP BY key

---
 src/DataStreams/TTLBlockInputStream.cpp  | 175 +++++++++--------------
 src/DataStreams/TTLBlockInputStream.h    |   6 +
 src/Parsers/ASTTTLElement.cpp            |  57 ++++----
 src/Parsers/ASTTTLElement.h              |  38 +++--
 src/Parsers/ExpressionElementParsers.cpp |  33 +++--
 src/Storages/MergeTree/MergeTreeData.cpp |  32 +++--
 src/Storages/MergeTree/MergeTreeData.h   |   8 +-
 src/Storages/MergeTree/TTLMode.h         |  14 ++
 8 files changed, 177 insertions(+), 186 deletions(-)
 create mode 100644 src/Storages/MergeTree/TTLMode.h

diff --git a/src/DataStreams/TTLBlockInputStream.cpp b/src/DataStreams/TTLBlockInputStream.cpp
index d816cba4b2f..d5e46a9929d 100644
--- a/src/DataStreams/TTLBlockInputStream.cpp
+++ b/src/DataStreams/TTLBlockInputStream.cpp
@@ -5,6 +5,7 @@
 #include <Interpreters/ExpressionAnalyzer.h>
 #include <Columns/ColumnConst.h>
 #include <Interpreters/addTypeConversionToAST.h>
+#include <Storages/MergeTree/TTLMode.h>
 
 namespace DB
 {
@@ -68,8 +69,10 @@ TTLBlockInputStream::TTLBlockInputStream(
         defaults_expression = ExpressionAnalyzer{default_expr_list, syntax_result, storage.global_context}.getActions(true);
     }
 
-    if (storage.hasRowsTTL() && !storage.rows_ttl_entry.group_by_keys.empty())
+    if (storage.hasRowsTTL() && storage.rows_ttl_entry.mode == TTLMode::GROUP_BY)
     {
+        current_key_value.resize(storage.rows_ttl_entry.group_by_keys.size());
+
         ColumnNumbers keys;
         for (const auto & key : storage.rows_ttl_entry.group_by_keys)
             keys.push_back(header.getPositionByName(key));
@@ -84,9 +87,11 @@ TTLBlockInputStream::TTLBlockInputStream(
 
         const Settings & settings = storage.global_context.getSettingsRef();
 
+        bool allow_to_use_two_level_group_by = false;  // settings.max_bytes_before_external_group_by != 0;
         Aggregator::Params params(header, keys, aggregates,
             false, settings.max_rows_to_group_by, settings.group_by_overflow_mode,
-            SettingUInt64(0), SettingUInt64(0),
+            allow_to_use_two_level_group_by ? settings.group_by_two_level_threshold : SettingUInt64(0),
+            allow_to_use_two_level_group_by ? settings.group_by_two_level_threshold_bytes : SettingUInt64(0),
             settings.max_bytes_before_external_group_by, settings.empty_result_for_aggregation_by_empty_set,
             storage.global_context.getTemporaryVolume(), settings.max_threads, settings.min_free_disk_space_for_temporary_data);
         aggregator = std::make_unique<Aggregator>(params);
@@ -102,7 +107,7 @@ Block TTLBlockInputStream::readImpl()
 {
     /// Skip all data if table ttl is expired for part
     if (storage.hasRowsTTL() && !storage.rows_ttl_entry.where_expression &&
-        storage.rows_ttl_entry.group_by_keys.empty() && isTTLExpired(old_ttl_infos.table_ttl.max))
+        storage.rows_ttl_entry.mode != TTLMode::GROUP_BY && isTTLExpired(old_ttl_infos.table_ttl.max))
     {
         rows_removed = data_part->rows_count;
         return {};
@@ -113,36 +118,9 @@ Block TTLBlockInputStream::readImpl()
     {
         if (aggregator && !agg_result.empty())
         {
-            MutableColumns result_columns;
-            const auto & column_names = header.getNames();
-            for (const auto & column_name : column_names)
-            {
-                const IColumn * values_column = header.getByName(column_name).column.get();
-                MutableColumnPtr result_column = values_column->cloneEmpty();
-                result_columns.emplace_back(std::move(result_column));
-            }
-
-            auto aggregated_res = aggregator->convertToBlocks(agg_result, true, 1);
-            for (auto & agg_block : aggregated_res)
-            {
-                for (const auto & [name, actions] : storage.rows_ttl_entry.group_by_aggregations)
-                    actions->execute(agg_block);
-                for (const auto & name : storage.rows_ttl_entry.group_by_keys)
-                {
-                    const IColumn * values_column = agg_block.getByName(name).column.get();
-                    auto & result_column = result_columns[header.getPositionByName(name)];
-                    result_column->insertRangeFrom(*values_column, 0, agg_block.rows());
-                }
-                for (const auto & [name, res_column] : storage.rows_ttl_entry.group_by_aggregations_res_column)
-                {
-                    const IColumn * values_column = agg_block.getByName(res_column).column.get();
-                    auto & result_column = result_columns[header.getPositionByName(name)];
-                    result_column->insertRangeFrom(*values_column, 0, agg_block.rows());
-                }
-            }
-
+            MutableColumns result_columns = header.cloneEmptyColumns();
+            finalizeAggregates(result_columns);
             block = header.cloneWithColumns(std::move(result_columns));
-            agg_result.invalidate();
         }
 
         return block;
@@ -181,8 +159,8 @@ void TTLBlockInputStream::removeRowsWithExpiredTableTTL(Block & block)
     const IColumn * ttl_column =
         block.getByName(storage.rows_ttl_entry.result_column).column.get();
 
-    const IColumn * where_filter_column = storage.rows_ttl_entry.where_expression ?
-        block.getByName(storage.rows_ttl_entry.where_filter_column).column.get() : nullptr;
+    const IColumn * where_result_column = storage.rows_ttl_entry.where_expression ?
+        block.getByName(storage.rows_ttl_entry.where_result_column).column.get() : nullptr;
 
     const auto & column_names = header.getNames();
 
@@ -199,7 +177,7 @@ void TTLBlockInputStream::removeRowsWithExpiredTableTTL(Block & block)
             for (size_t i = 0; i < block.rows(); ++i)
             {
                 UInt32 cur_ttl = getTimestampByIndex(ttl_column, i);
-                bool where_filter_passed = !where_filter_column || where_filter_column->getBool(i);
+                bool where_filter_passed = !where_result_column || where_result_column->getBool(i);
                 if (!isTTLExpired(cur_ttl) || !where_filter_passed)
                 {
                     new_ttl_infos.table_ttl.update(cur_ttl);
@@ -214,20 +192,8 @@ void TTLBlockInputStream::removeRowsWithExpiredTableTTL(Block & block)
     }
     else
     {
-        MutableColumns result_columns;
-        MutableColumns aggregate_columns;
-
-        for (const auto & column_name : column_names)
-        {
-            const IColumn * values_column = block.getByName(column_name).column.get();
-            MutableColumnPtr result_column = values_column->cloneEmpty();
-            result_column->reserve(block.rows());
-            result_columns.emplace_back(std::move(result_column));
-
-            MutableColumnPtr aggregate_column = values_column->cloneEmpty();
-            aggregate_column->reserve(block.rows());
-            aggregate_columns.emplace_back(std::move(aggregate_column));
-        }
+        MutableColumns result_columns = header.cloneEmptyColumns();
+        MutableColumns aggregate_columns = header.cloneEmptyColumns();
 
         size_t rows_aggregated = 0;
         size_t current_key_start = 0;
@@ -235,72 +201,30 @@ void TTLBlockInputStream::removeRowsWithExpiredTableTTL(Block & block)
         for (size_t i = 0; i < block.rows(); ++i)
         {
             UInt32 cur_ttl = getTimestampByIndex(ttl_column, i);
-            bool where_filter_passed = !where_filter_column || where_filter_column->getBool(i);
+            bool where_filter_passed = !where_result_column || where_result_column->getBool(i);
             bool ttl_expired = isTTLExpired(cur_ttl) && where_filter_passed;
 
             bool same_as_current = true;
-            if (current_key_value.empty())
-            {
-                same_as_current = false;
-                current_key_value.resize(storage.rows_ttl_entry.group_by_keys.size());
-            }
             for (size_t j = 0; j < storage.rows_ttl_entry.group_by_keys.size(); ++j)
             {
                 const String & key_column = storage.rows_ttl_entry.group_by_keys[j];
                 const IColumn * values_column = block.getByName(key_column).column.get();
-                if (!same_as_current)
-                    values_column->get(i, current_key_value[j]);
-                else
+                if (!same_as_current || (*values_column)[i] != current_key_value[j])
                 {
-                    Field value;
-                    values_column->get(i, value);
-                    if (value != current_key_value[j])
-                    {
-                        current_key_value[j] = value;
-                        same_as_current = false;
-                    }
+                    values_column->get(i, current_key_value[j]);
+                    same_as_current = false;
                 }
             }
             if (!same_as_current)
             {
                 if (rows_with_current_key)
-                {
-                    Columns aggregate_chunk;
-                    aggregate_chunk.reserve(aggregate_columns.size());
-                    for (const auto & name : column_names)
-                    {
-                        const auto & column = aggregate_columns[header.getPositionByName(name)];
-                        ColumnPtr chunk_column = column->cut(current_key_start, rows_with_current_key);
-                        aggregate_chunk.emplace_back(std::move(chunk_column));
-                    }
-                    aggregator->executeOnBlock(aggregate_chunk, rows_with_current_key, agg_result, agg_key_columns,
-                                            agg_aggregate_columns, agg_no_more_keys);
-                }
-                if (!agg_result.empty())
-                {
-                    auto aggregated_res = aggregator->convertToBlocks(agg_result, true, 1);
-                    for (auto & agg_block : aggregated_res)
-                    {
-                        for (const auto & [name, actions] : storage.rows_ttl_entry.group_by_aggregations)
-                            actions->execute(agg_block);
-                        for (const auto & name : storage.rows_ttl_entry.group_by_keys)
-                        {
-                            const IColumn * values_column = agg_block.getByName(name).column.get();
-                            auto & result_column = result_columns[header.getPositionByName(name)];
-                            result_column->insertRangeFrom(*values_column, 0, agg_block.rows());
-                        }
-                        for (const auto & [name, res_column] : storage.rows_ttl_entry.group_by_aggregations_res_column)
-                        {
-                            const IColumn * values_column = agg_block.getByName(res_column).column.get();
-                            auto & result_column = result_columns[header.getPositionByName(name)];
-                            result_column->insertRangeFrom(*values_column, 0, agg_block.rows());
-                        }
-                    }
-                }
-                agg_result.invalidate();
+                    calculateAggregates(aggregate_columns, current_key_start, rows_with_current_key);
+                finalizeAggregates(result_columns);
+
                 current_key_start = rows_aggregated;
                 rows_with_current_key = 0;
             }
+
             if (ttl_expired)
             {
                 ++rows_with_current_key;
@@ -325,23 +249,52 @@ void TTLBlockInputStream::removeRowsWithExpiredTableTTL(Block & block)
         }
 
         if (rows_with_current_key)
-        {
-            Columns aggregate_chunk;
-            aggregate_chunk.reserve(aggregate_columns.size());
-            for (const auto & name : column_names)
-            {
-                const auto & column = aggregate_columns[header.getPositionByName(name)];
-                ColumnPtr chunk_column = column->cut(current_key_start, rows_with_current_key);
-                aggregate_chunk.emplace_back(std::move(chunk_column));
-            }
-            aggregator->executeOnBlock(aggregate_chunk, rows_with_current_key, agg_result, agg_key_columns,
-                                       agg_aggregate_columns, agg_no_more_keys);
-        }
+            calculateAggregates(aggregate_columns, current_key_start, rows_with_current_key);
 
         block = header.cloneWithColumns(std::move(result_columns));
     }
 }
 
+void TTLBlockInputStream::calculateAggregates(const MutableColumns & aggregate_columns, size_t start_pos, size_t length)
+{
+    Columns aggregate_chunk;
+    aggregate_chunk.reserve(aggregate_columns.size());
+    for (const auto & name : header.getNames())
+    {
+        const auto & column = aggregate_columns[header.getPositionByName(name)];
+        ColumnPtr chunk_column = column->cut(start_pos, length);
+        aggregate_chunk.emplace_back(std::move(chunk_column));
+    }
+    aggregator->executeOnBlock(aggregate_chunk, length, agg_result, agg_key_columns,
+                               agg_aggregate_columns, agg_no_more_keys);
+}
+
+void TTLBlockInputStream::finalizeAggregates(MutableColumns & result_columns)
+{
+    if (!agg_result.empty())
+    {
+        auto aggregated_res = aggregator->convertToBlocks(agg_result, true, 1);
+        for (auto & agg_block : aggregated_res)
+        {
+            for (const auto & it : storage.rows_ttl_entry.group_by_aggregations)
+                std::get<2>(it)->execute(agg_block);
+            for (const auto & name : storage.rows_ttl_entry.group_by_keys)
+            {
+                const IColumn * values_column = agg_block.getByName(name).column.get();
+                auto & result_column = result_columns[header.getPositionByName(name)];
+                result_column->insertRangeFrom(*values_column, 0, agg_block.rows());
+            }
+            for (const auto & it : storage.rows_ttl_entry.group_by_aggregations)
+            {
+                const IColumn * values_column = agg_block.getByName(get<1>(it)).column.get();
+                auto & result_column = result_columns[header.getPositionByName(std::get<0>(it))];
+                result_column->insertRangeFrom(*values_column, 0, agg_block.rows());
+            }
+        }
+    }
+    agg_result.invalidate();
+}
+
 void TTLBlockInputStream::removeValuesWithExpiredColumnTTL(Block & block)
 {
     Block block_with_defaults;
diff --git a/src/DataStreams/TTLBlockInputStream.h b/src/DataStreams/TTLBlockInputStream.h
index 821f8dd9284..c5ec6596038 100644
--- a/src/DataStreams/TTLBlockInputStream.h
+++ b/src/DataStreams/TTLBlockInputStream.h
@@ -66,6 +66,12 @@ private:
 
     /// Removes rows with expired table ttl and computes new ttl_infos for part
     void removeRowsWithExpiredTableTTL(Block & block);
+    
+    // Calculate aggregates of aggregate_columns into agg_result
+    void calculateAggregates(const MutableColumns & aggregate_columns, size_t start_pos, size_t length);
+
+    /// Finalize agg_result into result_columns
+    void finalizeAggregates(MutableColumns & result_columns);
 
     /// Updates TTL for moves
     void updateMovesTTL(Block & block);
diff --git a/src/Parsers/ASTTTLElement.cpp b/src/Parsers/ASTTTLElement.cpp
index fb35266465b..35e9dc89a0d 100644
--- a/src/Parsers/ASTTTLElement.cpp
+++ b/src/Parsers/ASTTTLElement.cpp
@@ -11,10 +11,11 @@ ASTPtr ASTTTLElement::clone() const
 {
     auto clone = std::make_shared<ASTTTLElement>(*this);
     clone->children.clear();
-    clone->positions.clear();
+    clone->ttl_expr_pos = -1;
+    clone->where_expr_pos = -1;
     
-    for (auto expr : {Expression::TTL, Expression::WHERE})
-        clone->setExpression(expr, getExpression(expr, true));
+    clone->setExpression(clone->ttl_expr_pos, getExpression(ttl_expr_pos, true));
+    clone->setExpression(clone->where_expr_pos, getExpression(where_expr_pos, true));
 
     for (auto & [name, expr] : clone->group_by_aggregations)
         expr = expr->clone();
@@ -25,33 +26,33 @@ ASTPtr ASTTTLElement::clone() const
 void ASTTTLElement::formatImpl(const FormatSettings & settings, FormatState & state, FormatStateStacked frame) const
 {
     ttl()->formatImpl(settings, state, frame);
-    if (mode == Mode::MOVE && destination_type == PartDestinationType::DISK)
+    if (mode == TTLMode::MOVE && destination_type == PartDestinationType::DISK)
     {
         settings.ostr << " TO DISK " << quoteString(destination_name);
     }
-    else if (mode == Mode::MOVE && destination_type == PartDestinationType::VOLUME)
+    else if (mode == TTLMode::MOVE && destination_type == PartDestinationType::VOLUME)
     {
         settings.ostr << " TO VOLUME " << quoteString(destination_name);
     }
-    else if (mode == Mode::GROUP_BY)
+    else if (mode == TTLMode::GROUP_BY)
     {
         settings.ostr << " GROUP BY ";
-        for (size_t i = 0; i < group_by_key_columns.size(); ++i)
+        for (auto it = group_by_key_columns.begin(); it != group_by_key_columns.end(); ++it)
         {
-            settings.ostr << group_by_key_columns[i];
-            if (i + 1 != group_by_key_columns.size())
+            if (it != group_by_key_columns.begin())
                 settings.ostr << ", ";
+            settings.ostr << *it;
         }
         settings.ostr << " SET ";
-        for (size_t i =  0; i < group_by_aggregations.size(); ++i)
+        for (auto it = group_by_aggregations.begin(); it != group_by_aggregations.end(); ++it)
         {
-            settings.ostr << group_by_aggregations[i].first << " = ";
-            group_by_aggregations[i].second->formatImpl(settings, state, frame);
-            if (i + 1 != group_by_aggregations.size())
+            if (it != group_by_aggregations.begin())
                 settings.ostr << ", ";
+            settings.ostr << it->first << " = ";
+            it->second->formatImpl(settings, state, frame);
         }
     }
-    else if (mode == Mode::DELETE)
+    else if (mode == TTLMode::DELETE)
     {
         /// It would be better to output "DELETE" here but that will break compatibility with earlier versions.
     }
@@ -63,24 +64,28 @@ void ASTTTLElement::formatImpl(const FormatSettings & settings, FormatState & st
     }
 }
 
-void ASTTTLElement::setExpression(Expression expr, ASTPtr && ast) 
+void ASTTTLElement::setExpression(int & pos, ASTPtr && ast) 
 {
-    auto it = positions.find(expr);
-    if (it == positions.end())
+    if (ast)
     {
-        positions[expr] = children.size();
-        children.emplace_back(ast);
+        if (pos == -1)
+        {
+            pos = children.size();
+            children.emplace_back(ast);
+        }
+        else
+            children[pos] = ast;
+    }
+    else if (pos != -1)
+    {
+        children[pos] = ASTPtr{};
+        pos = -1;
     }
-    else
-        children[it->second] = ast;
 }
 
-ASTPtr ASTTTLElement::getExpression(Expression expr, bool clone) const
+ASTPtr ASTTTLElement::getExpression(int  pos, bool clone) const 
 {
-    auto it = positions.find(expr);
-    if (it != positions.end())
-        return clone ? children[it->second]->clone() : children[it->second];
-    return {};
+    return pos != -1 ? (clone ? children[pos]->clone() : children[pos]) : ASTPtr{};
 }
 
 }
diff --git a/src/Parsers/ASTTTLElement.h b/src/Parsers/ASTTTLElement.h
index 1a2d7d3c723..24b02860f32 100644
--- a/src/Parsers/ASTTTLElement.h
+++ b/src/Parsers/ASTTTLElement.h
@@ -2,6 +2,7 @@
 
 #include <Parsers/IAST.h>
 #include <Storages/MergeTree/PartDestinationType.h>
+#include <Storages/MergeTree/TTLMode.h>
 
 
 namespace DB
@@ -12,29 +13,19 @@ namespace DB
 class ASTTTLElement : public IAST
 {
 public:
-    enum class Expression : uint8_t
-    {
-        TTL,
-        WHERE
-    };
-
-    enum class Mode : uint8_t
-    {
-        DELETE,
-        MOVE,
-        GROUP_BY
-    };
-
-    Mode mode;
+    TTLMode mode;
     PartDestinationType destination_type;
     String destination_name;
-    std::vector<String> group_by_key_columns;
+
+    Strings group_by_key_columns;
     std::vector<std::pair<String, ASTPtr>> group_by_aggregations;
 
-    ASTTTLElement(Mode mode_, PartDestinationType destination_type_, const String & destination_name_)
+    ASTTTLElement(TTLMode mode_, PartDestinationType destination_type_, const String & destination_name_)
         : mode(mode_) 
         , destination_type(destination_type_)
         , destination_name(destination_name_)
+        , ttl_expr_pos(-1)
+        , where_expr_pos(-1)
     {
     }
 
@@ -42,17 +33,22 @@ public:
 
     ASTPtr clone() const override;
 
-    const ASTPtr ttl() const { return getExpression(Expression::TTL); }
-    const ASTPtr where() const { return getExpression(Expression::WHERE); }
+    const ASTPtr ttl() const { return getExpression(ttl_expr_pos); }
+    const ASTPtr where() const { return getExpression(where_expr_pos); }
 
-    void setExpression(Expression expr, ASTPtr && ast);
-    ASTPtr getExpression(Expression expr, bool clone = false) const;
+    void setTTL(ASTPtr && ast) { setExpression(ttl_expr_pos, std::forward<ASTPtr>(ast)); }
+    void setWhere(ASTPtr && ast) { setExpression(where_expr_pos, std::forward<ASTPtr>(ast)); }
 
 protected:
     void formatImpl(const FormatSettings & settings, FormatState & state, FormatStateStacked frame) const override;
 
+private: 
+    int ttl_expr_pos;
+    int where_expr_pos;
+
 private:
-    std::unordered_map<Expression, size_t> positions;
+    void setExpression(int & pos, ASTPtr && ast);    
+    ASTPtr getExpression(int pos, bool clone = false) const;
 };
 
 }
diff --git a/src/Parsers/ExpressionElementParsers.cpp b/src/Parsers/ExpressionElementParsers.cpp
index 2734aa67193..afd3814d913 100644
--- a/src/Parsers/ExpressionElementParsers.cpp
+++ b/src/Parsers/ExpressionElementParsers.cpp
@@ -1460,45 +1460,45 @@ bool ParserTTLElement::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
     ParserKeyword s_set("SET");
     ParserToken s_comma(TokenType::Comma);
     ParserToken s_eq(TokenType::Equals);
+
     ParserIdentifier parser_identifier;
     ParserStringLiteral parser_string_literal;
     ParserExpression parser_exp;
     ParserIdentifierList parser_identifier_list;
 
-
     ASTPtr ttl_expr;
     if (!parser_exp.parse(pos, ttl_expr, expected))
         return false;
 
-    ASTPtr where_expr;
-
-    std::vector<String> group_by_key_columns;
-    std::vector<std::pair<String, ASTPtr>> group_by_aggregations;
-
-    ASTTTLElement::Mode mode;
+    TTLMode mode;
     PartDestinationType destination_type = PartDestinationType::DELETE;
     String destination_name;
+
     if (s_to_disk.ignore(pos))
     {
-        mode = ASTTTLElement::Mode::MOVE;
+        mode = TTLMode::MOVE;
         destination_type = PartDestinationType::DISK;
     }
     else if (s_to_volume.ignore(pos))
     {
-        mode = ASTTTLElement::Mode::MOVE;
+        mode = TTLMode::MOVE;
         destination_type = PartDestinationType::VOLUME;
     }
     else if (s_group_by.ignore(pos)) 
     {
-        mode = ASTTTLElement::Mode::GROUP_BY;
+        mode = TTLMode::GROUP_BY;
     }
     else
     {
         s_delete.ignore(pos);
-        mode = ASTTTLElement::Mode::DELETE;
+        mode = TTLMode::DELETE;
     }
 
-    if (mode == ASTTTLElement::Mode::MOVE)
+    ASTPtr where_expr;
+    std::vector<String> group_by_key_columns;
+    std::vector<std::pair<String, ASTPtr>> group_by_aggregations;
+
+    if (mode == TTLMode::MOVE)
     {
         ASTPtr ast_space_name;
         if (!parser_string_literal.parse(pos, ast_space_name, expected))
@@ -1506,7 +1506,7 @@ bool ParserTTLElement::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
 
         destination_name = ast_space_name->as<ASTLiteral &>().value.get<const String &>();
     }
-    else if (mode == ASTTTLElement::Mode::GROUP_BY)
+    else if (mode == TTLMode::GROUP_BY)
     {
         ASTPtr ast_group_by_key_columns;
         if (!parser_identifier_list.parse(pos, ast_group_by_key_columns, expected))
@@ -1541,17 +1541,16 @@ bool ParserTTLElement::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
             group_by_aggregations.emplace_back(name_str, std::move(value));
         }
     }
-
-    if ((mode == ASTTTLElement::Mode::MOVE || mode == ASTTTLElement::Mode::DELETE) && s_where.ignore(pos))
+    else if (mode == TTLMode::DELETE && s_where.ignore(pos))
     {
         if (!parser_exp.parse(pos, where_expr, expected))
             return false;
     }
 
     auto ttl_element = std::make_shared<ASTTTLElement>(mode, destination_type, destination_name);
-    ttl_element->setExpression(ASTTTLElement::Expression::TTL, std::move(ttl_expr));
+    ttl_element->setTTL(std::move(ttl_expr));
     if (where_expr)
-        ttl_element->setExpression(ASTTTLElement::Expression::WHERE, std::move(where_expr));
+        ttl_element->setWhere(std::move(where_expr));
 
     ttl_element->group_by_key_columns = std::move(group_by_key_columns);
     ttl_element->group_by_aggregations = std::move(group_by_aggregations);
diff --git a/src/Storages/MergeTree/MergeTreeData.cpp b/src/Storages/MergeTree/MergeTreeData.cpp
index 9eb14ed5475..1f55f5ecff5 100644
--- a/src/Storages/MergeTree/MergeTreeData.cpp
+++ b/src/Storages/MergeTree/MergeTreeData.cpp
@@ -624,6 +624,7 @@ void MergeTreeData::setTTLExpressions(const ColumnsDescription & new_columns,
         result.result_column = ttl_expr_ast->getColumnName();
 
         result.destination_type = PartDestinationType::DELETE;
+        result.mode = TTLMode::DELETE;
 
         checkTTLExpression(result.expression, result.result_column);
         return result;
@@ -632,27 +633,43 @@ void MergeTreeData::setTTLExpressions(const ColumnsDescription & new_columns,
     auto create_rows_ttl_entry = [this, &new_columns, &create_ttl_entry](const ASTTTLElement * ttl_element)
     {
         auto result = create_ttl_entry(ttl_element->ttl());
-        if (ttl_element->mode == ASTTTLElement::Mode::DELETE || ttl_element->mode ==ASTTTLElement::Mode::MOVE)         
+        result.mode = ttl_element->mode;
+
+        if (ttl_element->mode == TTLMode::DELETE)         
         {
             if (ASTPtr where_expr_ast = ttl_element->where()) 
             {
                 auto where_syntax_result = SyntaxAnalyzer(global_context).analyze(where_expr_ast, new_columns.getAllPhysical());
                 result.where_expression = ExpressionAnalyzer(where_expr_ast, where_syntax_result, global_context).getActions(false);
-                result.where_filter_column = where_expr_ast->getColumnName();
+                result.where_result_column = where_expr_ast->getColumnName();
             }
         }
-        else if (ttl_element->mode == ASTTTLElement::Mode::GROUP_BY) 
+        else if (ttl_element->mode == TTLMode::GROUP_BY) 
         {
+            if (ttl_element->group_by_key_columns.size() > this->primary_key_columns.size())
+                throw Exception("TTL Expression GROUP BY key should be a prefix of primary key", ErrorCodes::BAD_TTL_EXPRESSION);
+            for (size_t i = 0; i < ttl_element->group_by_key_columns.size(); ++i)
+            {
+                if (ttl_element->group_by_key_columns[i] != this->primary_key_columns[i])
+                    throw Exception("TTL Expression GROUP BY key should be a prefix of primary key", ErrorCodes::BAD_TTL_EXPRESSION);
+            }
+
             result.group_by_keys = ttl_element->group_by_key_columns;
-            for (auto [name, value] : ttl_element->group_by_aggregations)
+
+            auto aggregations = ttl_element->group_by_aggregations;
+            for (size_t i = ttl_element->group_by_key_columns.size(); i < this->primary_key_columns.size(); ++i)
+            {
+                ASTPtr expr = makeASTFunction("max", std::make_shared<ASTIdentifier>(this->primary_key_columns[i]));
+                aggregations.emplace_back(this->primary_key_columns[i], std::move(expr));
+            }
+            for (auto [name, value] : aggregations)
             {
                 auto syntax_result = SyntaxAnalyzer(global_context).analyze(value, new_columns.getAllPhysical(), {}, true);
                 auto expr_analyzer = ExpressionAnalyzer(value, syntax_result, global_context);
 
-                result.group_by_aggregations.emplace_back(name, expr_analyzer.getActions(false));
-                result.group_by_aggregations_res_column.emplace_back(name, value->getColumnName());
+                result.group_by_aggregations.emplace_back(name, value->getColumnName(), expr_analyzer.getActions(false));
 
-                for (const auto descr : expr_analyzer.getAnalyzedData().aggregate_descriptions)
+                for (const auto & descr : expr_analyzer.getAnalyzedData().aggregate_descriptions)
                     result.aggregate_descriptions.push_back(descr);
             }
         }
@@ -703,7 +720,6 @@ void MergeTreeData::setTTLExpressions(const ColumnsDescription & new_columns,
                     throw Exception("More than one DELETE TTL expression is not allowed", ErrorCodes::BAD_TTL_EXPRESSION);
                 }
 
-                LOG_DEBUG(log, "ttl_element->size is " << ttl_element->size());
                 auto new_rows_ttl_entry = create_rows_ttl_entry(ttl_element);
                 if (!only_check)
                     update_rows_ttl_entry = new_rows_ttl_entry;
diff --git a/src/Storages/MergeTree/MergeTreeData.h b/src/Storages/MergeTree/MergeTreeData.h
index dfcddb40f3c..5272ee25314 100644
--- a/src/Storages/MergeTree/MergeTreeData.h
+++ b/src/Storages/MergeTree/MergeTreeData.h
@@ -21,6 +21,7 @@
 #include <Interpreters/PartLog.h>
 #include <Disks/StoragePolicy.h>
 #include <Interpreters/Aggregator.h>
+#include <Storages/MergeTree/TTLMode.h>
 
 #include <boost/multi_index_container.hpp>
 #include <boost/multi_index/ordered_index.hpp>
@@ -650,15 +651,16 @@ public:
 
     struct TTLEntry
     {
+        TTLMode mode;
+
         ExpressionActionsPtr expression;
         String result_column;
 
         ExpressionActionsPtr where_expression;
-        String where_filter_column;
+        String where_result_column;
 
         Names group_by_keys;
-        std::vector<std::pair<String, ExpressionActionsPtr>> group_by_aggregations;
-        std::vector<std::pair<String, String>> group_by_aggregations_res_column;
+        std::vector<std::tuple<String, String, ExpressionActionsPtr>> group_by_aggregations;
         AggregateDescriptions aggregate_descriptions;
 
         /// Name and type of a destination are only valid in table-level context.
diff --git a/src/Storages/MergeTree/TTLMode.h b/src/Storages/MergeTree/TTLMode.h
new file mode 100644
index 00000000000..99a57f46853
--- /dev/null
+++ b/src/Storages/MergeTree/TTLMode.h
@@ -0,0 +1,14 @@
+#pragma once
+
+
+namespace DB
+{
+
+enum class TTLMode 
+{
+    DELETE,
+    MOVE,
+    GROUP_BY
+};
+
+}

From 66496dc7e350971da8774f66517a4c3a1b62dbbe Mon Sep 17 00:00:00 2001
From: Nikolai Sorokin <sonikint@gmail.com>
Date: Tue, 12 May 2020 23:44:48 +0300
Subject: [PATCH 053/122] Fix trailing spaces

---
 src/DataStreams/TTLBlockInputStream.cpp  |  2 +-
 src/DataStreams/TTLBlockInputStream.h    |  4 ++--
 src/Interpreters/SyntaxAnalyzer.cpp      |  4 ++--
 src/Parsers/ASTTTLElement.cpp            | 10 +++++-----
 src/Parsers/ASTTTLElement.h              |  6 +++---
 src/Parsers/ExpressionElementParsers.cpp |  6 +++---
 src/Storages/MergeTree/MergeTreeData.cpp |  6 +++---
 src/Storages/MergeTree/TTLMode.h         |  2 +-
 8 files changed, 20 insertions(+), 20 deletions(-)

diff --git a/src/DataStreams/TTLBlockInputStream.cpp b/src/DataStreams/TTLBlockInputStream.cpp
index d5e46a9929d..106ae1fd5a0 100644
--- a/src/DataStreams/TTLBlockInputStream.cpp
+++ b/src/DataStreams/TTLBlockInputStream.cpp
@@ -197,7 +197,7 @@ void TTLBlockInputStream::removeRowsWithExpiredTableTTL(Block & block)
 
         size_t rows_aggregated = 0;
         size_t current_key_start = 0;
-        size_t rows_with_current_key = 0; 
+        size_t rows_with_current_key = 0;
         for (size_t i = 0; i < block.rows(); ++i)
         {
             UInt32 cur_ttl = getTimestampByIndex(ttl_column, i);
diff --git a/src/DataStreams/TTLBlockInputStream.h b/src/DataStreams/TTLBlockInputStream.h
index c5ec6596038..a8f86477b88 100644
--- a/src/DataStreams/TTLBlockInputStream.h
+++ b/src/DataStreams/TTLBlockInputStream.h
@@ -43,7 +43,7 @@ private:
     std::unique_ptr<Aggregator> aggregator;
     std::vector<Field> current_key_value;
     AggregatedDataVariants agg_result;
-    ColumnRawPtrs agg_key_columns; 
+    ColumnRawPtrs agg_key_columns;
     Aggregator::AggregateColumns agg_aggregate_columns;
     bool agg_no_more_keys;
 
@@ -66,7 +66,7 @@ private:
 
     /// Removes rows with expired table ttl and computes new ttl_infos for part
     void removeRowsWithExpiredTableTTL(Block & block);
-    
+
     // Calculate aggregates of aggregate_columns into agg_result
     void calculateAggregates(const MutableColumns & aggregate_columns, size_t start_pos, size_t length);
 
diff --git a/src/Interpreters/SyntaxAnalyzer.cpp b/src/Interpreters/SyntaxAnalyzer.cpp
index ec7f2154dad..54f369c9495 100644
--- a/src/Interpreters/SyntaxAnalyzer.cpp
+++ b/src/Interpreters/SyntaxAnalyzer.cpp
@@ -855,7 +855,7 @@ SyntaxAnalyzerResultPtr SyntaxAnalyzer::analyze(ASTPtr & query, const NamesAndTy
 
     optimizeIf(query, result.aliases, settings.optimize_if_chain_to_miltiif);
 
-    if (allow_aggregates) 
+    if (allow_aggregates)
     {
         GetAggregatesVisitor::Data data;
         GetAggregatesVisitor(data).visit(query);
@@ -866,7 +866,7 @@ SyntaxAnalyzerResultPtr SyntaxAnalyzer::analyze(ASTPtr & query, const NamesAndTy
                 assertNoAggregates(arg, "inside another aggregate function");
         result.aggregates = data.aggregates;
     }
-    else 
+    else
         assertNoAggregates(query, "in wrong place");
 
     result.collectUsedColumns(query);
diff --git a/src/Parsers/ASTTTLElement.cpp b/src/Parsers/ASTTTLElement.cpp
index 35e9dc89a0d..991274e8350 100644
--- a/src/Parsers/ASTTTLElement.cpp
+++ b/src/Parsers/ASTTTLElement.cpp
@@ -7,13 +7,13 @@
 namespace DB
 {
 
-ASTPtr ASTTTLElement::clone() const 
+ASTPtr ASTTTLElement::clone() const
 {
     auto clone = std::make_shared<ASTTTLElement>(*this);
     clone->children.clear();
     clone->ttl_expr_pos = -1;
     clone->where_expr_pos = -1;
-    
+
     clone->setExpression(clone->ttl_expr_pos, getExpression(ttl_expr_pos, true));
     clone->setExpression(clone->where_expr_pos, getExpression(where_expr_pos, true));
 
@@ -57,14 +57,14 @@ void ASTTTLElement::formatImpl(const FormatSettings & settings, FormatState & st
         /// It would be better to output "DELETE" here but that will break compatibility with earlier versions.
     }
 
-    if (where()) 
+    if (where())
     {
         settings.ostr << " WHERE ";
         where()->formatImpl(settings, state, frame);
     }
 }
 
-void ASTTTLElement::setExpression(int & pos, ASTPtr && ast) 
+void ASTTTLElement::setExpression(int & pos, ASTPtr && ast)
 {
     if (ast)
     {
@@ -83,7 +83,7 @@ void ASTTTLElement::setExpression(int & pos, ASTPtr && ast)
     }
 }
 
-ASTPtr ASTTTLElement::getExpression(int  pos, bool clone) const 
+ASTPtr ASTTTLElement::getExpression(int  pos, bool clone) const
 {
     return pos != -1 ? (clone ? children[pos]->clone() : children[pos]) : ASTPtr{};
 }
diff --git a/src/Parsers/ASTTTLElement.h b/src/Parsers/ASTTTLElement.h
index 24b02860f32..9b40adf2be4 100644
--- a/src/Parsers/ASTTTLElement.h
+++ b/src/Parsers/ASTTTLElement.h
@@ -21,7 +21,7 @@ public:
     std::vector<std::pair<String, ASTPtr>> group_by_aggregations;
 
     ASTTTLElement(TTLMode mode_, PartDestinationType destination_type_, const String & destination_name_)
-        : mode(mode_) 
+        : mode(mode_)
         , destination_type(destination_type_)
         , destination_name(destination_name_)
         , ttl_expr_pos(-1)
@@ -42,12 +42,12 @@ public:
 protected:
     void formatImpl(const FormatSettings & settings, FormatState & state, FormatStateStacked frame) const override;
 
-private: 
+private:
     int ttl_expr_pos;
     int where_expr_pos;
 
 private:
-    void setExpression(int & pos, ASTPtr && ast);    
+    void setExpression(int & pos, ASTPtr && ast);
     ASTPtr getExpression(int pos, bool clone = false) const;
 };
 
diff --git a/src/Parsers/ExpressionElementParsers.cpp b/src/Parsers/ExpressionElementParsers.cpp
index afd3814d913..65ff9e1da94 100644
--- a/src/Parsers/ExpressionElementParsers.cpp
+++ b/src/Parsers/ExpressionElementParsers.cpp
@@ -1484,7 +1484,7 @@ bool ParserTTLElement::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
         mode = TTLMode::MOVE;
         destination_type = PartDestinationType::VOLUME;
     }
-    else if (s_group_by.ignore(pos)) 
+    else if (s_group_by.ignore(pos))
     {
         mode = TTLMode::GROUP_BY;
     }
@@ -1517,7 +1517,7 @@ bool ParserTTLElement::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
             if (!tryGetIdentifierNameInto(identifier, identifier_str))
                 return false;
             group_by_key_columns.emplace_back(std::move(identifier_str));
-        }        
+        }
 
         if (!s_set.ignore(pos))
             return false;
@@ -1525,7 +1525,7 @@ bool ParserTTLElement::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
         {
             if (!group_by_aggregations.empty() && !s_comma.ignore(pos))
                 break;
-            
+
             ASTPtr name;
             ASTPtr value;
             if (!parser_identifier.parse(pos, name, expected))
diff --git a/src/Storages/MergeTree/MergeTreeData.cpp b/src/Storages/MergeTree/MergeTreeData.cpp
index 1f55f5ecff5..bb9d8ce186e 100644
--- a/src/Storages/MergeTree/MergeTreeData.cpp
+++ b/src/Storages/MergeTree/MergeTreeData.cpp
@@ -635,16 +635,16 @@ void MergeTreeData::setTTLExpressions(const ColumnsDescription & new_columns,
         auto result = create_ttl_entry(ttl_element->ttl());
         result.mode = ttl_element->mode;
 
-        if (ttl_element->mode == TTLMode::DELETE)         
+        if (ttl_element->mode == TTLMode::DELETE)
         {
-            if (ASTPtr where_expr_ast = ttl_element->where()) 
+            if (ASTPtr where_expr_ast = ttl_element->where())
             {
                 auto where_syntax_result = SyntaxAnalyzer(global_context).analyze(where_expr_ast, new_columns.getAllPhysical());
                 result.where_expression = ExpressionAnalyzer(where_expr_ast, where_syntax_result, global_context).getActions(false);
                 result.where_result_column = where_expr_ast->getColumnName();
             }
         }
-        else if (ttl_element->mode == TTLMode::GROUP_BY) 
+        else if (ttl_element->mode == TTLMode::GROUP_BY)
         {
             if (ttl_element->group_by_key_columns.size() > this->primary_key_columns.size())
                 throw Exception("TTL Expression GROUP BY key should be a prefix of primary key", ErrorCodes::BAD_TTL_EXPRESSION);
diff --git a/src/Storages/MergeTree/TTLMode.h b/src/Storages/MergeTree/TTLMode.h
index 99a57f46853..0681f10fc17 100644
--- a/src/Storages/MergeTree/TTLMode.h
+++ b/src/Storages/MergeTree/TTLMode.h
@@ -4,7 +4,7 @@
 namespace DB
 {
 
-enum class TTLMode 
+enum class TTLMode
 {
     DELETE,
     MOVE,

From a991dcf9f4f7d5552b9406548e414432fd237c41 Mon Sep 17 00:00:00 2001
From: Nikolai Sorokin <sonikint@gmail.com>
Date: Thu, 14 May 2020 13:48:45 +0300
Subject: [PATCH 054/122] Add tests

---
 .../01246_ttl_where_group_by.reference        | 15 +++++++
 .../0_stateless/01246_ttl_where_group_by.sql  | 41 +++++++++++++++++++
 2 files changed, 56 insertions(+)
 create mode 100644 tests/queries/0_stateless/01246_ttl_where_group_by.reference
 create mode 100644 tests/queries/0_stateless/01246_ttl_where_group_by.sql

diff --git a/tests/queries/0_stateless/01246_ttl_where_group_by.reference b/tests/queries/0_stateless/01246_ttl_where_group_by.reference
new file mode 100644
index 00000000000..7e276e2681e
--- /dev/null
+++ b/tests/queries/0_stateless/01246_ttl_where_group_by.reference
@@ -0,0 +1,15 @@
+1	1	0	4
+1	2	3	7
+1	3	0	5
+2	1	20	1
+2	1	0	1
+1	1	0	4
+1	1	10	6
+1	2	3	7
+1	3	0	5
+2	1	10	1
+3	1	0	8
+1	1	0	4
+1	3	10	6
+2	1	20	1
+3	1	0	8
diff --git a/tests/queries/0_stateless/01246_ttl_where_group_by.sql b/tests/queries/0_stateless/01246_ttl_where_group_by.sql
new file mode 100644
index 00000000000..fbbd0336e1b
--- /dev/null
+++ b/tests/queries/0_stateless/01246_ttl_where_group_by.sql
@@ -0,0 +1,41 @@
+drop table if exists ttl_01246_1;
+
+create table ttl_01246_1 (a Int, b Int, x Int, y Int, d DateTime) engine = MergeTree order by (a, b) ttl d + interval 1 second delete where x % 10 == 0 and y > 5;
+insert into ttl_01246_1 values (1, 1, 0, 4, now() + 10);
+insert into ttl_01246_1 values (1, 1, 10, 6, now());
+insert into ttl_01246_1 values (1, 2, 3, 7, now());
+insert into ttl_01246_1 values (1, 3, 0, 5, now());
+insert into ttl_01246_1 values (2, 1, 20, 1, now());
+insert into ttl_01246_1 values (2, 1, 0, 1, now());
+insert into ttl_01246_1 values (3, 1, 0, 8, now());
+select sleep(1.1) format Null;
+optimize table ttl_01246_1 final;
+select a, b, x, y from ttl_01246_1;
+
+drop table if exists ttl_01246_1;
+
+create table ttl_01246_1 (a Int, b Int, x Int32, y Double, d DateTime) engine = MergeTree order by (a, b) ttl d + interval 1 second group by a, b set x = cast(median(x) as Int32), y = avg(y), d = max(d);
+insert into ttl_01246_1 values (1, 1, 0, 4, now() + 10);
+insert into ttl_01246_1 values (1, 1, 10, 6, now());
+insert into ttl_01246_1 values (1, 2, 3, 7, now());
+insert into ttl_01246_1 values (1, 3, 0, 5, now());
+insert into ttl_01246_1 values (2, 1, 20, 1, now());
+insert into ttl_01246_1 values (2, 1, 0, 1, now());
+insert into ttl_01246_1 values (3, 1, 0, 8, now());
+select sleep(1.1) format Null;
+optimize table ttl_01246_1 final;
+select a, b, x, y from ttl_01246_1;
+
+drop table if exists ttl_01246_1;
+
+create table ttl_01246_1 (a Int, b Int, x Int32, y Int, d DateTime) engine = MergeTree order by (a, b) ttl d + interval 1 second group by a set x = max(x), y = cast(round(avg(y)) as Int), d = max(d);
+insert into ttl_01246_1 values (1, 1, 0, 4, now() + 10);
+insert into ttl_01246_1 values (1, 1, 10, 6, now());
+insert into ttl_01246_1 values (1, 2, 3, 7, now());
+insert into ttl_01246_1 values (1, 3, 0, 5, now());
+insert into ttl_01246_1 values (2, 1, 20, 1, now());
+insert into ttl_01246_1 values (2, 1, 0, 1, now());
+insert into ttl_01246_1 values (3, 1, 0, 8, now());
+select sleep(1.1) format Null;
+optimize table ttl_01246_1 final;
+select a, b, x, y from ttl_01246_1;
\ No newline at end of file

From 9269db072af72d5f8c5dbf6483d21b50bdad4ac9 Mon Sep 17 00:00:00 2001
From: Nikolai Sorokin <sonikint@gmail.com>
Date: Thu, 14 May 2020 17:07:36 +0300
Subject: [PATCH 055/122] Better tests

---
 .../0_stateless/01246_ttl_where_group_by.reference   | 11 +++++------
 .../queries/0_stateless/01246_ttl_where_group_by.sql | 12 ++++++++----
 2 files changed, 13 insertions(+), 10 deletions(-)

diff --git a/tests/queries/0_stateless/01246_ttl_where_group_by.reference b/tests/queries/0_stateless/01246_ttl_where_group_by.reference
index 7e276e2681e..69d3d77cd46 100644
--- a/tests/queries/0_stateless/01246_ttl_where_group_by.reference
+++ b/tests/queries/0_stateless/01246_ttl_where_group_by.reference
@@ -4,12 +4,11 @@
 2	1	20	1
 2	1	0	1
 1	1	0	4
-1	1	10	6
-1	2	3	7
-1	3	0	5
-2	1	10	1
+1	1	6	6.5
+1	3	4	5.666666666666667
+2	1	10	4
 3	1	0	8
 1	1	0	4
 1	3	10	6
-2	1	20	1
-3	1	0	8
+2	1	20	2
+3	5	8	4
diff --git a/tests/queries/0_stateless/01246_ttl_where_group_by.sql b/tests/queries/0_stateless/01246_ttl_where_group_by.sql
index fbbd0336e1b..ad969127eb1 100644
--- a/tests/queries/0_stateless/01246_ttl_where_group_by.sql
+++ b/tests/queries/0_stateless/01246_ttl_where_group_by.sql
@@ -17,9 +17,11 @@ drop table if exists ttl_01246_1;
 create table ttl_01246_1 (a Int, b Int, x Int32, y Double, d DateTime) engine = MergeTree order by (a, b) ttl d + interval 1 second group by a, b set x = cast(median(x) as Int32), y = avg(y), d = max(d);
 insert into ttl_01246_1 values (1, 1, 0, 4, now() + 10);
 insert into ttl_01246_1 values (1, 1, 10, 6, now());
-insert into ttl_01246_1 values (1, 2, 3, 7, now());
+insert into ttl_01246_1 values (1, 1, 3, 7, now());
 insert into ttl_01246_1 values (1, 3, 0, 5, now());
-insert into ttl_01246_1 values (2, 1, 20, 1, now());
+insert into ttl_01246_1 values (1, 3, 4, 9, now());
+insert into ttl_01246_1 values (1, 3, 6, 3, now());
+insert into ttl_01246_1 values (2, 1, 20, 7, now());
 insert into ttl_01246_1 values (2, 1, 0, 1, now());
 insert into ttl_01246_1 values (3, 1, 0, 8, now());
 select sleep(1.1) format Null;
@@ -34,8 +36,10 @@ insert into ttl_01246_1 values (1, 1, 10, 6, now());
 insert into ttl_01246_1 values (1, 2, 3, 7, now());
 insert into ttl_01246_1 values (1, 3, 0, 5, now());
 insert into ttl_01246_1 values (2, 1, 20, 1, now());
-insert into ttl_01246_1 values (2, 1, 0, 1, now());
-insert into ttl_01246_1 values (3, 1, 0, 8, now());
+insert into ttl_01246_1 values (2, 1, 0, 3, now());
+insert into ttl_01246_1 values (3, 1, 0, 3, now());
+insert into ttl_01246_1 values (3, 2, 8, 2, now());
+insert into ttl_01246_1 values (3, 5, 5, 8, now());
 select sleep(1.1) format Null;
 optimize table ttl_01246_1 final;
 select a, b, x, y from ttl_01246_1;
\ No newline at end of file

From 1f02ba9692b7152dc82c3ce8f0e6f62ec5aa120f Mon Sep 17 00:00:00 2001
From: Nikolai Sorokin <sonikint@gmail.com>
Date: Thu, 14 May 2020 21:46:15 +0300
Subject: [PATCH 056/122] Add eof

---
 src/Parsers/ExpressionListParsers.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/Parsers/ExpressionListParsers.h b/src/Parsers/ExpressionListParsers.h
index ba939d8d160..9cc36967bd7 100644
--- a/src/Parsers/ExpressionListParsers.h
+++ b/src/Parsers/ExpressionListParsers.h
@@ -430,4 +430,4 @@ protected:
     bool parseImpl(Pos & pos, ASTPtr & node, Expected & expected) override;
 };
 
-}
\ No newline at end of file
+}

From c6541b1dad1fe95644a9dc1c916dbd162be5d37e Mon Sep 17 00:00:00 2001
From: Nikolai Sorokin <sonikint@gmail.com>
Date: Fri, 15 May 2020 02:32:45 +0300
Subject: [PATCH 057/122] Up test id; Fix clang build

---
 src/Interpreters/SyntaxAnalyzer.cpp                           | 4 ++--
 src/Parsers/ExpressionElementParsers.cpp                      | 2 +-
 src/Parsers/ExpressionListParsers.cpp                         | 2 +-
 ..._group_by.reference => 01280_ttl_where_group_by.reference} | 0
 ...46_ttl_where_group_by.sql => 01280_ttl_where_group_by.sql} | 0
 5 files changed, 4 insertions(+), 4 deletions(-)
 rename tests/queries/0_stateless/{01246_ttl_where_group_by.reference => 01280_ttl_where_group_by.reference} (100%)
 rename tests/queries/0_stateless/{01246_ttl_where_group_by.sql => 01280_ttl_where_group_by.sql} (100%)

diff --git a/src/Interpreters/SyntaxAnalyzer.cpp b/src/Interpreters/SyntaxAnalyzer.cpp
index 54f369c9495..b5f86b87fdc 100644
--- a/src/Interpreters/SyntaxAnalyzer.cpp
+++ b/src/Interpreters/SyntaxAnalyzer.cpp
@@ -839,7 +839,7 @@ SyntaxAnalyzerResultPtr SyntaxAnalyzer::analyzeSelect(
     return std::make_shared<const SyntaxAnalyzerResult>(result);
 }
 
-SyntaxAnalyzerResultPtr SyntaxAnalyzer::analyze(ASTPtr & query, const NamesAndTypesList & source_columns, ConstStoragePtr storage, bool allow_aggregates) const
+SyntaxAnalyzerResultPtr SyntaxAnalyzer::analyze(ASTPtr & query, const NamesAndTypesList & source_columns, ConstStoragePtr storage, bool allow_aggregations) const
 {
     if (query->as<ASTSelectQuery>())
         throw Exception("Not select analyze for select asts.", ErrorCodes::LOGICAL_ERROR);
@@ -855,7 +855,7 @@ SyntaxAnalyzerResultPtr SyntaxAnalyzer::analyze(ASTPtr & query, const NamesAndTy
 
     optimizeIf(query, result.aliases, settings.optimize_if_chain_to_miltiif);
 
-    if (allow_aggregates)
+    if (allow_aggregations)
     {
         GetAggregatesVisitor::Data data;
         GetAggregatesVisitor(data).visit(query);
diff --git a/src/Parsers/ExpressionElementParsers.cpp b/src/Parsers/ExpressionElementParsers.cpp
index 65ff9e1da94..5b5d9f05b22 100644
--- a/src/Parsers/ExpressionElementParsers.cpp
+++ b/src/Parsers/ExpressionElementParsers.cpp
@@ -1511,7 +1511,7 @@ bool ParserTTLElement::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
         ASTPtr ast_group_by_key_columns;
         if (!parser_identifier_list.parse(pos, ast_group_by_key_columns, expected))
             return false;
-        for (const auto identifier : ast_group_by_key_columns->children)
+        for (const auto & identifier : ast_group_by_key_columns->children)
         {
             String identifier_str;
             if (!tryGetIdentifierNameInto(identifier, identifier_str))
diff --git a/src/Parsers/ExpressionListParsers.cpp b/src/Parsers/ExpressionListParsers.cpp
index 062d1dd4483..2561bf7ef83 100644
--- a/src/Parsers/ExpressionListParsers.cpp
+++ b/src/Parsers/ExpressionListParsers.cpp
@@ -751,4 +751,4 @@ bool ParserIdentifierList::parseImpl(Pos & pos, ASTPtr & node, Expected & expect
         .parse(pos, node, expected);
 }
 
-}
\ No newline at end of file
+}
diff --git a/tests/queries/0_stateless/01246_ttl_where_group_by.reference b/tests/queries/0_stateless/01280_ttl_where_group_by.reference
similarity index 100%
rename from tests/queries/0_stateless/01246_ttl_where_group_by.reference
rename to tests/queries/0_stateless/01280_ttl_where_group_by.reference
diff --git a/tests/queries/0_stateless/01246_ttl_where_group_by.sql b/tests/queries/0_stateless/01280_ttl_where_group_by.sql
similarity index 100%
rename from tests/queries/0_stateless/01246_ttl_where_group_by.sql
rename to tests/queries/0_stateless/01280_ttl_where_group_by.sql

From 6585571ee2e8d8b0a832e4b51f0e40762d096a4b Mon Sep 17 00:00:00 2001
From: Nikolai Sorokin <sonikint@gmail.com>
Date: Sat, 16 May 2020 02:53:47 +0300
Subject: [PATCH 058/122] Fix bug

---
 src/Interpreters/ExpressionAnalyzer.cpp | 89 +++++++++++++------------
 1 file changed, 45 insertions(+), 44 deletions(-)

diff --git a/src/Interpreters/ExpressionAnalyzer.cpp b/src/Interpreters/ExpressionAnalyzer.cpp
index c432b920a9f..b38af6feef9 100644
--- a/src/Interpreters/ExpressionAnalyzer.cpp
+++ b/src/Interpreters/ExpressionAnalyzer.cpp
@@ -192,61 +192,65 @@ void ExpressionAnalyzer::analyzeAggregation()
 
     if (has_aggregation)
     {
-        // getSelectQuery(); /// assertSelect()
 
         /// Find out aggregation keys.
-        if (select_query && select_query->groupBy())
+        if (select_query)
         {
-            NameSet unique_keys;
-            ASTs & group_asts = select_query->groupBy()->children;
-            for (ssize_t i = 0; i < ssize_t(group_asts.size()); ++i)
+            if (select_query->groupBy())
             {
-                ssize_t size = group_asts.size();
-                getRootActionsNoMakeSet(group_asts[i], true, temp_actions, false);
-
-                const auto & column_name = group_asts[i]->getColumnName();
-                const auto & block = temp_actions->getSampleBlock();
-
-                if (!block.has(column_name))
-                    throw Exception("Unknown identifier (in GROUP BY): " + column_name, ErrorCodes::UNKNOWN_IDENTIFIER);
-
-                const auto & col = block.getByName(column_name);
-
-                /// Constant expressions have non-null column pointer at this stage.
-                if (col.column && isColumnConst(*col.column))
+                NameSet unique_keys;
+                ASTs & group_asts = select_query->groupBy()->children;
+                for (ssize_t i = 0; i < ssize_t(group_asts.size()); ++i)
                 {
-                    /// But don't remove last key column if no aggregate functions, otherwise aggregation will not work.
-                    if (!aggregate_descriptions.empty() || size > 1)
+                    ssize_t size = group_asts.size();
+                    getRootActionsNoMakeSet(group_asts[i], true, temp_actions, false);
+
+                    const auto & column_name = group_asts[i]->getColumnName();
+                    const auto & block = temp_actions->getSampleBlock();
+
+                    if (!block.has(column_name))
+                        throw Exception("Unknown identifier (in GROUP BY): " + column_name, ErrorCodes::UNKNOWN_IDENTIFIER);
+
+                    const auto & col = block.getByName(column_name);
+
+                    /// Constant expressions have non-null column pointer at this stage.
+                    if (col.column && isColumnConst(*col.column))
                     {
-                        if (i + 1 < static_cast<ssize_t>(size))
-                            group_asts[i] = std::move(group_asts.back());
+                        /// But don't remove last key column if no aggregate functions, otherwise aggregation will not work.
+                        if (!aggregate_descriptions.empty() || size > 1)
+                        {
+                            if (i + 1 < static_cast<ssize_t>(size))
+                                group_asts[i] = std::move(group_asts.back());
 
-                        group_asts.pop_back();
+                            group_asts.pop_back();
 
-                        --i;
-                        continue;
+                            --i;
+                            continue;
+                        }
+                    }
+
+                    NameAndTypePair key{column_name, col.type};
+
+                    /// Aggregation keys are uniqued.
+                    if (!unique_keys.count(key.name))
+                    {
+                        unique_keys.insert(key.name);
+                        aggregation_keys.push_back(key);
+
+                        /// Key is no longer needed, therefore we can save a little by moving it.
+                        aggregated_columns.push_back(std::move(key));
                     }
                 }
 
-                NameAndTypePair key{column_name, col.type};
-
-                /// Aggregation keys are uniqued.
-                if (!unique_keys.count(key.name))
+                if (group_asts.empty())
                 {
-                    unique_keys.insert(key.name);
-                    aggregation_keys.push_back(key);
-
-                    /// Key is no longer needed, therefore we can save a little by moving it.
-                    aggregated_columns.push_back(std::move(key));
+                    select_query->setExpression(ASTSelectQuery::Expression::GROUP_BY, {});
+                    has_aggregation = select_query->having() || !aggregate_descriptions.empty();
                 }
             }
-
-            if (group_asts.empty())
-            {
-                select_query->setExpression(ASTSelectQuery::Expression::GROUP_BY, {});
-                has_aggregation = select_query->having() || !aggregate_descriptions.empty();
-            }
         }
+        else
+            aggregated_columns = temp_actions->getSampleBlock().getNamesAndTypesList();
 
         for (const auto & desc : aggregate_descriptions)
             aggregated_columns.emplace_back(desc.column_name, desc.function->getReturnType());
@@ -926,10 +930,7 @@ void ExpressionAnalyzer::appendExpression(ExpressionActionsChain & chain, const
 
 ExpressionActionsPtr ExpressionAnalyzer::getActions(bool add_aliases, bool project_result)
 {
-    NamesAndTypesList columns(sourceColumns());
-    for (const auto & col : aggregated_columns)
-        columns.push_back(col);
-    ExpressionActionsPtr actions = std::make_shared<ExpressionActions>(columns, context);
+    ExpressionActionsPtr actions = std::make_shared<ExpressionActions>(aggregated_columns, context);
     NamesWithAliases result_columns;
     Names result_names;
 

From ed09f68b8dbead260ee3506ff3533616259dca04 Mon Sep 17 00:00:00 2001
From: Nikolai Sorokin <sonikint@gmail.com>
Date: Sat, 16 May 2020 19:31:29 +0300
Subject: [PATCH 059/122] Better tests

---
 .../01280_ttl_where_group_by.reference        | 14 ++--
 .../0_stateless/01280_ttl_where_group_by.sql  | 76 +++++++++----------
 2 files changed, 45 insertions(+), 45 deletions(-)

diff --git a/tests/queries/0_stateless/01280_ttl_where_group_by.reference b/tests/queries/0_stateless/01280_ttl_where_group_by.reference
index 69d3d77cd46..572fc7731d3 100644
--- a/tests/queries/0_stateless/01280_ttl_where_group_by.reference
+++ b/tests/queries/0_stateless/01280_ttl_where_group_by.reference
@@ -3,12 +3,12 @@
 1	3	0	5
 2	1	20	1
 2	1	0	1
-1	1	0	4
-1	1	6	6.5
-1	3	4	5.666666666666667
-2	1	10	4
-3	1	0	8
+1	1	[0,2,3]	4
+1	1	[5,4,1]	13
+1	3	[1,0,1,0]	17
+2	1	[3,1,0,3]	8
+3	1	[2,4,5]	8
 1	1	0	4
 1	3	10	6
-2	1	20	2
-3	5	8	4
+2	1	0	3
+3	5	8	2
diff --git a/tests/queries/0_stateless/01280_ttl_where_group_by.sql b/tests/queries/0_stateless/01280_ttl_where_group_by.sql
index ad969127eb1..0ca06e77001 100644
--- a/tests/queries/0_stateless/01280_ttl_where_group_by.sql
+++ b/tests/queries/0_stateless/01280_ttl_where_group_by.sql
@@ -1,45 +1,45 @@
-drop table if exists ttl_01246_1;
+drop table if exists ttl_01280_1;
 
-create table ttl_01246_1 (a Int, b Int, x Int, y Int, d DateTime) engine = MergeTree order by (a, b) ttl d + interval 1 second delete where x % 10 == 0 and y > 5;
-insert into ttl_01246_1 values (1, 1, 0, 4, now() + 10);
-insert into ttl_01246_1 values (1, 1, 10, 6, now());
-insert into ttl_01246_1 values (1, 2, 3, 7, now());
-insert into ttl_01246_1 values (1, 3, 0, 5, now());
-insert into ttl_01246_1 values (2, 1, 20, 1, now());
-insert into ttl_01246_1 values (2, 1, 0, 1, now());
-insert into ttl_01246_1 values (3, 1, 0, 8, now());
+create table ttl_01280_1 (a Int, b Int, x Int, y Int, d DateTime) engine = MergeTree order by (a, b) ttl d + interval 1 second delete where x % 10 == 0 and y > 5;
+insert into ttl_01280_1 values (1, 1, 0, 4, now() + 10);
+insert into ttl_01280_1 values (1, 1, 10, 6, now());
+insert into ttl_01280_1 values (1, 2, 3, 7, now());
+insert into ttl_01280_1 values (1, 3, 0, 5, now());
+insert into ttl_01280_1 values (2, 1, 20, 1, now());
+insert into ttl_01280_1 values (2, 1, 0, 1, now());
+insert into ttl_01280_1 values (3, 1, 0, 8, now());
 select sleep(1.1) format Null;
-optimize table ttl_01246_1 final;
-select a, b, x, y from ttl_01246_1;
+optimize table ttl_01280_1 final;
+select a, b, x, y from ttl_01280_1;
 
-drop table if exists ttl_01246_1;
+drop table if exists ttl_01280_2;
 
-create table ttl_01246_1 (a Int, b Int, x Int32, y Double, d DateTime) engine = MergeTree order by (a, b) ttl d + interval 1 second group by a, b set x = cast(median(x) as Int32), y = avg(y), d = max(d);
-insert into ttl_01246_1 values (1, 1, 0, 4, now() + 10);
-insert into ttl_01246_1 values (1, 1, 10, 6, now());
-insert into ttl_01246_1 values (1, 1, 3, 7, now());
-insert into ttl_01246_1 values (1, 3, 0, 5, now());
-insert into ttl_01246_1 values (1, 3, 4, 9, now());
-insert into ttl_01246_1 values (1, 3, 6, 3, now());
-insert into ttl_01246_1 values (2, 1, 20, 7, now());
-insert into ttl_01246_1 values (2, 1, 0, 1, now());
-insert into ttl_01246_1 values (3, 1, 0, 8, now());
+create table ttl_01280_2 (a Int, b Int, x Array(Int32), y Double, d DateTime) engine = MergeTree order by (a, b) ttl d + interval 1 second group by a, b set x = minForEach(x), y = sum(y), d = max(d);
+insert into ttl_01280_2 values (1, 1, array(0, 2, 3), 4, now() + 10);
+insert into ttl_01280_2 values (1, 1, array(5, 4, 3), 6, now());
+insert into ttl_01280_2 values (1, 1, array(5, 5, 1), 7, now());
+insert into ttl_01280_2 values (1, 3, array(3, 0, 4), 5, now());
+insert into ttl_01280_2 values (1, 3, array(1, 1, 2, 1), 9, now());
+insert into ttl_01280_2 values (1, 3, array(3, 2, 1, 0), 3, now());
+insert into ttl_01280_2 values (2, 1, array(3, 3, 3), 7, now());
+insert into ttl_01280_2 values (2, 1, array(11, 1, 0, 3), 1, now());
+insert into ttl_01280_2 values (3, 1, array(2, 4, 5), 8, now());
 select sleep(1.1) format Null;
-optimize table ttl_01246_1 final;
-select a, b, x, y from ttl_01246_1;
+optimize table ttl_01280_2 final;
+select a, b, x, y from ttl_01280_2;
 
-drop table if exists ttl_01246_1;
+drop table if exists ttl_01280_3;
 
-create table ttl_01246_1 (a Int, b Int, x Int32, y Int, d DateTime) engine = MergeTree order by (a, b) ttl d + interval 1 second group by a set x = max(x), y = cast(round(avg(y)) as Int), d = max(d);
-insert into ttl_01246_1 values (1, 1, 0, 4, now() + 10);
-insert into ttl_01246_1 values (1, 1, 10, 6, now());
-insert into ttl_01246_1 values (1, 2, 3, 7, now());
-insert into ttl_01246_1 values (1, 3, 0, 5, now());
-insert into ttl_01246_1 values (2, 1, 20, 1, now());
-insert into ttl_01246_1 values (2, 1, 0, 3, now());
-insert into ttl_01246_1 values (3, 1, 0, 3, now());
-insert into ttl_01246_1 values (3, 2, 8, 2, now());
-insert into ttl_01246_1 values (3, 5, 5, 8, now());
-select sleep(1.1) format Null;
-optimize table ttl_01246_1 final;
-select a, b, x, y from ttl_01246_1;
\ No newline at end of file
+create table ttl_01280_3 (a Int, b Int, x Int64, y Int, d DateTime) engine = MergeTree order by (a, b) ttl d + interval 1 second group by a set x = argMax(x, d), y = argMax(y, d), d = max(d);
+insert into ttl_01280_3 values (1, 1, 0, 4, now() + 10);
+insert into ttl_01280_3 values (1, 1, 10, 6, now() + 1);
+insert into ttl_01280_3 values (1, 2, 3, 7, now());
+insert into ttl_01280_3 values (1, 3, 0, 5, now());
+insert into ttl_01280_3 values (2, 1, 20, 1, now());
+insert into ttl_01280_3 values (2, 1, 0, 3, now() + 1);
+insert into ttl_01280_3 values (3, 1, 0, 3, now());
+insert into ttl_01280_3 values (3, 2, 8, 2, now() + 1);
+insert into ttl_01280_3 values (3, 5, 5, 8, now());
+select sleep(2.1) format Null;
+optimize table ttl_01280_3 final;
+select a, b, x, y from ttl_01280_3;
\ No newline at end of file

From 7d937b43c7f0248069bdb7b1686244ce9d0c0f2b Mon Sep 17 00:00:00 2001
From: Nikolai Sorokin <sonikint@gmail.com>
Date: Sun, 17 May 2020 16:01:38 +0300
Subject: [PATCH 060/122] Init with correct value

---
 src/DataStreams/TTLBlockInputStream.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/DataStreams/TTLBlockInputStream.h b/src/DataStreams/TTLBlockInputStream.h
index a8f86477b88..c6ffa95cd75 100644
--- a/src/DataStreams/TTLBlockInputStream.h
+++ b/src/DataStreams/TTLBlockInputStream.h
@@ -45,7 +45,7 @@ private:
     AggregatedDataVariants agg_result;
     ColumnRawPtrs agg_key_columns;
     Aggregator::AggregateColumns agg_aggregate_columns;
-    bool agg_no_more_keys;
+    bool agg_no_more_keys = false;
 
     IMergeTreeDataPart::TTLInfos old_ttl_infos;
     IMergeTreeDataPart::TTLInfos new_ttl_infos;

From 141ed887515e6523c192690baf66dd1c2ca3a3b8 Mon Sep 17 00:00:00 2001
From: Nikolai Sorokin <sonikint@gmail.com>
Date: Sun, 24 May 2020 23:21:23 +0300
Subject: [PATCH 061/122] Allow functions in group by keys; Add default
 aggregate function; Add more tests

---
 src/Parsers/ASTTTLElement.cpp                 | 23 ++++---
 src/Parsers/ASTTTLElement.h                   |  2 +-
 src/Parsers/ExpressionElementParsers.cpp      | 58 ++++++++---------
 src/Parsers/ExpressionListParsers.cpp         |  9 ---
 src/Parsers/ExpressionListParsers.h           |  9 ---
 src/Storages/MergeTree/MergeTreeData.cpp      | 62 ++++++++++++++++---
 .../01280_ttl_where_group_by.reference        |  6 ++
 .../0_stateless/01280_ttl_where_group_by.sql  | 44 ++++++++++++-
 8 files changed, 146 insertions(+), 67 deletions(-)

diff --git a/src/Parsers/ASTTTLElement.cpp b/src/Parsers/ASTTTLElement.cpp
index 991274e8350..33f8a3ac906 100644
--- a/src/Parsers/ASTTTLElement.cpp
+++ b/src/Parsers/ASTTTLElement.cpp
@@ -17,6 +17,8 @@ ASTPtr ASTTTLElement::clone() const
     clone->setExpression(clone->ttl_expr_pos, getExpression(ttl_expr_pos, true));
     clone->setExpression(clone->where_expr_pos, getExpression(where_expr_pos, true));
 
+    for (auto & expr : clone->group_by_key)
+        expr = expr->clone();
     for (auto & [name, expr] : clone->group_by_aggregations)
         expr = expr->clone();
 
@@ -37,19 +39,22 @@ void ASTTTLElement::formatImpl(const FormatSettings & settings, FormatState & st
     else if (mode == TTLMode::GROUP_BY)
     {
         settings.ostr << " GROUP BY ";
-        for (auto it = group_by_key_columns.begin(); it != group_by_key_columns.end(); ++it)
+        for (auto it = group_by_key.begin(); it != group_by_key.end(); ++it)
         {
-            if (it != group_by_key_columns.begin())
+            if (it != group_by_key.begin())
                 settings.ostr << ", ";
-            settings.ostr << *it;
+            (*it)->formatImpl(settings, state, frame);
         }
-        settings.ostr << " SET ";
-        for (auto it = group_by_aggregations.begin(); it != group_by_aggregations.end(); ++it)
+        if (!group_by_aggregations.empty())
         {
-            if (it != group_by_aggregations.begin())
-                settings.ostr << ", ";
-            settings.ostr << it->first << " = ";
-            it->second->formatImpl(settings, state, frame);
+            settings.ostr << " SET ";
+            for (auto it = group_by_aggregations.begin(); it != group_by_aggregations.end(); ++it)
+            {
+                if (it != group_by_aggregations.begin())
+                    settings.ostr << ", ";
+                settings.ostr << it->first << " = ";
+                it->second->formatImpl(settings, state, frame);
+            }
         }
     }
     else if (mode == TTLMode::DELETE)
diff --git a/src/Parsers/ASTTTLElement.h b/src/Parsers/ASTTTLElement.h
index 9b40adf2be4..2004a4f927b 100644
--- a/src/Parsers/ASTTTLElement.h
+++ b/src/Parsers/ASTTTLElement.h
@@ -17,7 +17,7 @@ public:
     PartDestinationType destination_type;
     String destination_name;
 
-    Strings group_by_key_columns;
+    ASTs group_by_key;
     std::vector<std::pair<String, ASTPtr>> group_by_aggregations;
 
     ASTTTLElement(TTLMode mode_, PartDestinationType destination_type_, const String & destination_name_)
diff --git a/src/Parsers/ExpressionElementParsers.cpp b/src/Parsers/ExpressionElementParsers.cpp
index 5b5d9f05b22..2ea39946ff7 100644
--- a/src/Parsers/ExpressionElementParsers.cpp
+++ b/src/Parsers/ExpressionElementParsers.cpp
@@ -1464,7 +1464,7 @@ bool ParserTTLElement::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
     ParserIdentifier parser_identifier;
     ParserStringLiteral parser_string_literal;
     ParserExpression parser_exp;
-    ParserIdentifierList parser_identifier_list;
+    ParserExpressionList parser_expression_list(false);
 
     ASTPtr ttl_expr;
     if (!parser_exp.parse(pos, ttl_expr, expected))
@@ -1495,7 +1495,7 @@ bool ParserTTLElement::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
     }
 
     ASTPtr where_expr;
-    std::vector<String> group_by_key_columns;
+    ASTPtr ast_group_by_key;
     std::vector<std::pair<String, ASTPtr>> group_by_aggregations;
 
     if (mode == TTLMode::MOVE)
@@ -1508,37 +1508,30 @@ bool ParserTTLElement::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
     }
     else if (mode == TTLMode::GROUP_BY)
     {
-        ASTPtr ast_group_by_key_columns;
-        if (!parser_identifier_list.parse(pos, ast_group_by_key_columns, expected))
+        if (!parser_expression_list.parse(pos, ast_group_by_key, expected))
             return false;
-        for (const auto & identifier : ast_group_by_key_columns->children)
+
+        if (s_set.ignore(pos))
         {
-            String identifier_str;
-            if (!tryGetIdentifierNameInto(identifier, identifier_str))
-                return false;
-            group_by_key_columns.emplace_back(std::move(identifier_str));
-        }
+            while (true)
+            {
+                if (!group_by_aggregations.empty() && !s_comma.ignore(pos))
+                    break;
 
-        if (!s_set.ignore(pos))
-            return false;
-        while (true)
-        {
-            if (!group_by_aggregations.empty() && !s_comma.ignore(pos))
-                break;
+                ASTPtr name;
+                ASTPtr value;
+                if (!parser_identifier.parse(pos, name, expected))
+                    return false;
+                if (!s_eq.ignore(pos))
+                    return false;
+                if (!parser_exp.parse(pos, value, expected))
+                    return false;
 
-            ASTPtr name;
-            ASTPtr value;
-            if (!parser_identifier.parse(pos, name, expected))
-                return false;
-            if (!s_eq.ignore(pos))
-                return false;
-            if (!parser_exp.parse(pos, value, expected))
-                return false;
-
-            String name_str;
-            if (!tryGetIdentifierNameInto(name, name_str))
-                return false;
-            group_by_aggregations.emplace_back(name_str, std::move(value));
+                String name_str;
+                if (!tryGetIdentifierNameInto(name, name_str))
+                    return false;
+                group_by_aggregations.emplace_back(name_str, std::move(value));
+            }
         }
     }
     else if (mode == TTLMode::DELETE && s_where.ignore(pos))
@@ -1552,8 +1545,11 @@ bool ParserTTLElement::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
     if (where_expr)
         ttl_element->setWhere(std::move(where_expr));
 
-    ttl_element->group_by_key_columns = std::move(group_by_key_columns);
-    ttl_element->group_by_aggregations = std::move(group_by_aggregations);
+    if (mode == TTLMode::GROUP_BY)
+    {
+        ttl_element->group_by_key = std::move(ast_group_by_key->children);
+        ttl_element->group_by_aggregations = std::move(group_by_aggregations);
+    }
 
     node = ttl_element;
     return true;
diff --git a/src/Parsers/ExpressionListParsers.cpp b/src/Parsers/ExpressionListParsers.cpp
index 2561bf7ef83..a967ae19691 100644
--- a/src/Parsers/ExpressionListParsers.cpp
+++ b/src/Parsers/ExpressionListParsers.cpp
@@ -742,13 +742,4 @@ bool ParserKeyValuePairsList::parseImpl(Pos & pos, ASTPtr & node, Expected & exp
     return parser.parse(pos, node, expected);
 }
 
-
-bool ParserIdentifierList::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
-{
-    return ParserList(
-        std::make_unique<ParserIdentifier>(),
-        std::make_unique<ParserToken>(TokenType::Comma))
-        .parse(pos, node, expected);
-}
-
 }
diff --git a/src/Parsers/ExpressionListParsers.h b/src/Parsers/ExpressionListParsers.h
index 9cc36967bd7..0cef29b6d67 100644
--- a/src/Parsers/ExpressionListParsers.h
+++ b/src/Parsers/ExpressionListParsers.h
@@ -421,13 +421,4 @@ protected:
     bool parseImpl(Pos & pos, ASTPtr & node, Expected & expected) override;
 };
 
-
-/** A comma-separated list of identifiers, probably empty. */
-class ParserIdentifierList : public IParserBase
-{
-protected:
-    const char * getName() const override { return "list of identifiers"; }
-    bool parseImpl(Pos & pos, ASTPtr & node, Expected & expected) override;
-};
-
 }
diff --git a/src/Storages/MergeTree/MergeTreeData.cpp b/src/Storages/MergeTree/MergeTreeData.cpp
index bb9d8ce186e..5de3d204f16 100644
--- a/src/Storages/MergeTree/MergeTreeData.cpp
+++ b/src/Storages/MergeTree/MergeTreeData.cpp
@@ -646,22 +646,70 @@ void MergeTreeData::setTTLExpressions(const ColumnsDescription & new_columns,
         }
         else if (ttl_element->mode == TTLMode::GROUP_BY)
         {
-            if (ttl_element->group_by_key_columns.size() > this->primary_key_columns.size())
+            if (ttl_element->group_by_key.size() > this->primary_key_columns.size())
                 throw Exception("TTL Expression GROUP BY key should be a prefix of primary key", ErrorCodes::BAD_TTL_EXPRESSION);
-            for (size_t i = 0; i < ttl_element->group_by_key_columns.size(); ++i)
+
+            NameSet primary_key_columns_set(this->primary_key_columns.begin(), this->primary_key_columns.end());
+            NameSet aggregation_columns_set;
+
+            for (const auto & column : this->primary_key_expr->getRequiredColumns())
+                primary_key_columns_set.insert(column);
+
+            for (size_t i = 0; i < ttl_element->group_by_key.size(); ++i)
             {
-                if (ttl_element->group_by_key_columns[i] != this->primary_key_columns[i])
+                if (ttl_element->group_by_key[i]->getColumnName() != this->primary_key_columns[i])
                     throw Exception("TTL Expression GROUP BY key should be a prefix of primary key", ErrorCodes::BAD_TTL_EXPRESSION);
             }
+            for (const auto & [name, value] : ttl_element->group_by_aggregations)
+            {
+                if (primary_key_columns_set.contains(name))
+                    throw Exception("Can not set custom aggregation for column in primary key in TTL Expression", ErrorCodes::BAD_TTL_EXPRESSION);
+                aggregation_columns_set.insert(name);
+            }
+            if (aggregation_columns_set.size() != ttl_element->group_by_aggregations.size())
+                throw Exception("Multiple aggregations set for one column in TTL Expression", ErrorCodes::BAD_TTL_EXPRESSION);
 
-            result.group_by_keys = ttl_element->group_by_key_columns;
+            result.group_by_keys = Names(this->primary_key_columns.begin(), this->primary_key_columns.begin() + ttl_element->group_by_key.size());
 
             auto aggregations = ttl_element->group_by_aggregations;
-            for (size_t i = ttl_element->group_by_key_columns.size(); i < this->primary_key_columns.size(); ++i)
+            for (size_t i = 0; i < this->primary_key_columns.size(); ++i)
             {
-                ASTPtr expr = makeASTFunction("max", std::make_shared<ASTIdentifier>(this->primary_key_columns[i]));
-                aggregations.emplace_back(this->primary_key_columns[i], std::move(expr));
+                ASTPtr value = this->primary_key_expr_ast->children[i]->clone();
+
+                if (i >= ttl_element->group_by_key.size())
+                {
+                    ASTPtr value_max = makeASTFunction("max", value->clone());
+                    aggregations.emplace_back(value->getColumnName(), std::move(value_max));
+                }
+
+                if (value->as<ASTFunction>())
+                {
+                    auto syntax_result = SyntaxAnalyzer(global_context).analyze(value, new_columns.getAllPhysical(), {}, true);
+                    auto expr_actions = ExpressionAnalyzer(value, syntax_result, global_context).getActions(false);
+                    for (const auto & column : expr_actions->getRequiredColumns())
+                    {
+                        if (i < ttl_element->group_by_key.size())
+                        {
+                            ASTPtr expr = makeASTFunction("any", std::make_shared<ASTIdentifier>(column));
+                            aggregations.emplace_back(column, std::move(expr));
+                        }
+                        else
+                        {
+                            ASTPtr expr = makeASTFunction("argMax", std::make_shared<ASTIdentifier>(column), value->clone());
+                            aggregations.emplace_back(column, std::move(expr));
+                        }
+                    }
+                }
             }
+            for (const auto & column : new_columns.getAllPhysical())
+            {
+                if (!primary_key_columns_set.contains(column.name) && !aggregation_columns_set.contains(column.name))
+                {
+                    ASTPtr expr = makeASTFunction("any", std::make_shared<ASTIdentifier>(column.name));
+                    aggregations.emplace_back(column.name, std::move(expr));
+                }
+            }
+
             for (auto [name, value] : aggregations)
             {
                 auto syntax_result = SyntaxAnalyzer(global_context).analyze(value, new_columns.getAllPhysical(), {}, true);
diff --git a/tests/queries/0_stateless/01280_ttl_where_group_by.reference b/tests/queries/0_stateless/01280_ttl_where_group_by.reference
index 572fc7731d3..dead0a5aac3 100644
--- a/tests/queries/0_stateless/01280_ttl_where_group_by.reference
+++ b/tests/queries/0_stateless/01280_ttl_where_group_by.reference
@@ -12,3 +12,9 @@
 1	3	10	6
 2	1	0	3
 3	5	8	2
+1	1	0	4
+3	3	13	9
+1	2	7	5
+2	3	6	5
+1	2	3	5
+2	3	3	5
diff --git a/tests/queries/0_stateless/01280_ttl_where_group_by.sql b/tests/queries/0_stateless/01280_ttl_where_group_by.sql
index 0ca06e77001..e61716cfe81 100644
--- a/tests/queries/0_stateless/01280_ttl_where_group_by.sql
+++ b/tests/queries/0_stateless/01280_ttl_where_group_by.sql
@@ -42,4 +42,46 @@ insert into ttl_01280_3 values (3, 2, 8, 2, now() + 1);
 insert into ttl_01280_3 values (3, 5, 5, 8, now());
 select sleep(2.1) format Null;
 optimize table ttl_01280_3 final;
-select a, b, x, y from ttl_01280_3;
\ No newline at end of file
+select a, b, x, y from ttl_01280_3;
+
+drop table if exists ttl_01280_4;
+
+create table ttl_01280_4 (a Int, b Int, x Int64, y Int64, d DateTime) engine = MergeTree order by (toDate(d), -(a + b)) ttl d + interval 1 second group by toDate(d) set x = sum(x), y = max(y);
+insert into ttl_01280_4 values (1, 1, 0, 4, now() + 10);
+insert into ttl_01280_4 values (10, 2, 3, 3, now());
+insert into ttl_01280_4 values (2, 10, 1, 7, now());
+insert into ttl_01280_4 values (3, 3, 5, 2, now());
+insert into ttl_01280_4 values (1, 5, 4, 9, now());
+select sleep(1.1) format Null;
+optimize table ttl_01280_4 final;
+select a, b, x, y from ttl_01280_4;
+
+drop table if exists ttl_01280_5;
+
+create table ttl_01280_5 (a Int, b Int, x Int64, y Int64, d DateTime) engine = MergeTree order by (toDate(d), a, -b) ttl d + interval 1 second group by toDate(d), a set x = sum(x);
+insert into ttl_01280_5 values (1, 2, 3, 5, now());
+insert into ttl_01280_5 values (2, 10, 1, 5, now());
+insert into ttl_01280_5 values (2, 3, 5, 5, now());
+insert into ttl_01280_5 values (1, 5, 4, 5, now());
+select sleep(1.1) format Null;
+optimize table ttl_01280_5 final;
+select a, b, x, y from ttl_01280_5;
+
+drop table if exists ttl_01280_6;
+
+create table ttl_01280_6 (a Int, b Int, x Int64, y Int64, d DateTime) engine = MergeTree order by (toDate(d), a, -b) ttl d + interval 1 second group by toDate(d), a;
+insert into ttl_01280_6 values (1, 2, 3, 5, now());
+insert into ttl_01280_6 values (2, 10, 3, 5, now());
+insert into ttl_01280_6 values (2, 3, 3, 5, now());
+insert into ttl_01280_6 values (1, 5, 3, 5, now());
+select sleep(1.1) format Null;
+optimize table ttl_01280_6 final;
+select a, b, x, y from ttl_01280_6;
+
+create table ttl_01280_error (a Int, b Int, x Int64, y Int64, d DateTime) engine = MergeTree order by (a, b) ttl d + interval 1 second group by x set y = max(y); -- { serverError 450}
+create table ttl_01280_error (a Int, b Int, x Int64, y Int64, d DateTime) engine = MergeTree order by (a, b) ttl d + interval 1 second group by b set y = max(y); -- { serverError 450}
+create table ttl_01280_error (a Int, b Int, x Int64, y Int64, d DateTime) engine = MergeTree order by (a, b) ttl d + interval 1 second group by a, b, x set y = max(y); -- { serverError 450}
+create table ttl_01280_error (a Int, b Int, x Int64, y Int64, d DateTime) engine = MergeTree order by (a, b) ttl d + interval 1 second group by a set b = min(b), y = max(y); -- { serverError 450}
+create table ttl_01280_error (a Int, b Int, x Int64, y Int64, d DateTime) engine = MergeTree order by (a, b) ttl d + interval 1 second group by a, b set y = max(y), y = max(y); -- { serverError 450}
+create table ttl_01280_error (a Int, b Int, x Int64, y Int64, d DateTime) engine = MergeTree order by (toDate(d), a) ttl d + interval 1 second group by toDate(d), a set d = min(d), b = max(b); -- { serverError 450}
+create table ttl_01280_error (a Int, b Int, x Int64, y Int64, d DateTime) engine = MergeTree order by (d, -(a + b)) ttl d + interval 1 second group by d, -(a + b) set a = sum(a), b = min(b); -- { serverError 450}

From e182d4df2ef95c59527bb06b162d92c9af3fd341 Mon Sep 17 00:00:00 2001
From: Nikolai Sorokin <sonikint@gmail.com>
Date: Mon, 25 May 2020 01:52:18 +0300
Subject: [PATCH 062/122] Fix context usage

---
 src/DataStreams/TTLBlockInputStream.cpp | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/src/DataStreams/TTLBlockInputStream.cpp b/src/DataStreams/TTLBlockInputStream.cpp
index 106ae1fd5a0..f11f6ea38cd 100644
--- a/src/DataStreams/TTLBlockInputStream.cpp
+++ b/src/DataStreams/TTLBlockInputStream.cpp
@@ -6,6 +6,7 @@
 #include <Columns/ColumnConst.h>
 #include <Interpreters/addTypeConversionToAST.h>
 #include <Storages/MergeTree/TTLMode.h>
+#include <Interpreters/Context.h>
 
 namespace DB
 {
@@ -87,11 +88,9 @@ TTLBlockInputStream::TTLBlockInputStream(
 
         const Settings & settings = storage.global_context.getSettingsRef();
 
-        bool allow_to_use_two_level_group_by = false;  // settings.max_bytes_before_external_group_by != 0;
         Aggregator::Params params(header, keys, aggregates,
             false, settings.max_rows_to_group_by, settings.group_by_overflow_mode,
-            allow_to_use_two_level_group_by ? settings.group_by_two_level_threshold : SettingUInt64(0),
-            allow_to_use_two_level_group_by ? settings.group_by_two_level_threshold_bytes : SettingUInt64(0),
+            SettingUInt64(0), SettingUInt64(0),
             settings.max_bytes_before_external_group_by, settings.empty_result_for_aggregation_by_empty_set,
             storage.global_context.getTemporaryVolume(), settings.max_threads, settings.min_free_disk_space_for_temporary_data);
         aggregator = std::make_unique<Aggregator>(params);

From f62adfd588b579e51f76ac78fe19e66df250ff8b Mon Sep 17 00:00:00 2001
From: Anton Popov <pad11rus@gmail.com>
Date: Wed, 27 May 2020 04:52:02 +0300
Subject: [PATCH 063/122] Fix Arcadia build

---
 src/Storages/MergeTree/MergeTreeData.cpp | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/src/Storages/MergeTree/MergeTreeData.cpp b/src/Storages/MergeTree/MergeTreeData.cpp
index 5de3d204f16..9a1b4f27872 100644
--- a/src/Storages/MergeTree/MergeTreeData.cpp
+++ b/src/Storages/MergeTree/MergeTreeData.cpp
@@ -662,7 +662,7 @@ void MergeTreeData::setTTLExpressions(const ColumnsDescription & new_columns,
             }
             for (const auto & [name, value] : ttl_element->group_by_aggregations)
             {
-                if (primary_key_columns_set.contains(name))
+                if (primary_key_columns_set.count(name))
                     throw Exception("Can not set custom aggregation for column in primary key in TTL Expression", ErrorCodes::BAD_TTL_EXPRESSION);
                 aggregation_columns_set.insert(name);
             }
@@ -703,7 +703,7 @@ void MergeTreeData::setTTLExpressions(const ColumnsDescription & new_columns,
             }
             for (const auto & column : new_columns.getAllPhysical())
             {
-                if (!primary_key_columns_set.contains(column.name) && !aggregation_columns_set.contains(column.name))
+                if (!primary_key_columns_set.count(column.name) && !aggregation_columns_set.count(column.name))
                 {
                     ASTPtr expr = makeASTFunction("any", std::make_shared<ASTIdentifier>(column.name));
                     aggregations.emplace_back(column.name, std::move(expr));

From 146370934a295759f342c310ab74eb660d675423 Mon Sep 17 00:00:00 2001
From: Alexey Milovidov <milovidov@yandex-team.ru>
Date: Wed, 27 May 2020 13:07:38 +0300
Subject: [PATCH 064/122] Keep the value of DistributedFilesToInsert metric on
 exceptions

---
 src/Common/CurrentMetrics.h                   |  6 ++++
 src/Storages/Distributed/DirectoryMonitor.cpp | 32 +++++++++++++------
 src/Storages/Distributed/DirectoryMonitor.h   |  8 +++--
 3 files changed, 34 insertions(+), 12 deletions(-)

diff --git a/src/Common/CurrentMetrics.h b/src/Common/CurrentMetrics.h
index b87504ef49a..a3bac96a16c 100644
--- a/src/Common/CurrentMetrics.h
+++ b/src/Common/CurrentMetrics.h
@@ -94,6 +94,12 @@ namespace CurrentMetrics
             amount = new_amount;
         }
 
+        void sub(Value value = 1)
+        {
+            what->fetch_sub(value, std::memory_order_relaxed);
+            amount -= value;
+        }
+
         /// Subtract value before destructor.
         void destroy()
         {
diff --git a/src/Storages/Distributed/DirectoryMonitor.cpp b/src/Storages/Distributed/DirectoryMonitor.cpp
index fdcbe8e69f9..4dd62db0965 100644
--- a/src/Storages/Distributed/DirectoryMonitor.cpp
+++ b/src/Storages/Distributed/DirectoryMonitor.cpp
@@ -110,8 +110,9 @@ void StorageDistributedDirectoryMonitor::flushAllData()
 {
     if (!quit)
     {
+        CurrentMetrics::Increment metric_pending_files{CurrentMetrics::DistributedFilesToInsert, 0};
         std::unique_lock lock{mutex};
-        processFiles();
+        processFiles(metric_pending_files);
     }
 }
 
@@ -131,6 +132,9 @@ void StorageDistributedDirectoryMonitor::run()
 {
     std::unique_lock lock{mutex};
 
+    /// This metric will be updated with the number of pending files later.
+    CurrentMetrics::Increment metric_pending_files{CurrentMetrics::DistributedFilesToInsert, 0};
+
     bool do_sleep = false;
     while (!quit)
     {
@@ -139,7 +143,7 @@ void StorageDistributedDirectoryMonitor::run()
         {
             try
             {
-                do_sleep = !processFiles();
+                do_sleep = !processFiles(metric_pending_files);
             }
             catch (...)
             {
@@ -222,7 +226,7 @@ ConnectionPoolPtr StorageDistributedDirectoryMonitor::createPool(const std::stri
 }
 
 
-bool StorageDistributedDirectoryMonitor::processFiles()
+bool StorageDistributedDirectoryMonitor::processFiles(CurrentMetrics::Increment & metric_pending_files)
 {
     std::map<UInt64, std::string> files;
 
@@ -236,14 +240,16 @@ bool StorageDistributedDirectoryMonitor::processFiles()
             files[parse<UInt64>(file_path.getBaseName())] = file_path_str;
     }
 
+    /// Note: the value of this metric will be kept if this function will throw an exception.
+    /// This is needed, because in case of exception, files still pending.
+    metric_pending_files.changeTo(files.size());
+
     if (files.empty())
         return false;
 
-    CurrentMetrics::Increment metric_increment{CurrentMetrics::DistributedFilesToInsert, CurrentMetrics::Value(files.size())};
-
     if (should_batch_inserts)
     {
-        processFilesWithBatching(files);
+        processFilesWithBatching(files, metric_pending_files);
     }
     else
     {
@@ -252,14 +258,14 @@ bool StorageDistributedDirectoryMonitor::processFiles()
             if (quit)
                 return true;
 
-            processFile(file.second);
+            processFile(file.second, metric_pending_files);
         }
     }
 
     return true;
 }
 
-void StorageDistributedDirectoryMonitor::processFile(const std::string & file_path)
+void StorageDistributedDirectoryMonitor::processFile(const std::string & file_path, CurrentMetrics::Increment & metric_pending_files)
 {
     LOG_TRACE(log, "Started processing `{}`", file_path);
     auto timeouts = ConnectionTimeouts::getTCPTimeoutsWithFailover(storage.global_context->getSettingsRef());
@@ -289,6 +295,7 @@ void StorageDistributedDirectoryMonitor::processFile(const std::string & file_pa
     }
 
     Poco::File{file_path}.remove();
+    metric_pending_files.sub();
 
     LOG_TRACE(log, "Finished processing `{}`", file_path);
 }
@@ -584,7 +591,9 @@ bool StorageDistributedDirectoryMonitor::scheduleAfter(size_t ms)
     return task_handle->scheduleAfter(ms, false);
 }
 
-void StorageDistributedDirectoryMonitor::processFilesWithBatching(const std::map<UInt64, std::string> & files)
+void StorageDistributedDirectoryMonitor::processFilesWithBatching(
+    const std::map<UInt64, std::string> & files,
+    CurrentMetrics::Increment & metric_pending_files)
 {
     std::unordered_set<UInt64> file_indices_to_skip;
 
@@ -596,6 +605,7 @@ void StorageDistributedDirectoryMonitor::processFilesWithBatching(const std::map
         batch.readText(in);
         file_indices_to_skip.insert(batch.file_indices.begin(), batch.file_indices.end());
         batch.send();
+        metric_pending_files.sub(batch.file_indices.size());
     }
 
     std::unordered_map<BatchHeader, Batch, BatchHeader::Hash> header_to_batch;
@@ -656,13 +666,17 @@ void StorageDistributedDirectoryMonitor::processFilesWithBatching(const std::map
         batch.total_bytes += total_bytes;
 
         if (batch.isEnoughSize())
+        {
             batch.send();
+            metric_pending_files.sub(batch.file_indices.size());
+        }
     }
 
     for (auto & kv : header_to_batch)
     {
         Batch & batch = kv.second;
         batch.send();
+        metric_pending_files.sub(batch.file_indices.size());
     }
 
     /// current_batch.txt will not exist if there was no send
diff --git a/src/Storages/Distributed/DirectoryMonitor.h b/src/Storages/Distributed/DirectoryMonitor.h
index 77abf35630c..e2a913ee1ef 100644
--- a/src/Storages/Distributed/DirectoryMonitor.h
+++ b/src/Storages/Distributed/DirectoryMonitor.h
@@ -9,6 +9,8 @@
 #include <IO/ReadBufferFromFile.h>
 
 
+namespace CurrentMetrics { class Increment; }
+
 namespace DB
 {
 
@@ -37,9 +39,9 @@ public:
     bool scheduleAfter(size_t ms);
 private:
     void run();
-    bool processFiles();
-    void processFile(const std::string & file_path);
-    void processFilesWithBatching(const std::map<UInt64, std::string> & files);
+    bool processFiles(CurrentMetrics::Increment & metric_pending_files);
+    void processFile(const std::string & file_path, CurrentMetrics::Increment & metric_pending_files);
+    void processFilesWithBatching(const std::map<UInt64, std::string> & files, CurrentMetrics::Increment & metric_pending_files);
 
     static bool isFileBrokenErrorCode(int code);
     void markAsBroken(const std::string & file_path) const;

From 57555dbabf11780ca751e8f89a8e8f3787294d90 Mon Sep 17 00:00:00 2001
From: Nikolai Sorokin <sonikint@gmail.com>
Date: Wed, 27 May 2020 14:00:17 +0300
Subject: [PATCH 065/122] Fix after rebase

---
 src/Storages/MergeTree/MergeTreeData.cpp | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/src/Storages/MergeTree/MergeTreeData.cpp b/src/Storages/MergeTree/MergeTreeData.cpp
index 9a1b4f27872..da351fc7750 100644
--- a/src/Storages/MergeTree/MergeTreeData.cpp
+++ b/src/Storages/MergeTree/MergeTreeData.cpp
@@ -646,18 +646,18 @@ void MergeTreeData::setTTLExpressions(const ColumnsDescription & new_columns,
         }
         else if (ttl_element->mode == TTLMode::GROUP_BY)
         {
-            if (ttl_element->group_by_key.size() > this->primary_key_columns.size())
+            if (ttl_element->group_by_key.size() > this->getPrimaryKey().column_names.size())
                 throw Exception("TTL Expression GROUP BY key should be a prefix of primary key", ErrorCodes::BAD_TTL_EXPRESSION);
 
-            NameSet primary_key_columns_set(this->primary_key_columns.begin(), this->primary_key_columns.end());
+            NameSet primary_key_columns_set(this->getPrimaryKey().column_names.begin(), this->getPrimaryKey().column_names.end());
             NameSet aggregation_columns_set;
 
-            for (const auto & column : this->primary_key_expr->getRequiredColumns())
+            for (const auto & column : this->getPrimaryKey().expression->getRequiredColumns())
                 primary_key_columns_set.insert(column);
 
             for (size_t i = 0; i < ttl_element->group_by_key.size(); ++i)
             {
-                if (ttl_element->group_by_key[i]->getColumnName() != this->primary_key_columns[i])
+                if (ttl_element->group_by_key[i]->getColumnName() != this->getPrimaryKey().column_names[i])
                     throw Exception("TTL Expression GROUP BY key should be a prefix of primary key", ErrorCodes::BAD_TTL_EXPRESSION);
             }
             for (const auto & [name, value] : ttl_element->group_by_aggregations)
@@ -669,12 +669,12 @@ void MergeTreeData::setTTLExpressions(const ColumnsDescription & new_columns,
             if (aggregation_columns_set.size() != ttl_element->group_by_aggregations.size())
                 throw Exception("Multiple aggregations set for one column in TTL Expression", ErrorCodes::BAD_TTL_EXPRESSION);
 
-            result.group_by_keys = Names(this->primary_key_columns.begin(), this->primary_key_columns.begin() + ttl_element->group_by_key.size());
+            result.group_by_keys = Names(this->getPrimaryKey().column_names.begin(), this->getPrimaryKey().column_names.begin() + ttl_element->group_by_key.size());
 
             auto aggregations = ttl_element->group_by_aggregations;
-            for (size_t i = 0; i < this->primary_key_columns.size(); ++i)
+            for (size_t i = 0; i < this->getPrimaryKey().column_names.size(); ++i)
             {
-                ASTPtr value = this->primary_key_expr_ast->children[i]->clone();
+                ASTPtr value = this->getPrimaryKey().expression_list_ast->children[i]->clone();
 
                 if (i >= ttl_element->group_by_key.size())
                 {

From 8bfe21f10194ae05f4f1fa74409560622f5c0460 Mon Sep 17 00:00:00 2001
From: alexey-milovidov <milovidov@yandex-team.ru>
Date: Wed, 27 May 2020 14:09:12 +0300
Subject: [PATCH 066/122] Update 01281_group_by_limit_memory_tracking.sh

---
 .../queries/0_stateless/01281_group_by_limit_memory_tracking.sh  | 1 +
 1 file changed, 1 insertion(+)

diff --git a/tests/queries/0_stateless/01281_group_by_limit_memory_tracking.sh b/tests/queries/0_stateless/01281_group_by_limit_memory_tracking.sh
index 79a731b9273..5922b8d74d2 100755
--- a/tests/queries/0_stateless/01281_group_by_limit_memory_tracking.sh
+++ b/tests/queries/0_stateless/01281_group_by_limit_memory_tracking.sh
@@ -27,6 +27,7 @@ function execute_group_by()
     execute_null "${opts[@]}" <<<'SELECT uniq(number) FROM numbers_mt(toUInt64(1e6)) GROUP BY number % 5e5'
 }
 
+# This is needed to keep at least one running query for user for the time of test.
 execute_null <<<'SELECT sleep(3)' &
 execute_group_by
 # if memory accounting will be incorrect, the second query will be failed with MEMORY_LIMIT_EXCEEDED

From d24286d5da2ce890872de6f0837b076032673339 Mon Sep 17 00:00:00 2001
From: Albert Kidrachev <kidrachev@sandbox-410109003>
Date: Wed, 27 May 2020 14:52:45 +0300
Subject: [PATCH 067/122] fix

---
 src/Common/SharedBlockRowRef.h                | 29 -------
 src/Interpreters/InterpreterSelectQuery.cpp   |  3 +-
 .../OptimizedPartialSortingTransform.cpp      | 83 -------------------
 .../OptimizedPartialSortingTransform.h        | 34 --------
 .../Transforms/PartialSortingTransform.cpp    | 66 +++++++++++++++
 .../Transforms/PartialSortingTransform.h      |  2 +
 6 files changed, 69 insertions(+), 148 deletions(-)
 delete mode 100644 src/Processors/Transforms/OptimizedPartialSortingTransform.cpp
 delete mode 100644 src/Processors/Transforms/OptimizedPartialSortingTransform.h

diff --git a/src/Common/SharedBlockRowRef.h b/src/Common/SharedBlockRowRef.h
index e9fd076da07..957d66243dd 100644
--- a/src/Common/SharedBlockRowRef.h
+++ b/src/Common/SharedBlockRowRef.h
@@ -87,33 +87,4 @@ struct SharedBlockRowRef
     }
 };
 
-struct SharedBlockRowWithSortDescriptionRef : SharedBlockRowRef
-{
-    SortDescription * description = nullptr;
-
-    void set(SharedBlockPtr & shared_block_, ColumnRawPtrs * columns_, size_t row_num_) = delete;
-
-    bool operator< (const SharedBlockRowRef & other) const
-    {
-        size_t size = columns->size();
-        for (size_t i = 0; i < size; ++i)
-        {
-            int res = (*description)[i].direction * (*columns)[i]->compareAt(row_num, other.row_num, *(*other.columns)[i], 1);
-            if (res < 0)
-                return true;
-            else if (res > 0)
-                return false;
-        }
-        return false;
-    }
-
-    void set(SharedBlockPtr & shared_block_, ColumnRawPtrs * columns_, size_t row_num_, SortDescription * description_)
-    {
-        shared_block = shared_block_;
-        columns = columns_;
-        row_num = row_num_;
-        description = description_;
-    }
-};
-
 }
diff --git a/src/Interpreters/InterpreterSelectQuery.cpp b/src/Interpreters/InterpreterSelectQuery.cpp
index c3fce45ac61..c8d842dfa53 100644
--- a/src/Interpreters/InterpreterSelectQuery.cpp
+++ b/src/Interpreters/InterpreterSelectQuery.cpp
@@ -84,7 +84,6 @@
 #include <Processors/Transforms/MergingAggregatedMemoryEfficientTransform.h>
 #include <Processors/Transforms/TotalsHavingTransform.h>
 #include <Processors/Transforms/PartialSortingTransform.h>
-#include <Processors/Transforms/OptimizedPartialSortingTransform.h>
 #include <Processors/Transforms/LimitsCheckingTransform.h>
 #include <Processors/Transforms/MergeSortingTransform.h>
 #include <Processors/Transforms/DistinctTransform.h>
@@ -2109,7 +2108,7 @@ void InterpreterSelectQuery::executeOrder(QueryPipeline & pipeline, InputSorting
         if (stream_type != QueryPipeline::StreamType::Main)
             return nullptr;
 
-        return std::make_shared<OptimizedPartialSortingTransform>(header, output_order_descr, limit);
+        return std::make_shared<PartialSortingTransform>(header, output_order_descr, limit);
     });
 
     pipeline.addSimpleTransform([&](const Block & header, QueryPipeline::StreamType stream_type) -> ProcessorPtr
diff --git a/src/Processors/Transforms/OptimizedPartialSortingTransform.cpp b/src/Processors/Transforms/OptimizedPartialSortingTransform.cpp
deleted file mode 100644
index e753052a2b3..00000000000
--- a/src/Processors/Transforms/OptimizedPartialSortingTransform.cpp
+++ /dev/null
@@ -1,83 +0,0 @@
-#include <Processors/Transforms/OptimizedPartialSortingTransform.h>
-#include <Interpreters/sortBlock.h>
-#include <Common/PODArray.h>
-
-namespace DB
-{
-
-OptimizedPartialSortingTransform::OptimizedPartialSortingTransform(
-    const Block & header_, SortDescription & description_, UInt64 limit_)
-    : ISimpleTransform(header_, header_, false)
-    , description(description_), limit(limit_)
-    , threshold_shared_block(nullptr)
-{
-}
-
-static ColumnRawPtrs extractColumns(const Block & block, const SortDescription& description)
-{
-    size_t size = description.size();
-    ColumnRawPtrs res;
-    res.reserve(size);
-
-    for (size_t i = 0; i < size; ++i)
-    {
-        const IColumn * column = !description[i].column_name.empty()
-            ? block.getByName(description[i].column_name).column.get()
-            : block.safeGetByPosition(description[i].column_number).column.get();
-        res.emplace_back(column);
-    }
-
-    return res;
-}
-
-void OptimizedPartialSortingTransform::transform(Chunk & chunk)
-{
-    if (read_rows)
-        read_rows->add(chunk.getNumRows());
-
-    auto block = getInputPort().getHeader().cloneWithColumns(chunk.detachColumns());
-    chunk.clear();
-
-    SharedBlockPtr shared_block = new detail::SharedBlock(std::move(block));
-    UInt64 rows_num = shared_block->rows();
-
-
-    if (threshold_shared_block) {
-        SharedBlockRowWithSortDescriptionRef row;
-        IColumn::Filter filter(rows_num);
-        ColumnRawPtrs shared_block_columns = extractColumns(*shared_block, description);
-        size_t filtered_count = 0;
-
-        for (UInt64 i = 0; i < rows_num; ++i) {
-            row.set(shared_block, &shared_block_columns, i, &description);
-
-            if (threshold_row < row)
-            {
-                ++filtered_count;
-                filter[i] = 1;
-            }
-        }
-
-        if (filtered_count)
-        {
-            for (auto & column : shared_block->getColumns())
-            {
-                column = column->filter(filter, filtered_count);
-            }
-        }
-    }
-
-    sortBlock(*shared_block, description, limit);
-
-    if (!threshold_shared_block && limit && limit < rows_num)
-    {
-        Block threshold_block = shared_block->cloneWithColumns(shared_block->getColumns());
-        threshold_shared_block = new detail::SharedBlock(std::move(threshold_block));
-        threshold_block_columns = extractColumns(*threshold_shared_block, description);
-        threshold_row.set(threshold_shared_block, &threshold_block_columns, limit - 1, &description);
-    }
-
-    chunk.setColumns(shared_block->getColumns(), shared_block->rows());
-}
-
-}
diff --git a/src/Processors/Transforms/OptimizedPartialSortingTransform.h b/src/Processors/Transforms/OptimizedPartialSortingTransform.h
deleted file mode 100644
index 20e72bd836f..00000000000
--- a/src/Processors/Transforms/OptimizedPartialSortingTransform.h
+++ /dev/null
@@ -1,34 +0,0 @@
-#pragma once
-#include <Processors/ISimpleTransform.h>
-#include <Processors/RowsBeforeLimitCounter.h>
-#include <Common/SharedBlockRowRef.h>
-#include <Core/SortDescription.h>
-
-
-namespace DB
-{
-class OptimizedPartialSortingTransform : public ISimpleTransform
-{
-public:
-    OptimizedPartialSortingTransform(
-        const Block & header_,
-        SortDescription & description_,
-        UInt64 limit_ = 0);
-
-    String getName() const override { return "OptimizedPartialSortingTransform"; }
-
-    void setRowsBeforeLimitCounter(RowsBeforeLimitCounterPtr counter) { read_rows.swap(counter); }
-
-protected:
-    void transform(Chunk & chunk) override;
-
-private:
-    SortDescription description;
-    UInt64 limit;
-    RowsBeforeLimitCounterPtr read_rows;
-    SharedBlockRowWithSortDescriptionRef threshold_row;
-    SharedBlockPtr threshold_shared_block;
-    ColumnRawPtrs threshold_block_columns;
-};
-
-}
diff --git a/src/Processors/Transforms/PartialSortingTransform.cpp b/src/Processors/Transforms/PartialSortingTransform.cpp
index 018614f0165..062064c0fd6 100644
--- a/src/Processors/Transforms/PartialSortingTransform.cpp
+++ b/src/Processors/Transforms/PartialSortingTransform.cpp
@@ -1,5 +1,6 @@
 #include <Processors/Transforms/PartialSortingTransform.h>
 #include <Interpreters/sortBlock.h>
+#include <Common/PODArray.h>
 
 namespace DB
 {
@@ -11,6 +12,38 @@ PartialSortingTransform::PartialSortingTransform(
 {
 }
 
+static ColumnRawPtrs extractColumns(const Block & block, const SortDescription & description)
+{
+    size_t size = description.size();
+    ColumnRawPtrs res;
+    res.reserve(size);
+
+    for (size_t i = 0; i < size; ++i)
+    {
+        const IColumn * column = !description[i].column_name.empty()
+            ? block.getByName(description[i].column_name).column.get()
+            : block.safeGetByPosition(description[i].column_number).column.get();
+        res.emplace_back(column);
+    }
+
+    return res;
+}
+
+bool less(const ColumnRawPtrs & lhs, UInt64 lhs_row_num,
+          const ColumnRawPtrs & rhs, UInt64 rhs_row_num, const SortDescription & description)
+{
+    size_t size = description.size();
+    for (size_t i = 0; i < size; ++i)
+    {
+        int res = description[i].direction * lhs[i]->compareAt(lhs_row_num, rhs_row_num, *rhs[i], 1);
+        if (res < 0)
+            return true;
+        else if (res > 0)
+            return false;
+    }
+    return false;
+}
+
 void PartialSortingTransform::transform(Chunk & chunk)
 {
     if (read_rows)
@@ -19,7 +52,40 @@ void PartialSortingTransform::transform(Chunk & chunk)
     auto block = getInputPort().getHeader().cloneWithColumns(chunk.detachColumns());
     chunk.clear();
 
+    UInt64 rows_num = block.rows();
+
+    if (!threshold_block_columns.empty())
+    {
+        IColumn::Filter filter(rows_num, 0);
+        ColumnRawPtrs block_columns = extractColumns(block, description);
+        size_t filtered_count = 0;
+
+        for (UInt64 i = 0; i < rows_num; ++i)
+        {
+            if (less(threshold_block_columns, limit - 1, block_columns, i, description))
+            {
+                ++filtered_count;
+                filter[i] = 1;
+            }
+        }
+
+        if (filtered_count)
+        {
+            for (auto & column : block.getColumns())
+            {
+                column = column->filter(filter, filtered_count);
+            }
+        }
+    }
+
     sortBlock(block, description, limit);
+
+    if (threshold_block_columns.empty() && limit && limit < rows_num)
+    {
+        threshold_block = block.cloneWithColumns(block.getColumns());
+        threshold_block_columns = extractColumns(threshold_block, description);
+    }
+
     chunk.setColumns(block.getColumns(), block.rows());
 }
 
diff --git a/src/Processors/Transforms/PartialSortingTransform.h b/src/Processors/Transforms/PartialSortingTransform.h
index 47ac90c6904..d6749e4dfad 100644
--- a/src/Processors/Transforms/PartialSortingTransform.h
+++ b/src/Processors/Transforms/PartialSortingTransform.h
@@ -29,6 +29,8 @@ private:
     SortDescription description;
     UInt64 limit;
     RowsBeforeLimitCounterPtr read_rows;
+    Block threshold_block;
+    ColumnRawPtrs threshold_block_columns;
 };
 
 }

From a0f99fde700d702f29ad2411ee53571498900e41 Mon Sep 17 00:00:00 2001
From: Albert Kidrachev <kidrachev@sandbox-410109003>
Date: Wed, 27 May 2020 14:56:01 +0300
Subject: [PATCH 068/122] fix

---
 src/Common/SharedBlockRowRef.h | 1 -
 src/Processors/ya.make         | 1 -
 2 files changed, 2 deletions(-)

diff --git a/src/Common/SharedBlockRowRef.h b/src/Common/SharedBlockRowRef.h
index 957d66243dd..193f7e4dd05 100644
--- a/src/Common/SharedBlockRowRef.h
+++ b/src/Common/SharedBlockRowRef.h
@@ -2,7 +2,6 @@
 
 #include <algorithm>
 #include <Core/Block.h>
-#include <Core/SortDescription.h>
 #include <Columns/IColumn.h>
 #include <boost/smart_ptr/intrusive_ptr.hpp>
 
diff --git a/src/Processors/ya.make b/src/Processors/ya.make
index 5952341527b..62320f1c147 100644
--- a/src/Processors/ya.make
+++ b/src/Processors/ya.make
@@ -129,7 +129,6 @@ SRCS(
     Transforms/MergeSortingTransform.cpp
     Transforms/MergingAggregatedMemoryEfficientTransform.cpp
     Transforms/MergingAggregatedTransform.cpp
-    Transforms/OptimizedPartialSortingTransform.cpp
     Transforms/PartialSortingTransform.cpp
     Transforms/ReverseTransform.cpp
     Transforms/RollupTransform.cpp

From 81a5af1938afbd6168500b02579b259393b8a987 Mon Sep 17 00:00:00 2001
From: Alexander Kuzmenkov <36882414+akuzm@users.noreply.github.com>
Date: Wed, 27 May 2020 15:41:45 +0300
Subject: [PATCH 069/122] Update
 src/Processors/Formats/Impl/ArrowBlockInputFormat.h

---
 src/Processors/Formats/Impl/ArrowBlockInputFormat.h | 1 +
 1 file changed, 1 insertion(+)

diff --git a/src/Processors/Formats/Impl/ArrowBlockInputFormat.h b/src/Processors/Formats/Impl/ArrowBlockInputFormat.h
index 5ad112efde9..7599182bb3a 100644
--- a/src/Processors/Formats/Impl/ArrowBlockInputFormat.h
+++ b/src/Processors/Formats/Impl/ArrowBlockInputFormat.h
@@ -25,6 +25,7 @@ protected:
     Chunk generate() override;
 
 private:
+    // Whether to use ArrowStream format
     // Whether to use ArrowStream format
     bool stream;
     // This field is only used for ArrowStream format

From 61f833efc11f6dfb109343e86721327bf83cd742 Mon Sep 17 00:00:00 2001
From: Alexander Kuzmenkov <36882414+akuzm@users.noreply.github.com>
Date: Wed, 27 May 2020 15:41:52 +0300
Subject: [PATCH 070/122] Update
 src/Processors/Formats/Impl/ArrowBlockInputFormat.h

---
 src/Processors/Formats/Impl/ArrowBlockInputFormat.h | 1 +
 1 file changed, 1 insertion(+)

diff --git a/src/Processors/Formats/Impl/ArrowBlockInputFormat.h b/src/Processors/Formats/Impl/ArrowBlockInputFormat.h
index 7599182bb3a..92cced2ead8 100644
--- a/src/Processors/Formats/Impl/ArrowBlockInputFormat.h
+++ b/src/Processors/Formats/Impl/ArrowBlockInputFormat.h
@@ -31,6 +31,7 @@ private:
     // This field is only used for ArrowStream format
     std::shared_ptr<arrow::RecordBatchReader> stream_reader;
     // The following fields are used only for Arrow format
+    // The following fields are used only for Arrow format
     std::shared_ptr<arrow::ipc::RecordBatchFileReader> file_reader;
 
     int record_batch_total = 0;

From f54c892b49a3f1f52135d0ba9008cf2641f19898 Mon Sep 17 00:00:00 2001
From: Alexander Kuzmenkov <36882414+akuzm@users.noreply.github.com>
Date: Wed, 27 May 2020 15:41:59 +0300
Subject: [PATCH 071/122] Update
 src/Processors/Formats/Impl/ArrowBlockInputFormat.h

---
 src/Processors/Formats/Impl/ArrowBlockInputFormat.h | 1 +
 1 file changed, 1 insertion(+)

diff --git a/src/Processors/Formats/Impl/ArrowBlockInputFormat.h b/src/Processors/Formats/Impl/ArrowBlockInputFormat.h
index 92cced2ead8..b1ee21a51e4 100644
--- a/src/Processors/Formats/Impl/ArrowBlockInputFormat.h
+++ b/src/Processors/Formats/Impl/ArrowBlockInputFormat.h
@@ -29,6 +29,7 @@ private:
     // Whether to use ArrowStream format
     bool stream;
     // This field is only used for ArrowStream format
+    // This field is only used for ArrowStream format
     std::shared_ptr<arrow::RecordBatchReader> stream_reader;
     // The following fields are used only for Arrow format
     // The following fields are used only for Arrow format

From 935524096545a1d0ed7dc7980609c2b98dddfc97 Mon Sep 17 00:00:00 2001
From: alesapin <alesapin@gmail.com>
Date: Wed, 27 May 2020 15:43:25 +0300
Subject: [PATCH 072/122] Better path detection

---
 contrib/cctz-cmake/CMakeLists.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/contrib/cctz-cmake/CMakeLists.txt b/contrib/cctz-cmake/CMakeLists.txt
index 2c44b25b17a..0837a366f20 100644
--- a/contrib/cctz-cmake/CMakeLists.txt
+++ b/contrib/cctz-cmake/CMakeLists.txt
@@ -623,7 +623,7 @@ if (USE_INTERNAL_CCTZ)
         # libraries in linker command. To avoid this we hardcode whole-archive
         # library into single string.
         add_dependencies(cctz tzdata)
-        target_link_libraries(cctz INTERFACE "-Wl,--whole-archive contrib/cctz-cmake/libtzdata.a -Wl,--no-whole-archive")
+        target_link_libraries(cctz INTERFACE "-Wl,--whole-archive $<TARGET_FILE:tzdata> -Wl,--no-whole-archive")
     endif ()
 
 else ()

From c1cfc68cd611bf5aa7f625185867aec18dfa02b4 Mon Sep 17 00:00:00 2001
From: Alexander Kuzmenkov <akuzm@yandex-team.ru>
Date: Wed, 27 May 2020 15:46:49 +0300
Subject: [PATCH 073/122] revert accidental changes

---
 src/Processors/Formats/Impl/ArrowBlockInputFormat.h | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/src/Processors/Formats/Impl/ArrowBlockInputFormat.h b/src/Processors/Formats/Impl/ArrowBlockInputFormat.h
index b1ee21a51e4..5ad112efde9 100644
--- a/src/Processors/Formats/Impl/ArrowBlockInputFormat.h
+++ b/src/Processors/Formats/Impl/ArrowBlockInputFormat.h
@@ -25,14 +25,11 @@ protected:
     Chunk generate() override;
 
 private:
-    // Whether to use ArrowStream format
     // Whether to use ArrowStream format
     bool stream;
     // This field is only used for ArrowStream format
-    // This field is only used for ArrowStream format
     std::shared_ptr<arrow::RecordBatchReader> stream_reader;
     // The following fields are used only for Arrow format
-    // The following fields are used only for Arrow format
     std::shared_ptr<arrow::ipc::RecordBatchFileReader> file_reader;
 
     int record_batch_total = 0;

From ad936442547d69ef3d7cf39c708c0f2f78072110 Mon Sep 17 00:00:00 2001
From: Alexander Kuzmenkov <akuzm@yandex-team.ru>
Date: Wed, 27 May 2020 15:50:12 +0300
Subject: [PATCH 074/122] make single if

---
 src/Processors/Formats/Impl/ArrowBlockInputFormat.cpp | 10 +++-------
 1 file changed, 3 insertions(+), 7 deletions(-)

diff --git a/src/Processors/Formats/Impl/ArrowBlockInputFormat.cpp b/src/Processors/Formats/Impl/ArrowBlockInputFormat.cpp
index e0eb0e21567..2873a5417ea 100644
--- a/src/Processors/Formats/Impl/ArrowBlockInputFormat.cpp
+++ b/src/Processors/Formats/Impl/ArrowBlockInputFormat.cpp
@@ -30,13 +30,6 @@ Chunk ArrowBlockInputFormat::generate()
 {
     Chunk res;
     const Block & header = getPort().getHeader();
-
-    if (!stream)
-    {
-        if (record_batch_current >= record_batch_total)
-            return res;
-    }
-
     std::vector<std::shared_ptr<arrow::RecordBatch>> single_batch(1);
     arrow::Status read_status;
 
@@ -48,6 +41,9 @@ Chunk ArrowBlockInputFormat::generate()
     }
     else
     {
+        if (record_batch_current >= record_batch_total)
+            return res;
+
         read_status = file_reader->ReadRecordBatch(record_batch_current, &single_batch[0]);
     }
 

From a727e1fb1b0a0d51caa2c578e0dfdd30dce18140 Mon Sep 17 00:00:00 2001
From: Albert Kidrachev <kidrachev@sandbox-410109003>
Date: Wed, 27 May 2020 15:57:14 +0300
Subject: [PATCH 075/122] more relaxation

---
 src/Processors/Transforms/PartialSortingTransform.cpp | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/src/Processors/Transforms/PartialSortingTransform.cpp b/src/Processors/Transforms/PartialSortingTransform.cpp
index 062064c0fd6..ad45159f77e 100644
--- a/src/Processors/Transforms/PartialSortingTransform.cpp
+++ b/src/Processors/Transforms/PartialSortingTransform.cpp
@@ -52,12 +52,13 @@ void PartialSortingTransform::transform(Chunk & chunk)
     auto block = getInputPort().getHeader().cloneWithColumns(chunk.detachColumns());
     chunk.clear();
 
+    ColumnRawPtrs block_columns;
     UInt64 rows_num = block.rows();
 
     if (!threshold_block_columns.empty())
     {
         IColumn::Filter filter(rows_num, 0);
-        ColumnRawPtrs block_columns = extractColumns(block, description);
+        block_columns = extractColumns(block, description);
         size_t filtered_count = 0;
 
         for (UInt64 i = 0; i < rows_num; ++i)
@@ -80,7 +81,8 @@ void PartialSortingTransform::transform(Chunk & chunk)
 
     sortBlock(block, description, limit);
 
-    if (threshold_block_columns.empty() && limit && limit < rows_num)
+    if (limit && limit < rows_num &&
+        (threshold_block_columns.empty() || less(block_columns, limit - 1, threshold_block_columns, limit - 1, description)))
     {
         threshold_block = block.cloneWithColumns(block.getColumns());
         threshold_block_columns = extractColumns(threshold_block, description);

From c34f1ed6fdbfaffca28aa2d61d8b3271fea9f0a8 Mon Sep 17 00:00:00 2001
From: BayoNet <da-daos@yandex.ru>
Date: Wed, 27 May 2020 17:13:37 +0300
Subject: [PATCH 076/122] DOCS-631: Settings partial_merge_join_optimizations,
 partial_merge_join_rows_in_right_blocks (#11130)

* CLICKHOUSEDOCS-631: Settings partial_merge_join_optimizations, partial_merge_join_rows_in_right_blocks

* CLICKHOUSEDOCS-631: Updated by comments.

* CLICKHOUSEDOCS-631: Fixed some grammar.

Co-authored-by: Sergei Shtykov <bayonet@yandex-team.ru>
---
 docs/en/operations/settings/settings.md | 29 +++++++++++++++++++++++++
 1 file changed, 29 insertions(+)

diff --git a/docs/en/operations/settings/settings.md b/docs/en/operations/settings/settings.md
index 909ba7bf876..7103819e499 100644
--- a/docs/en/operations/settings/settings.md
+++ b/docs/en/operations/settings/settings.md
@@ -404,6 +404,35 @@ Possible values:
 
 Default value: 0.
 
+## partial_merge_join_optimizations {#partial_merge_join_optimizations}
+
+Disables optimizations in partial merge join algorithm for [JOIN](../../sql-reference/statements/select/join.md) queries.
+
+By default, this setting enables improvements that could lead to wrong results. If you see suspicious results in your queries, disable optimizations by this setting. Optimizations can be different in different versions of the ClickHouse server. 
+
+Possible values:
+
+-   0 — Optimizations disabled.
+-   1 — Optimizations enabled.
+
+Default value: 1.
+
+## partial_merge_join_rows_in_right_blocks {#partial_merge_join_rows_in_right_blocks}
+
+Limits sizes of right-hand join data blocks in partial merge join algorithm for [JOIN](../../sql-reference/statements/select/join.md) queries.
+
+ClickHouse server:
+
+1. Splits right-hand join data into blocks with up to the specified number of rows.
+2. Indexes each block with their minimum and maximum values
+3. Unloads prepared blocks to disk if possible.
+
+Possible values:
+
+- Any positive integer. Recommended range of values: [1000, 100000].
+
+Default value: 65536.
+
 ## any_join_distinct_right_table_keys {#any_join_distinct_right_table_keys}
 
 Enables legacy ClickHouse server behavior in `ANY INNER|LEFT JOIN` operations.

From 4fa3628097180dced023b35a4e83aef10beaa6f5 Mon Sep 17 00:00:00 2001
From: Artem Zuikov <chertus@gmail.com>
Date: Wed, 27 May 2020 17:25:11 +0300
Subject: [PATCH 077/122] test for probably fixed issue (#11221)

---
 .../01138_join_on_distributed_and_tmp.reference  |  0
 .../01138_join_on_distributed_and_tmp.sql        | 16 ++++++++++++++++
 2 files changed, 16 insertions(+)
 create mode 100644 tests/queries/0_stateless/01138_join_on_distributed_and_tmp.reference
 create mode 100644 tests/queries/0_stateless/01138_join_on_distributed_and_tmp.sql

diff --git a/tests/queries/0_stateless/01138_join_on_distributed_and_tmp.reference b/tests/queries/0_stateless/01138_join_on_distributed_and_tmp.reference
new file mode 100644
index 00000000000..e69de29bb2d
diff --git a/tests/queries/0_stateless/01138_join_on_distributed_and_tmp.sql b/tests/queries/0_stateless/01138_join_on_distributed_and_tmp.sql
new file mode 100644
index 00000000000..67492e7c683
--- /dev/null
+++ b/tests/queries/0_stateless/01138_join_on_distributed_and_tmp.sql
@@ -0,0 +1,16 @@
+DROP TABLE IF EXISTS foo_local;
+DROP TABLE IF EXISTS foo_distributed;
+
+CREATE TABLE foo_local (bar UInt64)
+ENGINE = MergeTree()
+ORDER BY tuple();
+
+CREATE TABLE foo_distributed AS foo_local
+ENGINE = Distributed('test_cluster_two_shards_localhost', currentDatabase(), foo_local);
+
+CREATE TEMPORARY TABLE _tmp_baz (qux UInt64);
+
+SELECT * FROM foo_distributed JOIN _tmp_baz ON foo_distributed.bar = _tmp_baz.qux;
+
+DROP TABLE foo_local;
+DROP TABLE foo_distributed;

From 1658705f7c82839906b9d9b40afde08f2d37f428 Mon Sep 17 00:00:00 2001
From: Alexander Kuzmenkov <36882414+akuzm@users.noreply.github.com>
Date: Wed, 27 May 2020 18:29:22 +0300
Subject: [PATCH 078/122] Add URL/port.cpp to ya.make (a follow-up for #11120)

---
 src/Functions/ya.make | 1 +
 1 file changed, 1 insertion(+)

diff --git a/src/Functions/ya.make b/src/Functions/ya.make
index ddb1d27b398..0b4776853e9 100644
--- a/src/Functions/ya.make
+++ b/src/Functions/ya.make
@@ -425,6 +425,7 @@ SRCS(
     URL/path.cpp
     URL/pathFull.cpp
     URL/protocol.cpp
+    URL/port.cpp
     URL/queryStringAndFragment.cpp
     URL/queryString.cpp
     URL/registerFunctionsURL.cpp

From 4c5324600b6c847009da083d185d97a9c425b0f6 Mon Sep 17 00:00:00 2001
From: Albert Kidrachev <kidrachev@sandbox-410109003>
Date: Wed, 27 May 2020 18:38:29 +0300
Subject: [PATCH 079/122] fix tests

---
 src/Processors/Transforms/PartialSortingTransform.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/Processors/Transforms/PartialSortingTransform.cpp b/src/Processors/Transforms/PartialSortingTransform.cpp
index ad45159f77e..f68a415117b 100644
--- a/src/Processors/Transforms/PartialSortingTransform.cpp
+++ b/src/Processors/Transforms/PartialSortingTransform.cpp
@@ -81,7 +81,7 @@ void PartialSortingTransform::transform(Chunk & chunk)
 
     sortBlock(block, description, limit);
 
-    if (limit && limit < rows_num &&
+    if (limit && limit < block.rows() &&
         (threshold_block_columns.empty() || less(block_columns, limit - 1, threshold_block_columns, limit - 1, description)))
     {
         threshold_block = block.cloneWithColumns(block.getColumns());

From 788e32822adc97ebb4538d335049c25174011b2f Mon Sep 17 00:00:00 2001
From: Michael Smitasin <46496263+michaelsmitasin@users.noreply.github.com>
Date: Wed, 27 May 2020 09:29:34 -0700
Subject: [PATCH 080/122] Added LBNL (#11229)

---
 docs/en/introduction/adopters.md | 1 +
 1 file changed, 1 insertion(+)

diff --git a/docs/en/introduction/adopters.md b/docs/en/introduction/adopters.md
index 8868522e977..4daadf32be6 100644
--- a/docs/en/introduction/adopters.md
+++ b/docs/en/introduction/adopters.md
@@ -41,6 +41,7 @@ toc_title: Adopters
 | [Integros](https://integros.com){.favicon}                                    | Platform for video services     | Analytics             | —                                                          | —                                                                            | [Slides in Russian, May 2019](https://github.com/ClickHouse/clickhouse-presentations/blob/master/meetup22/strategies.pdf)                                                                                               |
 | [Kodiak Data](https://www.kodiakdata.com/){.favicon}                          | Clouds                          | Main product          | —                                                          | —                                                                            | [Slides in Engish, April 2018](https://github.com/ClickHouse/clickhouse-presentations/blob/master/meetup13/kodiak_data.pdf)                                                                                             |
 | [Kontur](https://kontur.ru){.favicon}                                         | Software Development            | Metrics               | —                                                          | —                                                                            | [Talk in Russian, November 2018](https://www.youtube.com/watch?v=U4u4Bd0FtrY)                                                                                                                                           |
+| [Lawrence Berkeley National Laboratory](https://www.lbl.gov){.favicon}        | Research                        | Traffic analysis      | 1 server                                                   | 11.8 TiB                                                                     | [Slides in English, April 2019](https://www.smitasin.com/presentations/2019-04-17_DOE-NSM.pdf)                                                                                                                          |
 | [LifeStreet](https://lifestreet.com/){.favicon}                               | Ad network                      | Main product          | 75 servers (3 replicas)                                    | 5.27 PiB                                                                     | [Blog post in Russian, February 2017](https://habr.com/en/post/322620/)                                                                                                                                                 |
 | [Mail.ru Cloud Solutions](https://mcs.mail.ru/){.favicon}                     | Cloud services                  | Main product          | —                                                          | —                                                                            | [Article in Russian](https://mcs.mail.ru/help/db-create/clickhouse#)                                                                                                                               |
 | [MessageBird](https://www.messagebird.com){.favicon}                          | Telecommunications              | Statistics            | —                                                          | —                                                                            | [Slides in English, November 2018](https://github.com/ClickHouse/clickhouse-presentations/blob/master/meetup20/messagebird.pdf)                                                                                         |

From 7b03e36c2a6c390344ce855eac7b6dd95a6b9e08 Mon Sep 17 00:00:00 2001
From: Vitaly Baranov <vitbar@yandex-team.ru>
Date: Wed, 27 May 2020 17:03:38 +0300
Subject: [PATCH 081/122] Improve build scripts related to protobuf and gRPC a
 little more.

---
 cmake/protobuf_generate_cpp.cmake | 17 +++++++++++------
 1 file changed, 11 insertions(+), 6 deletions(-)

diff --git a/cmake/protobuf_generate_cpp.cmake b/cmake/protobuf_generate_cpp.cmake
index 40ec3c8e681..7ee9e8d7c81 100644
--- a/cmake/protobuf_generate_cpp.cmake
+++ b/cmake/protobuf_generate_cpp.cmake
@@ -112,16 +112,21 @@ if (PROTOBUF_GENERATE_CPP_SCRIPT_MODE)
     set (intermediate_dir ${DIR}/intermediate)
     set (intermediate_output "${intermediate_dir}/${FILENAME}")
 
-    if (COMPILER_ID STREQUAL "GNU")
+    if (COMPILER_ID STREQUAL "Clang")
+        set (pragma_push "#pragma clang diagnostic push\n")
+        set (pragma_pop "#pragma clang diagnostic pop\n")
+        set (pragma_disable_warnings "#pragma clang diagnostic ignored \"-Weverything\"\n")
+    elseif (COMPILER_ID MATCHES "GNU")
         set (pragma_push "#pragma GCC diagnostic push\n")
         set (pragma_pop "#pragma GCC diagnostic pop\n")
         set (pragma_disable_warnings "#pragma GCC diagnostic ignored \"-Wall\"\n"
                                      "#pragma GCC diagnostic ignored \"-Wextra\"\n"
-                                     "#pragma GCC diagnostic ignored \"-Warray-bounds\"\n")
-    elseif (COMPILER_ID MATCHES "Clang")
-        set (pragma_push "#pragma clang diagnostic push\n")
-        set (pragma_pop "#pragma clang diagnostic pop\n")
-        set (pragma_disable_warnings "#pragma clang diagnostic ignored \"-Weverything\"\n")
+                                     "#pragma GCC diagnostic ignored \"-Warray-bounds\"\n"
+                                     "#pragma GCC diagnostic ignored \"-Wold-style-cast\"\n"
+                                     "#pragma GCC diagnostic ignored \"-Wshadow\"\n"
+                                     "#pragma GCC diagnostic ignored \"-Wsuggest-override\"\n"
+                                     "#pragma GCC diagnostic ignored \"-Wcast-qual\"\n"
+                                     "#pragma GCC diagnostic ignored \"-Wunused-parameter\"\n")
     endif()
 
     if (${FILENAME} MATCHES ".*\\.h")

From fd64d391666a06fe737fde6be951185b5af0c74e Mon Sep 17 00:00:00 2001
From: Azat Khuzhin <a3at.mail@gmail.com>
Date: Wed, 27 May 2020 20:23:09 +0300
Subject: [PATCH 082/122] Fix style check for POPCNT check

---
 programs/main.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/programs/main.cpp b/programs/main.cpp
index b1bff1de1b0..746b624e065 100644
--- a/programs/main.cpp
+++ b/programs/main.cpp
@@ -198,7 +198,7 @@ void checkRequiredInstructionsImpl(volatile InstructionFail & fail)
     {
         uint64_t a = 0;
         uint64_t b = 0;
-        __asm__ volatile ("popcnt %1, %0" : "=r"(a) :"r"(b) : );
+        __asm__ volatile ("popcnt %1, %0" : "=r"(a) :"r"(b) :);
     }
 #endif
 

From bb0045a242e5f23908ddbaf84706fa10da6bcf0a Mon Sep 17 00:00:00 2001
From: Vitaly Baranov <vitbar@yandex-team.ru>
Date: Wed, 27 May 2020 16:27:51 +0300
Subject: [PATCH 083/122] Add test for row policy that policy defined in
 users.xml affects only the user it assigned to.

---
 tests/integration/test_row_policy/test.py | 20 ++++++++++++++++----
 1 file changed, 16 insertions(+), 4 deletions(-)

diff --git a/tests/integration/test_row_policy/test.py b/tests/integration/test_row_policy/test.py
index a1884d059c7..71496c6dbf2 100644
--- a/tests/integration/test_row_policy/test.py
+++ b/tests/integration/test_row_policy/test.py
@@ -42,8 +42,13 @@ def started_cluster():
 
                 CREATE TABLE mydb.`.filtered_table4` (a UInt8, b UInt8, c UInt16 ALIAS a + b) ENGINE MergeTree ORDER BY a;
                 INSERT INTO mydb.`.filtered_table4` values (0, 0), (0, 1), (1, 0), (1, 1);
+                
+                CREATE TABLE mydb.local (a UInt8, b UInt8) ENGINE MergeTree ORDER BY a;
             ''')
 
+        node.query("INSERT INTO mydb.local values (2, 0), (2, 1), (1, 0), (1, 1)")
+        node2.query("INSERT INTO mydb.local values (3, 0), (3, 1), (1, 0), (1, 1)")
+
         yield cluster
 
     finally:
@@ -122,6 +127,17 @@ def test_single_table_name():
     assert node.query("SELECT a + b = 1 FROM mydb.filtered_table3") == TSV([[1], [1]])
 
 
+def test_policy_from_users_xml_affects_only_user_assigned():
+    assert node.query("SELECT * FROM mydb.filtered_table1") == TSV([[1,0], [1, 1]])
+    assert node.query("SELECT * FROM mydb.filtered_table1", user="another") == TSV([[0, 0], [0, 1], [1, 0], [1, 1]])
+
+    assert node.query("SELECT * FROM mydb.filtered_table2") == TSV([[0, 0, 0, 0], [0, 0, 6, 0]])
+    assert node.query("SELECT * FROM mydb.filtered_table2", user="another") == TSV([[0, 0, 0, 0], [0, 0, 6, 0], [1, 2, 3, 4], [4, 3, 2, 1]])
+
+    assert node.query("SELECT * FROM mydb.local") == TSV([[1,0], [1, 1], [2, 0], [2, 1]])
+    assert node.query("SELECT * FROM mydb.local", user="another") == TSV([[1, 0], [1, 1]])
+
+
 def test_custom_table_name():
     copy_policy_xml('multiple_tags_with_table_names.xml')
     assert node.query("SELECT * FROM mydb.table") == TSV([[1, 0], [1, 1]])
@@ -286,9 +302,5 @@ def test_miscellaneous_engines():
     # DistributedMergeTree
     node.query("DROP TABLE IF EXISTS mydb.not_filtered_table")
     node.query("CREATE TABLE mydb.not_filtered_table (a UInt8, b UInt8) ENGINE Distributed('test_local_cluster', mydb, local)")
-    node.query("CREATE TABLE mydb.local (a UInt8, b UInt8) ENGINE MergeTree ORDER BY a")
-    node2.query("CREATE TABLE mydb.local (a UInt8, b UInt8) ENGINE MergeTree ORDER BY a")
-    node.query("INSERT INTO mydb.local values (2, 0), (2, 1), (1, 0), (1, 1)")
-    node2.query("INSERT INTO mydb.local values (3, 0), (3, 1), (1, 0), (1, 1)")
     assert node.query("SELECT * FROM mydb.not_filtered_table", user="another") == TSV([[1, 0], [1, 1], [1, 0], [1, 1]])
     assert node.query("SELECT sum(a), b FROM mydb.not_filtered_table GROUP BY b ORDER BY b", user="another") == TSV([[2, 0], [2, 1]])

From 5b858de374b1de5d4c1a53d64142c70dbf4841ec Mon Sep 17 00:00:00 2001
From: Ivan <5627721+abyss7@users.noreply.github.com>
Date: Wed, 27 May 2020 20:52:52 +0300
Subject: [PATCH 084/122] Split programs/server into actual program and library
 (#11186)

* Split programs/server into actual program and library
---
 programs/server/CMakeLists.txt                | 19 ++--------------
 programs/server/Server.cpp                    |  6 ++---
 programs/server/Server.h                      |  2 +-
 programs/server/ya.make                       | 11 ----------
 src/CMakeLists.txt                            |  2 ++
 src/Server/CMakeLists.txt                     |  0
 .../server => src/Server}/HTTPHandler.cpp     |  0
 {programs/server => src/Server}/HTTPHandler.h |  3 ++-
 .../Server}/HTTPHandlerFactory.cpp            |  1 +
 .../Server}/HTTPHandlerFactory.h              |  0
 .../Server}/HTTPHandlerRequestFilter.h        |  1 +
 {programs/server => src/Server}/IServer.h     | 14 +++++++++---
 .../Server}/InterserverIOHTTPHandler.cpp      |  1 +
 .../Server}/InterserverIOHTTPHandler.h        |  0
 .../server => src/Server}/MySQLHandler.cpp    |  0
 .../server => src/Server}/MySQLHandler.h      |  0
 .../Server}/MySQLHandlerFactory.cpp           |  3 +--
 .../Server}/MySQLHandlerFactory.h             |  3 ++-
 .../server => src/Server}/NotFoundHandler.cpp |  0
 .../server => src/Server}/NotFoundHandler.h   |  0
 .../Server}/PrometheusMetricsWriter.cpp       |  0
 .../Server}/PrometheusMetricsWriter.h         |  0
 .../Server}/PrometheusRequestHandler.cpp      |  1 +
 .../Server}/PrometheusRequestHandler.h        |  0
 .../Server}/ReplicasStatusHandler.cpp         |  0
 .../Server}/ReplicasStatusHandler.h           |  0
 .../Server}/StaticRequestHandler.cpp          |  0
 .../Server}/StaticRequestHandler.h            |  2 ++
 .../server => src/Server}/TCPHandler.cpp      |  0
 {programs/server => src/Server}/TCPHandler.h  |  0
 .../server => src/Server}/TCPHandlerFactory.h |  4 ++--
 src/Server/ya.make                            | 22 +++++++++++++++++++
 src/ya.make                                   |  1 +
 33 files changed, 55 insertions(+), 41 deletions(-)
 create mode 100644 src/Server/CMakeLists.txt
 rename {programs/server => src/Server}/HTTPHandler.cpp (100%)
 rename {programs/server => src/Server}/HTTPHandler.h (97%)
 rename {programs/server => src/Server}/HTTPHandlerFactory.cpp (99%)
 rename {programs/server => src/Server}/HTTPHandlerFactory.h (100%)
 rename {programs/server => src/Server}/HTTPHandlerRequestFilter.h (99%)
 rename {programs/server => src/Server}/IServer.h (82%)
 rename {programs/server => src/Server}/InterserverIOHTTPHandler.cpp (99%)
 rename {programs/server => src/Server}/InterserverIOHTTPHandler.h (100%)
 rename {programs/server => src/Server}/MySQLHandler.cpp (100%)
 rename {programs/server => src/Server}/MySQLHandler.h (100%)
 rename {programs/server => src/Server}/MySQLHandlerFactory.cpp (99%)
 rename {programs/server => src/Server}/MySQLHandlerFactory.h (95%)
 rename {programs/server => src/Server}/NotFoundHandler.cpp (100%)
 rename {programs/server => src/Server}/NotFoundHandler.h (100%)
 rename {programs/server => src/Server}/PrometheusMetricsWriter.cpp (100%)
 rename {programs/server => src/Server}/PrometheusMetricsWriter.h (100%)
 rename {programs/server => src/Server}/PrometheusRequestHandler.cpp (95%)
 rename {programs/server => src/Server}/PrometheusRequestHandler.h (100%)
 rename {programs/server => src/Server}/ReplicasStatusHandler.cpp (100%)
 rename {programs/server => src/Server}/ReplicasStatusHandler.h (100%)
 rename {programs/server => src/Server}/StaticRequestHandler.cpp (100%)
 rename {programs/server => src/Server}/StaticRequestHandler.h (93%)
 rename {programs/server => src/Server}/TCPHandler.cpp (100%)
 rename {programs/server => src/Server}/TCPHandler.h (100%)
 rename {programs/server => src/Server}/TCPHandlerFactory.h (95%)
 create mode 100644 src/Server/ya.make

diff --git a/programs/server/CMakeLists.txt b/programs/server/CMakeLists.txt
index 026bb0bfeb2..1563f5ac51e 100644
--- a/programs/server/CMakeLists.txt
+++ b/programs/server/CMakeLists.txt
@@ -1,21 +1,6 @@
 set(CLICKHOUSE_SERVER_SOURCES
-    ${CMAKE_CURRENT_SOURCE_DIR}/HTTPHandler.cpp
-    ${CMAKE_CURRENT_SOURCE_DIR}/HTTPHandlerFactory.cpp
-    ${CMAKE_CURRENT_SOURCE_DIR}/InterserverIOHTTPHandler.cpp
-    ${CMAKE_CURRENT_SOURCE_DIR}/MetricsTransmitter.cpp
-    ${CMAKE_CURRENT_SOURCE_DIR}/NotFoundHandler.cpp
-    ${CMAKE_CURRENT_SOURCE_DIR}/PrometheusMetricsWriter.cpp
-    ${CMAKE_CURRENT_SOURCE_DIR}/PrometheusRequestHandler.cpp
-    ${CMAKE_CURRENT_SOURCE_DIR}/ReplicasStatusHandler.cpp
-    ${CMAKE_CURRENT_SOURCE_DIR}/StaticRequestHandler.cpp
-    ${CMAKE_CURRENT_SOURCE_DIR}/Server.cpp
-    ${CMAKE_CURRENT_SOURCE_DIR}/TCPHandler.cpp
-)
-
-set(CLICKHOUSE_SERVER_SOURCES
-    ${CLICKHOUSE_SERVER_SOURCES}
-    ${CMAKE_CURRENT_SOURCE_DIR}/MySQLHandler.cpp
-    ${CMAKE_CURRENT_SOURCE_DIR}/MySQLHandlerFactory.cpp
+    MetricsTransmitter.cpp
+    Server.cpp
 )
 
 set (CLICKHOUSE_SERVER_LINK
diff --git a/programs/server/Server.cpp b/programs/server/Server.cpp
index c1a520030f4..ce1d35e65d4 100644
--- a/programs/server/Server.cpp
+++ b/programs/server/Server.cpp
@@ -53,13 +53,13 @@
 #include <Dictionaries/registerDictionaries.h>
 #include <Disks/registerDisks.h>
 #include <Common/Config/ConfigReloader.h>
-#include "HTTPHandlerFactory.h"
+#include <Server/HTTPHandlerFactory.h>
 #include "MetricsTransmitter.h"
 #include <Common/StatusFile.h>
-#include "TCPHandlerFactory.h"
+#include <Server/TCPHandlerFactory.h>
 #include <Common/SensitiveDataMasker.h>
 #include <Common/ThreadFuzzer.h>
-#include "MySQLHandlerFactory.h"
+#include <Server/MySQLHandlerFactory.h>
 
 #if !defined(ARCADIA_BUILD)
 #   include "config_core.h"
diff --git a/programs/server/Server.h b/programs/server/Server.h
index ffd89df6af4..ad9e51c881c 100644
--- a/programs/server/Server.h
+++ b/programs/server/Server.h
@@ -1,6 +1,6 @@
 #pragma once
 
-#include "IServer.h"
+#include <Server/IServer.h>
 
 #include <daemon/BaseDaemon.h>
 
diff --git a/programs/server/ya.make b/programs/server/ya.make
index 2c74c01c7cb..2e13267f715 100644
--- a/programs/server/ya.make
+++ b/programs/server/ya.make
@@ -11,19 +11,8 @@ PEERDIR(
 SRCS(
     clickhouse-server.cpp
 
-    HTTPHandler.cpp
-    HTTPHandlerFactory.cpp
-    InterserverIOHTTPHandler.cpp
     MetricsTransmitter.cpp
-    MySQLHandler.cpp
-    MySQLHandlerFactory.cpp
-    NotFoundHandler.cpp
-    PrometheusMetricsWriter.cpp
-    PrometheusRequestHandler.cpp
-    ReplicasStatusHandler.cpp
-    StaticRequestHandler.cpp
     Server.cpp
-    TCPHandler.cpp
 )
 
 END()
diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
index 222a3e486f9..baa0fbcb883 100644
--- a/src/CMakeLists.txt
+++ b/src/CMakeLists.txt
@@ -58,6 +58,7 @@ add_subdirectory (TableFunctions)
 add_subdirectory (Processors)
 add_subdirectory (Formats)
 add_subdirectory (Compression)
+add_subdirectory (Server)
 
 
 set(dbms_headers)
@@ -145,6 +146,7 @@ add_object_library(clickhouse_storages_distributed Storages/Distributed)
 add_object_library(clickhouse_storages_mergetree Storages/MergeTree)
 add_object_library(clickhouse_storages_liveview Storages/LiveView)
 add_object_library(clickhouse_client Client)
+add_object_library(clickhouse_server Server)
 add_object_library(clickhouse_formats Formats)
 add_object_library(clickhouse_processors Processors)
 add_object_library(clickhouse_processors_executors Processors/Executors)
diff --git a/src/Server/CMakeLists.txt b/src/Server/CMakeLists.txt
new file mode 100644
index 00000000000..e69de29bb2d
diff --git a/programs/server/HTTPHandler.cpp b/src/Server/HTTPHandler.cpp
similarity index 100%
rename from programs/server/HTTPHandler.cpp
rename to src/Server/HTTPHandler.cpp
diff --git a/programs/server/HTTPHandler.h b/src/Server/HTTPHandler.h
similarity index 97%
rename from programs/server/HTTPHandler.h
rename to src/Server/HTTPHandler.h
index 6228523d343..b1a6355d281 100644
--- a/programs/server/HTTPHandler.h
+++ b/src/Server/HTTPHandler.h
@@ -6,6 +6,7 @@
 
 #include <Common/CurrentMetrics.h>
 #include <Common/HTMLForm.h>
+#include <Core/Names.h>
 
 #include <re2/re2.h>
 
@@ -21,7 +22,7 @@ namespace DB
 
 class WriteBufferFromHTTPServerResponse;
 
-typedef std::shared_ptr<const re2::RE2> CompiledRegexPtr;
+using CompiledRegexPtr = std::shared_ptr<const re2::RE2>;
 
 class HTTPHandler : public Poco::Net::HTTPRequestHandler
 {
diff --git a/programs/server/HTTPHandlerFactory.cpp b/src/Server/HTTPHandlerFactory.cpp
similarity index 99%
rename from programs/server/HTTPHandlerFactory.cpp
rename to src/Server/HTTPHandlerFactory.cpp
index f302216e22b..e916070be22 100644
--- a/programs/server/HTTPHandlerFactory.cpp
+++ b/src/Server/HTTPHandlerFactory.cpp
@@ -4,6 +4,7 @@
 #include <re2/stringpiece.h>
 #include <common/find_symbols.h>
 #include <Poco/StringTokenizer.h>
+#include <Poco/Util/LayeredConfiguration.h>
 
 #include "HTTPHandler.h"
 #include "NotFoundHandler.h"
diff --git a/programs/server/HTTPHandlerFactory.h b/src/Server/HTTPHandlerFactory.h
similarity index 100%
rename from programs/server/HTTPHandlerFactory.h
rename to src/Server/HTTPHandlerFactory.h
diff --git a/programs/server/HTTPHandlerRequestFilter.h b/src/Server/HTTPHandlerRequestFilter.h
similarity index 99%
rename from programs/server/HTTPHandlerRequestFilter.h
rename to src/Server/HTTPHandlerRequestFilter.h
index b0b748506e5..f952efd7653 100644
--- a/programs/server/HTTPHandlerRequestFilter.h
+++ b/src/Server/HTTPHandlerRequestFilter.h
@@ -6,6 +6,7 @@
 #include <re2/stringpiece.h>
 #include <Poco/StringTokenizer.h>
 #include <Poco/Net/HTTPServerRequest.h>
+#include <Poco/Util/LayeredConfiguration.h>
 
 #include <common/find_symbols.h>
 
diff --git a/programs/server/IServer.h b/src/Server/IServer.h
similarity index 82%
rename from programs/server/IServer.h
rename to src/Server/IServer.h
index 29e9bc16a75..131e7443646 100644
--- a/programs/server/IServer.h
+++ b/src/Server/IServer.h
@@ -1,14 +1,22 @@
 #pragma once
 
-#include <Poco/Logger.h>
-#include <Poco/Util/LayeredConfiguration.h>
+namespace Poco
+{
 
-#include <Interpreters/Context.h>
+namespace Util
+{
+class LayeredConfiguration;
+}
+class Logger;
+
+}
 
 
 namespace DB
 {
 
+class Context;
+
 class IServer
 {
 public:
diff --git a/programs/server/InterserverIOHTTPHandler.cpp b/src/Server/InterserverIOHTTPHandler.cpp
similarity index 99%
rename from programs/server/InterserverIOHTTPHandler.cpp
rename to src/Server/InterserverIOHTTPHandler.cpp
index 4b733c7f1fd..062721a01aa 100644
--- a/programs/server/InterserverIOHTTPHandler.cpp
+++ b/src/Server/InterserverIOHTTPHandler.cpp
@@ -3,6 +3,7 @@
 #include <Poco/Net/HTTPBasicCredentials.h>
 #include <Poco/Net/HTTPServerRequest.h>
 #include <Poco/Net/HTTPServerResponse.h>
+#include <Poco/Util/LayeredConfiguration.h>
 #include <common/logger_useful.h>
 #include <Common/HTMLForm.h>
 #include <Common/setThreadName.h>
diff --git a/programs/server/InterserverIOHTTPHandler.h b/src/Server/InterserverIOHTTPHandler.h
similarity index 100%
rename from programs/server/InterserverIOHTTPHandler.h
rename to src/Server/InterserverIOHTTPHandler.h
diff --git a/programs/server/MySQLHandler.cpp b/src/Server/MySQLHandler.cpp
similarity index 100%
rename from programs/server/MySQLHandler.cpp
rename to src/Server/MySQLHandler.cpp
diff --git a/programs/server/MySQLHandler.h b/src/Server/MySQLHandler.h
similarity index 100%
rename from programs/server/MySQLHandler.h
rename to src/Server/MySQLHandler.h
diff --git a/programs/server/MySQLHandlerFactory.cpp b/src/Server/MySQLHandlerFactory.cpp
similarity index 99%
rename from programs/server/MySQLHandlerFactory.cpp
rename to src/Server/MySQLHandlerFactory.cpp
index 022167fe766..5d78ed81068 100644
--- a/programs/server/MySQLHandlerFactory.cpp
+++ b/src/Server/MySQLHandlerFactory.cpp
@@ -4,8 +4,7 @@
 #include <Poco/Util/Application.h>
 #include <common/logger_useful.h>
 #include <ext/scope_guard.h>
-#include "IServer.h"
-#include "MySQLHandler.h"
+#include <Server/MySQLHandler.h>
 
 #if USE_SSL
 #    include <Poco/Net/SSLManager.h>
diff --git a/programs/server/MySQLHandlerFactory.h b/src/Server/MySQLHandlerFactory.h
similarity index 95%
rename from programs/server/MySQLHandlerFactory.h
rename to src/Server/MySQLHandlerFactory.h
index 74f0bb35a40..df7bd794b16 100644
--- a/programs/server/MySQLHandlerFactory.h
+++ b/src/Server/MySQLHandlerFactory.h
@@ -2,7 +2,8 @@
 
 #include <Poco/Net/TCPServerConnectionFactory.h>
 #include <atomic>
-#include "IServer.h"
+#include <memory>
+#include <Server/IServer.h>
 
 #if !defined(ARCADIA_BUILD)
 #    include <Common/config.h>
diff --git a/programs/server/NotFoundHandler.cpp b/src/Server/NotFoundHandler.cpp
similarity index 100%
rename from programs/server/NotFoundHandler.cpp
rename to src/Server/NotFoundHandler.cpp
diff --git a/programs/server/NotFoundHandler.h b/src/Server/NotFoundHandler.h
similarity index 100%
rename from programs/server/NotFoundHandler.h
rename to src/Server/NotFoundHandler.h
diff --git a/programs/server/PrometheusMetricsWriter.cpp b/src/Server/PrometheusMetricsWriter.cpp
similarity index 100%
rename from programs/server/PrometheusMetricsWriter.cpp
rename to src/Server/PrometheusMetricsWriter.cpp
diff --git a/programs/server/PrometheusMetricsWriter.h b/src/Server/PrometheusMetricsWriter.h
similarity index 100%
rename from programs/server/PrometheusMetricsWriter.h
rename to src/Server/PrometheusMetricsWriter.h
diff --git a/programs/server/PrometheusRequestHandler.cpp b/src/Server/PrometheusRequestHandler.cpp
similarity index 95%
rename from programs/server/PrometheusRequestHandler.cpp
rename to src/Server/PrometheusRequestHandler.cpp
index b5a48d13b64..43f39e36de8 100644
--- a/programs/server/PrometheusRequestHandler.cpp
+++ b/src/Server/PrometheusRequestHandler.cpp
@@ -6,6 +6,7 @@
 
 #include <Poco/Net/HTTPServerRequest.h>
 #include <Poco/Net/HTTPServerResponse.h>
+#include <Poco/Util/LayeredConfiguration.h>
 
 #include <Common/ProfileEvents.h>
 #include <Common/CurrentMetrics.h>
diff --git a/programs/server/PrometheusRequestHandler.h b/src/Server/PrometheusRequestHandler.h
similarity index 100%
rename from programs/server/PrometheusRequestHandler.h
rename to src/Server/PrometheusRequestHandler.h
diff --git a/programs/server/ReplicasStatusHandler.cpp b/src/Server/ReplicasStatusHandler.cpp
similarity index 100%
rename from programs/server/ReplicasStatusHandler.cpp
rename to src/Server/ReplicasStatusHandler.cpp
diff --git a/programs/server/ReplicasStatusHandler.h b/src/Server/ReplicasStatusHandler.h
similarity index 100%
rename from programs/server/ReplicasStatusHandler.h
rename to src/Server/ReplicasStatusHandler.h
diff --git a/programs/server/StaticRequestHandler.cpp b/src/Server/StaticRequestHandler.cpp
similarity index 100%
rename from programs/server/StaticRequestHandler.cpp
rename to src/Server/StaticRequestHandler.cpp
diff --git a/programs/server/StaticRequestHandler.h b/src/Server/StaticRequestHandler.h
similarity index 93%
rename from programs/server/StaticRequestHandler.h
rename to src/Server/StaticRequestHandler.h
index bdbc17f5e00..707087df24d 100644
--- a/programs/server/StaticRequestHandler.h
+++ b/src/Server/StaticRequestHandler.h
@@ -4,6 +4,8 @@
 
 #include <Poco/Net/HTTPRequestHandler.h>
 #include <Common/StringUtils/StringUtils.h>
+#include <Core/Types.h>
+#include <IO/WriteBuffer.h>
 
 
 namespace DB
diff --git a/programs/server/TCPHandler.cpp b/src/Server/TCPHandler.cpp
similarity index 100%
rename from programs/server/TCPHandler.cpp
rename to src/Server/TCPHandler.cpp
diff --git a/programs/server/TCPHandler.h b/src/Server/TCPHandler.h
similarity index 100%
rename from programs/server/TCPHandler.h
rename to src/Server/TCPHandler.h
diff --git a/programs/server/TCPHandlerFactory.h b/src/Server/TCPHandlerFactory.h
similarity index 95%
rename from programs/server/TCPHandlerFactory.h
rename to src/Server/TCPHandlerFactory.h
index 68652540192..a5532a8dc02 100644
--- a/programs/server/TCPHandlerFactory.h
+++ b/src/Server/TCPHandlerFactory.h
@@ -3,8 +3,8 @@
 #include <Poco/Net/TCPServerConnectionFactory.h>
 #include <Poco/Net/NetException.h>
 #include <common/logger_useful.h>
-#include "IServer.h"
-#include "TCPHandler.h"
+#include <Server/IServer.h>
+#include <Server/TCPHandler.h>
 
 namespace Poco { class Logger; }
 
diff --git a/src/Server/ya.make b/src/Server/ya.make
new file mode 100644
index 00000000000..1d689ee73b8
--- /dev/null
+++ b/src/Server/ya.make
@@ -0,0 +1,22 @@
+LIBRARY()
+
+PEERDIR(
+    clickhouse/src/Common
+    contrib/libs/poco/Util
+)
+
+SRCS(
+    HTTPHandler.cpp
+    HTTPHandlerFactory.cpp
+    InterserverIOHTTPHandler.cpp
+    MySQLHandler.cpp
+    MySQLHandlerFactory.cpp
+    NotFoundHandler.cpp
+    PrometheusMetricsWriter.cpp
+    PrometheusRequestHandler.cpp
+    ReplicasStatusHandler.cpp
+    StaticRequestHandler.cpp
+    TCPHandler.cpp
+)
+
+END()
diff --git a/src/ya.make b/src/ya.make
index 95e5914bbe7..eedd98cb178 100644
--- a/src/ya.make
+++ b/src/ya.make
@@ -19,6 +19,7 @@ PEERDIR(
     clickhouse/src/IO
     clickhouse/src/Parsers
     clickhouse/src/Processors
+    clickhouse/src/Server
     clickhouse/src/Storages
     clickhouse/src/TableFunctions
 )

From 0456302355d6eedc1c502ba60497b80941e75d91 Mon Sep 17 00:00:00 2001
From: Alexander Kuzmenkov <36882414+akuzm@users.noreply.github.com>
Date: Wed, 27 May 2020 20:58:05 +0300
Subject: [PATCH 085/122] Update formats.md

---
 docs/en/interfaces/formats.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/docs/en/interfaces/formats.md b/docs/en/interfaces/formats.md
index ed411a1b0d4..a08f2db7149 100644
--- a/docs/en/interfaces/formats.md
+++ b/docs/en/interfaces/formats.md
@@ -1143,13 +1143,13 @@ $ clickhouse-client --query="SELECT * FROM {some_table} FORMAT Parquet" > {some_
 
 To exchange data with Hadoop, you can use [HDFS table engine](../engines/table-engines/integrations/hdfs.md).
 
-## Arrow {data-format-arrow}
+## Arrow {#data-format-arrow}
 
 [Apache Arrow](https://arrow.apache.org/) comes with two built-in columnar storage formats. ClickHouse supports read and write operations for these formats.
 
 `Arrow` is Apache Arrow's "file mode" format. It is designed for in-memory random access.
 
-## ArrowStream {data-format-arrow-stream}
+## ArrowStream {#data-format-arrow-stream}
 
 `ArrowStream` is Apache Arrow's "stream mode" format. It is designed for in-memory stream processing.
 

From e93882c977ba53486e6bd06eac2d4b16cb7d3ecc Mon Sep 17 00:00:00 2001
From: Nikolai Kochetov <nik-kochetov@yandex-team.ru>
Date: Wed, 27 May 2020 21:20:26 +0300
Subject: [PATCH 086/122] Insert select using PipelineExecutor.

---
 programs/server/TCPHandler.cpp                |   4 +-
 .../NullAndDoCopyBlockInputStream.h           |  24 +---
 src/DataStreams/copyData.cpp                  |  81 ------------
 src/DataStreams/copyData.h                    |   2 -
 src/Interpreters/InterpreterInsertQuery.cpp   |  78 +++++++-----
 src/Interpreters/executeQuery.cpp             |  48 ++++---
 .../PullingAsyncPipelineExecutor.cpp          |  51 ++++++--
 .../Executors/PullingPipelineExecutor.cpp     |   2 +-
 src/Processors/QueryPipeline.cpp              | 117 +++++++++++++-----
 src/Processors/QueryPipeline.h                |  16 ++-
 10 files changed, 222 insertions(+), 201 deletions(-)

diff --git a/programs/server/TCPHandler.cpp b/programs/server/TCPHandler.cpp
index e64d7799dcb..21a4a389b9e 100644
--- a/programs/server/TCPHandler.cpp
+++ b/programs/server/TCPHandler.cpp
@@ -263,8 +263,8 @@ void TCPHandler::runImpl()
             else if (state.need_receive_data_for_input)
             {
                 /// It is special case for input(), all works for reading data from client will be done in callbacks.
-                /// state.io.in is NullAndDoCopyBlockInputStream so read it once.
-                state.io.in->read();
+                auto executor = state.io.pipeline.execute();
+                executor->execute(state.io.pipeline.getNumThreads());
                 state.io.onFinish();
             }
             else if (state.io.pipeline.initialized())
diff --git a/src/DataStreams/NullAndDoCopyBlockInputStream.h b/src/DataStreams/NullAndDoCopyBlockInputStream.h
index 8fe05c387a3..8bfb3538f3a 100644
--- a/src/DataStreams/NullAndDoCopyBlockInputStream.h
+++ b/src/DataStreams/NullAndDoCopyBlockInputStream.h
@@ -21,19 +21,10 @@ class NullAndDoCopyBlockInputStream : public IBlockInputStream
 {
 public:
     NullAndDoCopyBlockInputStream(const BlockInputStreamPtr & input_, BlockOutputStreamPtr output_)
+        : input(std::move(input_))
+        , output(std::move(output_))
     {
-        input_streams.push_back(input_);
-        output_streams.push_back(output_);
-
-        for (auto & input_stream : input_streams)
-            children.push_back(input_stream);
-    }
-
-    NullAndDoCopyBlockInputStream(const BlockInputStreams & input_, BlockOutputStreams & output_)
-        : input_streams(input_), output_streams(output_)
-    {
-        for (auto & input_stream : input_)
-            children.push_back(input_stream);
+        children.push_back(input);
     }
 
     /// Suppress readPrefix and readSuffix, because they are called by copyData.
@@ -53,16 +44,13 @@ protected:
         /// If query was cancelled, it will be processed by child streams.
         /// Part of the data will be processed.
 
-        if (input_streams.size() == 1 && output_streams.size() == 1)
-            copyData(*input_streams.at(0), *output_streams.at(0));
-        else
-            copyData(input_streams, output_streams);
+        copyData(*input, *output);
         return Block();
     }
 
 private:
-    BlockInputStreams input_streams;
-    BlockOutputStreams output_streams;
+    BlockInputStreamPtr input;
+    BlockOutputStreamPtr output;
 };
 
 }
diff --git a/src/DataStreams/copyData.cpp b/src/DataStreams/copyData.cpp
index fd4bfab28d8..a0651999034 100644
--- a/src/DataStreams/copyData.cpp
+++ b/src/DataStreams/copyData.cpp
@@ -1,9 +1,6 @@
-#include <thread>
 #include <DataStreams/IBlockInputStream.h>
 #include <DataStreams/IBlockOutputStream.h>
 #include <DataStreams/copyData.h>
-#include <DataStreams/ParallelInputsProcessor.h>
-#include <Common/ConcurrentBoundedQueue.h>
 #include <Common/ThreadPool.h>
 
 
@@ -55,79 +52,6 @@ void copyDataImpl(IBlockInputStream & from, IBlockOutputStream & to, TCancelCall
 
 inline void doNothing(const Block &) {}
 
-namespace
-{
-
-
-struct ParallelInsertsHandler
-{
-    using CencellationHook = std::function<void()>;
-
-    explicit ParallelInsertsHandler(BlockOutputStreams & output_streams, CencellationHook cancellation_hook_, size_t num_threads)
-        : outputs(output_streams.size()), cancellation_hook(std::move(cancellation_hook_))
-    {
-        exceptions.resize(num_threads);
-
-        for (auto & output : output_streams)
-            outputs.push(output.get());
-    }
-
-    void onBlock(Block & block, size_t /*thread_num*/)
-    {
-        IBlockOutputStream * out = nullptr;
-
-        outputs.pop(out);
-        out->write(block);
-        outputs.push(out);
-    }
-
-    void onFinishThread(size_t /*thread_num*/) {}
-    void onFinish() {}
-
-    void onException(std::exception_ptr & exception, size_t thread_num)
-    {
-        exceptions[thread_num] = exception;
-        cancellation_hook();
-    }
-
-    void rethrowFirstException()
-    {
-        for (auto & exception : exceptions)
-            if (exception)
-                std::rethrow_exception(exception);
-    }
-
-    ConcurrentBoundedQueue<IBlockOutputStream *> outputs;
-    std::vector<std::exception_ptr> exceptions;
-    CencellationHook cancellation_hook;
-};
-
-}
-
-static void copyDataImpl(BlockInputStreams & inputs, BlockOutputStreams & outputs)
-{
-    for (auto & output : outputs)
-        output->writePrefix();
-
-    using Processor = ParallelInputsProcessor<ParallelInsertsHandler>;
-    Processor * processor_ptr = nullptr;
-
-    ParallelInsertsHandler handler(outputs, [&processor_ptr]() { processor_ptr->cancel(false); }, inputs.size());
-    ParallelInputsProcessor<ParallelInsertsHandler> processor(inputs, nullptr, inputs.size(), handler);
-    processor_ptr = &processor;
-
-    processor.process();
-    processor.wait();
-    handler.rethrowFirstException();
-
-    /// readPrefix is called in ParallelInputsProcessor.
-    for (auto & input : inputs)
-        input->readSuffix();
-
-    for (auto & output : outputs)
-        output->writeSuffix();
-}
-
 void copyData(IBlockInputStream & from, IBlockOutputStream & to, std::atomic<bool> * is_cancelled)
 {
     auto is_cancelled_pred = [is_cancelled] ()
@@ -138,11 +62,6 @@ void copyData(IBlockInputStream & from, IBlockOutputStream & to, std::atomic<boo
     copyDataImpl(from, to, is_cancelled_pred, doNothing);
 }
 
-void copyData(BlockInputStreams & inputs, BlockOutputStreams & outputs)
-{
-    copyDataImpl(inputs, outputs);
-}
-
 void copyData(IBlockInputStream & from, IBlockOutputStream & to, const std::function<bool()> & is_cancelled)
 {
     copyDataImpl(from, to, is_cancelled, doNothing);
diff --git a/src/DataStreams/copyData.h b/src/DataStreams/copyData.h
index ae72dbd2421..f2bce8f411b 100644
--- a/src/DataStreams/copyData.h
+++ b/src/DataStreams/copyData.h
@@ -16,8 +16,6 @@ class Block;
   */
 void copyData(IBlockInputStream & from, IBlockOutputStream & to, std::atomic<bool> * is_cancelled = nullptr);
 
-void copyData(BlockInputStreams & inputs, BlockOutputStreams & outputs);
-
 void copyData(IBlockInputStream & from, IBlockOutputStream & to, const std::function<bool()> & is_cancelled);
 
 void copyData(IBlockInputStream & from, IBlockOutputStream & to, const std::function<bool()> & is_cancelled,
diff --git a/src/Interpreters/InterpreterInsertQuery.cpp b/src/Interpreters/InterpreterInsertQuery.cpp
index f3b116e490c..b6efa5d6d46 100644
--- a/src/Interpreters/InterpreterInsertQuery.cpp
+++ b/src/Interpreters/InterpreterInsertQuery.cpp
@@ -28,6 +28,11 @@
 #include <Storages/StorageDistributed.h>
 #include <TableFunctions/TableFunctionFactory.h>
 #include <Common/checkStackSize.h>
+#include <Processors/Sources/SourceFromInputStream.h>
+#include <Processors/NullSink.h>
+#include <Processors/Transforms/ConvertingTransform.h>
+#include <Processors/Sources/SinkToOutputStream.h>
+#include <Processors/ConcatProcessor.h>
 
 
 namespace DB
@@ -117,8 +122,6 @@ BlockIO InterpreterInsertQuery::execute()
     if (!query.table_function)
         context.checkAccess(AccessType::INSERT, query.table_id, query_sample_block.getNames());
 
-    BlockInputStreams in_streams;
-    BlockOutputStreams out_streams;
     bool is_distributed_insert_select = false;
 
     if (query.select && table->isRemote() && settings.parallel_distributed_insert_select)
@@ -159,6 +162,8 @@ BlockIO InterpreterInsertQuery::execute()
             const auto & cluster = storage_src->getCluster();
             const auto & shards_info = cluster->getShardsInfo();
 
+            std::vector<QueryPipeline> pipelines;
+
             String new_query_str = queryToString(new_query);
             for (size_t shard_index : ext::range(0, shards_info.size()))
             {
@@ -166,8 +171,7 @@ BlockIO InterpreterInsertQuery::execute()
                 if (shard_info.isLocal())
                 {
                     InterpreterInsertQuery interpreter(new_query, context);
-                    auto block_io = interpreter.execute();
-                    in_streams.push_back(block_io.in);
+                    pipelines.emplace_back(interpreter.execute().pipeline);
                 }
                 else
                 {
@@ -179,13 +183,20 @@ BlockIO InterpreterInsertQuery::execute()
 
                     ///  INSERT SELECT query returns empty block
                     auto in_stream = std::make_shared<RemoteBlockInputStream>(std::move(connections), new_query_str, Block{}, context);
-                    in_streams.push_back(in_stream);
+                    pipelines.emplace_back();
+                    pipelines.back().init(Pipe(std::make_shared<SourceFromInputStream>(std::move(in_stream))));
+                    pipelines.back().setSinks([](const Block & header, QueryPipeline::StreamType) -> ProcessorPtr
+                    {
+                        return std::make_shared<EmptySink>(header);
+                    });
                 }
-                out_streams.push_back(std::make_shared<NullBlockOutputStream>(Block()));
             }
+
+            res.pipeline.unitePipelines(std::move(pipelines), {});
         }
     }
 
+    BlockOutputStreams out_streams;
     if (!is_distributed_insert_select || query.watch)
     {
         size_t out_streams_size = 1;
@@ -193,27 +204,21 @@ BlockIO InterpreterInsertQuery::execute()
         {
             /// Passing 1 as subquery_depth will disable limiting size of intermediate result.
             InterpreterSelectWithUnionQuery interpreter_select{ query.select, context, SelectQueryOptions(QueryProcessingStage::Complete, 1)};
+            res.pipeline = interpreter_select.executeWithProcessors();
 
             if (table->supportsParallelInsert() && settings.max_insert_threads > 1)
-            {
-                in_streams = interpreter_select.executeWithMultipleStreams(res.pipeline);
-                out_streams_size = std::min(size_t(settings.max_insert_threads), in_streams.size());
-            }
+                out_streams_size = std::min(size_t(settings.max_insert_threads), res.pipeline.getNumStreams());
+
+            if (out_streams_size == 1)
+                res.pipeline.addPipe({std::make_shared<ConcatProcessor>(res.pipeline.getHeader(), res.pipeline.getNumStreams())});
             else
-            {
-                res = interpreter_select.execute();
-                in_streams.emplace_back(res.in);
-                res.in = nullptr;
-                res.out = nullptr;
-            }
+                res.pipeline.resize(out_streams_size);
         }
         else if (query.watch)
         {
             InterpreterWatchQuery interpreter_watch{ query.watch, context };
             res = interpreter_watch.execute();
-            in_streams.emplace_back(res.in);
-            res.in = nullptr;
-            res.out = nullptr;
+            res.pipeline.init(Pipe(std::make_shared<SourceFromInputStream>(std::move(res.in))));
         }
 
         for (size_t i = 0; i < out_streams_size; i++)
@@ -256,27 +261,35 @@ BlockIO InterpreterInsertQuery::execute()
     }
 
     /// What type of query: INSERT or INSERT SELECT or INSERT WATCH?
-    if (query.select || query.watch)
+    if (is_distributed_insert_select)
     {
-        for (auto & in_stream : in_streams)
-        {
-            in_stream = std::make_shared<ConvertingBlockInputStream>(
-                in_stream, out_streams.at(0)->getHeader(), ConvertingBlockInputStream::MatchColumnsMode::Position);
-        }
+        /// Pipeline was already built.
+    }
+    else if (query.select || query.watch)
+    {
+        const auto & header = out_streams.at(0)->getHeader();
 
-        Block in_header = in_streams.at(0)->getHeader();
-        if (in_streams.size() > 1)
+        res.pipeline.addSimpleTransform([&](const Block & in_header) -> ProcessorPtr
         {
-            for (size_t i = 1; i < in_streams.size(); ++i)
-                assertBlocksHaveEqualStructure(in_streams[i]->getHeader(), in_header, query.select ? "INSERT SELECT" : "INSERT WATCH");
-        }
+            return std::make_shared<ConvertingTransform>(in_header, header,
+                    ConvertingTransform::MatchColumnsMode::Position);
+        });
 
-        res.in = std::make_shared<NullAndDoCopyBlockInputStream>(in_streams, out_streams);
+        res.pipeline.setSinks([&](const Block &, QueryPipeline::StreamType type) -> ProcessorPtr
+        {
+            if (type != QueryPipeline::StreamType::Main)
+                return nullptr;
+
+            auto stream = std::move(out_streams.back());
+            out_streams.pop_back();
+
+            return std::make_shared<SinkToOutputStream>(std::move(stream));
+        });
 
         if (!allow_materialized)
         {
             for (const auto & column : table->getColumns())
-                if (column.default_desc.kind == ColumnDefaultKind::Materialized && in_header.has(column.name))
+                if (column.default_desc.kind == ColumnDefaultKind::Materialized && header.has(column.name))
                     throw Exception("Cannot insert column " + column.name + ", because it is MATERIALIZED column.", ErrorCodes::ILLEGAL_COLUMN);
         }
     }
@@ -288,6 +301,7 @@ BlockIO InterpreterInsertQuery::execute()
     }
     else
         res.out = std::move(out_streams.at(0));
+
     res.pipeline.addStorageHolder(table);
 
     return res;
diff --git a/src/Interpreters/executeQuery.cpp b/src/Interpreters/executeQuery.cpp
index 2d8f901d3bb..f90f37b88c3 100644
--- a/src/Interpreters/executeQuery.cpp
+++ b/src/Interpreters/executeQuery.cpp
@@ -342,6 +342,9 @@ static std::tuple<ASTPtr, BlockIO> executeQueryImpl(
         else
             res = interpreter->execute();
 
+        if (res.pipeline.initialized())
+            use_processors = true;
+
         if (const auto * insert_interpreter = typeid_cast<const InterpreterInsertQuery *>(&*interpreter))
         {
             /// Save insertion table (not table function). TODO: support remote() table function.
@@ -369,7 +372,7 @@ static std::tuple<ASTPtr, BlockIO> executeQueryImpl(
             /// Limits apply only to the final result.
             pipeline.setProgressCallback(context.getProgressCallback());
             pipeline.setProcessListElement(context.getProcessListElement());
-            if (stage == QueryProcessingStage::Complete)
+            if (stage == QueryProcessingStage::Complete && !pipeline.isCompleted())
             {
                 pipeline.resize(1);
                 pipeline.addSimpleTransform([&](const Block & header)
@@ -740,29 +743,36 @@ void executeQuery(
             if (ast_query_with_output && ast_query_with_output->settings_ast)
                 InterpreterSetQuery(ast_query_with_output->settings_ast, context).executeForCurrentContext();
 
-            pipeline.addSimpleTransform([](const Block & header)
+            if (!pipeline.isCompleted())
             {
-                return std::make_shared<MaterializingTransform>(header);
-            });
+                pipeline.addSimpleTransform([](const Block & header)
+                {
+                    return std::make_shared<MaterializingTransform>(header);
+                });
 
-            auto out = context.getOutputFormatProcessor(format_name, *out_buf, pipeline.getHeader());
-            out->setAutoFlush();
+                auto out = context.getOutputFormatProcessor(format_name, *out_buf, pipeline.getHeader());
+                out->setAutoFlush();
 
-            /// Save previous progress callback if any. TODO Do it more conveniently.
-            auto previous_progress_callback = context.getProgressCallback();
+                /// Save previous progress callback if any. TODO Do it more conveniently.
+                auto previous_progress_callback = context.getProgressCallback();
 
-            /// NOTE Progress callback takes shared ownership of 'out'.
-            pipeline.setProgressCallback([out, previous_progress_callback] (const Progress & progress)
+                /// NOTE Progress callback takes shared ownership of 'out'.
+                pipeline.setProgressCallback([out, previous_progress_callback] (const Progress & progress)
+                {
+                    if (previous_progress_callback)
+                        previous_progress_callback(progress);
+                    out->onProgress(progress);
+                });
+
+                if (set_result_details)
+                    set_result_details(context.getClientInfo().current_query_id, out->getContentType(), format_name, DateLUT::instance().getTimeZone());
+
+                pipeline.setOutputFormat(std::move(out));
+            }
+            else
             {
-                if (previous_progress_callback)
-                    previous_progress_callback(progress);
-                out->onProgress(progress);
-            });
-
-            if (set_result_details)
-                set_result_details(context.getClientInfo().current_query_id, out->getContentType(), format_name, DateLUT::instance().getTimeZone());
-
-            pipeline.setOutput(std::move(out));
+                pipeline.setProgressCallback(context.getProgressCallback());
+            }
 
             {
                 auto executor = pipeline.execute();
diff --git a/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp b/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp
index 003508ab86f..49e85fcc6d3 100644
--- a/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp
+++ b/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp
@@ -14,9 +14,10 @@ struct PullingAsyncPipelineExecutor::Data
 {
     PipelineExecutorPtr executor;
     std::exception_ptr exception;
-    std::atomic_bool is_executed = false;
+    std::atomic_bool is_finished = false;
     std::atomic_bool has_exception = false;
     ThreadFromGlobalPool thread;
+    Poco::Event finish_event;
 
     ~Data()
     {
@@ -36,8 +37,11 @@ struct PullingAsyncPipelineExecutor::Data
 
 PullingAsyncPipelineExecutor::PullingAsyncPipelineExecutor(QueryPipeline & pipeline_) : pipeline(pipeline_)
 {
-    lazy_format = std::make_shared<LazyOutputFormat>(pipeline.getHeader());
-    pipeline.setOutput(lazy_format);
+    if (!pipeline.isCompleted())
+    {
+        lazy_format = std::make_shared<LazyOutputFormat>(pipeline.getHeader());
+        pipeline.setOutputFormat(lazy_format);
+    }
 }
 
 PullingAsyncPipelineExecutor::~PullingAsyncPipelineExecutor()
@@ -54,7 +58,8 @@ PullingAsyncPipelineExecutor::~PullingAsyncPipelineExecutor()
 
 const Block & PullingAsyncPipelineExecutor::getHeader() const
 {
-    return lazy_format->getPort(IOutputFormat::PortKind::Main).getHeader();
+    return lazy_format ? lazy_format->getPort(IOutputFormat::PortKind::Main).getHeader()
+                       : pipeline.getHeader(); /// Empty.
 }
 
 static void threadFunction(PullingAsyncPipelineExecutor::Data & data, ThreadGroupStatusPtr thread_group, size_t num_threads)
@@ -78,6 +83,9 @@ static void threadFunction(PullingAsyncPipelineExecutor::Data & data, ThreadGrou
         data.exception = std::current_exception();
         data.has_exception = true;
     }
+
+    data.is_finished = true;
+    data.finish_event.set();
 }
 
 
@@ -99,20 +107,33 @@ bool PullingAsyncPipelineExecutor::pull(Chunk & chunk, uint64_t milliseconds)
     if (data->has_exception)
     {
         /// Finish lazy format in case of exception. Otherwise thread.join() may hung.
-        lazy_format->finish();
+        if (lazy_format)
+            lazy_format->finish();
+
         data->has_exception = false;
         std::rethrow_exception(std::move(data->exception));
     }
 
-    if (lazy_format->isFinished())
+    bool is_execution_finished = lazy_format ? lazy_format->isFinished()
+                                             : data->is_finished.load();
+
+    if (is_execution_finished)
     {
-        data->is_executed = true;
+        /// If lazy format is finished, we don't cancel pipeline but wait for main thread to be finished.
+        data->is_finished = true;
         /// Wait thread ant rethrow exception if any.
         cancel();
         return false;
     }
 
-    chunk = lazy_format->getChunk(milliseconds);
+    if (lazy_format)
+    {
+        chunk = lazy_format->getChunk(milliseconds);
+        return true;
+    }
+
+    chunk.clear();
+    data->finish_event.tryWait(milliseconds);
     return true;
 }
 
@@ -147,11 +168,11 @@ bool PullingAsyncPipelineExecutor::pull(Block & block, uint64_t milliseconds)
 void PullingAsyncPipelineExecutor::cancel()
 {
     /// Cancel execution if it wasn't finished.
-    if (data && !data->is_executed && data->executor)
+    if (data && !data->is_finished && data->executor)
         data->executor->cancel();
 
     /// Finish lazy format. Otherwise thread.join() may hung.
-    if (!lazy_format->isFinished())
+    if (lazy_format && !lazy_format->isFinished())
         lazy_format->finish();
 
     /// Join thread here to wait for possible exception.
@@ -165,12 +186,14 @@ void PullingAsyncPipelineExecutor::cancel()
 
 Chunk PullingAsyncPipelineExecutor::getTotals()
 {
-    return lazy_format->getTotals();
+    return lazy_format ? lazy_format->getTotals()
+                       : Chunk();
 }
 
 Chunk PullingAsyncPipelineExecutor::getExtremes()
 {
-    return lazy_format->getExtremes();
+    return lazy_format ? lazy_format->getExtremes()
+                       : Chunk();
 }
 
 Block PullingAsyncPipelineExecutor::getTotalsBlock()
@@ -197,7 +220,9 @@ Block PullingAsyncPipelineExecutor::getExtremesBlock()
 
 BlockStreamProfileInfo & PullingAsyncPipelineExecutor::getProfileInfo()
 {
-    return lazy_format->getProfileInfo();
+    static BlockStreamProfileInfo profile_info;
+    return lazy_format ? lazy_format->getProfileInfo()
+                       : profile_info;
 }
 
 }
diff --git a/src/Processors/Executors/PullingPipelineExecutor.cpp b/src/Processors/Executors/PullingPipelineExecutor.cpp
index 375f6c9ed0e..af061a373cc 100644
--- a/src/Processors/Executors/PullingPipelineExecutor.cpp
+++ b/src/Processors/Executors/PullingPipelineExecutor.cpp
@@ -9,7 +9,7 @@ namespace DB
 PullingPipelineExecutor::PullingPipelineExecutor(QueryPipeline & pipeline_) : pipeline(pipeline_)
 {
     pulling_format = std::make_shared<PullingOutputFormat>(pipeline.getHeader(), has_data_flag);
-    pipeline.setOutput(pulling_format);
+    pipeline.setOutputFormat(pulling_format);
 }
 
 PullingPipelineExecutor::~PullingPipelineExecutor()
diff --git a/src/Processors/QueryPipeline.cpp b/src/Processors/QueryPipeline.cpp
index 13787a3fd3b..146061e8a47 100644
--- a/src/Processors/QueryPipeline.cpp
+++ b/src/Processors/QueryPipeline.cpp
@@ -34,6 +34,14 @@ void QueryPipeline::checkInitialized()
         throw Exception("QueryPipeline wasn't initialized.", ErrorCodes::LOGICAL_ERROR);
 }
 
+void QueryPipeline::checkInitializedAndNotCompleted()
+{
+    checkInitialized();
+
+    if (streams.empty())
+        throw Exception("QueryPipeline was already completed.", ErrorCodes::LOGICAL_ERROR);
+}
+
 void QueryPipeline::checkSource(const ProcessorPtr & source, bool can_have_totals)
 {
     if (!source->getInputs().empty())
@@ -194,11 +202,11 @@ static ProcessorPtr callProcessorGetter(
 template <typename TProcessorGetter>
 void QueryPipeline::addSimpleTransformImpl(const TProcessorGetter & getter)
 {
-    checkInitialized();
+    checkInitializedAndNotCompleted();
 
     Block header;
 
-    auto add_transform = [&](OutputPort *& stream, StreamType stream_type, size_t stream_num [[maybe_unused]] = IProcessor::NO_STREAM)
+    auto add_transform = [&](OutputPort *& stream, StreamType stream_type)
     {
         if (!stream)
             return;
@@ -231,17 +239,14 @@ void QueryPipeline::addSimpleTransformImpl(const TProcessorGetter & getter)
 
         if (transform)
         {
-//            if (stream_type == StreamType::Main)
-//                transform->setStream(stream_num);
-
             connect(*stream, transform->getInputs().front());
             stream = &transform->getOutputs().front();
             processors.emplace_back(std::move(transform));
         }
     };
 
-    for (size_t stream_num = 0; stream_num < streams.size(); ++stream_num)
-        add_transform(streams[stream_num], StreamType::Main, stream_num);
+    for (auto & stream : streams)
+        add_transform(stream, StreamType::Main);
 
     add_transform(totals_having_port, StreamType::Totals);
     add_transform(extremes_port, StreamType::Extremes);
@@ -259,9 +264,50 @@ void QueryPipeline::addSimpleTransform(const ProcessorGetterWithStreamKind & get
     addSimpleTransformImpl(getter);
 }
 
+void QueryPipeline::setSinks(const ProcessorGetterWithStreamKind & getter)
+{
+    checkInitializedAndNotCompleted();
+
+    auto add_transform = [&](OutputPort *& stream, StreamType stream_type)
+    {
+        if (!stream)
+            return;
+
+        auto transform = getter(stream->getHeader(), stream_type);
+
+        if (transform)
+        {
+            if (transform->getInputs().size() != 1)
+                throw Exception("Sink for query pipeline transform should have single input, "
+                                "but " + transform->getName() + " has " +
+                                toString(transform->getInputs().size()) + " inputs.", ErrorCodes::LOGICAL_ERROR);
+
+            if (!transform->getOutputs().empty())
+                throw Exception("Sink for query pipeline transform should have no outputs, "
+                                "but " + transform->getName() + " has " +
+                                toString(transform->getOutputs().size()) + " outputs.", ErrorCodes::LOGICAL_ERROR);
+        }
+
+        if (!transform)
+            transform = std::make_shared<NullSink>(stream->getHeader());
+
+        connect(*stream, transform->getInputs().front());
+        processors.emplace_back(std::move(transform));
+    };
+
+    for (auto & stream : streams)
+        add_transform(stream, StreamType::Main);
+
+    add_transform(totals_having_port, StreamType::Totals);
+    add_transform(extremes_port, StreamType::Extremes);
+
+    streams.clear();
+    current_header.clear();
+}
+
 void QueryPipeline::addPipe(Processors pipe)
 {
-    checkInitialized();
+    checkInitializedAndNotCompleted();
 
     if (pipe.empty())
         throw Exception("Can't add empty processors list to QueryPipeline.", ErrorCodes::LOGICAL_ERROR);
@@ -298,7 +344,7 @@ void QueryPipeline::addPipe(Processors pipe)
 
 void QueryPipeline::addDelayedStream(ProcessorPtr source)
 {
-    checkInitialized();
+    checkInitializedAndNotCompleted();
 
     checkSource(source, false);
     assertBlocksHaveEqualStructure(current_header, source->getOutputs().front().getHeader(), "QueryPipeline");
@@ -313,7 +359,7 @@ void QueryPipeline::addDelayedStream(ProcessorPtr source)
 
 void QueryPipeline::resize(size_t num_streams, bool force, bool strict)
 {
-    checkInitialized();
+    checkInitializedAndNotCompleted();
 
     if (!force && num_streams == getNumStreams())
         return;
@@ -347,7 +393,7 @@ void QueryPipeline::enableQuotaForCurrentStreams()
 
 void QueryPipeline::addTotalsHavingTransform(ProcessorPtr transform)
 {
-    checkInitialized();
+    checkInitializedAndNotCompleted();
 
     if (!typeid_cast<const TotalsHavingTransform *>(transform.get()))
         throw Exception("TotalsHavingTransform expected for QueryPipeline::addTotalsHavingTransform.",
@@ -370,7 +416,7 @@ void QueryPipeline::addTotalsHavingTransform(ProcessorPtr transform)
 
 void QueryPipeline::addDefaultTotals()
 {
-    checkInitialized();
+    checkInitializedAndNotCompleted();
 
     if (totals_having_port)
         throw Exception("Totals having transform was already added to pipeline.", ErrorCodes::LOGICAL_ERROR);
@@ -392,7 +438,7 @@ void QueryPipeline::addDefaultTotals()
 
 void QueryPipeline::addTotals(ProcessorPtr source)
 {
-    checkInitialized();
+    checkInitializedAndNotCompleted();
 
     if (totals_having_port)
         throw Exception("Totals having transform was already added to pipeline.", ErrorCodes::LOGICAL_ERROR);
@@ -423,7 +469,7 @@ void QueryPipeline::dropTotalsAndExtremes()
 
 void QueryPipeline::addExtremesTransform()
 {
-    checkInitialized();
+    checkInitializedAndNotCompleted();
 
     if (extremes_port)
         throw Exception("Extremes transform was already added to pipeline.", ErrorCodes::LOGICAL_ERROR);
@@ -450,7 +496,7 @@ void QueryPipeline::addExtremesTransform()
 
 void QueryPipeline::addCreatingSetsTransform(ProcessorPtr transform)
 {
-    checkInitialized();
+    checkInitializedAndNotCompleted();
 
     if (!typeid_cast<const CreatingSetsTransform *>(transform.get()))
         throw Exception("CreatingSetsTransform expected for QueryPipeline::addExtremesTransform.",
@@ -467,14 +513,14 @@ void QueryPipeline::addCreatingSetsTransform(ProcessorPtr transform)
     processors.emplace_back(std::move(concat));
 }
 
-void QueryPipeline::setOutput(ProcessorPtr output)
+void QueryPipeline::setOutputFormat(ProcessorPtr output)
 {
-    checkInitialized();
+    checkInitializedAndNotCompleted();
 
     auto * format = dynamic_cast<IOutputFormat * >(output.get());
 
     if (!format)
-        throw Exception("IOutputFormat processor expected for QueryPipeline::setOutput.", ErrorCodes::LOGICAL_ERROR);
+        throw Exception("IOutputFormat processor expected for QueryPipeline::setOutputFormat.", ErrorCodes::LOGICAL_ERROR);
 
     if (output_format)
         throw Exception("QueryPipeline already has output.", ErrorCodes::LOGICAL_ERROR);
@@ -507,19 +553,25 @@ void QueryPipeline::setOutput(ProcessorPtr output)
     connect(*totals_having_port, totals);
     connect(*extremes_port, extremes);
 
+    streams.clear();
+    current_header.clear();
+    extremes_port = nullptr;
+    totals_having_port = nullptr;
+
     initRowsBeforeLimit();
 }
 
 void QueryPipeline::unitePipelines(
     std::vector<QueryPipeline> && pipelines, const Block & common_header)
 {
-    checkInitialized();
-
-    addSimpleTransform([&](const Block & header)
+    if (initialized())
     {
-        return std::make_shared<ConvertingTransform>(
-                header, common_header, ConvertingTransform::MatchColumnsMode::Position);
-    });
+        addSimpleTransform([&](const Block & header)
+        {
+            return std::make_shared<ConvertingTransform>(
+                    header, common_header, ConvertingTransform::MatchColumnsMode::Position);
+        });
+    }
 
     std::vector<OutputPort *> extremes;
     std::vector<OutputPort *> totals;
@@ -534,11 +586,14 @@ void QueryPipeline::unitePipelines(
     {
         pipeline.checkInitialized();
 
-        pipeline.addSimpleTransform([&](const Block & header)
+        if (!pipeline.isCompleted())
         {
-           return std::make_shared<ConvertingTransform>(
-                   header, common_header, ConvertingTransform::MatchColumnsMode::Position);
-        });
+            pipeline.addSimpleTransform([&](const Block & header)
+            {
+               return std::make_shared<ConvertingTransform>(
+                       header, common_header, ConvertingTransform::MatchColumnsMode::Position);
+            });
+        }
 
         if (pipeline.extremes_port)
         {
@@ -726,10 +781,8 @@ Pipe QueryPipeline::getPipe() &&
 
 PipelineExecutorPtr QueryPipeline::execute()
 {
-    checkInitialized();
-
-    if (!output_format)
-        throw Exception("Cannot execute pipeline because it doesn't have output.", ErrorCodes::LOGICAL_ERROR);
+    if (!isCompleted())
+        throw Exception("Cannot execute pipeline because it is not completed.", ErrorCodes::LOGICAL_ERROR);
 
     return std::make_shared<PipelineExecutor>(processors, process_list_element);
 }
diff --git a/src/Processors/QueryPipeline.h b/src/Processors/QueryPipeline.h
index 45e38ffa715..4b2b32c3258 100644
--- a/src/Processors/QueryPipeline.h
+++ b/src/Processors/QueryPipeline.h
@@ -28,6 +28,7 @@ private:
     {
     public:
         auto size() const { return data.size(); }
+        bool empty() const { return size() == 0; }
         auto begin() { return data.begin(); }
         auto end() { return data.end(); }
         auto & front() { return data.front(); }
@@ -81,6 +82,7 @@ public:
     void init(Pipes pipes);
     void init(Pipe pipe); /// Simple init for single pipe
     bool initialized() { return !processors.empty(); }
+    bool isCompleted() { return initialized() && streams.empty(); }
 
     /// Type of logical data stream for simple transform.
     /// Sometimes it's important to know which part of pipeline we are working for.
@@ -95,13 +97,23 @@ public:
     using ProcessorGetter = std::function<ProcessorPtr(const Block & header)>;
     using ProcessorGetterWithStreamKind = std::function<ProcessorPtr(const Block & header, StreamType stream_type)>;
 
+    /// Add transform with simple input and simple output for each port.
     void addSimpleTransform(const ProcessorGetter & getter);
     void addSimpleTransform(const ProcessorGetterWithStreamKind & getter);
+    /// Add several processors. They must have same header for inputs and same for outputs.
+    /// Total number of inputs must be the same as the number of streams. Output ports will become new streams.
     void addPipe(Processors pipe);
+    /// Add TotalsHavingTransform. Resize pipeline to single input. Adds totals port.
     void addTotalsHavingTransform(ProcessorPtr transform);
+    /// Add transform which calculates extremes. This transform adds extremes port and doesn't change inputs number.
     void addExtremesTransform();
+    /// Adds transform which creates sets. It will be executed before reading any data from input ports.
     void addCreatingSetsTransform(ProcessorPtr transform);
-    void setOutput(ProcessorPtr output);
+    /// Resize pipeline to single output and add IOutputFormat. Pipeline will be completed after this transformation.
+    void setOutputFormat(ProcessorPtr output);
+    /// Sink is a processor with single input port and no output ports. Creates sink for each output port.
+    /// Pipeline will be completed after this transformation.
+    void setSinks(const ProcessorGetterWithStreamKind & getter);
 
     /// Add totals which returns one chunk with single row with defaults.
     void addDefaultTotals();
@@ -118,6 +130,7 @@ public:
     /// Check if resize transform was used. (In that case another distinct transform will be added).
     bool hasMixedStreams() const { return has_resize || hasMoreThanOneStream(); }
 
+    /// Changes the number of input ports if needed. Adds ResizeTransform.
     void resize(size_t num_streams, bool force = false, bool strict = false);
 
     void enableQuotaForCurrentStreams();
@@ -193,6 +206,7 @@ private:
     QueryStatus * process_list_element = nullptr;
 
     void checkInitialized();
+    void checkInitializedAndNotCompleted();
     static void checkSource(const ProcessorPtr & source, bool can_have_totals);
 
     template <typename TProcessorGetter>

From ec479839d93919f6f604cf5ca1fa4546353f7154 Mon Sep 17 00:00:00 2001
From: Nikita Mikhaylov <mikhaylovnikitka@gmail.com>
Date: Wed, 27 May 2020 22:44:22 +0300
Subject: [PATCH 087/122] fixed

---
 src/Dictionaries/CacheDictionary.inc.h        | 48 +++++++--------
 .../cache_strings_default_settings.xml        | 35 +++++++++++
 .../test_default_string.py                    | 61 +++++++++++++++++++
 3 files changed, 118 insertions(+), 26 deletions(-)
 create mode 100644 tests/integration/test_dictionary_allow_read_expired_keys/configs/dictionaries/cache_strings_default_settings.xml
 create mode 100644 tests/integration/test_dictionary_allow_read_expired_keys/test_default_string.py

diff --git a/src/Dictionaries/CacheDictionary.inc.h b/src/Dictionaries/CacheDictionary.inc.h
index 71f3f1857ce..27064d113e6 100644
--- a/src/Dictionaries/CacheDictionary.inc.h
+++ b/src/Dictionaries/CacheDictionary.inc.h
@@ -302,37 +302,33 @@ void CacheDictionary::getItemsString(
 
     /// Request new values sync.
     /// We have request both cache_not_found_ids and cache_expired_ids.
-    if (!cache_not_found_ids.empty())
+    std::vector<Key> required_ids;
+    required_ids.reserve(cache_not_found_ids.size() + cache_expired_ids.size());
+    std::transform(
+        std::begin(cache_not_found_ids), std::end(cache_not_found_ids),
+        std::back_inserter(required_ids), [](auto & pair) { return pair.first; });
+    std::transform(
+        std::begin(cache_expired_ids), std::end(cache_expired_ids),
+        std::back_inserter(required_ids), [](auto & pair) { return pair.first; });
+
+    auto on_cell_updated = [&] (const auto id, const auto cell_idx)
     {
-        std::vector<Key> required_ids;
-        required_ids.reserve(cache_not_found_ids.size() + cache_expired_ids.size());
-        std::transform(
-                std::begin(cache_not_found_ids), std::end(cache_not_found_ids),
-                std::back_inserter(required_ids), [](auto & pair) { return pair.first; });
-        std::transform(
-                std::begin(cache_expired_ids), std::end(cache_expired_ids),
-                std::back_inserter(required_ids), [](auto & pair) { return pair.first; });
+        const auto attribute_value = attribute_array[cell_idx];
 
-        auto on_cell_updated = [&] (const auto id, const auto cell_idx)
-        {
-            const auto attribute_value = attribute_array[cell_idx];
+        map[id] = String{attribute_value};
+        total_length += (attribute_value.size + 1) * cache_not_found_ids[id].size();
+    };
 
-            map[id] = String{attribute_value};
-            total_length += (attribute_value.size + 1) * cache_not_found_ids[id].size();
-        };
+    auto on_id_not_found = [&] (const auto id, const auto)
+    {
+        for (const auto row : cache_not_found_ids[id])
+            total_length += get_default(row).size + 1;
+    };
 
-        auto on_id_not_found = [&] (const auto id, const auto)
-        {
-            for (const auto row : cache_not_found_ids[id])
-                total_length += get_default(row).size + 1;
-        };
-
-        auto update_unit_ptr = std::make_shared<UpdateUnit>(required_ids, on_cell_updated, on_id_not_found);
-
-        tryPushToUpdateQueueOrThrow(update_unit_ptr);
-        waitForCurrentUpdateFinish(update_unit_ptr);
-    }
+    auto update_unit_ptr = std::make_shared<UpdateUnit>(required_ids, on_cell_updated, on_id_not_found);
 
+    tryPushToUpdateQueueOrThrow(update_unit_ptr);
+    waitForCurrentUpdateFinish(update_unit_ptr);
     out->getChars().reserve(total_length);
 
     for (const auto row : ext::range(0, ext::size(ids)))
diff --git a/tests/integration/test_dictionary_allow_read_expired_keys/configs/dictionaries/cache_strings_default_settings.xml b/tests/integration/test_dictionary_allow_read_expired_keys/configs/dictionaries/cache_strings_default_settings.xml
new file mode 100644
index 00000000000..11807bc1ad6
--- /dev/null
+++ b/tests/integration/test_dictionary_allow_read_expired_keys/configs/dictionaries/cache_strings_default_settings.xml
@@ -0,0 +1,35 @@
+<yandex>
+<dictionary>
+    <name>default_string</name>
+    <source>
+        <clickhouse>
+            <host>dictionary_node</host>
+            <port>9000</port>
+            <user>default</user>
+            <password></password>
+            <db>test</db>
+            <table>strings</table>
+        </clickhouse>
+    </source>
+    <lifetime>
+        <max>2</max>
+        <min>1</min>
+    </lifetime>
+    <layout>
+        <cache>
+            <size_in_cells>1000</size_in_cells>
+            <max_update_queue_size>10000</max_update_queue_size>
+        </cache>
+    </layout>
+    <structure>
+        <id>
+            <name>key</name>
+        </id>
+        <attribute>
+            <name>value</name>
+            <type>String</type>
+            <null_value></null_value>>
+        </attribute>
+    </structure>
+</dictionary>
+</yandex>
\ No newline at end of file
diff --git a/tests/integration/test_dictionary_allow_read_expired_keys/test_default_string.py b/tests/integration/test_dictionary_allow_read_expired_keys/test_default_string.py
new file mode 100644
index 00000000000..7d762db2a6d
--- /dev/null
+++ b/tests/integration/test_dictionary_allow_read_expired_keys/test_default_string.py
@@ -0,0 +1,61 @@
+from __future__ import print_function
+import pytest
+import os
+import random
+import string
+import time
+
+from helpers.cluster import ClickHouseCluster
+from helpers.test_tools import TSV
+
+SCRIPT_DIR = os.path.dirname(os.path.realpath(__file__))
+cluster = ClickHouseCluster(__file__, base_configs_dir=os.path.join(SCRIPT_DIR, 'configs'))
+
+dictionary_node = cluster.add_instance('dictionary_node', stay_alive=True)
+main_node = cluster.add_instance('main_node', main_configs=['configs/dictionaries/cache_strings_default_settings.xml'])
+
+
+def get_random_string(string_length=8):
+    alphabet = string.ascii_letters + string.digits
+    return ''.join((random.choice(alphabet) for _ in range(string_length)))
+
+@pytest.fixture(scope="module")
+def started_cluster():
+    try:
+        cluster.start()
+        dictionary_node.query("CREATE DATABASE IF NOT EXISTS test;")
+        dictionary_node.query("DROP TABLE IF EXISTS test.strings;")
+        dictionary_node.query("""
+                             CREATE TABLE test.strings 
+                             (key UInt64, value String)
+                             ENGINE = Memory;
+                             """)
+
+        values_to_insert = ", ".join(["({}, '{}')".format(1000000 + number, get_random_string()) for number in range(100)])
+        dictionary_node.query("INSERT INTO test.strings VALUES {}".format(values_to_insert))
+
+        yield cluster
+    finally:
+        cluster.shutdown()
+
+# @pytest.mark.skip(reason="debugging")
+def test_return_real_values(started_cluster):
+    assert None != dictionary_node.get_process_pid("clickhouse"), "ClickHouse must be alive"
+
+    first_batch = """
+    SELECT count(*)
+    FROM
+    (
+    SELECT
+        arrayJoin(arrayMap(x -> (x + 1000000), range(100))) AS id,
+        dictGetString('default_string', 'value', toUInt64(id)) AS value
+    )
+    WHERE value = '';
+    """
+
+    assert TSV("0") == TSV(main_node.query(first_batch))
+
+    # Waiting for cache to become expired
+    time.sleep(5)
+
+    assert TSV("0") == TSV(main_node.query(first_batch))

From 2432ea2c4b5a0307c175f0a5dcc7e95639e1da68 Mon Sep 17 00:00:00 2001
From: Ivan Lezhankin <ilezhankin@yandex-team.ru>
Date: Thu, 28 May 2020 00:58:46 +0300
Subject: [PATCH 088/122] Build universal binary using ya make

---
 programs/client/CMakeLists.txt                | 11 ++---
 programs/client/Client.cpp                    |  5 ++-
 programs/client/ConnectionParameters.cpp      |  2 +-
 programs/client/config_client.h.in            |  3 --
 programs/client/readpassphrase/CMakeLists.txt | 12 ++----
 .../{includes.h.in => includes.h}             |  2 +-
 .../client/readpassphrase/readpassphrase.c    | 16 +------
 .../client/readpassphrase/readpassphrase.h    | 39 +++++------------
 programs/client/readpassphrase/ya.make        |  7 +++
 programs/main.cpp                             | 43 +++++++++----------
 programs/ya.make                              | 28 +++++++++++-
 src/Common/ya.make                            |  3 ++
 12 files changed, 80 insertions(+), 91 deletions(-)
 delete mode 100644 programs/client/config_client.h.in
 rename programs/client/readpassphrase/{includes.h.in => includes.h} (79%)
 create mode 100644 programs/client/readpassphrase/ya.make

diff --git a/programs/client/CMakeLists.txt b/programs/client/CMakeLists.txt
index 11ade559a8d..e273123afe0 100644
--- a/programs/client/CMakeLists.txt
+++ b/programs/client/CMakeLists.txt
@@ -6,14 +6,9 @@ set(CLICKHOUSE_CLIENT_SOURCES
 
 set(CLICKHOUSE_CLIENT_LINK PRIVATE clickhouse_common_config clickhouse_functions clickhouse_aggregate_functions clickhouse_common_io clickhouse_parsers string_utils ${Boost_PROGRAM_OPTIONS_LIBRARY})
 
-include(CheckSymbolExists)
-check_symbol_exists(readpassphrase readpassphrase.h HAVE_READPASSPHRASE)
-configure_file(config_client.h.in ${ConfigIncludePath}/config_client.h)
-
-if(NOT HAVE_READPASSPHRASE)
-    add_subdirectory(readpassphrase)
-    list(APPEND CLICKHOUSE_CLIENT_LINK PRIVATE readpassphrase)
-endif()
+# Always use internal readpassphrase
+add_subdirectory(readpassphrase)
+list(APPEND CLICKHOUSE_CLIENT_LINK PRIVATE readpassphrase)
 
 clickhouse_program_add(client)
 
diff --git a/programs/client/Client.cpp b/programs/client/Client.cpp
index d6cac7a7b02..afc8f9a72b1 100644
--- a/programs/client/Client.cpp
+++ b/programs/client/Client.cpp
@@ -39,7 +39,6 @@
 #include <Common/StringUtils/StringUtils.h>
 #include <Common/typeid_cast.h>
 #include <Common/Config/ConfigProcessor.h>
-#include <Common/config_version.h>
 #include <Core/Types.h>
 #include <Core/QueryProcessingStage.h>
 #include <Core/ExternalTable.h>
@@ -77,6 +76,10 @@
 #include <common/argsToConfig.h>
 #include <Common/TerminalSize.h>
 
+#if !defined(ARCADIA_BUILD)
+#    include <Common/config_version.h>
+#endif
+
 #ifndef __clang__
 #pragma GCC optimize("-fno-var-tracking-assignments")
 #endif
diff --git a/programs/client/ConnectionParameters.cpp b/programs/client/ConnectionParameters.cpp
index f0ef3ae5694..d8b4d0f1add 100644
--- a/programs/client/ConnectionParameters.cpp
+++ b/programs/client/ConnectionParameters.cpp
@@ -9,7 +9,7 @@
 #include <Common/Exception.h>
 #include <common/setTerminalEcho.h>
 #include <ext/scope_guard.h>
-#include <readpassphrase.h>
+#include "readpassphrase/readpassphrase.h"
 
 namespace DB
 {
diff --git a/programs/client/config_client.h.in b/programs/client/config_client.h.in
deleted file mode 100644
index 5ad788ff54c..00000000000
--- a/programs/client/config_client.h.in
+++ /dev/null
@@ -1,3 +0,0 @@
-#pragma once
-
-#cmakedefine HAVE_READPASSPHRASE
diff --git a/programs/client/readpassphrase/CMakeLists.txt b/programs/client/readpassphrase/CMakeLists.txt
index a10b54c377d..dd1bf2c91b9 100644
--- a/programs/client/readpassphrase/CMakeLists.txt
+++ b/programs/client/readpassphrase/CMakeLists.txt
@@ -1,13 +1,7 @@
-
 # wget https://raw.githubusercontent.com/openssh/openssh-portable/master/openbsd-compat/readpassphrase.c
 # wget https://raw.githubusercontent.com/openssh/openssh-portable/master/openbsd-compat/readpassphrase.h
 
-set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Wno-unused-result -Wno-reserved-id-macro")
+add_library(readpassphrase readpassphrase.c)
 
-configure_file(includes.h.in ${CMAKE_CURRENT_BINARY_DIR}/include/includes.h)
-add_library(readpassphrase ${CMAKE_CURRENT_SOURCE_DIR}/readpassphrase.c)
-set_target_properties(readpassphrase
-                      PROPERTIES LINKER_LANGUAGE C
-                      )
-# . to allow #include <readpassphrase.h>
-target_include_directories(readpassphrase PUBLIC . ${CMAKE_CURRENT_BINARY_DIR}/include)
+set_target_properties(readpassphrase PROPERTIES LINKER_LANGUAGE C)
+target_compile_options(readpassphrase PRIVATE -Wno-unused-result -Wno-reserved-id-macro)
diff --git a/programs/client/readpassphrase/includes.h.in b/programs/client/readpassphrase/includes.h
similarity index 79%
rename from programs/client/readpassphrase/includes.h.in
rename to programs/client/readpassphrase/includes.h
index 44580d1ed95..3ca5eb2bff8 100644
--- a/programs/client/readpassphrase/includes.h.in
+++ b/programs/client/readpassphrase/includes.h
@@ -1,6 +1,6 @@
 #pragma once
 
-#cmakedefine HAVE_READPASSPHRASE
+/* #undef HAVE_READPASSPHRASE */
 
 #if !defined(HAVE_READPASSPHRASE)
 #    ifndef _PATH_TTY
diff --git a/programs/client/readpassphrase/readpassphrase.c b/programs/client/readpassphrase/readpassphrase.c
index 243701239bf..21b48e7efc3 100644
--- a/programs/client/readpassphrase/readpassphrase.c
+++ b/programs/client/readpassphrase/readpassphrase.c
@@ -25,13 +25,11 @@
 
 #include "includes.h"
 
-#ifndef HAVE_READPASSPHRASE
-
 #include <termios.h>
 #include <signal.h>
 #include <ctype.h>
 #include <fcntl.h>
-#include <readpassphrase.h>
+#include "readpassphrase.h"
 #include <errno.h>
 #include <string.h>
 #include <unistd.h>
@@ -193,19 +191,7 @@ restart:
 }
 //DEF_WEAK(readpassphrase);
 
-#if 0
-char *
-getpass(const char *prompt)
-{
-    static char buf[_PASSWORD_LEN + 1];
-
-    return(readpassphrase(prompt, buf, sizeof(buf), RPP_ECHO_OFF));
-}
-#endif
-
 static void handler(int s)
 {
-
     signo[s] = 1;
 }
-#endif /* HAVE_READPASSPHRASE */
diff --git a/programs/client/readpassphrase/readpassphrase.h b/programs/client/readpassphrase/readpassphrase.h
index 0782a1773ea..399eb7b062c 100644
--- a/programs/client/readpassphrase/readpassphrase.h
+++ b/programs/client/readpassphrase/readpassphrase.h
@@ -23,39 +23,22 @@
 /* OPENBSD ORIGINAL: include/readpassphrase.h */
 
 #pragma once
-// #ifndef _READPASSPHRASE_H_
-// #define _READPASSPHRASE_H_
 
-//#include "includes.h"
-#include "config_client.h"
-
-// Should not be included on BSD systems, but if it happen...
-#ifdef HAVE_READPASSPHRASE
-#   include_next <readpassphrase.h>
+#if defined(__cplusplus)
+extern "C" {
 #endif
 
-#ifndef HAVE_READPASSPHRASE
 
-#    ifdef __cplusplus
-extern "C" {
-#    endif
-
-
-#    define RPP_ECHO_OFF 0x00 /* Turn off echo (default). */
-#    define RPP_ECHO_ON 0x01 /* Leave echo on. */
-#    define RPP_REQUIRE_TTY 0x02 /* Fail if there is no tty. */
-#    define RPP_FORCELOWER 0x04 /* Force input to lower case. */
-#    define RPP_FORCEUPPER 0x08 /* Force input to upper case. */
-#    define RPP_SEVENBIT 0x10 /* Strip the high bit from input. */
-#    define RPP_STDIN 0x20 /* Read from stdin, not /dev/tty */
+#define RPP_ECHO_OFF 0x00 /* Turn off echo (default). */
+#define RPP_ECHO_ON 0x01 /* Leave echo on. */
+#define RPP_REQUIRE_TTY 0x02 /* Fail if there is no tty. */
+#define RPP_FORCELOWER 0x04 /* Force input to lower case. */
+#define RPP_FORCEUPPER 0x08 /* Force input to upper case. */
+#define RPP_SEVENBIT 0x10 /* Strip the high bit from input. */
+#define RPP_STDIN 0x20 /* Read from stdin, not /dev/tty */
 
 char * readpassphrase(const char *, char *, size_t, int);
 
-#    ifdef __cplusplus
+#if defined(__cplusplus)
 }
-#    endif
-
-
-#endif /* HAVE_READPASSPHRASE */
-
-// #endif /* !_READPASSPHRASE_H_ */
+#endif
diff --git a/programs/client/readpassphrase/ya.make b/programs/client/readpassphrase/ya.make
new file mode 100644
index 00000000000..80ad197e5d4
--- /dev/null
+++ b/programs/client/readpassphrase/ya.make
@@ -0,0 +1,7 @@
+LIBRARY()
+
+SRCS(
+    readpassphrase.c
+)
+
+END()
diff --git a/programs/main.cpp b/programs/main.cpp
index 2eb226d3c00..6b34504ce37 100644
--- a/programs/main.cpp
+++ b/programs/main.cpp
@@ -8,11 +8,8 @@
 #include <string>
 #include <utility> /// pair
 
-#if __has_include("config_tools.h")
-#include "config_tools.h"
-#endif
-#if __has_include("config_core.h")
-#include "config_core.h"
+#if !defined(ARCADIA_BUILD)
+#    include "config_tools.h"
 #endif
 
 #include <Common/StringUtils/StringUtils.h>
@@ -22,31 +19,31 @@
 
 
 /// Universal executable for various clickhouse applications
-#if ENABLE_CLICKHOUSE_SERVER || !defined(ENABLE_CLICKHOUSE_SERVER)
+#if ENABLE_CLICKHOUSE_SERVER
 int mainEntryClickHouseServer(int argc, char ** argv);
 #endif
-#if ENABLE_CLICKHOUSE_CLIENT || !defined(ENABLE_CLICKHOUSE_CLIENT)
+#if ENABLE_CLICKHOUSE_CLIENT
 int mainEntryClickHouseClient(int argc, char ** argv);
 #endif
-#if ENABLE_CLICKHOUSE_LOCAL || !defined(ENABLE_CLICKHOUSE_LOCAL)
+#if ENABLE_CLICKHOUSE_LOCAL
 int mainEntryClickHouseLocal(int argc, char ** argv);
 #endif
-#if ENABLE_CLICKHOUSE_BENCHMARK || !defined(ENABLE_CLICKHOUSE_BENCHMARK)
+#if ENABLE_CLICKHOUSE_BENCHMARK
 int mainEntryClickHouseBenchmark(int argc, char ** argv);
 #endif
-#if ENABLE_CLICKHOUSE_EXTRACT_FROM_CONFIG || !defined(ENABLE_CLICKHOUSE_EXTRACT_FROM_CONFIG)
+#if ENABLE_CLICKHOUSE_EXTRACT_FROM_CONFIG
 int mainEntryClickHouseExtractFromConfig(int argc, char ** argv);
 #endif
-#if ENABLE_CLICKHOUSE_COMPRESSOR || !defined(ENABLE_CLICKHOUSE_COMPRESSOR)
+#if ENABLE_CLICKHOUSE_COMPRESSOR
 int mainEntryClickHouseCompressor(int argc, char ** argv);
 #endif
-#if ENABLE_CLICKHOUSE_FORMAT || !defined(ENABLE_CLICKHOUSE_FORMAT)
+#if ENABLE_CLICKHOUSE_FORMAT
 int mainEntryClickHouseFormat(int argc, char ** argv);
 #endif
-#if ENABLE_CLICKHOUSE_COPIER || !defined(ENABLE_CLICKHOUSE_COPIER)
+#if ENABLE_CLICKHOUSE_COPIER
 int mainEntryClickHouseClusterCopier(int argc, char ** argv);
 #endif
-#if ENABLE_CLICKHOUSE_OBFUSCATOR || !defined(ENABLE_CLICKHOUSE_OBFUSCATOR)
+#if ENABLE_CLICKHOUSE_OBFUSCATOR
 int mainEntryClickHouseObfuscator(int argc, char ** argv);
 #endif
 
@@ -60,31 +57,31 @@ using MainFunc = int (*)(int, char**);
 /// Add an item here to register new application
 std::pair<const char *, MainFunc> clickhouse_applications[] =
 {
-#if ENABLE_CLICKHOUSE_LOCAL || !defined(ENABLE_CLICKHOUSE_LOCAL)
+#if ENABLE_CLICKHOUSE_LOCAL
     {"local", mainEntryClickHouseLocal},
 #endif
-#if ENABLE_CLICKHOUSE_CLIENT || !defined(ENABLE_CLICKHOUSE_CLIENT)
+#if ENABLE_CLICKHOUSE_CLIENT
     {"client", mainEntryClickHouseClient},
 #endif
-#if ENABLE_CLICKHOUSE_BENCHMARK || !defined(ENABLE_CLICKHOUSE_BENCHMARK)
+#if ENABLE_CLICKHOUSE_BENCHMARK
     {"benchmark", mainEntryClickHouseBenchmark},
 #endif
-#if ENABLE_CLICKHOUSE_SERVER || !defined(ENABLE_CLICKHOUSE_SERVER)
+#if ENABLE_CLICKHOUSE_SERVER
     {"server", mainEntryClickHouseServer},
 #endif
-#if ENABLE_CLICKHOUSE_EXTRACT_FROM_CONFIG || !defined(ENABLE_CLICKHOUSE_EXTRACT_FROM_CONFIG)
+#if ENABLE_CLICKHOUSE_EXTRACT_FROM_CONFIG
     {"extract-from-config", mainEntryClickHouseExtractFromConfig},
 #endif
-#if ENABLE_CLICKHOUSE_COMPRESSOR || !defined(ENABLE_CLICKHOUSE_COMPRESSOR)
+#if ENABLE_CLICKHOUSE_COMPRESSOR
     {"compressor", mainEntryClickHouseCompressor},
 #endif
-#if ENABLE_CLICKHOUSE_FORMAT || !defined(ENABLE_CLICKHOUSE_FORMAT)
+#if ENABLE_CLICKHOUSE_FORMAT
     {"format", mainEntryClickHouseFormat},
 #endif
-#if ENABLE_CLICKHOUSE_COPIER || !defined(ENABLE_CLICKHOUSE_COPIER)
+#if ENABLE_CLICKHOUSE_COPIER
     {"copier", mainEntryClickHouseClusterCopier},
 #endif
-#if ENABLE_CLICKHOUSE_OBFUSCATOR || !defined(ENABLE_CLICKHOUSE_OBFUSCATOR)
+#if ENABLE_CLICKHOUSE_OBFUSCATOR
     {"obfuscator", mainEntryClickHouseObfuscator},
 #endif
 };
diff --git a/programs/ya.make b/programs/ya.make
index 6c773c312b8..f4a61850212 100644
--- a/programs/ya.make
+++ b/programs/ya.make
@@ -1,3 +1,27 @@
-RECURSE(
-    server
+PROGRAM(clickhouse)
+
+CFLAGS(
+    -DENABLE_CLICKHOUSE_CLIENT
+    -DENABLE_CLICKHOUSE_EXTRACT_FROM_CONFIG
+    -DENABLE_CLICKHOUSE_SERVER
 )
+
+PEERDIR(
+    clickhouse/base/daemon
+    clickhouse/base/loggers
+    clickhouse/programs/client/readpassphrase
+    clickhouse/src
+)
+
+SRCS(
+    main.cpp
+
+    client/Client.cpp
+    client/ConnectionParameters.cpp
+    client/Suggest.cpp
+    extract-from-config/ExtractFromConfig.cpp
+    server/Server.cpp
+    server/MetricsTransmitter.cpp
+)
+
+END()
diff --git a/src/Common/ya.make b/src/Common/ya.make
index c0178f3d310..99659aa2aa0 100644
--- a/src/Common/ya.make
+++ b/src/Common/ya.make
@@ -25,6 +25,9 @@ CFLAGS (GLOBAL -DDBMS_VERSION_MINOR=0)
 CFLAGS (GLOBAL -DDBMS_VERSION_PATCH=0)
 CFLAGS (GLOBAL -DVERSION_FULL=\"ClickHouse\")
 CFLAGS (GLOBAL -DVERSION_INTEGER=0)
+CFLAGS (GLOBAL -DVERSION_MAJOR=0)
+CFLAGS (GLOBAL -DVERSION_MINOR=0)
+CFLAGS (GLOBAL -DVERSION_PATCH=0)
 CFLAGS (GLOBAL -DVERSION_NAME=\"ClickHouse\")
 CFLAGS (GLOBAL -DVERSION_OFFICIAL=\"\\\(arcadia\\\)\")
 CFLAGS (GLOBAL -DVERSION_REVISION=0)

From 98172deffc81b1bd962ae5213aa3e62dbf031e07 Mon Sep 17 00:00:00 2001
From: Vitaly Baranov <vitbar@yandex-team.ru>
Date: Wed, 20 May 2020 23:31:55 +0300
Subject: [PATCH 089/122] Show authentication type in table system.users and
 while executing SHOW CREATE USER query.

---
 src/Access/Authentication.cpp                 |  12 +-
 src/Access/Authentication.h                   |  61 +++++++-
 ...InterpreterShowCreateAccessEntityQuery.cpp |   4 +-
 src/Parsers/ASTCreateUserQuery.cpp            |  68 ++++++---
 src/Parsers/ASTCreateUserQuery.h              |   5 +-
 src/Parsers/ParserCreateUserQuery.cpp         | 138 +++++++-----------
 src/Parsers/ParserCreateUserQuery.h           |   4 +-
 src/Storages/System/StorageSystemUsers.cpp    |  23 ++-
 .../test_disk_access_storage/test.py          |   4 +-
 .../integration/test_grant_and_revoke/test.py |   6 +-
 10 files changed, 207 insertions(+), 118 deletions(-)

diff --git a/src/Access/Authentication.cpp b/src/Access/Authentication.cpp
index f435d6e6336..cb2c9e7a256 100644
--- a/src/Access/Authentication.cpp
+++ b/src/Access/Authentication.cpp
@@ -7,8 +7,8 @@ namespace DB
 {
 namespace ErrorCodes
 {
-    extern const int LOGICAL_ERROR;
     extern const int BAD_ARGUMENTS;
+    extern const int NOT_IMPLEMENTED;
 }
 
 
@@ -36,8 +36,11 @@ Authentication::Digest Authentication::getPasswordDoubleSHA1() const
 
         case DOUBLE_SHA1_PASSWORD:
             return password_hash;
+
+        case MAX_TYPE:
+            break;
     }
-    throw Exception("Unknown authentication type: " + std::to_string(static_cast<int>(type)), ErrorCodes::LOGICAL_ERROR);
+    throw Exception("getPasswordDoubleSHA1(): authentication type " + toString(type) + " not supported", ErrorCodes::NOT_IMPLEMENTED);
 }
 
 
@@ -71,8 +74,11 @@ bool Authentication::isCorrectPassword(const String & password_) const
 
             return encodeSHA1(first_sha1) == password_hash;
         }
+
+        case MAX_TYPE:
+            break;
     }
-    throw Exception("Unknown authentication type: " + std::to_string(static_cast<int>(type)), ErrorCodes::LOGICAL_ERROR);
+    throw Exception("Cannot check if the password is correct for authentication type " + toString(type), ErrorCodes::NOT_IMPLEMENTED);
 }
 
 }
diff --git a/src/Access/Authentication.h b/src/Access/Authentication.h
index 3f16dc56de3..c410a101cdd 100644
--- a/src/Access/Authentication.h
+++ b/src/Access/Authentication.h
@@ -5,6 +5,7 @@
 #include <Common/OpenSSLHelpers.h>
 #include <Poco/SHA1Engine.h>
 #include <boost/algorithm/hex.hpp>
+#include <boost/algorithm/string/case_conv.hpp>
 
 
 namespace DB
@@ -14,6 +15,7 @@ namespace ErrorCodes
     extern const int SUPPORT_IS_DISABLED;
     extern const int BAD_ARGUMENTS;
     extern const int LOGICAL_ERROR;
+    extern const int NOT_IMPLEMENTED;
 }
 
 
@@ -35,6 +37,15 @@ public:
         /// SHA1(SHA1(password)).
         /// This kind of hash is used by the `mysql_native_password` authentication plugin.
         DOUBLE_SHA1_PASSWORD,
+
+        MAX_TYPE,
+    };
+
+    struct TypeInfo
+    {
+        const char * const raw_name;
+        const String name; /// Lowercased with underscores, e.g. "sha256_password".
+        static const TypeInfo & get(Type type_);
     };
 
     using Digest = std::vector<uint8_t>;
@@ -85,6 +96,48 @@ private:
 };
 
 
+inline const Authentication::TypeInfo & Authentication::TypeInfo::get(Type type_)
+{
+    static constexpr auto make_info = [](const char * raw_name_)
+    {
+        String init_name = raw_name_;
+        boost::to_lower(init_name);
+        return TypeInfo{raw_name_, std::move(init_name)};
+    };
+
+    switch (type_)
+    {
+        case NO_PASSWORD:
+        {
+            static const auto info = make_info("NO_PASSWORD");
+            return info;
+        }
+        case PLAINTEXT_PASSWORD:
+        {
+            static const auto info = make_info("PLAINTEXT_PASSWORD");
+            return info;
+        }
+        case SHA256_PASSWORD:
+        {
+            static const auto info = make_info("SHA256_PASSWORD");
+            return info;
+        }
+        case DOUBLE_SHA1_PASSWORD:
+        {
+            static const auto info = make_info("DOUBLE_SHA1_PASSWORD");
+            return info;
+        }
+        case MAX_TYPE: break;
+    }
+    throw Exception("Unknown authentication type: " + std::to_string(static_cast<int>(type_)), ErrorCodes::LOGICAL_ERROR);
+}
+
+inline String toString(Authentication::Type type_)
+{
+    return Authentication::TypeInfo::get(type_).raw_name;
+}
+
+
 inline Authentication::Digest Authentication::encodeSHA256(const std::string_view & text [[maybe_unused]])
 {
 #if USE_SSL
@@ -122,8 +175,10 @@ inline void Authentication::setPassword(const String & password_)
 
         case DOUBLE_SHA1_PASSWORD:
             return setPasswordHashBinary(encodeDoubleSHA1(password_));
+
+        case MAX_TYPE: break;
     }
-    throw Exception("Unknown authentication type: " + std::to_string(static_cast<int>(type)), ErrorCodes::LOGICAL_ERROR);
+    throw Exception("setPassword(): authentication type " + toString(type) + " not supported", ErrorCodes::NOT_IMPLEMENTED);
 }
 
 
@@ -186,8 +241,10 @@ inline void Authentication::setPasswordHashBinary(const Digest & hash)
             password_hash = hash;
             return;
         }
+
+        case MAX_TYPE: break;
     }
-    throw Exception("Unknown authentication type: " + std::to_string(static_cast<int>(type)), ErrorCodes::LOGICAL_ERROR);
+    throw Exception("setPasswordHashBinary(): authentication type " + toString(type) + " not supported", ErrorCodes::NOT_IMPLEMENTED);
 }
 
 }
diff --git a/src/Interpreters/InterpreterShowCreateAccessEntityQuery.cpp b/src/Interpreters/InterpreterShowCreateAccessEntityQuery.cpp
index 87e1265f793..e37c31aab22 100644
--- a/src/Interpreters/InterpreterShowCreateAccessEntityQuery.cpp
+++ b/src/Interpreters/InterpreterShowCreateAccessEntityQuery.cpp
@@ -56,10 +56,10 @@ namespace
                 query->default_roles = user.default_roles.toASTWithNames(*manager);
         }
 
-        if (attach_mode && (user.authentication.getType() != Authentication::NO_PASSWORD))
+        if (user.authentication.getType() != Authentication::NO_PASSWORD)
         {
-            /// We don't show password unless it's an ATTACH statement.
             query->authentication = user.authentication;
+            query->show_password = attach_mode; /// We don't show password unless it's an ATTACH statement.
         }
 
         if (!user.settings.empty())
diff --git a/src/Parsers/ASTCreateUserQuery.cpp b/src/Parsers/ASTCreateUserQuery.cpp
index c8e2a76dfa2..e5c1178285b 100644
--- a/src/Parsers/ASTCreateUserQuery.cpp
+++ b/src/Parsers/ASTCreateUserQuery.cpp
@@ -6,6 +6,12 @@
 
 namespace DB
 {
+namespace ErrorCodes
+{
+    extern const int LOGICAL_ERROR;
+}
+
+
 namespace
 {
     void formatRenameTo(const String & new_name, const IAST::FormatSettings & settings)
@@ -15,27 +21,51 @@ namespace
     }
 
 
-    void formatAuthentication(const Authentication & authentication, const IAST::FormatSettings & settings)
+    void formatAuthentication(const Authentication & authentication, bool show_password, const IAST::FormatSettings & settings)
     {
-        settings.ostr << (settings.hilite ? IAST::hilite_keyword : "") << " IDENTIFIED WITH " << (settings.hilite ? IAST::hilite_none : "");
-        switch (authentication.getType())
+        auto authentication_type = authentication.getType();
+        if (authentication_type == Authentication::NO_PASSWORD)
         {
-            case Authentication::Type::NO_PASSWORD:
-                settings.ostr << (settings.hilite ? IAST::hilite_keyword : "") << "no_password" << (settings.hilite ? IAST::hilite_none : "");
-                break;
-            case Authentication::Type::PLAINTEXT_PASSWORD:
-                settings.ostr << (settings.hilite ? IAST::hilite_keyword : "") << "plaintext_password BY " << (settings.hilite ? IAST::hilite_none : "")
-                              << quoteString(authentication.getPassword());
-                break;
-            case Authentication::Type::SHA256_PASSWORD:
-                settings.ostr << (settings.hilite ? IAST::hilite_keyword : "") << "sha256_hash BY " << (settings.hilite ? IAST::hilite_none : "")
-                              << quoteString(authentication.getPasswordHashHex());
-                break;
-            case Authentication::Type::DOUBLE_SHA1_PASSWORD:
-                settings.ostr << (settings.hilite ? IAST::hilite_keyword : "") << "double_sha1_hash BY " << (settings.hilite ? IAST::hilite_none : "")
-                              << quoteString(authentication.getPasswordHashHex());
-                break;
+            settings.ostr << (settings.hilite ? IAST::hilite_keyword : "") << " NOT IDENTIFIED"
+                          << (settings.hilite ? IAST::hilite_none : "");
+            return;
         }
+
+        String authentication_type_name = Authentication::TypeInfo::get(authentication_type).name;
+        std::optional<String> password;
+
+        if (show_password)
+        {
+            switch (authentication_type)
+            {
+                case Authentication::PLAINTEXT_PASSWORD:
+                {
+                    password = authentication.getPassword();
+                    break;
+                }
+                case Authentication::SHA256_PASSWORD:
+                {
+                    authentication_type_name = "sha256_hash";
+                    password = authentication.getPasswordHashHex();
+                    break;
+                }
+                case Authentication::DOUBLE_SHA1_PASSWORD:
+                {
+                    authentication_type_name = "double_sha1_hash";
+                    password = authentication.getPasswordHashHex();
+                    break;
+                }
+
+                case Authentication::NO_PASSWORD: [[fallthrough]];
+                case Authentication::MAX_TYPE:
+                    throw Exception("AST: Unexpected authentication type " + toString(authentication_type), ErrorCodes::LOGICAL_ERROR);
+            }
+        }
+
+        settings.ostr << (settings.hilite ? IAST::hilite_keyword : "") << " IDENTIFIED WITH " << authentication_type_name
+                      << (settings.hilite ? IAST::hilite_none : "");
+        if (password)
+            settings.ostr << (settings.hilite ? IAST::hilite_keyword : "") << " BY " << quoteString(*password);
     }
 
 
@@ -190,7 +220,7 @@ void ASTCreateUserQuery::formatImpl(const FormatSettings & format, FormatState &
         formatRenameTo(new_name, format);
 
     if (authentication)
-        formatAuthentication(*authentication, format);
+        formatAuthentication(*authentication, show_password, format);
 
     if (hosts)
         formatHosts(nullptr, *hosts, format);
diff --git a/src/Parsers/ASTCreateUserQuery.h b/src/Parsers/ASTCreateUserQuery.h
index 54dc51d783b..28ef6c059da 100644
--- a/src/Parsers/ASTCreateUserQuery.h
+++ b/src/Parsers/ASTCreateUserQuery.h
@@ -12,14 +12,14 @@ class ASTExtendedRoleSet;
 class ASTSettingsProfileElements;
 
 /** CREATE USER [IF NOT EXISTS | OR REPLACE] name
-  *     [IDENTIFIED [WITH {NO_PASSWORD|PLAINTEXT_PASSWORD|SHA256_PASSWORD|SHA256_HASH|DOUBLE_SHA1_PASSWORD|DOUBLE_SHA1_HASH}] BY {'password'|'hash'}]
+  *     [NOT IDENTIFIED | IDENTIFIED [WITH {no_password|plaintext_password|sha256_password|sha256_hash|double_sha1_password|double_sha1_hash}] BY {'password'|'hash'}]
   *     [HOST {LOCAL | NAME 'name' | REGEXP 'name_regexp' | IP 'address' | LIKE 'pattern'} [,...] | ANY | NONE]
   *     [DEFAULT ROLE role [,...]]
   *     [SETTINGS variable [= value] [MIN [=] min_value] [MAX [=] max_value] [READONLY|WRITABLE] | PROFILE 'profile_name'] [,...]
   *
   * ALTER USER [IF EXISTS] name
   *      [RENAME TO new_name]
-  *      [IDENTIFIED [WITH {PLAINTEXT_PASSWORD|SHA256_PASSWORD|DOUBLE_SHA1_PASSWORD}] BY {'password'|'hash'}]
+  *      [NOT IDENTIFIED | IDENTIFIED [WITH {no_password|plaintext_password|sha256_password|sha256_hash|double_sha1_password|double_sha1_hash}] BY {'password'|'hash'}]
   *      [[ADD|DROP] HOST {LOCAL | NAME 'name' | REGEXP 'name_regexp' | IP 'address' | LIKE 'pattern'} [,...] | ANY | NONE]
   *      [DEFAULT ROLE role [,...] | ALL | ALL EXCEPT role [,...] ]
   *      [SETTINGS variable [= value] [MIN [=] min_value] [MAX [=] max_value] [READONLY|WRITABLE] | PROFILE 'profile_name'] [,...]
@@ -38,6 +38,7 @@ public:
     String new_name;
 
     std::optional<Authentication> authentication;
+    bool show_password = true; /// formatImpl() will show the password or hash.
 
     std::optional<AllowedClientHosts> hosts;
     std::optional<AllowedClientHosts> add_hosts;
diff --git a/src/Parsers/ParserCreateUserQuery.cpp b/src/Parsers/ParserCreateUserQuery.cpp
index 76a06a0282f..2f73e7fa75a 100644
--- a/src/Parsers/ParserCreateUserQuery.cpp
+++ b/src/Parsers/ParserCreateUserQuery.cpp
@@ -35,100 +35,74 @@ namespace
     }
 
 
-    bool parseByPassword(IParserBase::Pos & pos, Expected & expected, String & password)
-    {
-        return IParserBase::wrapParseImpl(pos, [&]
-        {
-            if (!ParserKeyword{"BY"}.ignore(pos, expected))
-                return false;
-
-            ASTPtr ast;
-            if (!ParserStringLiteral{}.parse(pos, ast, expected))
-                return false;
-
-            password = ast->as<const ASTLiteral &>().value.safeGet<String>();
-            return true;
-        });
-    }
-
-
     bool parseAuthentication(IParserBase::Pos & pos, Expected & expected, std::optional<Authentication> & authentication)
     {
         return IParserBase::wrapParseImpl(pos, [&]
         {
+            if (ParserKeyword{"NOT IDENTIFIED"}.ignore(pos, expected))
+            {
+                authentication = Authentication{Authentication::NO_PASSWORD};
+                return true;
+            }
+
             if (!ParserKeyword{"IDENTIFIED"}.ignore(pos, expected))
                 return false;
 
-            if (!ParserKeyword{"WITH"}.ignore(pos, expected))
+            std::optional<Authentication::Type> type;
+            bool expect_password = false;
+            bool expect_hash = false;
+
+            if (ParserKeyword{"WITH"}.ignore(pos, expected))
             {
-                String password;
-                if (!parseByPassword(pos, expected, password))
+                for (auto check_type : ext::range(Authentication::MAX_TYPE))
+                {
+                    if (ParserKeyword{Authentication::TypeInfo::get(check_type).raw_name}.ignore(pos, expected))
+                    {
+                        type = check_type;
+                        expect_password = (check_type != Authentication::NO_PASSWORD);
+                        break;
+                    }
+                }
+
+                if (!type)
+                {
+                    if (ParserKeyword{"SHA256_HASH"}.ignore(pos, expected))
+                    {
+                        type = Authentication::SHA256_PASSWORD;
+                        expect_hash = true;
+                    }
+                    else if (ParserKeyword{"DOUBLE_SHA1_HASH"}.ignore(pos, expected))
+                    {
+                        type = Authentication::DOUBLE_SHA1_PASSWORD;
+                        expect_hash = true;
+                    }
+                    else
+                        return false;
+                }
+            }
+
+            if (!type)
+            {
+                type = Authentication::SHA256_PASSWORD;
+                expect_password = true;
+            }
+
+            String password;
+            if (expect_password || expect_hash)
+            {
+                ASTPtr ast;
+                if (!ParserKeyword{"BY"}.ignore(pos, expected) || !ParserStringLiteral{}.parse(pos, ast, expected))
                     return false;
 
-                authentication = Authentication{Authentication::SHA256_PASSWORD};
+                password = ast->as<const ASTLiteral &>().value.safeGet<String>();
+            }
+
+            authentication = Authentication{*type};
+            if (expect_password)
                 authentication->setPassword(password);
-                return true;
-            }
+            else if (expect_hash)
+                authentication->setPasswordHashHex(password);
 
-            if (ParserKeyword{"PLAINTEXT_PASSWORD"}.ignore(pos, expected))
-            {
-                String password;
-                if (!parseByPassword(pos, expected, password))
-                    return false;
-
-                authentication = Authentication{Authentication::PLAINTEXT_PASSWORD};
-                authentication->setPassword(password);
-                return true;
-            }
-
-            if (ParserKeyword{"SHA256_PASSWORD"}.ignore(pos, expected))
-            {
-                String password;
-                if (!parseByPassword(pos, expected, password))
-                    return false;
-
-                authentication = Authentication{Authentication::SHA256_PASSWORD};
-                authentication->setPassword(password);
-                return true;
-            }
-
-            if (ParserKeyword{"SHA256_HASH"}.ignore(pos, expected))
-            {
-                String hash;
-                if (!parseByPassword(pos, expected, hash))
-                    return false;
-
-                authentication = Authentication{Authentication::SHA256_PASSWORD};
-                authentication->setPasswordHashHex(hash);
-                return true;
-            }
-
-            if (ParserKeyword{"DOUBLE_SHA1_PASSWORD"}.ignore(pos, expected))
-            {
-                String password;
-                if (!parseByPassword(pos, expected, password))
-                    return false;
-
-                authentication = Authentication{Authentication::DOUBLE_SHA1_PASSWORD};
-                authentication->setPassword(password);
-                return true;
-            }
-
-            if (ParserKeyword{"DOUBLE_SHA1_HASH"}.ignore(pos, expected))
-            {
-                String hash;
-                if (!parseByPassword(pos, expected, hash))
-                    return false;
-
-                authentication = Authentication{Authentication::DOUBLE_SHA1_PASSWORD};
-                authentication->setPasswordHashHex(hash);
-                return true;
-            }
-
-            if (!ParserKeyword{"NO_PASSWORD"}.ignore(pos, expected))
-                return false;
-
-            authentication = Authentication{Authentication::NO_PASSWORD};
             return true;
         });
     }
diff --git a/src/Parsers/ParserCreateUserQuery.h b/src/Parsers/ParserCreateUserQuery.h
index d609894a7ec..2a890f41060 100644
--- a/src/Parsers/ParserCreateUserQuery.h
+++ b/src/Parsers/ParserCreateUserQuery.h
@@ -7,13 +7,13 @@ namespace DB
 {
 /** Parses queries like
   * CREATE USER [IF NOT EXISTS | OR REPLACE] name
-  *     [IDENTIFIED [WITH {NO_PASSWORD|PLAINTEXT_PASSWORD|SHA256_PASSWORD|SHA256_HASH|DOUBLE_SHA1_PASSWORD|DOUBLE_SHA1_HASH}] BY {'password'|'hash'}]
+  *     [NOT IDENTIFIED | IDENTIFIED [WITH {no_password|plaintext_password|sha256_password|sha256_hash|double_sha1_password|double_sha1_hash}] BY {'password'|'hash'}]
   *     [HOST {LOCAL | NAME 'name' | REGEXP 'name_regexp' | IP 'address' | LIKE 'pattern'} [,...] | ANY | NONE]
   *     [SETTINGS variable [= value] [MIN [=] min_value] [MAX [=] max_value] [READONLY|WRITABLE] | PROFILE 'profile_name'] [,...]
   *
   * ALTER USER [IF EXISTS] name
   *     [RENAME TO new_name]
-  *     [IDENTIFIED [WITH {PLAINTEXT_PASSWORD|SHA256_PASSWORD|DOUBLE_SHA1_PASSWORD}] BY {'password'|'hash'}]
+  *     [NOT IDENTIFIED | IDENTIFIED [WITH {no_password|plaintext_password|sha256_password|sha256_hash|double_sha1_password|double_sha1_hash}] BY {'password'|'hash'}]
   *     [[ADD|DROP] HOST {LOCAL | NAME 'name' | REGEXP 'name_regexp' | IP 'address' | LIKE 'pattern'} [,...] | ANY | NONE]
   *     [SETTINGS variable [= value] [MIN [=] min_value] [MAX [=] max_value] [READONLY|WRITABLE] | PROFILE 'profile_name'] [,...]
   */
diff --git a/src/Storages/System/StorageSystemUsers.cpp b/src/Storages/System/StorageSystemUsers.cpp
index d0e042d054f..e0755fe59ab 100644
--- a/src/Storages/System/StorageSystemUsers.cpp
+++ b/src/Storages/System/StorageSystemUsers.cpp
@@ -3,6 +3,7 @@
 #include <DataTypes/DataTypesNumber.h>
 #include <DataTypes/DataTypeUUID.h>
 #include <DataTypes/DataTypeArray.h>
+#include <DataTypes/DataTypeEnum.h>
 #include <Columns/ColumnArray.h>
 #include <Columns/ColumnString.h>
 #include <Columns/ColumnsNumber.h>
@@ -15,12 +16,26 @@
 
 namespace DB
 {
+namespace
+{
+    DataTypeEnum8::Values getAuthenticationTypeEnumValues()
+    {
+        DataTypeEnum8::Values enum_values;
+        for (auto type : ext::range(Authentication::MAX_TYPE))
+            enum_values.emplace_back(Authentication::TypeInfo::get(type).name, static_cast<Int8>(type));
+        return enum_values;
+    }
+}
+
+
 NamesAndTypesList StorageSystemUsers::getNamesAndTypes()
 {
     NamesAndTypesList names_and_types{
         {"name", std::make_shared<DataTypeString>()},
         {"id", std::make_shared<DataTypeUUID>()},
         {"storage", std::make_shared<DataTypeString>()},
+        {"auth_type", std::make_shared<DataTypeEnum8>(getAuthenticationTypeEnumValues())},
+        {"auth_params", std::make_shared<DataTypeArray>(std::make_shared<DataTypeString>())},
         {"host_ip", std::make_shared<DataTypeArray>(std::make_shared<DataTypeString>())},
         {"host_names", std::make_shared<DataTypeArray>(std::make_shared<DataTypeString>())},
         {"host_names_regexp", std::make_shared<DataTypeArray>(std::make_shared<DataTypeString>())},
@@ -43,6 +58,9 @@ void StorageSystemUsers::fillData(MutableColumns & res_columns, const Context &
     auto & column_name = assert_cast<ColumnString &>(*res_columns[column_index++]);
     auto & column_id = assert_cast<ColumnUInt128 &>(*res_columns[column_index++]).getData();
     auto & column_storage = assert_cast<ColumnString &>(*res_columns[column_index++]);
+    auto & column_auth_type = assert_cast<ColumnInt8 &>(*res_columns[column_index++]).getData();
+    auto & column_auth_params = assert_cast<ColumnString &>(assert_cast<ColumnArray &>(*res_columns[column_index]).getData());
+    auto & column_auth_params_offsets = assert_cast<ColumnArray &>(*res_columns[column_index++]).getOffsets();
     auto & column_host_ip = assert_cast<ColumnString &>(assert_cast<ColumnArray &>(*res_columns[column_index]).getData());
     auto & column_host_ip_offsets = assert_cast<ColumnArray &>(*res_columns[column_index++]).getOffsets();
     auto & column_host_names = assert_cast<ColumnString &>(assert_cast<ColumnArray &>(*res_columns[column_index]).getData());
@@ -60,12 +78,15 @@ void StorageSystemUsers::fillData(MutableColumns & res_columns, const Context &
     auto add_row = [&](const String & name,
                        const UUID & id,
                        const String & storage_name,
+                       const Authentication & authentication,
                        const AllowedClientHosts & allowed_hosts,
                        const ExtendedRoleSet & default_roles)
     {
         column_name.insertData(name.data(), name.length());
         column_id.push_back(id);
         column_storage.insertData(storage_name.data(), storage_name.length());
+        column_auth_type.push_back(static_cast<Int8>(authentication.getType()));
+        column_auth_params_offsets.push_back(column_auth_params.size());
 
         if (allowed_hosts.containsAnyHost())
         {
@@ -128,7 +149,7 @@ void StorageSystemUsers::fillData(MutableColumns & res_columns, const Context &
         if (!storage)
             continue;
 
-        add_row(user->getName(), id, storage->getStorageName(), user->allowed_client_hosts, user->default_roles);
+        add_row(user->getName(), id, storage->getStorageName(), user->authentication, user->allowed_client_hosts, user->default_roles);
     }
 }
 
diff --git a/tests/integration/test_disk_access_storage/test.py b/tests/integration/test_disk_access_storage/test.py
index 315440b4358..a47af5ad5b8 100644
--- a/tests/integration/test_disk_access_storage/test.py
+++ b/tests/integration/test_disk_access_storage/test.py
@@ -39,7 +39,7 @@ def test_create():
 
     def check():
         assert instance.query("SHOW CREATE USER u1") == "CREATE USER u1 SETTINGS PROFILE s1\n"
-        assert instance.query("SHOW CREATE USER u2") == "CREATE USER u2 HOST LOCAL DEFAULT ROLE rx\n"
+        assert instance.query("SHOW CREATE USER u2") == "CREATE USER u2 IDENTIFIED WITH sha256_password HOST LOCAL DEFAULT ROLE rx\n"
         assert instance.query("SHOW CREATE ROW POLICY p ON mydb.mytable") == "CREATE ROW POLICY p ON mydb.mytable FOR SELECT USING a < 1000 TO u1, u2\n"
         assert instance.query("SHOW CREATE QUOTA q") == "CREATE QUOTA q KEYED BY \\'none\\' FOR INTERVAL 1 HOUR MAX QUERIES 100 TO ALL EXCEPT rx\n"
         assert instance.query("SHOW GRANTS FOR u1") == ""
@@ -69,7 +69,7 @@ def test_alter():
 
     def check():
         assert instance.query("SHOW CREATE USER u1") == "CREATE USER u1 SETTINGS PROFILE s1\n"
-        assert instance.query("SHOW CREATE USER u2") == "CREATE USER u2 HOST LOCAL DEFAULT ROLE ry\n"
+        assert instance.query("SHOW CREATE USER u2") == "CREATE USER u2 IDENTIFIED WITH sha256_password HOST LOCAL DEFAULT ROLE ry\n"
         assert instance.query("SHOW GRANTS FOR u1") == "GRANT SELECT ON mydb.mytable TO u1\n"
         assert instance.query("SHOW GRANTS FOR u2") == "GRANT rx, ry TO u2\n"
         assert instance.query("SHOW CREATE ROLE rx") == "CREATE ROLE rx SETTINGS PROFILE s2\n"
diff --git a/tests/integration/test_grant_and_revoke/test.py b/tests/integration/test_grant_and_revoke/test.py
index 8df68547f38..7054ce28e59 100644
--- a/tests/integration/test_grant_and_revoke/test.py
+++ b/tests/integration/test_grant_and_revoke/test.py
@@ -155,9 +155,9 @@ def test_introspection():
     assert instance.query("SHOW ENABLED ROLES", user='A') == TSV([[ "R1", 0, 1, 1 ]])
     assert instance.query("SHOW ENABLED ROLES", user='B') == TSV([[ "R2", 1, 1, 1 ]])
 
-    assert instance.query("SELECT name, storage, host_ip, host_names, host_names_regexp, host_names_like, default_roles_all, default_roles_list, default_roles_except from system.users WHERE name IN ('A', 'B') ORDER BY name") ==\
-           TSV([[ "A", "disk", "['::/0']", "[]", "[]", "[]", 1, "[]", "[]" ],
-                [ "B", "disk", "['::/0']", "[]", "[]", "[]", 1, "[]", "[]" ]])
+    assert instance.query("SELECT name, storage, auth_type, auth_params, host_ip, host_names, host_names_regexp, host_names_like, default_roles_all, default_roles_list, default_roles_except from system.users WHERE name IN ('A', 'B') ORDER BY name") ==\
+           TSV([[ "A", "disk", "no_password", "[]", "['::/0']", "[]", "[]", "[]", 1, "[]", "[]" ],
+                [ "B", "disk", "no_password", "[]", "['::/0']", "[]", "[]", "[]", 1, "[]", "[]" ]])
     
     assert instance.query("SELECT name, storage from system.roles WHERE name IN ('R1', 'R2') ORDER BY name") ==\
            TSV([[ "R1", "disk" ],

From aabf8d479eba263860b814437237a1c77fa34262 Mon Sep 17 00:00:00 2001
From: "dependabot-preview[bot]"
 <27856297+dependabot-preview[bot]@users.noreply.github.com>
Date: Thu, 28 May 2020 07:33:06 +0000
Subject: [PATCH 090/122] Bump protobuf from 3.12.1 to 3.12.2 in /docs/tools

Bumps [protobuf](https://github.com/protocolbuffers/protobuf) from 3.12.1 to 3.12.2.
- [Release notes](https://github.com/protocolbuffers/protobuf/releases)
- [Changelog](https://github.com/protocolbuffers/protobuf/blob/master/generate_changelog.py)
- [Commits](https://github.com/protocolbuffers/protobuf/compare/v3.12.1...v3.12.2)

Signed-off-by: dependabot-preview[bot] <support@dependabot.com>
---
 docs/tools/requirements.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/docs/tools/requirements.txt b/docs/tools/requirements.txt
index 4d4e9f98780..d9ea19ff389 100644
--- a/docs/tools/requirements.txt
+++ b/docs/tools/requirements.txt
@@ -21,7 +21,7 @@ mkdocs-htmlproofer-plugin==0.0.3
 mkdocs-macros-plugin==0.4.9
 nltk==3.5
 nose==1.3.7
-protobuf==3.12.1
+protobuf==3.12.2
 numpy==1.18.4
 Pygments==2.5.2
 pymdown-extensions==7.1

From ca5d7cf5bb0d06224015881dcf4171be8adaffae Mon Sep 17 00:00:00 2001
From: Alexander Kuzmenkov <akuzm@yandex-team.ru>
Date: Thu, 28 May 2020 10:45:03 +0300
Subject: [PATCH 091/122] fix some broken performance tests

---
 docker/test/performance-comparison/compare.sh | 4 ++--
 docker/test/performance-comparison/report.py  | 2 +-
 tests/performance/leftpad.xml                 | 1 -
 tests/performance/sort_radix_trivial.xml      | 6 +++---
 4 files changed, 6 insertions(+), 7 deletions(-)

diff --git a/docker/test/performance-comparison/compare.sh b/docker/test/performance-comparison/compare.sh
index bf4599acb9a..209b36f59af 100755
--- a/docker/test/performance-comparison/compare.sh
+++ b/docker/test/performance-comparison/compare.sh
@@ -317,7 +317,7 @@ function report
 rm -r report ||:
 mkdir report report/tmp ||:
 
-rm ./*.{rep,svg} test-times.tsv test-dump.tsv unstable.tsv unstable-query-ids.tsv unstable-query-metrics.tsv changed-perf.tsv unstable-tests.tsv unstable-queries.tsv bad-tests.tsv slow-on-client.tsv all-queries.tsv ||:
+rm ./*.{rep,svg} test-times.tsv test-dump.tsv unstable.tsv unstable-query-ids.tsv unstable-query-metrics.tsv changed-perf.tsv unstable-tests.tsv unstable-queries.tsv bad-tests.tsv slow-on-client.tsv all-queries.tsv run-errors.tsv ||:
 
 build_log_column_definitions
 
@@ -434,7 +434,7 @@ create table wall_clock engine Memory as select *
     from file('wall-clock-times.tsv', TSV, 'test text, real float, user float, system float');
 
 create table slow_on_client_tsv engine File(TSV, 'report/slow-on-client.tsv') as
-    select client, server, floor(client/server, 3) p, query_display_name
+    select client, server, floor(client/server, 3) p, test, query_display_name
     from query_time left join query_display_names using (test, query_index)
     where p > 1.02 order by p desc;
 
diff --git a/docker/test/performance-comparison/report.py b/docker/test/performance-comparison/report.py
index b171603700d..866e78da098 100755
--- a/docker/test/performance-comparison/report.py
+++ b/docker/test/performance-comparison/report.py
@@ -189,7 +189,7 @@ if args.report == 'main':
     slow_on_client_rows = tsvRows('report/slow-on-client.tsv')
     error_tests += len(slow_on_client_rows)
     printSimpleTable('Slow on client',
-                     ['Client time, s', 'Server time, s', 'Ratio', 'Query'],
+                     ['Client time, s', 'Server time, s', 'Ratio', 'Test', 'Query'],
                      slow_on_client_rows)
 
     def print_changes():
diff --git a/tests/performance/leftpad.xml b/tests/performance/leftpad.xml
index 4349c6ac0e5..199f9224b89 100644
--- a/tests/performance/leftpad.xml
+++ b/tests/performance/leftpad.xml
@@ -6,7 +6,6 @@
     <query><![CDATA[SELECT max(length(MobilePhoneModel)) FROM hits_100m_single]]></query>
     <query><![CDATA[SELECT max(length(Params)) FROM hits_100m_single]]></query>
     <query><![CDATA[SELECT max(length(Title)) FROM hits_100m_single]]></query>
-    <query><![CDATA[SELECT max(length(URLDomain)) FROM hits_100m_single]]></query>
     <query><![CDATA[SELECT max(length(PageCharset)) FROM hits_100m_single]]></query>
     <query><![CDATA[SELECT max(length(Referer)) FROM hits_100m_single]]></query>
     <query><![CDATA[SELECT max(length(URL)) FROM hits_100m_single]]></query>
diff --git a/tests/performance/sort_radix_trivial.xml b/tests/performance/sort_radix_trivial.xml
index fb3dc838a9c..096c3dd0854 100644
--- a/tests/performance/sort_radix_trivial.xml
+++ b/tests/performance/sort_radix_trivial.xml
@@ -1,5 +1,5 @@
 <test>
-    <query>SELECT rand32() AS x FROM numbers(1000000) ORDER BY x</query>
-    <query>SELECT rand64() AS x FROM numbers(1000000) ORDER BY x</query>
-    <query>SELECT 1 / rand64() AS x FROM numbers(1000000) ORDER BY x</query>
+    <query>SELECT rand32() AS x FROM numbers(1000000) ORDER BY x FORMAT Null</query>
+    <query>SELECT rand64() AS x FROM numbers(1000000) ORDER BY x FORMAT Null</query>
+    <query>SELECT 1 / rand64() AS x FROM numbers(1000000) ORDER BY x FORMAT Null</query>
 </test>

From bc773be0e4dab18c45e67e053f82457e5c107534 Mon Sep 17 00:00:00 2001
From: Ilya <is.homutov@gmail.com>
Date: Thu, 28 May 2020 13:30:29 +0300
Subject: [PATCH 092/122] Fix package name in docker file: tzata -> tzdata

---
 docker/server/Dockerfile | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/docker/server/Dockerfile b/docker/server/Dockerfile
index 3ad20a9479c..93f192c3f3c 100644
--- a/docker/server/Dockerfile
+++ b/docker/server/Dockerfile
@@ -21,7 +21,7 @@ RUN apt-get update \
             locales \
             ca-certificates \
             wget \
-            tzata \
+            tzdata \
     && rm -rf \
         /var/lib/apt/lists/* \
         /var/cache/debconf \

From 4f6b12457b8b390823765912d43fad98846e9506 Mon Sep 17 00:00:00 2001
From: Nikita Mikhaylov <mikhaylovnikitka@gmail.com>
Date: Thu, 28 May 2020 13:53:40 +0300
Subject: [PATCH 093/122] trigger


From b71d520a3f85f1d6474d5cec88b57404f03521c2 Mon Sep 17 00:00:00 2001
From: Nikolai Kochetov <nik-kochetov@yandex-team.ru>
Date: Thu, 28 May 2020 15:30:42 +0300
Subject: [PATCH 094/122] Fix race for description in MemoryTracker.

---
 src/Common/MemoryTracker.cpp | 6 ++++--
 src/Common/MemoryTracker.h   | 6 +++---
 2 files changed, 7 insertions(+), 5 deletions(-)

diff --git a/src/Common/MemoryTracker.cpp b/src/Common/MemoryTracker.cpp
index 082b78682a2..16d166c191b 100644
--- a/src/Common/MemoryTracker.cpp
+++ b/src/Common/MemoryTracker.cpp
@@ -49,11 +49,13 @@ MemoryTracker::~MemoryTracker()
 
 void MemoryTracker::logPeakMemoryUsage() const
 {
+    const auto * description = description_ptr.load(std::memory_order_relaxed);
     LOG_DEBUG(&Logger::get("MemoryTracker"), "Peak memory usage{}: {}.", (description ? " " + std::string(description) : ""), formatReadableSizeWithBinarySuffix(peak));
 }
 
 void MemoryTracker::logMemoryUsage(Int64 current) const
 {
+    const auto * description = description_ptr.load(std::memory_order_relaxed);
     LOG_DEBUG(&Logger::get("MemoryTracker"), "Current memory usage{}: {}.", (description ? " " + std::string(description) : ""), formatReadableSizeWithBinarySuffix(current));
 }
 
@@ -85,7 +87,7 @@ void MemoryTracker::alloc(Int64 size)
 
         std::stringstream message;
         message << "Memory tracker";
-        if (description)
+        if (const auto * description = description_ptr.load(std::memory_order_relaxed))
             message << " " << description;
         message << ": fault injected. Would use " << formatReadableSizeWithBinarySuffix(will_be)
             << " (attempt to allocate chunk of " << size << " bytes)"
@@ -117,7 +119,7 @@ void MemoryTracker::alloc(Int64 size)
 
         std::stringstream message;
         message << "Memory limit";
-        if (description)
+        if (const auto * description = description_ptr.load(std::memory_order_relaxed))
             message << " " << description;
         message << " exceeded: would use " << formatReadableSizeWithBinarySuffix(will_be)
             << " (attempt to allocate chunk of " << size << " bytes)"
diff --git a/src/Common/MemoryTracker.h b/src/Common/MemoryTracker.h
index 23f32b051b2..8af683ae790 100644
--- a/src/Common/MemoryTracker.h
+++ b/src/Common/MemoryTracker.h
@@ -35,7 +35,7 @@ private:
     CurrentMetrics::Metric metric = CurrentMetrics::end();
 
     /// This description will be used as prefix into log messages (if isn't nullptr)
-    const char * description = nullptr;
+    std::atomic<const char *> description_ptr = nullptr;
 
     void updatePeak(Int64 will_be);
     void logMemoryUsage(Int64 current) const;
@@ -114,9 +114,9 @@ public:
         metric = metric_;
     }
 
-    void setDescription(const char * description_)
+    void setDescription(const char * description)
     {
-        description = description_;
+        description_ptr.store(description, std::memory_order_relaxed);
     }
 
     /// Reset the accumulated data

From f5b1ab8e4a7556644f3d2598ef108ecd635abe37 Mon Sep 17 00:00:00 2001
From: Maxim Akhmedov <max42@yandex-team.ru>
Date: Thu, 28 May 2020 16:46:56 +0300
Subject: [PATCH 095/122] Use proper versions in Arcadia build.

---
 cmake/autogenerated_versions.txt  |  9 +++++++++
 cmake/version.cmake               | 10 +---------
 cmake/yandex/ya.make.versions.inc | 22 ++++++++++++++++++++++
 src/Common/ya.make                | 15 +--------------
 utils/release/release_lib.sh      | 28 ++++++++++++++--------------
 5 files changed, 47 insertions(+), 37 deletions(-)
 create mode 100644 cmake/autogenerated_versions.txt
 create mode 100644 cmake/yandex/ya.make.versions.inc

diff --git a/cmake/autogenerated_versions.txt b/cmake/autogenerated_versions.txt
new file mode 100644
index 00000000000..72959d3c1d8
--- /dev/null
+++ b/cmake/autogenerated_versions.txt
@@ -0,0 +1,9 @@
+# This strings autochanged from release_lib.sh:
+SET(VERSION_REVISION 54435)
+SET(VERSION_MAJOR 20)
+SET(VERSION_MINOR 5)
+SET(VERSION_PATCH 1)
+SET(VERSION_GITHASH 91df18a906dcffdbee6816e5389df6c65f86e35f)
+SET(VERSION_DESCRIBE v20.5.1.1-prestable)
+SET(VERSION_STRING 20.5.1.1)
+# end of autochange
diff --git a/cmake/version.cmake b/cmake/version.cmake
index 3f51f59cf0f..eea17f68c47 100644
--- a/cmake/version.cmake
+++ b/cmake/version.cmake
@@ -1,12 +1,4 @@
-# This strings autochanged from release_lib.sh:
-set(VERSION_REVISION 54435)
-set(VERSION_MAJOR 20)
-set(VERSION_MINOR 5)
-set(VERSION_PATCH 1)
-set(VERSION_GITHASH 91df18a906dcffdbee6816e5389df6c65f86e35f)
-set(VERSION_DESCRIBE v20.5.1.1-prestable)
-set(VERSION_STRING 20.5.1.1)
-# end of autochange
+include(${CMAKE_SOURCE_DIR}/cmake/autogenerated_versions.txt)
 
 set(VERSION_EXTRA "" CACHE STRING "")
 set(VERSION_TWEAK "" CACHE STRING "")
diff --git a/cmake/yandex/ya.make.versions.inc b/cmake/yandex/ya.make.versions.inc
new file mode 100644
index 00000000000..5c3c5eabb60
--- /dev/null
+++ b/cmake/yandex/ya.make.versions.inc
@@ -0,0 +1,22 @@
+INCLUDE(${ARCADIA_ROOT}/clickhouse/cmake/autogenerated_versions.txt)
+
+# TODO: not sure if this is customizable per-binary
+SET(VERSION_NAME "ClickHouse")
+
+# TODO: not quite sure how to replace dash with space in ya.make
+SET(VERSION_FULL "${VERSION_NAME}-${VERSION_STRING}")
+
+CFLAGS (GLOBAL -DDBMS_NAME=\"ClickHouse\")
+CFLAGS (GLOBAL -DDBMS_VERSION_MAJOR=${VERSION_MAJOR})
+CFLAGS (GLOBAL -DDBMS_VERSION_MINOR=${VERSION_MINOR})
+CFLAGS (GLOBAL -DDBMS_VERSION_PATCH=${VERSION_PATCH})
+CFLAGS (GLOBAL -DVERSION_FULL=\"\\\"${VERSION_FULL}\\\"\")
+
+# TODO: not supported yet, not sure if ya.make supports arithmetics.
+CFLAGS (GLOBAL -DVERSION_INTEGER=0)
+
+CFLAGS (GLOBAL -DVERSION_NAME=\"\\\"${VERSION_NAME}\\\"\")
+CFLAGS (GLOBAL -DVERSION_OFFICIAL=\"-arcadia\")
+CFLAGS (GLOBAL -DVERSION_REVISION=${VERSION_REVISION})
+CFLAGS (GLOBAL -DVERSION_STRING=\"\\\"${VERSION_STRING}\\\"\")
+
diff --git a/src/Common/ya.make b/src/Common/ya.make
index 99659aa2aa0..83a419212bd 100644
--- a/src/Common/ya.make
+++ b/src/Common/ya.make
@@ -18,20 +18,7 @@ PEERDIR(
     contrib/restricted/ryu
 )
 
-# TODO: stub for config_version.h
-CFLAGS (GLOBAL -DDBMS_NAME=\"ClickHouse\")
-CFLAGS (GLOBAL -DDBMS_VERSION_MAJOR=0)
-CFLAGS (GLOBAL -DDBMS_VERSION_MINOR=0)
-CFLAGS (GLOBAL -DDBMS_VERSION_PATCH=0)
-CFLAGS (GLOBAL -DVERSION_FULL=\"ClickHouse\")
-CFLAGS (GLOBAL -DVERSION_INTEGER=0)
-CFLAGS (GLOBAL -DVERSION_MAJOR=0)
-CFLAGS (GLOBAL -DVERSION_MINOR=0)
-CFLAGS (GLOBAL -DVERSION_PATCH=0)
-CFLAGS (GLOBAL -DVERSION_NAME=\"ClickHouse\")
-CFLAGS (GLOBAL -DVERSION_OFFICIAL=\"\\\(arcadia\\\)\")
-CFLAGS (GLOBAL -DVERSION_REVISION=0)
-CFLAGS (GLOBAL -DVERSION_STRING=\"Unknown\")
+INCLUDE(${ARCADIA_ROOT}/clickhouse/cmake/yandex/ya.make.versions.inc)
 
 SRCS(
     ActionLock.cpp
diff --git a/utils/release/release_lib.sh b/utils/release/release_lib.sh
index 2e700593523..569e1dfb9f2 100644
--- a/utils/release/release_lib.sh
+++ b/utils/release/release_lib.sh
@@ -12,10 +12,10 @@ function gen_version_string {
 function get_version {
     if [ -z "$VERSION_MAJOR" ] && [ -z "$VERSION_MINOR" ] && [ -z "$VERSION_PATCH" ]; then
         BASEDIR=$(dirname "${BASH_SOURCE[0]}")/../../
-        VERSION_REVISION=`grep "set(VERSION_REVISION" ${BASEDIR}/cmake/version.cmake | sed 's/^.*VERSION_REVISION \(.*\)$/\1/' | sed 's/[) ].*//'`
-        VERSION_MAJOR=`grep "set(VERSION_MAJOR" ${BASEDIR}/cmake/version.cmake | sed 's/^.*VERSION_MAJOR \(.*\)/\1/' | sed 's/[) ].*//'`
-        VERSION_MINOR=`grep "set(VERSION_MINOR" ${BASEDIR}/cmake/version.cmake | sed 's/^.*VERSION_MINOR \(.*\)/\1/' | sed 's/[) ].*//'`
-        VERSION_PATCH=`grep "set(VERSION_PATCH" ${BASEDIR}/cmake/version.cmake | sed 's/^.*VERSION_PATCH \(.*\)/\1/' | sed 's/[) ].*//'`
+        VERSION_REVISION=`grep "SET(VERSION_REVISION" ${BASEDIR}/cmake/autogenerated_versions.txt | sed 's/^.*VERSION_REVISION \(.*\)$/\1/' | sed 's/[) ].*//'`
+        VERSION_MAJOR=`grep "SET(VERSION_MAJOR" ${BASEDIR}/cmake/autogenerated_versions.txt | sed 's/^.*VERSION_MAJOR \(.*\)/\1/' | sed 's/[) ].*//'`
+        VERSION_MINOR=`grep "SET(VERSION_MINOR" ${BASEDIR}/cmake/autogenerated_versions.txt | sed 's/^.*VERSION_MINOR \(.*\)/\1/' | sed 's/[) ].*//'`
+        VERSION_PATCH=`grep "SET(VERSION_PATCH" ${BASEDIR}/cmake/autogenerated_versions.txt | sed 's/^.*VERSION_PATCH \(.*\)/\1/' | sed 's/[) ].*//'`
     fi
     VERSION_PREFIX="${VERSION_PREFIX:-v}"
     VERSION_POSTFIX_TAG="${VERSION_POSTFIX:--testing}"
@@ -90,28 +90,28 @@ function gen_revision_author {
 
             git_describe=`git describe`
             git_hash=`git rev-parse HEAD`
-            sed -i -e "s/set(VERSION_REVISION [^) ]*/set(VERSION_REVISION $VERSION_REVISION/g;" \
-                -e "s/set(VERSION_DESCRIBE [^) ]*/set(VERSION_DESCRIBE $git_describe/g;" \
-                -e "s/set(VERSION_GITHASH [^) ]*/set(VERSION_GITHASH $git_hash/g;" \
-                -e "s/set(VERSION_MAJOR [^) ]*/set(VERSION_MAJOR $VERSION_MAJOR/g;" \
-                -e "s/set(VERSION_MINOR [^) ]*/set(VERSION_MINOR $VERSION_MINOR/g;" \
-                -e "s/set(VERSION_PATCH [^) ]*/set(VERSION_PATCH $VERSION_PATCH/g;" \
-                -e "s/set(VERSION_STRING [^) ]*/set(VERSION_STRING $VERSION_STRING/g;" \
-                cmake/version.cmake
+            sed -i -e "s/SET(VERSION_REVISION [^) ]*/SET(VERSION_REVISION $VERSION_REVISION/g;" \
+                -e "s/SET(VERSION_DESCRIBE [^) ]*/SET(VERSION_DESCRIBE $git_describe/g;" \
+                -e "s/SET(VERSION_GITHASH [^) ]*/SET(VERSION_GITHASH $git_hash/g;" \
+                -e "s/SET(VERSION_MAJOR [^) ]*/SET(VERSION_MAJOR $VERSION_MAJOR/g;" \
+                -e "s/SET(VERSION_MINOR [^) ]*/SET(VERSION_MINOR $VERSION_MINOR/g;" \
+                -e "s/SET(VERSION_PATCH [^) ]*/SET(VERSION_PATCH $VERSION_PATCH/g;" \
+                -e "s/SET(VERSION_STRING [^) ]*/SET(VERSION_STRING $VERSION_STRING/g;" \
+                cmake/autogenerated_versions.txt
 
             gen_changelog "$VERSION_STRING" "" "$AUTHOR" ""
             gen_dockerfiles "$VERSION_STRING"
             src/Storages/System/StorageSystemContributors.sh ||:
             utils/list-versions/list-versions.sh > utils/list-versions/version_date.tsv
 
-            git commit -m "$auto_message [$VERSION_STRING] [$VERSION_REVISION]" cmake/version.cmake debian/changelog docker/*/Dockerfile src/Storages/System/StorageSystemContributors.generated.cpp utils/list-versions/version_date.tsv
+            git commit -m "$auto_message [$VERSION_STRING] [$VERSION_REVISION]" cmake/autogenerated_versions.txt debian/changelog docker/*/Dockerfile src/Storages/System/StorageSystemContributors.generated.cpp utils/list-versions/version_date.tsv
             if [ -z $NO_PUSH ]; then
                 git push
             fi
 
             echo "Generated version: ${VERSION_STRING}, revision: ${VERSION_REVISION}."
 
-            # Second tag for correct version information in version.cmake inside tag
+            # Second tag for correct version information in autogenerated_versions.txt inside tag
             if git tag --force -a "$tag" -m "$tag"
             then
                 if [ -z $NO_PUSH ]; then

From 7853722519cf594a6a940e61151689fea967b706 Mon Sep 17 00:00:00 2001
From: Alexey Milovidov <milovidov@yandex-team.ru>
Date: Thu, 28 May 2020 17:16:56 +0300
Subject: [PATCH 096/122] Update tests

---
 .../queries/0_stateless/00342_escape_sequences.reference  | 1 -
 tests/queries/0_stateless/00926_multimatch.sql            | 2 +-
 .../0_stateless/01075_allowed_client_hosts.reference      | 8 ++++----
 3 files changed, 5 insertions(+), 6 deletions(-)
 delete mode 100644 tests/queries/0_stateless/00342_escape_sequences.reference

diff --git a/tests/queries/0_stateless/00342_escape_sequences.reference b/tests/queries/0_stateless/00342_escape_sequences.reference
deleted file mode 100644
index b7ecd06be78..00000000000
--- a/tests/queries/0_stateless/00342_escape_sequences.reference
+++ /dev/null
@@ -1 +0,0 @@
-07080C0A0D090B5C27223FAA
diff --git a/tests/queries/0_stateless/00926_multimatch.sql b/tests/queries/0_stateless/00926_multimatch.sql
index d54e4fd2280..bf67533bb6c 100644
--- a/tests/queries/0_stateless/00926_multimatch.sql
+++ b/tests/queries/0_stateless/00926_multimatch.sql
@@ -89,4 +89,4 @@ SELECT [1, 2, 3, 11] = arraySort(multiMatchAllIndices('фабрикант', ['',
 SELECT [1] = multiMatchAllIndices(materialize('/odezhda-dlya-bega/'), ['/odezhda-dlya-bega/', 'kurtki-i-vetrovki-dlya-bega', 'futbolki-i-mayki-dlya-bega']);
 SELECT [] = multiMatchAllIndices(materialize('aaaa'), ['.*aa.*aaa.*', 'aaaaaa{2}', '\(aa\){3}']);
 SELECT 'All tests above must return 1, all tests below return something.';
-SELECT arraySort(multiMatchAllIndices(arrayJoin(['aaaa', 'aaaaaa', 'bbbb', 'aaaaaaaaaaaaaa']), ['.*aa.*aaa.*', 'aaaaaa{2}', '\(aa\){3}']));
+SELECT arraySort(multiMatchAllIndices(arrayJoin(['aaaa', 'aaaaaa', 'bbbb', 'aaaaaaaaaaaaaa']), ['.*aa.*aaa.*', 'aaaaaa{2}', '(aa){3}']));
diff --git a/tests/queries/0_stateless/01075_allowed_client_hosts.reference b/tests/queries/0_stateless/01075_allowed_client_hosts.reference
index 73f54c6027a..3fdea9d1cda 100644
--- a/tests/queries/0_stateless/01075_allowed_client_hosts.reference
+++ b/tests/queries/0_stateless/01075_allowed_client_hosts.reference
@@ -8,10 +8,10 @@ CREATE USER test_user_01075 HOST LOCAL, IP \'2001:db8:11a3:9d7:1f34:8a2e:7a0:765
 CREATE USER test_user_01075 HOST LOCAL
 CREATE USER test_user_01075 HOST NONE
 CREATE USER test_user_01075 HOST LIKE \'@.somesite.com\'
-CREATE USER test_user_01075 HOST REGEXP \'.*.anothersite.com\'
-CREATE USER test_user_01075 HOST REGEXP \'.*.anothersite.com\', \'.*.anothersite.org\'
-CREATE USER test_user_01075 HOST REGEXP \'.*.anothersite2.com\', \'.*.anothersite2.org\'
-CREATE USER test_user_01075 HOST REGEXP \'.*.anothersite3.com\', \'.*.anothersite3.org\'
+CREATE USER test_user_01075 HOST REGEXP \'.*\\\\.anothersite\\\\.com\'
+CREATE USER test_user_01075 HOST REGEXP \'.*\\\\.anothersite\\\\.com\', \'.*\\\\.anothersite\\\\.org\'
+CREATE USER test_user_01075 HOST REGEXP \'.*\\\\.anothersite2\\\\.com\', \'.*\\\\.anothersite2\\\\.org\'
+CREATE USER test_user_01075 HOST REGEXP \'.*\\\\.anothersite3\\\\.com\', \'.*\\\\.anothersite3\\\\.org\'
 CREATE USER `test_user_01075_x@localhost` HOST LOCAL
 CREATE USER test_user_01075_x
 CREATE USER `test_user_01075_x@192.168.23.15` HOST LIKE \'192.168.23.15\'

From a452d037efe71a7b9d200ad42c06267279176d99 Mon Sep 17 00:00:00 2001
From: Vitaly Baranov <vitbar@yandex-team.ru>
Date: Thu, 28 May 2020 18:35:35 +0300
Subject: [PATCH 097/122] Fix handling bad data while reading in Protobuf
 format.

---
 src/Formats/ProtobufReader.cpp | 29 ++++++++++++++++++++---------
 src/Formats/ProtobufReader.h   | 10 +++++-----
 2 files changed, 25 insertions(+), 14 deletions(-)

diff --git a/src/Formats/ProtobufReader.cpp b/src/Formats/ProtobufReader.cpp
index d8302771002..d76b048c1e9 100644
--- a/src/Formats/ProtobufReader.cpp
+++ b/src/Formats/ProtobufReader.cpp
@@ -32,11 +32,12 @@ namespace
         BITS32 = 5,
     };
 
-    // The following condition must always be true:
-    // any_cursor_position < min(END_OF_VARINT, END_OF_GROUP)
-    // This inequation helps to check conditions in SimpleReader.
-    constexpr UInt64 END_OF_VARINT = static_cast<UInt64>(-1);
-    constexpr UInt64 END_OF_GROUP = static_cast<UInt64>(-2);
+    // The following conditions must always be true:
+    // any_cursor_position > END_OF_VARINT
+    // any_cursor_position > END_OF_GROUP
+    // Those inequations helps checking conditions in ProtobufReader::SimpleReader.
+    constexpr Int64 END_OF_VARINT = -1;
+    constexpr Int64 END_OF_GROUP = -2;
 
     Int64 decodeZigZag(UInt64 n) { return static_cast<Int64>((n >> 1) ^ (~(n & 1) + 1)); }
 
@@ -77,7 +78,7 @@ void ProtobufReader::SimpleReader::endMessage(bool ignore_errors)
     if (!current_message_level)
         return;
 
-    UInt64 root_message_end = (current_message_level == 1) ? current_message_end : parent_message_ends.front();
+    Int64 root_message_end = (current_message_level == 1) ? current_message_end : parent_message_ends.front();
     if (cursor != root_message_end)
     {
         if (cursor < root_message_end)
@@ -95,6 +96,9 @@ void ProtobufReader::SimpleReader::endMessage(bool ignore_errors)
 void ProtobufReader::SimpleReader::startNestedMessage()
 {
     assert(current_message_level >= 1);
+    if ((cursor > field_end) && (field_end != END_OF_GROUP))
+        throwUnknownFormat();
+
     // Start reading a nested message which is located inside a length-delimited field
     // of another message.
     parent_message_ends.emplace_back(current_message_end);
@@ -146,7 +150,7 @@ bool ProtobufReader::SimpleReader::readFieldNumber(UInt32 & field_number)
             throwUnknownFormat();
     }
 
-    if (cursor >= current_message_end)
+    if ((cursor >= current_message_end) && (current_message_end != END_OF_GROUP))
         return false;
 
     UInt64 varint = readVarint();
@@ -196,11 +200,17 @@ bool ProtobufReader::SimpleReader::readFieldNumber(UInt32 & field_number)
 
 bool ProtobufReader::SimpleReader::readUInt(UInt64 & value)
 {
+    if (field_end == END_OF_VARINT)
+    {
+        value = readVarint();
+        field_end = cursor;
+        return true;
+    }
+
     if (unlikely(cursor >= field_end))
         return false;
+
     value = readVarint();
-    if (field_end == END_OF_VARINT)
-        field_end = cursor;
     return true;
 }
 
@@ -227,6 +237,7 @@ bool ProtobufReader::SimpleReader::readFixed(T & value)
 {
     if (unlikely(cursor >= field_end))
         return false;
+
     readBinary(&value, sizeof(T));
     return true;
 }
diff --git a/src/Formats/ProtobufReader.h b/src/Formats/ProtobufReader.h
index a50c2f202f0..7c3c19ba376 100644
--- a/src/Formats/ProtobufReader.h
+++ b/src/Formats/ProtobufReader.h
@@ -124,12 +124,12 @@ private:
         void ignoreGroup();
 
         ReadBuffer & in;
-        UInt64 cursor;
+        Int64 cursor;
         size_t current_message_level;
-        UInt64 current_message_end;
-        std::vector<UInt64> parent_message_ends;
-        UInt64 field_end;
-        UInt64 last_string_pos;
+        Int64 current_message_end;
+        std::vector<Int64> parent_message_ends;
+        Int64 field_end;
+        Int64 last_string_pos;
     };
 
     class IConverter

From 8e63fd34b91c608feccc9ea2f803043deb73d72e Mon Sep 17 00:00:00 2001
From: Maxim Akhmedov <max42@yandex-team.ru>
Date: Thu, 28 May 2020 18:59:19 +0300
Subject: [PATCH 098/122] Bump build.


From ce52a2c820b60d8a44476b42d261248e809c7c62 Mon Sep 17 00:00:00 2001
From: Maxim Akhmedov <max42@yandex-team.ru>
Date: Thu, 28 May 2020 19:44:30 +0300
Subject: [PATCH 099/122] Add lost VERSION_* defines to ya.make.versions.inc

---
 cmake/yandex/ya.make.versions.inc | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/cmake/yandex/ya.make.versions.inc b/cmake/yandex/ya.make.versions.inc
index 5c3c5eabb60..3ac401cb108 100644
--- a/cmake/yandex/ya.make.versions.inc
+++ b/cmake/yandex/ya.make.versions.inc
@@ -11,6 +11,9 @@ CFLAGS (GLOBAL -DDBMS_VERSION_MAJOR=${VERSION_MAJOR})
 CFLAGS (GLOBAL -DDBMS_VERSION_MINOR=${VERSION_MINOR})
 CFLAGS (GLOBAL -DDBMS_VERSION_PATCH=${VERSION_PATCH})
 CFLAGS (GLOBAL -DVERSION_FULL=\"\\\"${VERSION_FULL}\\\"\")
+CFLAGS (GLOBAL -DVERSION_MAJOR=${VERSION_MAJOR})	
+CFLAGS (GLOBAL -DVERSION_MINOR=${VERSION_MINOR})	
+CFLAGS (GLOBAL -DVERSION_PATCH=${VERSION_PATCH})
 
 # TODO: not supported yet, not sure if ya.make supports arithmetics.
 CFLAGS (GLOBAL -DVERSION_INTEGER=0)

From 99626a5bd3fa3ba0470142793ce93faec1a0efc5 Mon Sep 17 00:00:00 2001
From: Vitaly Baranov <vitbar@yandex-team.ru>
Date: Thu, 28 May 2020 20:07:47 +0300
Subject: [PATCH 100/122] Add test for handling malformed protobuf data.

---
 .../0_stateless/00825_protobuf_format_input.reference    | 1 +
 tests/queries/0_stateless/00825_protobuf_format_input.sh | 9 ++++++++-
 2 files changed, 9 insertions(+), 1 deletion(-)

diff --git a/tests/queries/0_stateless/00825_protobuf_format_input.reference b/tests/queries/0_stateless/00825_protobuf_format_input.reference
index 884cc74c4e5..0c56bc4ebf0 100644
--- a/tests/queries/0_stateless/00825_protobuf_format_input.reference
+++ b/tests/queries/0_stateless/00825_protobuf_format_input.reference
@@ -8,3 +8,4 @@ a7522158-3d41-4b77-ad69-6c598ee55c49	Ivan	Petrov	male	1980-12-29	png	+7495123456
 0	0
 2	4
 3	9
+ok
diff --git a/tests/queries/0_stateless/00825_protobuf_format_input.sh b/tests/queries/0_stateless/00825_protobuf_format_input.sh
index d28b70bb002..1c915bc3f24 100755
--- a/tests/queries/0_stateless/00825_protobuf_format_input.sh
+++ b/tests/queries/0_stateless/00825_protobuf_format_input.sh
@@ -3,7 +3,7 @@
 CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
 . $CURDIR/../shell_config.sh
 
-set -e -o pipefail
+set -eo pipefail
 
 # Run the client.
 $CLICKHOUSE_CLIENT --multiquery <<'EOF'
@@ -48,5 +48,12 @@ source $CURDIR/00825_protobuf_format_input.insh
 $CLICKHOUSE_CLIENT --query "SELECT * FROM in_persons_00825 ORDER BY uuid;"
 $CLICKHOUSE_CLIENT --query "SELECT * FROM in_squares_00825 ORDER BY number;"
 
+# Try to input malformed data.
+set +eo pipefail
+echo -ne '\xe0\x80\x3f\x0b' \
+    | $CLICKHOUSE_CLIENT --query="INSERT INTO in_persons_00825 FORMAT Protobuf SETTINGS format_schema = '$CURDIR/00825_protobuf_format:Person'" 2>&1 \
+    | grep -qF "Protobuf messages are corrupted" && echo "ok" || echo "fail"
+set -eo pipefail
+
 $CLICKHOUSE_CLIENT --query "DROP TABLE IF EXISTS in_persons_00825;"
 $CLICKHOUSE_CLIENT --query "DROP TABLE IF EXISTS in_squares_00825;"

From a95469020adf75ca6e07e2294113539588a81989 Mon Sep 17 00:00:00 2001
From: alesapin <alesapin@gmail.com>
Date: Thu, 28 May 2020 21:02:41 +0300
Subject: [PATCH 101/122] Better error codes

---
 src/Common/ErrorCodes.cpp                       | 1 +
 src/DataStreams/CheckSortedBlockInputStream.cpp | 6 +++---
 2 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/src/Common/ErrorCodes.cpp b/src/Common/ErrorCodes.cpp
index cb4c591041c..5898e956400 100644
--- a/src/Common/ErrorCodes.cpp
+++ b/src/Common/ErrorCodes.cpp
@@ -498,6 +498,7 @@ namespace ErrorCodes
     extern const int ALTER_OF_COLUMN_IS_FORBIDDEN = 524;
     extern const int INCORRECT_DISK_INDEX = 525;
     extern const int UNKNOWN_VOLUME_TYPE = 526;
+    extern const int SORT_ORDER_VIOLATED = 527;
 
     extern const int KEEPER_EXCEPTION = 999;
     extern const int POCO_EXCEPTION = 1000;
diff --git a/src/DataStreams/CheckSortedBlockInputStream.cpp b/src/DataStreams/CheckSortedBlockInputStream.cpp
index a56738a9a9f..db4794b917f 100644
--- a/src/DataStreams/CheckSortedBlockInputStream.cpp
+++ b/src/DataStreams/CheckSortedBlockInputStream.cpp
@@ -6,7 +6,7 @@ namespace DB
 
 namespace ErrorCodes
 {
-extern const int LOGICAL_ERROR;
+extern const int SORT_ORDER_VIOLATED;
 }
 
 CheckSortedBlockInputStream::CheckSortedBlockInputStream(
@@ -72,12 +72,12 @@ Block CheckSortedBlockInputStream::readImpl()
 
     auto block_columns = block.getColumns();
     if (!last_row.empty() && !less(last_row, 0, block_columns, 0))
-        throw Exception("Sort order of blocks violated", ErrorCodes::LOGICAL_ERROR);
+        throw Exception("Sort order of blocks violated", ErrorCodes::SORT_ORDER_VIOLATED);
 
     size_t rows = block.rows();
     for (size_t i = 1; i < rows; ++i)
         if (!less(block_columns, i - 1, block_columns, i))
-            throw Exception("Sort order of blocks violated", ErrorCodes::LOGICAL_ERROR);
+            throw Exception("Sort order of blocks violated", ErrorCodes::SORT_ORDER_VIOLATED);
 
     last_row.clear();
     for (size_t i = 0; i < block.columns(); ++i)

From 93801a1a9dcfc0c8f801d465ce9611b30ace5976 Mon Sep 17 00:00:00 2001
From: alesapin <alesapin@gmail.com>
Date: Thu, 28 May 2020 21:52:08 +0300
Subject: [PATCH 102/122] Fix unit tests

---
 src/Common/tests/gtest_global_context.h       |  9 +++---
 src/Storages/tests/gtest_storage_log.cpp      |  6 ++--
 ..._transform_query_for_external_database.cpp | 28 +++++++++----------
 3 files changed, 22 insertions(+), 21 deletions(-)

diff --git a/src/Common/tests/gtest_global_context.h b/src/Common/tests/gtest_global_context.h
index 05f60e01774..959d51c6e0b 100644
--- a/src/Common/tests/gtest_global_context.h
+++ b/src/Common/tests/gtest_global_context.h
@@ -1,6 +1,7 @@
 #pragma once
 
 #include <Interpreters/Context.h>
+#include <Functions/registerFunctions.h>
 
 struct ContextHolder
 {
@@ -11,15 +12,15 @@ struct ContextHolder
         : shared_context(DB::Context::createShared())
         , context(DB::Context::createGlobal(shared_context.get()))
     {
+        context.makeGlobalContext();
+        context.setPath("./");
     }
 
     ContextHolder(ContextHolder &&) = default;
 };
 
-inline ContextHolder getContext()
+inline const ContextHolder & getContext()
 {
-    ContextHolder holder;
-    holder.context.makeGlobalContext();
-    holder.context.setPath("./");
+    static ContextHolder holder;
     return holder;
 }
diff --git a/src/Storages/tests/gtest_storage_log.cpp b/src/Storages/tests/gtest_storage_log.cpp
index 203104801aa..fff352210e7 100644
--- a/src/Storages/tests/gtest_storage_log.cpp
+++ b/src/Storages/tests/gtest_storage_log.cpp
@@ -68,7 +68,7 @@ using DiskImplementations = testing::Types<DB::DiskMemory, DB::DiskLocal>;
 TYPED_TEST_SUITE(StorageLogTest, DiskImplementations);
 
 // Returns data written to table in Values format.
-std::string writeData(int rows, DB::StoragePtr & table, DB::Context & context)
+std::string writeData(int rows, DB::StoragePtr & table, const DB::Context & context)
 {
     using namespace DB;
 
@@ -104,7 +104,7 @@ std::string writeData(int rows, DB::StoragePtr & table, DB::Context & context)
 }
 
 // Returns all table data in Values format.
-std::string readData(DB::StoragePtr & table, DB::Context & context)
+std::string readData(DB::StoragePtr & table, const DB::Context & context)
 {
     using namespace DB;
 
@@ -136,7 +136,7 @@ std::string readData(DB::StoragePtr & table, DB::Context & context)
 TYPED_TEST(StorageLogTest, testReadWrite)
 {
     using namespace DB;
-    auto context_holder = getContext();
+    const auto & context_holder = getContext();
 
     std::string data;
 
diff --git a/src/Storages/tests/gtest_transform_query_for_external_database.cpp b/src/Storages/tests/gtest_transform_query_for_external_database.cpp
index df6ee2bd0a5..f51e38d32a9 100644
--- a/src/Storages/tests/gtest_transform_query_for_external_database.cpp
+++ b/src/Storages/tests/gtest_transform_query_for_external_database.cpp
@@ -8,7 +8,6 @@
 #include <Interpreters/Context.h>
 #include <Databases/DatabaseMemory.h>
 #include <Storages/StorageMemory.h>
-#include <Functions/registerFunctions.h>
 #include <Common/tests/gtest_global_context.h>
 
 
@@ -18,7 +17,7 @@ using namespace DB;
 /// NOTE How to do better?
 struct State
 {
-    Context & context;
+    Context context;
     NamesAndTypesList columns{
         {"column", std::make_shared<DataTypeUInt8>()},
         {"apply_id", std::make_shared<DataTypeUInt64>()},
@@ -27,9 +26,10 @@ struct State
         {"create_time", std::make_shared<DataTypeDateTime>()},
     };
 
-    explicit State(Context & context_) : context(context_)
+    explicit State()
+        : context(getContext().context)
     {
-        registerFunctions();
+        DB::registerFunctions();
         DatabasePtr database = std::make_shared<DatabaseMemory>("test");
         database->attachTable("table", StorageMemory::create(StorageID("test", "table"), ColumnsDescription{columns}, ConstraintsDescription{}));
         context.makeGlobalContext();
@@ -38,6 +38,11 @@ struct State
     }
 };
 
+State getState()
+{
+    static State state;
+    return state;
+}
 
 static void check(const std::string & query, const std::string & expected, const Context & context, const NamesAndTypesList & columns)
 {
@@ -54,8 +59,7 @@ static void check(const std::string & query, const std::string & expected, const
 
 TEST(TransformQueryForExternalDatabase, InWithSingleElement)
 {
-    auto context_holder = getContext();
-    State state(context_holder.context);
+    const State & state = getState();
 
     check("SELECT column FROM test.table WHERE 1 IN (1)",
           R"(SELECT "column" FROM "test"."table" WHERE 1)",
@@ -70,8 +74,7 @@ TEST(TransformQueryForExternalDatabase, InWithSingleElement)
 
 TEST(TransformQueryForExternalDatabase, Like)
 {
-    auto context_holder = getContext();
-    State state(context_holder.context);
+    const State & state = getState();
 
     check("SELECT column FROM test.table WHERE column LIKE '%hello%'",
           R"(SELECT "column" FROM "test"."table" WHERE "column" LIKE '%hello%')",
@@ -83,8 +86,7 @@ TEST(TransformQueryForExternalDatabase, Like)
 
 TEST(TransformQueryForExternalDatabase, Substring)
 {
-    auto context_holder = getContext();
-    State state(context_holder.context);
+    const State & state = getState();
 
     check("SELECT column FROM test.table WHERE left(column, 10) = RIGHT(column, 10) AND SUBSTRING(column FROM 1 FOR 2) = 'Hello'",
           R"(SELECT "column" FROM "test"."table")",
@@ -93,8 +95,7 @@ TEST(TransformQueryForExternalDatabase, Substring)
 
 TEST(TransformQueryForExternalDatabase, MultipleAndSubqueries)
 {
-    auto context_holder = getContext();
-    State state(context_holder.context);
+    const State & state = getState();
 
     check("SELECT column FROM test.table WHERE 1 = 1 AND toString(column) = '42' AND column = 42 AND left(column, 10) = RIGHT(column, 10) AND column IN (1, 42) AND SUBSTRING(column FROM 1 FOR 2) = 'Hello' AND column != 4",
           R"(SELECT "column" FROM "test"."table" WHERE 1 AND ("column" = 42) AND ("column" IN (1, 42)) AND ("column" != 4))",
@@ -106,8 +107,7 @@ TEST(TransformQueryForExternalDatabase, MultipleAndSubqueries)
 
 TEST(TransformQueryForExternalDatabase, Issue7245)
 {
-    auto context_holder = getContext();
-    State state(context_holder.context);
+    const State & state = getState();
 
     check("select apply_id from test.table where apply_type = 2 and create_time > addDays(toDateTime('2019-01-01 01:02:03'),-7) and apply_status in (3,4)",
           R"(SELECT "apply_id", "apply_type", "apply_status", "create_time" FROM "test"."table" WHERE ("apply_type" = 2) AND ("create_time" > '2018-12-25 01:02:03') AND ("apply_status" IN (3, 4)))",

From e9c1e1a7e4913fafb1164b8633b79bf76a4fd48a Mon Sep 17 00:00:00 2001
From: alesapin <alesapin@gmail.com>
Date: Thu, 28 May 2020 21:54:42 +0300
Subject: [PATCH 103/122] Remove redundant header

---
 src/Common/tests/gtest_global_context.h | 1 -
 1 file changed, 1 deletion(-)

diff --git a/src/Common/tests/gtest_global_context.h b/src/Common/tests/gtest_global_context.h
index 959d51c6e0b..b6529f09b46 100644
--- a/src/Common/tests/gtest_global_context.h
+++ b/src/Common/tests/gtest_global_context.h
@@ -1,7 +1,6 @@
 #pragma once
 
 #include <Interpreters/Context.h>
-#include <Functions/registerFunctions.h>
 
 struct ContextHolder
 {

From 026244b07f796a5cd36c80ea20f3484b629b68ee Mon Sep 17 00:00:00 2001
From: alesapin <alesapin@gmail.com>
Date: Thu, 28 May 2020 21:57:09 +0300
Subject: [PATCH 104/122] Better

---
 .../tests/gtest_transform_query_for_external_database.cpp      | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/src/Storages/tests/gtest_transform_query_for_external_database.cpp b/src/Storages/tests/gtest_transform_query_for_external_database.cpp
index f51e38d32a9..6caf759da42 100644
--- a/src/Storages/tests/gtest_transform_query_for_external_database.cpp
+++ b/src/Storages/tests/gtest_transform_query_for_external_database.cpp
@@ -8,6 +8,7 @@
 #include <Interpreters/Context.h>
 #include <Databases/DatabaseMemory.h>
 #include <Storages/StorageMemory.h>
+#include <Functions/registerFunctions.h>
 #include <Common/tests/gtest_global_context.h>
 
 
@@ -29,7 +30,7 @@ struct State
     explicit State()
         : context(getContext().context)
     {
-        DB::registerFunctions();
+        registerFunctions();
         DatabasePtr database = std::make_shared<DatabaseMemory>("test");
         database->attachTable("table", StorageMemory::create(StorageID("test", "table"), ColumnsDescription{columns}, ConstraintsDescription{}));
         context.makeGlobalContext();

From db72f3959489ef54261071390e20703dea5caf51 Mon Sep 17 00:00:00 2001
From: alesapin <alesapin@gmail.com>
Date: Thu, 28 May 2020 22:07:20 +0300
Subject: [PATCH 105/122] Remove error code

---
 src/Common/ErrorCodes.cpp                       | 1 -
 src/DataStreams/CheckSortedBlockInputStream.cpp | 6 +++---
 2 files changed, 3 insertions(+), 4 deletions(-)

diff --git a/src/Common/ErrorCodes.cpp b/src/Common/ErrorCodes.cpp
index 5898e956400..cb4c591041c 100644
--- a/src/Common/ErrorCodes.cpp
+++ b/src/Common/ErrorCodes.cpp
@@ -498,7 +498,6 @@ namespace ErrorCodes
     extern const int ALTER_OF_COLUMN_IS_FORBIDDEN = 524;
     extern const int INCORRECT_DISK_INDEX = 525;
     extern const int UNKNOWN_VOLUME_TYPE = 526;
-    extern const int SORT_ORDER_VIOLATED = 527;
 
     extern const int KEEPER_EXCEPTION = 999;
     extern const int POCO_EXCEPTION = 1000;
diff --git a/src/DataStreams/CheckSortedBlockInputStream.cpp b/src/DataStreams/CheckSortedBlockInputStream.cpp
index db4794b917f..a56738a9a9f 100644
--- a/src/DataStreams/CheckSortedBlockInputStream.cpp
+++ b/src/DataStreams/CheckSortedBlockInputStream.cpp
@@ -6,7 +6,7 @@ namespace DB
 
 namespace ErrorCodes
 {
-extern const int SORT_ORDER_VIOLATED;
+extern const int LOGICAL_ERROR;
 }
 
 CheckSortedBlockInputStream::CheckSortedBlockInputStream(
@@ -72,12 +72,12 @@ Block CheckSortedBlockInputStream::readImpl()
 
     auto block_columns = block.getColumns();
     if (!last_row.empty() && !less(last_row, 0, block_columns, 0))
-        throw Exception("Sort order of blocks violated", ErrorCodes::SORT_ORDER_VIOLATED);
+        throw Exception("Sort order of blocks violated", ErrorCodes::LOGICAL_ERROR);
 
     size_t rows = block.rows();
     for (size_t i = 1; i < rows; ++i)
         if (!less(block_columns, i - 1, block_columns, i))
-            throw Exception("Sort order of blocks violated", ErrorCodes::SORT_ORDER_VIOLATED);
+            throw Exception("Sort order of blocks violated", ErrorCodes::LOGICAL_ERROR);
 
     last_row.clear();
     for (size_t i = 0; i < block.columns(); ++i)

From d7ec509f337a88b0125b5af48d981bb4b98ba604 Mon Sep 17 00:00:00 2001
From: alexey-milovidov <milovidov@yandex-team.ru>
Date: Thu, 28 May 2020 22:26:20 +0300
Subject: [PATCH 106/122] Update extended-roadmap.md

---
 docs/ru/whats-new/extended-roadmap.md | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/docs/ru/whats-new/extended-roadmap.md b/docs/ru/whats-new/extended-roadmap.md
index b7756f7fb79..9b92dbfa717 100644
--- a/docs/ru/whats-new/extended-roadmap.md
+++ b/docs/ru/whats-new/extended-roadmap.md
@@ -203,10 +203,11 @@ Upd. SharedContext вынесен из Context.
 
 Upd. В очереди. Иван Лежанкин.
 
-### 2.9. Логгировние в format-стиле {#loggirovnie-v-format-stile}
+### 2.9. + Логгировние в format-стиле {#loggirovnie-v-format-stile}
 
-Делает [Иван Лежанкин](https://github.com/abyss7). Низкий приоритет.
-[\#6049](https://github.com/ClickHouse/ClickHouse/issues/6049#issuecomment-570836998)
+[#6049](https://github.com/ClickHouse/ClickHouse/issues/6049#issuecomment-570836998)
+
+Сделано.
 
 ### 2.10. Запрашивать у таблиц не столбцы, а срезы {#zaprashivat-u-tablits-ne-stolbtsy-a-srezy}
 

From d8049be204ed7325c5725bdb0b2a49fef9cd8464 Mon Sep 17 00:00:00 2001
From: alexey-milovidov <milovidov@yandex-team.ru>
Date: Thu, 28 May 2020 22:27:31 +0300
Subject: [PATCH 107/122] Update extended-roadmap.md

---
 docs/ru/whats-new/extended-roadmap.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/docs/ru/whats-new/extended-roadmap.md b/docs/ru/whats-new/extended-roadmap.md
index 9b92dbfa717..0bedfda57b1 100644
--- a/docs/ru/whats-new/extended-roadmap.md
+++ b/docs/ru/whats-new/extended-roadmap.md
@@ -73,10 +73,10 @@ Upd. Включено для системных таблиц.
 Q1. Закоммичено, но есть технический долг, который исправляется сейчас.
 Готово. Нет, не готово - там всё ещё технический долг.
 
-### 1.9. Использование TTL для прореживания данных {#ispolzovanie-ttl-dlia-prorezhivaniia-dannykh}
+### 1.9. + Использование TTL для прореживания данных {#ispolzovanie-ttl-dlia-prorezhivaniia-dannykh}
 
 Будет делать Сорокин Николай, ВШЭ и Яндекс.
-Upd. Есть pull request.
+Upd. Есть pull request. Upd. Сделано.
 
 Сейчас пользователь может задать в таблице выражение, которое определяет, сколько времени хранятся данные. Обычно это выражение задаётся относительно значения столбца с датой - например: удалять данные через три месяца. https://clickhouse.tech/docs/ru/operations/table_engines/mergetree/\#table_engine-mergetree-ttl
 

From 24fb6266daa54d4aa8a7ad9e7bcfa98da9afe34b Mon Sep 17 00:00:00 2001
From: Alexey Milovidov <milovidov@yandex-team.ru>
Date: Thu, 28 May 2020 22:28:44 +0300
Subject: [PATCH 108/122] Fix wrong markup

---
 website/css/docs.css | 1 -
 1 file changed, 1 deletion(-)

diff --git a/website/css/docs.css b/website/css/docs.css
index 18bf40de13b..2c175b18db6 100644
--- a/website/css/docs.css
+++ b/website/css/docs.css
@@ -69,7 +69,6 @@ summary {
 #content code {
     color: #111;
     background: #eee;
-    padding: 2px;
 }
 
 

From 05642d2a84bae35313be9050070425158e8b596e Mon Sep 17 00:00:00 2001
From: alexey-milovidov <milovidov@yandex-team.ru>
Date: Thu, 28 May 2020 23:53:13 +0300
Subject: [PATCH 109/122] Update extended-roadmap.md

---
 docs/ru/whats-new/extended-roadmap.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/docs/ru/whats-new/extended-roadmap.md b/docs/ru/whats-new/extended-roadmap.md
index 0bedfda57b1..091e7f37c66 100644
--- a/docs/ru/whats-new/extended-roadmap.md
+++ b/docs/ru/whats-new/extended-roadmap.md
@@ -124,7 +124,7 @@ Q2.
 Upd. Олег будет делать только часть про HDFS.
 Upd. Реализация поверх S3 является рабочей на уровне PoC.
 
-### 1.13. Ускорение запросов с FINAL {#uskorenie-zaprosov-s-final}
+### 1.13. + Ускорение запросов с FINAL {#uskorenie-zaprosov-s-final}
 
 Требует 2.1. Делает [Николай Кочетов](https://github.com/KochetovNicolai). Нужно для Яндекс.Метрики. Q2.
 Upd: PR [#10463](https://github.com/ClickHouse/ClickHouse/pull/10463)

From 2918677a9e3ece71b5403e4d72b9275bbcba71ff Mon Sep 17 00:00:00 2001
From: alexey-milovidov <milovidov@yandex-team.ru>
Date: Thu, 28 May 2020 23:58:25 +0300
Subject: [PATCH 110/122] Update extended-roadmap.md

---
 docs/ru/whats-new/extended-roadmap.md | 13 ++++++-------
 1 file changed, 6 insertions(+), 7 deletions(-)

diff --git a/docs/ru/whats-new/extended-roadmap.md b/docs/ru/whats-new/extended-roadmap.md
index 091e7f37c66..895138e1a33 100644
--- a/docs/ru/whats-new/extended-roadmap.md
+++ b/docs/ru/whats-new/extended-roadmap.md
@@ -283,24 +283,20 @@ Upd. Сейчас обсуждается, как сделать другую з
 
 ### 4.3. Ограничение числа одновременных скачиваний с реплик {#ogranichenie-chisla-odnovremennykh-skachivanii-s-replik}
 
-Дмитрий Григорьев, ВШЭ.
 Изначально делал Олег Алексеенков, но пока решение не готово, хотя там не так уж много доделывать.
 
 ### 4.4. Ограничение сетевой полосы при репликации {#ogranichenie-setevoi-polosy-pri-replikatsii}
 
-Дмитрий Григорьев, ВШЭ. Нужно для Метрики.
+Нужно для Метрики.
 
 ### 4.5. Возможность продолжить передачу куска данных при репликации после сбоя {#vozmozhnost-prodolzhit-peredachu-kuska-dannykh-pri-replikatsii-posle-sboia}
 
-Дмитрий Григорьев, ВШЭ.
-
 ### 4.6. p2p передача для GLOBAL подзапросов {#p2p-peredacha-dlia-global-podzaprosov}
 
 ### 4.7. Ленивая загрузка множеств для IN и JOIN с помощью k/v запросов {#lenivaia-zagruzka-mnozhestv-dlia-in-i-join-s-pomoshchiu-kv-zaprosov}
 
 ### 4.8. Разделить background pool для fetch и merge {#razdelit-background-pool-dlia-fetch-i-merge}
 
-Дмитрий Григорьев, ВШЭ.
 В очереди. Исправить проблему, что восстанавливающаяся реплика перестаёт мержить. Частично компенсируется 4.3.
 
 
@@ -330,6 +326,7 @@ Upd. Сделано. Эффективность работы под вопрос
 Метрика, БК, Маркет, Altinity уже используют более свежие версии чем LTS.
 Upd. Появилась вторая версия LTS - 20.3.
 
+
 ## 6. Инструментирование {#instrumentirovanie}
 
 ### 6.1. + Исправления сэмплирующего профайлера запросов {#ispravleniia-sempliruiushchego-profailera-zaprosov}
@@ -426,11 +423,11 @@ Upd. Рассмотрели все проверки подряд.
 
 UBSan включен в функциональных тестах, но не включен в интеграционных тестах. Требует 7.7.
 
-### 7.11. Включение \*San в unit тестах {#vkliuchenie-san-v-unit-testakh}
+### 7.11. + Включение \*San в unit тестах {#vkliuchenie-san-v-unit-testakh}
 
 У нас мало unit тестов по сравнению с функциональными тестами и их использование не обязательно. Но они всё-равно важны и нет причин не запускать их под всеми видами sanitizers.
 
-Илья Яцишин.
+Илья Яцишин. Сделано.
 
 ### 7.12. Показывать тестовое покрытие нового кода в PR {#pokazyvat-testovoe-pokrytie-novogo-koda-v-pr}
 
@@ -529,6 +526,8 @@ Upd. Есть сборки, [пример](https://clickhouse-builds.s3.yandex.n
 
 Дарья Петрова, УрФУ.
 
+Рабочий прототип: https://pulls-dashboard-demo.herokuapp.com/dashboard/ClickHouse/ClickHouse
+
 Над ClickHouse одновременно работает большое количество разработчиков, которые оформляют свои изменения в виде pull requests. Когда непомерженных pull requests много, то возникает сложность с организацией работы - непонятно, на какой pull request смотреть в первую очередь.
 
 Предлагается реализовать простое одностраничное веб-приложение, в котором отображается список pull requests со следующей информацией:

From f0f7b7cc7cafba1376e2ac33f142f973686ea136 Mon Sep 17 00:00:00 2001
From: alexey-milovidov <milovidov@yandex-team.ru>
Date: Fri, 29 May 2020 00:21:03 +0300
Subject: [PATCH 111/122] Update extended-roadmap.md

---
 docs/ru/whats-new/extended-roadmap.md | 1 +
 1 file changed, 1 insertion(+)

diff --git a/docs/ru/whats-new/extended-roadmap.md b/docs/ru/whats-new/extended-roadmap.md
index 895138e1a33..df4bbb4ca57 100644
--- a/docs/ru/whats-new/extended-roadmap.md
+++ b/docs/ru/whats-new/extended-roadmap.md
@@ -627,6 +627,7 @@ Upd. Готово (все директории кроме contrib).
 ### 7.32. Обфускация продакшен запросов {#obfuskatsiia-prodakshen-zaprosov}
 
 Роман Ильговский. Нужно для Яндекс.Метрики.
+Есть pull request, почти готово: https://github.com/ClickHouse/ClickHouse/pull/10973
 
 Имея SQL запрос, требуется вывести структуру таблиц, на которых этот запрос будет выполнен, и заполнить эти таблицы случайными данными, такими, что результат этого запроса зависит от выбора подмножества данных.
 

From 9d5b1b69f7c3d6279e09ab593c22f6b1ec3ec7bf Mon Sep 17 00:00:00 2001
From: alexey-milovidov <milovidov@yandex-team.ru>
Date: Fri, 29 May 2020 00:22:18 +0300
Subject: [PATCH 112/122] Update extended-roadmap.md

---
 docs/ru/whats-new/extended-roadmap.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/docs/ru/whats-new/extended-roadmap.md b/docs/ru/whats-new/extended-roadmap.md
index df4bbb4ca57..20ebe28fe5b 100644
--- a/docs/ru/whats-new/extended-roadmap.md
+++ b/docs/ru/whats-new/extended-roadmap.md
@@ -1398,11 +1398,11 @@ Constraints позволяют задать выражение, истиннос
 Василий Морозов, Арслан Гумеров, Альберт Кидрачев, ВШЭ.
 В прошлом году задачу начинал делать другой человек, но не добился достаточного прогресса.
 
-1.  Оптимизация top sort.
++ 1.  Оптимизация top sort.
 
 В ClickHouse используется неоптимальный вариант top sort. Суть его в том, что из каждого блока достаётся top N записей, а затем, все блоки мержатся. Но доставание top N записей у каждого следующего блока бессмысленно, если мы знаем, что из них в глобальный top N войдёт меньше. Конечно нужно реализовать вариацию на тему priority queue (heap) с быстрым пропуском целых блоков, если ни одна строка не попадёт в накопленный top.
 
-2.  Рекурсивный вариант сортировки по кортежам.
++ 2.  Рекурсивный вариант сортировки по кортежам.
 
 Для сортировки по кортежам используется обычная сортировка с компаратором, который в цикле по элементам кортежа делает виртуальные вызовы `IColumn::compareAt`. Это неоптимально - как из-за короткого цикла по неизвестному в compile-time количеству элементов, так и из-за виртуальных вызовов. Чтобы обойтись без виртуальных вызовов, есть метод `IColumn::getPermutation`. Он используется в случае сортировки по одному столбцу. Есть вариант, что в случае сортировки по кортежу, что-то похожее тоже можно применить… например, сделать метод `updatePermutation`, принимающий аргументы offset и limit, и допереставляющий перестановку в диапазоне значений, в которых предыдущий столбец имел равные значения.
 

From 2b4d9a6349a6c530d1574577254bde216ccced84 Mon Sep 17 00:00:00 2001
From: Alexey Milovidov <milovidov@yandex-team.ru>
Date: Fri, 29 May 2020 01:00:01 +0300
Subject: [PATCH 113/122] Add a test from @donmikel

---
 .../01285_engine_join_donmikel.reference      |   1 +
 .../0_stateless/01285_engine_join_donmikel.sh |  65 ++++++++++++
 .../01285_engine_join_donmikel.tsv            | 100 ++++++++++++++++++
 3 files changed, 166 insertions(+)
 create mode 100644 tests/queries/0_stateless/01285_engine_join_donmikel.reference
 create mode 100755 tests/queries/0_stateless/01285_engine_join_donmikel.sh
 create mode 100644 tests/queries/0_stateless/01285_engine_join_donmikel.tsv

diff --git a/tests/queries/0_stateless/01285_engine_join_donmikel.reference b/tests/queries/0_stateless/01285_engine_join_donmikel.reference
new file mode 100644
index 00000000000..573541ac970
--- /dev/null
+++ b/tests/queries/0_stateless/01285_engine_join_donmikel.reference
@@ -0,0 +1 @@
+0
diff --git a/tests/queries/0_stateless/01285_engine_join_donmikel.sh b/tests/queries/0_stateless/01285_engine_join_donmikel.sh
new file mode 100755
index 00000000000..8a867921dbe
--- /dev/null
+++ b/tests/queries/0_stateless/01285_engine_join_donmikel.sh
@@ -0,0 +1,65 @@
+#!/usr/bin/env bash
+
+CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
+. $CURDIR/../shell_config.sh
+
+$CLICKHOUSE_CLIENT --multiquery --query "
+DROP TABLE IF EXISTS NmSubj;
+DROP TABLE IF EXISTS events;
+
+create table NmSubj
+(
+    NmId      UInt32,
+    SubjectId UInt32
+)
+    engine = Join(All, inner, NmId);
+
+create table events
+(
+    EventDate       Date,
+    EventDateTime   DateTime,
+    EventId         String,
+    SessionId       FixedString(36),
+    PageViewId      FixedString(36),
+    UserId          UInt64,
+    UniqUserId      FixedString(36),
+    UrlReferrer     String,
+    Param1          String,
+    Param2          String,
+    Param3          String,
+    Param4          String,
+    Param5          String,
+    Param6          String,
+    Param7          String,
+    Param8          String,
+    Param9          String,
+    Param10         String,
+    ApplicationType UInt8,
+    Locale          String,
+    Lang            String,
+    Version         String,
+    Path            String,
+    QueryString     String,
+    UserHostAddress UInt32
+)
+    engine = MergeTree()
+        PARTITION BY (toYYYYMM(EventDate), EventId)
+        ORDER BY (EventId, EventDate, Locale, ApplicationType, intHash64(UserId))
+        SAMPLE BY intHash64(UserId)
+        SETTINGS index_granularity = 8192;
+
+insert into NmSubj values (1, 1), (2, 2), (3, 3);
+"
+
+$CLICKHOUSE_CLIENT --query "INSERT INTO events FORMAT TSV" < ${CURDIR}/01285_engine_join_donmikel.tsv
+
+$CLICKHOUSE_CLIENT --query "
+SELECT toInt32(count() / 24) as Count
+FROM events as e INNER JOIN NmSubj as ns
+ON ns.NmId = toUInt32(e.Param1)
+WHERE e.EventDate = today() - 7 AND e.EventId = 'GCO' AND ns.SubjectId = 2073"
+
+$CLICKHOUSE_CLIENT --multiquery --query "
+DROP TABLE NmSubj;
+DROP TABLE events;
+"
diff --git a/tests/queries/0_stateless/01285_engine_join_donmikel.tsv b/tests/queries/0_stateless/01285_engine_join_donmikel.tsv
new file mode 100644
index 00000000000..a6f47144ff2
--- /dev/null
+++ b/tests/queries/0_stateless/01285_engine_join_donmikel.tsv
@@ -0,0 +1,100 @@
+2020-02-17	2020-02-17 19:40:09	EMC	wFxiX8iZ3Vns8pO6b�x0Wv�YftGEOM9WMff	Ku5ofnKw�BaQ0CtcdZ�a6sA�nSQbyufu0zfd	0	X16qWFb_�h8JO�NLGxs7oiXFZTdBFfTniRMG		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=33&userId=274574634&email=Z21haWwucnU%3D&source=mn	657583482
+2020-02-17	2020-02-17 20:23:53	EMC	foco1fj5Li6ey>MSfn?jKHBBdiwKXiBmy7Ni	ICZBjN7en5snnszHKbTVUU*xyOhuZEgI1EK	0	I7XAKnCMsD9tdFHmm�cCrI~KqiKQdo7Gxd5		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=E17727463464LnJ1&source=mn	2992671877
+2020-02-17	2020-02-17 19:49:04	EMC	q3AxEdMQhsr1NewVSK�jHzYvh63XdxtGcBZX	GVGMPZhe�nJd8�3Jar�I7Qicq5BKRfijT0ls	18536997	D_myvD3U4bmy5anwdNRFaxm KNoDfcCDL9U6	http&o=mail.ru/redir?u=http&o=mail.ru/redir?u=https%3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6userId	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=AD0581918974621haWwuY29t&source=mn	1656424571
+2020-02-17	2020-02-17 21:55:32	EMC	ef7tm7k7�Yvpb4XIfvf7ON-s4Xv3ztD7SkB	KIDkUzEg�cAMu%ObAQ�AkI3�jmm3638bSnX0	22341123	6NcB26yJrLhWg�ldC_HvV2f�UOHNLRFvkLbA		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=E943634627468871D581940274695&userId=2746887723D&source=mn	1429586160
+2020-02-17	2020-02-17 21:37:47	EMC	3Y_KZLEJgn4bw?omv4�EizIs1lUMz0Li0Ws0	I3QaVG9U�vXWv�WReLSHuDC�s3DdgTbX8av6	28213800	Knoofunr�qFKdIta8�pMuR!FNX2yu1I6gNo	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https://click.mail&v=2&s=3745772	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=5YWxpbC5jb20%3D&source=mn	872136829
+2020-02-17	2020-02-17 19:15:34	EMC	JcHEtSzM�b70K}04FW#ShpZ�w1ot_fb7wxj8	KG9bmRo6�yweN�2EX4�8CUBQcsOqzbHn6g3u	30047107	UKb5rJ5D\bFncBHxCioYswtJKAlv6Ua5MVSrj		mn										48	Ru	ru	7.0.32	/confirm/vipstatus	?key=75ydQ%3D%3D&source=mn	626057004
+2020-02-17	2020-02-17 18:15:04	EMC	_sjptcVU�5CF4�sTnX�h0C6�DCYcOo0rNowf	0a0AcabO�llf7\rN9X9�aa7r�VmJ0J1oZGJY2	30648090	Z2K_4kff�e21iW1hl2�jby1RXAAh5XtPLP3		mn										48	Ru	ru	7.0.32	/confirm/vipstatus	?key=5&ts=15819420%3D&source=mn	3682305993
+2020-02-17	2020-02-17 16:10:20	EMC	WM44OUvE�Zse_9hQ0NJsn0�k47pX6_UnaS7	LgCWwTMZ+GwiQEbNT95zUn�kMCXYqPs1Xt1	0	QUuSVpqq#SgMdveJDW�0bQ6�s8jn8MLXzL1w		mn										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=CCD&source=mn	726391852
+2020-02-17	2020-02-17 17:30:44	EMC	PGfQVZHu�AY0A�_kUA\04ceY�RjWqgPJtfhCn	4TR9oRKCNrouy:zJBLBi8J4=u7Mkt9jgzgmB	0	xbCoBX9pTDaW�iRxW@F3fO�gKD9lX58oWsc		mn										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=DB3D&source=pcap	2498835913
+2020-02-17	2020-02-17 19:10:51	EMC	gnd66xWl�Kf2H�2tiZ�dbiv�joz4ILmJ4lp_	ysgw6WLd�bNWS�HQry�bWjh�H9uJqnlZMWAv	0	nDQDqsnm/dlpbFMPJE93b6�EETHzFxUSzVv		pcap										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=4&userId=274634688&userId=274688&userId=27465D85CD4F634627468877266792YXJ1&source=mn	816668334
+2020-02-17	2020-02-17 20:52:00	EMC	pBzkcTpn50WpO]BZH7^js6TpPGp7Uid7mFg	jQVdKDl4+Sp_j�5CZcPtVo�MGTIGOfYO6Q2	0	aQGyrF9s3qUMPq_dUn�OiM6�QVMX5PNQ_eiV		pcap										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=50&email=YWlsLmNvbQ%3D%3D&source=pcap	1770779506
+2020-02-17	2020-02-17 21:31:14	EMC	cyTu3wNR�CI1TzkxCy�bFNz\rCkrdDZQxYtlW	gmtJF_hWE8M79bWSaK�kn8UT5Xdwfb0uK8T	0	RSJUU0SB�DIXj�pYOR�na59�cWouc20W5VIP		mn										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=4CD%3D&source=mn	3448881839
+2020-02-17	2020-02-17 17:15:46	EMC	I3bvr8ce�2SU_&UuDX�4J6K�RJl5uipnL2Or	myxD8nQX�FYnK�qBoP�GGo4Hy6hBTXRL04K	0	ntHjLMvh�nb7b�09Tu�7jRkM8KgS5ni0w4oH		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=36274695&userId=2560&userId=27463465&userId=27468872EDB908&ts=158193819382	2447621965
+2020-02-17	2020-02-17 18:21:21	EMC	_468tOVLPpIk�gKvP�GX8y�pjfTZV09k8w5	0KakGIWk�9uaL�0Rv3]1T5wPeHPklVWzhytg	0	QAJxAuQA�DoEQ�qY0H�UT56�P0O4K5y55K_L	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https://mail&v=2&s=92Ymail&v=2&s=3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D66email&v=2&s=93D158192YWlsLnJ1%26source%3Dmn&c=swm&r=https%3A%2F%2Fwww.wildberries.ru	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=5819392158193634&userId=274688&ts=15819408BD1haWwucnU%3D&source=mn	1547300784
+2020-02-17	2020-02-17 19:07:33	EMC	Y8yBoZ7C�p2GU�TQ8srLMZJ�FIZ85WgiBeau	wmIgwW88aSGvvjKwV,8Eoo�QonI_90taLrh	0	KXVqerLQyaMaMtuT1FzUjr�Lev8Fw11mkfO		pcap										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=72846239707&userId=274688&ts=1581921haWwucnU%3D&source=pcap	2777789095
+2020-02-17	2020-02-17 11:27:36	EMC	n66D1ZN9�C6y7�LASQ�ucHNdkw_QseVRnt32	yeqFeZFa,7cQ_JyTO3yJJKSQaCKEWERPx4Z	24388614	6h91HeiP�yIKw8yV0p\tcaK99w_zmnmXU77RI	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https://click.mail.ru/redir?u=https://click.mail&v=2&s=3D158192YWlsLnJ1%26source%3Dmn&c=swm&r=http&o=mail.ru/redir?u=https://click.mail.ru/redir?u=https://mail%3DYWlsLnJ1%26source	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=6167462E4CD570A0312D6BF1BtYWlsLnJ1&source=mn	662531688
+2020-02-17	2020-02-17 14:11:06	EMC	jamANU2z�ei_lBkfEk_Im5ow0gWB5G0gL1S	So7Jc6gheYgclfoq8�l0hq�sPExrMNSXPLV	20685668	nwRXATVEY2PaYr34qw�tZX0�Q6ah_sceXjbL	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=http&o=mail&v=2&s=9444%26email.ru/redir?u=https%3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69103Dmn&c=swm&r=http&o=mail&v=2&s=92YWlsLnJ1%26source%3Dpc&c=swm&r=https	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=420%3D&source=mn	3233174651
+2020-02-17	2020-02-17 14:32:04	EMC	lsruHZ0Ywg1yDdrio.j5pi�NMTnqNxeaOQJ	G6K62xDb�m9aJ_kNlLqEo4AhVf46wDtWi	32046518	wgpf1yqV�tVhLL7yT9sp1RbqaXGF81GmSiQ6		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=1581943D&source=pcap	3183159832
+2020-02-17	2020-02-17 15:11:37	EMC	5EW8o4vD�850H>TQam�3NA2@egQm32jpASIF	KJKN5QAHdoqw_�RGol�5hLv�Sq6753cKLswM	32971244	8Vh3DVlC�jfXs[5hqV�MGDDP2iBjoiqhoACB	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https://mail&v=2&s=3D27457AD3D274576userId%3D27461haWwucnU%253D%26source%3Dmn&c=swm&r=https%3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6ts%3D158193Fkey%3D669%26email&v=2&s=3Fkey%3D69D01	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=5634&userId=274695&ts=15819440446274695&userId=27465&email=Z21haWwucnU%3D&source=pcap	574798405
+2020-02-17	2020-02-17 13:59:25	EMC	i6LAlVL3U0GUF�Ftji�RgfPyjyattjzds6y2	IYv_jZ2k+bNO3�3nbL�KRRG�cJvTHVIueEDI	21779069	eUgOCUGbEZM15\'Of0V�z6zQmAT0T0zX3XSO	http&o=mail.ru/redir?u=http&o=mail.ru/redir?u=https%3A%2F%2Fwww.wildberries	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=D1F9sail=YWxpbC5ydQ%3D%3D&source=mn	3475274476
+2020-02-17	2020-02-17 06:40:47	EMC	A_3372QT5xhTm�jNrokk3y�ZCtuOm52FfZT	oOD7es9h�GUs7�jB3_)jVKh�3Oo10IQvAu01	27168564	UYEHX0it�bLKZ�jChW�xLsP�4a7LUXQLX4iB		mn										48	Ru	ru	7.0.32	/confirm/vipstatus	?key=467468874634&email=dG92&userId=274695&email=email=a2B7AD60&email=bWFpbC5ydQ%3D%3D&source=mn	3084609784
+2020-02-17	2020-02-17 14:10:13	EMC	WSlfONgz�mGj8�kylb�ItxLW8MX4u60oapxy	U8TMovlAcJTcA~u3KoOFG7�pchP3CBPbcB4	28904171	nBfRyelBAkOfR�70tIzBusu�S0M8_KbRsOGI		pcap										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=4462F5B37274695&email=Ymail=Z21haWwucnU%3D&source=mn	2020344465
+2020-02-17	2020-02-17 01:43:57	EMC	khjsNWq7Iq56�4D1u\'vWuK�ZHN4kfq0Kyfv	hcAK5RGO�jZHY�bFJw�ufN7\tZeXCRWAEKoT2	0	G0Zi9TAk`t0aU�cHRi�Ocb1|RgoT8xMOH4KV		mn										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=1581940095&ts=158193819382&userId=27465D8B8F1936256C35B4B466&userId=2746887C1A97&userId=274695&userId=274634&userId	902363005
+2020-02-17	2020-02-17 02:17:35	EMC	H56Q4qwa\'eYiF�CmGR40Pp7,m6fDiE2nJSmB	Efk7W3ZrgNsEi.anvkrteuN>lF2UlX0RZFzr	0	QhyRjVGEsD_4YaQiaL�orQ6�y0fbCxkEVzQ5		pcap										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=44695&email=YW5hcnU%3D&source=mn	2161496531
+2020-02-17	2020-02-17 03:20:48	EMC	1wBv9Q7r�DQuImuif>0v1t�8OXWSIXymKrI	sNujW7ia�tZLPxEOpI1fZdT4p0QRaotsXPZ	0	LSKNxAcr�TMsADK9VI7AMqm�05blzLWXCJ05		pcap										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=AE1FAC2BFA1E75hce=mn	985607445
+2020-02-17	2020-02-17 06:18:16	EMC	6dqrbyyKHWPBE�yPZG�bJRKjxtmt0Aq8IsON	k57VHYX59Ws9u[L3nN�EruO�GCVgGH_l27J1	0	_lwuZQyc�qGeyMxkRSM7pR1�lHNU6_V0nVWG		pcap										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=B27465D88&userId=274572BE3274577273AF0&email=a274578&userId=274692&userId=274688&userId=27465D8192	130983489
+2020-02-17	2020-02-17 09:20:13	EMC	UCLTP56_�YcQ9�rWVJa8rif2Aq5NNqVThlwH	7Rx9OsBs�7Bem�VxuA�rwZ7N5BqcFwK7CSI	0	JOIbkOPd�PPbP�CtqRV6wQH�sdWYNnmarCxm		mn										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=A6E819420%3D&source=pcap	2567294448
+2020-02-17	2020-02-17 13:05:00	EMC	28AZ4n0zpsVuwzenZ6*gd0V�NMjnT8Kmfncv	PFiGZRLT�jK1CxfLJajtGVFOATfWnTE00	0	QYPUzvwV=w8ev0E77B|RK6Z<66ZX5y1YIK1J		mn										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=C19213C0060312E9E28A82&email=bmFA2095&userId=2746346&userId=27457455266&ts=15819213D&source=pcap	2881908356
+2020-02-17	2020-02-17 13:40:53	EMC	uDOkWKoT�LCbB2SFYoepda0itJJ9zJgyZVY	d50Y87RQCO70�ci0u{QDZc�A8dU31qm0IS1	0	0MuqyXoT�tKYp�mheI�D1Ai�XWz9sZ0zroDj		mn										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=A3274695&userId=2746346470&userId=274634&userId=256AEA56926&ts=158194407&userId=27463467720%3D&source=pcap	700256219
+2020-02-17	2020-02-17 13:56:10	EMC	rOXnSk45QcoEkw03yMYbGUhD_YZ_yCx9OU	DyGXuVg8;P6Tq�8CDp2_0XI`zPz4wuG8iJJh	0	3IM74UyhDyv2Jamik02pDQJ�u_L10vfPCKTm		mn										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=459&email=bmFA81942095&userId=27468872095&userId=27457F1944672Y29t&source=pcap	607085190
+2020-02-17	2020-02-17 13:57:55	EMC	xo09Zqxj�uPBo�_2aY�OgnG\\sH3HDRtPH7xW	1I3ceRxY�eGTFgNy30�ojqJ]qrOgqu1Ag8yO	0	PwQkOkG2�N1_p*ZFjOwmM6FviVATMGY5_rbK		mn										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=158193979967F628BBA48E274634&email=dGV4LnJ1&source=mn	3851759707
+2020-02-17	2020-02-17 14:09:55	EMC	vmlDYHpC�g270cyeLJ�ZBJT�wBBnnJHH99Av	l9FPMTHyyKlHijgeO]2oJK90OO8hWvntXS8	0	0Xu7Ha8e�qQdwcG3pm#GNMw6HzEsxoAszdK		mn										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=BtYWlsLnJ1&source=mn	2783694582
+2020-02-17	2020-02-17 15:31:52	EMC	WIzjCT0n�1XqAlOuRi2YskE�x619ZjBaFe4g	z_NGqjbLB0IhQ�V1Sx�DRMNG0DlMUxPlqaQ6	0	bv75_7V5�B4pBB2aNK\fRwnB�DQm0ngrcFPLN		mn										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=158193628FA603193692&email=bmFA934&userId=265&ts=158192YW5haWwucnU%3D&source=pcap	722494088
+2020-02-17	2020-02-17 06:28:37	EMC	434gE67E�ZuzQ�L3ss<YSbs�lSKkc3m3x_yv	z1Pq6pVwzBbbooT7uGVDX_}tTYKMpBkc4sR	0	yDB73mVP�4_s4�5wcA�Fidijmoz9Z8Z42YgR		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=EAC5jb20%3D&source=mn	1715283655
+2020-02-17	2020-02-17 13:00:54	EMC	YtB5N7m6�fG2T�Yx7FdUcwmexJvYphKtmqW	hAd5dqRmE0nQI4zKofhKFll6AVX08xJOYk	0	2nUgIwJh�wYTt�eNxh1H_zOJcbKA6y_ECSO		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=BtYWlsLnJ1&source=mn	3600915089
+2020-02-17	2020-02-17 11:52:28	EMC	zTqvvbns�qeoO�JoDN�bX6d�C4WFm5s0aN0J	62Nwk85B�aoJb
T5QS�a16QQGhmItxS9DiK	0	LAdBkCNA�3XVYEazFQlM65a�HvJLa2Uwcbou		mn										48	Ru	ru	7.0.32	/confirm/vipstatus	?key=AA4D&source=pcap	3757909189
+2020-02-17	2020-02-17 10:47:40	EMC	j72d0hbmaSx8z�DgG3tQJ4I2GADb2ODpcjr	OXHTehM1qzfV�kDM3L820Dj9W9jM1w6M8rn	0	c_q9hJgP�HrdU�lhlHhiA4uycDO_boRaNMT_		mn										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=462DF2921355274634&userId=274692&ts=15819397A7C5ydQ%3D%3D&source=mn	1740722989
+2020-02-17	2020-02-17 11:04:34	EMC	ym0RU_o2�2n0hkvvZO�OvxP4LYfXtzgmsjF	sEUZx6WdxQKfQ#p8Yg�X07I[kfwWziyMCQ0A	0	JsbtDnKA�5QYP�wgZo�nSmRUk4td1meycR2y		mn										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=666&ts=15819382&email=YW5hce=pcap	3685826953
+2020-02-17	2020-02-17 15:35:55	EMC	cvPUJNgQ�b6omlvnDp/jZQtbpAmtLBDbrkvl	Uof6vXhj�CJ2rWMT7P�SR1in0v1NOLW9fBhf	0	noQAt7IALqk4WfBYO�spAi�OqvCOMDBX0It		mn										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=EF31940BE3974634&userId=27457095&ts=158193634634&email=dGdtYWlsLnJ1&source=mn	2225930849
+2020-02-17	2020-02-17 15:36:22	EMC	bAiQYrkn�NTsK�x2Xzz3Y0Y|htnTleAO6ykM	6g6IaOBb2ySm�BtlcY29n�Fc0tkHBDefJ5	0	dyVlnui8k6b7eWHdw�g8q9�AHZaMvK_nvcJ		mn										187	Ru	ru	7.0.32	/confirm/vipstatus	?key=E760&userId=27465&userId=274688&email=Z21haWwuY29t&source=mn	1995672768
+2020-02-17	2020-02-17 12:06:44	EMC	hK6YNczKhnANLb_v1B�Ac02�mmJ4v3eHlP1v	2zBNHZW3�bhf0�26Se�hB6b�Ns3ibmJjnHZq	30807093	a7HUuGsA\rHDZo�0qB6\\eEiI.mhFXMDkuWVok	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https://click.mail%3DYWlsLnJ1%26source%3Dmn&c=swm&r=https://click.mail%3DYWlsLnJ1%26source%3Dmn&c=swm&r=http&o=mail%3DZG1haWwucnU%253D%26source%3Dmn&c=swm&r=http&o=mail.ru/redir	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=EA1942095&userId=274552095&email=Z21haWwuY29t&source=mn	1131763241
+2020-02-17	2020-02-17 13:29:00	EMC	yL4jFOc_2M63V�zEuQyNNN:l3yf54Az0gBM	y8Mk0a0v�xOHa)7mp0)npYs�JTlKq73zNSH0	32458442	_Q8lQMng�PyH2qGZvp_yjq�F2x5K0Mm_Pxe		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=C6346887465&ts=1581940C5ydQ%3D%3D&source=mn	3476824011
+2020-02-17	2020-02-17 13:35:20	EMC	nQ8H1CE0�zu5Z�dAVjsr16YBwMvcC2RrTSE3	xxeQBboi\fChONZrrlz�LwJv�bKcTM0BhtRT0	23232160	GbEGq80s�BaiA�Qc0v�jpyp�88CmkMJjyHVq		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=4E9BC8&email=a2B634&email=YWlsLnJ1&source=mn	1803353690
+2020-02-17	2020-02-17 14:35:33	EMC	6q7D5tuGBcVx4�971w�eVnB�R2wSldXMEi9i	ws0bhhM0!yYYb�OD33iTffOtXR7Z3SQyWu5	26234412	diKZMgVo�uQYP\'GVQybabDRGHgyfohtzfmK		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=5D82&ts=1581921haWwuY29t&source=pcap	1534869198
+2020-02-17	2020-02-17 14:56:20	EMC	6LPYg86G�0SWz�c45KEYeM0�35bisDLk9BVq	fQqF75EIUhQxVRWgBkvGeqIkEQ_ev_2ko3	32251365	HrU8iJjt|_h3i�nvMb7psYV�sJJVsYLvLZs4	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=http&o=mail&v=2&s=3c	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=5ydQ%3D%3D&source=mn	1107116720
+2020-02-17	2020-02-17 15:57:00	EMC	8bVyLB64�Z_fCUL3R�68gI\t818cRZJwctr7	4ANTHKEQeWVL�Cyl8�bT3s�pOxYtr4uS0FA	26326848	hjLpN8IV�D8gubNzOKAc4�UltehFJvxrN6	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https%3A%2F%2Fwwwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6email&v=2&s=3D2746126source%3Dmn&c=swm&r=https%3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6ACA7746194441581944461%26source%3Dmn&c=swm&r=https%3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6userId%3D27457461581937BBE5ydQ	pcap										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=758193634&email=bWFpbC5ydQ%3D%3D&source=pcap	3192390078
+2020-02-17	2020-02-17 06:37:18	EMC	vkWez1lP�PK87�iyEyvFzKR�rsZ6gXCXwsPh	Cvezv2mmySmX6DKK[EIL7�uccdaX1Z6rGG	18846619	2ChPFtaO+dQB7�JLRa�vUOt�dtB1gaeAJYK7	http&o=mail.ru/redir?u=http&o=mail.ru/redir?u=https://click.mail&v=2	pcap										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=42095&email=bW95&ts=1581940FCCAD&source=pcap	1825815078
+2020-02-17	2020-02-17 17:28:02	EMC	rXmItx0Za8A6z�3nJx�qyAw�FYgntazShou0	k3f6e_5EhvLY`rFcr0LiOi�Cjjo_MxbCOhq	26234806	CKG16UI8Qxp3�jVz2�r7mS\\_Ka0_r838J0r	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https://click.mail.ru/redir?u=http&o=mail.ru/redir?u=https://click.mail%3DZG1haWwucnU%253D%26source%3Dmn&c=swm&r=https://click.mail.ru/redir?u=http&o=mail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6186	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=EBE3D&source=mn	3849914033
+2020-02-17	2020-02-17 17:04:50	EMC	0y0AXkG7�Fyv_)diHn3nL6pxxlxN3YIbDCX	QuY2xKik�HKCO�O37j�dJQ8�tIY0xSj1P9xU	27942721	rcOVMm5600B3�jju3�msnk�mTmNXN8pc37H		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=15819189&email=Z21haWwucnU%3D&source=mn	108142745
+2020-02-17	2020-02-17 23:25:00	EMC	A8tzIc0j�yNY2�mplQ|hXDF�FLKSpHo5Qs0w	Mb_hLrLH�v1Wd�W02e�zAxfFHOy5zxu0dE2	31886688	a7ISoIAk�QuUy�_jr7�B0TObOs7oy4iBp00d		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=158194406&userId=27457B2566&ts=158193629t&source=mn	132315033
+2020-02-17	2020-02-17 09:59:38	EMC	znVyLmnCxmBmPwbI1�QRlw�Nepu6ZgHbcLz	pTD62K17�AHgoJvdK�Hoe4�0d1ciQ8lZSVj	18096770	I_lmWWYs�T2AX�KjMO�Db72Hh2hqzjbS4t6	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https%3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6userId%3D27461005A%2F%2Fwwwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D68FDDberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D65	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=EBE327455274634&email=c2Vsa2CAZ21haWwuY29t&source=mn	1574091904
+2020-02-17	2020-02-17 07:48:43	EMC	X_kKbaxz�j5D1c5eaa�l9Mpr79AQyZnhsOL	Lu_EhlFv�evSA�n38B�PcyP�UVBjxhkjLpuf	28018695	XWs5eS1b�RTB2�KjgK.MABz�URMSDAc6NXLn	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=http&o=mail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6703D27457079D1581938194461106ts%3D158194446158193DbW94457D91haWwucnU%253D%253D%253D%26source%3Dmn&c=swm&r=https%3A%2F	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=C70&email=Ymail=Z21haWwucnU%3D&source=pcap	1963271726
+2020-02-17	2020-02-17 04:41:36	EMC	G4IywqjH�ANSgABFcx�yX92JkyRRaypfrhNb	QKpsJQlY�dfkN�plib�b7yxuxWXsqoG6PWz	22669596	A4B8xM1JXGrwV�YZdlvdgMt0itiUFqoiE15	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https://click.mail%3DZG1haWwucnU%253D%26source%3Dmn&c=swm&r=http&o=mail.ru/redir?u=https%3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6ts%3D1581944E6email.ru/redir?u=https%3A%2F%2Fwwwwwwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6A%2F%2Fwwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6203D158192Ymail&v=2&s=9BBE4D3D27461haWwucnU%253D%253D%26source%3Dmn&c=swm&r=https%3A%2F%2Fwww	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=7465&userId=274695&userId=27465D86887B6&userId=274552416&ts=15819266	2384793579
+2020-02-17	2020-02-17 07:58:50	EMC	kWh5m_gA�YWjwmGYpp0yKohX3jXUWnK0ZMg	iyE6SVGecMrS9i00F�pqUG�TAcsinxktMxS	32418178	kHtD_Sba�x6eseNNnK�VTgH�tGA3wUEpO_vJ	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https%3A%2F%2Fwwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6source%3Dmn&c=swm&r=https%3A%2F%2Fwwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D620Fkey%3D6792YWlsLnJ1%26source%3Dmn&c=swm&r=https%3A%2F%2Fwwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6901	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=15819392YWlsLnJ1&source=pcap	1365119965
+2020-02-17	2020-02-17 08:36:56	EMC	ChsfAa0k�2S4f�4gG4nolq�UEHuuF9DnjUL	xXit4wGEobkT7ebgNO�CqAWiSkw5zvcQr1k	23623655	PgaTdfnr�E2Le�agJl�UWlb�D_tSsAastO4l	http&o=mail.ru/redir?u=https	pcap										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=15819363468874695&ts=1581944095&userId=27465&email=bWFpbC5ydQ%3D%3D&source=mn	1139408171
+2020-02-17	2020-02-17 00:25:53	EMC	ExIWQbTf�aZ2j�Ojla_vwzCe0CKomREUDEex	FWdWk09I�PqPWiqJh5�bi5o�6xjTF0D7N9cL	23644610	ZzTIlkXB>U31NIt0eo�eEpm8YplE7RGkXuG	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https%3A%2F%2Fwwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6source%3Dpc&c=swm&r=http&o=mail.ru/redir?u=https://click.mail.ru/redir?u=http&o=mail&v=2&s=974578Fvipstatus%3Fkey	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=1581938C8193634&email=a29t&source=mn	4267472950
+2020-02-17	2020-02-17 03:43:33	EMC	TAl7tSOQ�aATL�UUMZTx90iv0sqRoADhEn2r	zvCR7ost�HdMiQuVI8uxOyYGf0mm6k3ckHSL	32488481	9kFERwTF"AWKx�n0YP�OUv_tA8c5X2cdy4bA		pcap										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=1581936292&email=a256F095&ts=158192&userId=274695&userId=27465D84D65D8B8E7EA692YW5haWwucnU%3D&source	1722142578
+2020-02-17	2020-02-17 04:06:30	EMC	kL0p32eE J8hj�51lG�JQdDpzRytjcc2_AnA	eO7MVT5A47viQ�fb3A01Jap�oD001J0l_UAt	23078127	KDZSswsocsumo�OtSx\\phTv�uffYOTHGNoUa		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=7724D&source=mn	2669624134
+2020-02-17	2020-02-17 04:23:38	EMC	q3PyeDUQPYw7�QN9BwE1eP\nDzWsZDAv3GTN	u3GnfRRy>bOqp�qFvS�Dike
B8V0zHacc0ur	26965394	CxxiDtfVM9T6W�uleL�1P4m�MbzF9k0TLWLF		pcap										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=58193695&ts=1581942095&email=bWFpbC5ydQ%3D%3D&source=mn	2552851257
+2020-02-17	2020-02-17 05:36:35	EMC	ipti_lG9�4H8P�5RgF�wKza�oxUwWNEV9beK	zolMT7vt�_ttY|BmlD�2PSe�jtg6O7WzJ2S_	31107154	ha6rDRxktOKP�jl2g�zz7R�2TpQu2KKta9Q		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=4942095&ts=15819369266&userId=274692158193634&userId=2745523552D695	1459132007
+2020-02-17	2020-02-17 00:45:31	EMC	hXw0ETb7piXc>PKMz�qxhV�_F_UdWZnFB5O	VXB5E4w2�8Orr�ZT0X�1Sn9SrTWuJhjff7I	19656120	v06fJehG\'_m05�7sFduiQBx(mS5JnuSD6jph		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=B193695&ts=15819346165D85jb20%3D&source=pcap	1300966980
+2020-02-17	2020-02-17 03:44:40	EMC	0Q0EMlWcxIACIHUsZ�X4mE�hJnxPpCJV6Qe	ZqVPZ1Yb>rFSB\tzEMNdTEjoWIHLaQ2o5vI_E	17361913	AuRaBJOGvceRXWiMb9\'wolU�nAGR0hxXQgOS		pcap										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=725634&email=bWFpbC5ydQ%3D%3D&source=mn	1552491257
+2020-02-17	2020-02-17 02:35:40	EMC	mSk0tnq0soJQs�lU0K�lkSm�fnXFu6fqRw0A	hjlyHsr0&gybjsAJ0WRqdLJ�z32l3lmiEgDN	32664890	YAoxYQxl"Ce8H�vY2j�90VWzKQxlGCTNRAyJ		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=6459&ts=158194464089&userId=27465D82&userId=27455229t&source=mn	3051884197
+2020-02-17	2020-02-17 13:44:36	EMC	znOwy_e4�iHG8(NdCh�2qgz�1tjBR0ZjsFus	eNSFOPi9
oax3Ru7x0�1JA4TWeX66Ev8T9U	18979240	LosZ3fZj�VY8h�XjH8AJgKx:O3FNAMVDRGlZ		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=71581944027465D819384019215819397327457AbWFpbC5ydQ%3D%3D&source=mn	1828619156
+2020-02-17	2020-02-17 02:07:55	EMC	yvDmRiaY�km43�iWRNnnW3Y�DPBqq55XWVXr	7gpmg2Pi�rkSp]TYhcY7i6]DawnwV9tqC0b	22858229	yDybPZi3Oc7m,JVxfltuRt�Wm8TmiHQJOjY	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https://click.mail%3DZG1haWwucnU%253D%26source%3Dmn&c=swm&r=http&o=mail%3DZG1haWwucnU%253D%26source%3Dmn&c=swm&r=http&o=mail&v=2&s=9158192YWlsLnJ1%26source%3Dmn&c=swm&r=https://click.mail&v=2&s=3Dmn&c=swm&r=https://click.mail.ru/redir?u=http&o=mail%3DYm_ce%3Dmn	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=158189721581939265D89&email=dG1haWwuY29t&source=pcap	1735264694
+2020-02-17	2020-02-17 00:39:49	EMC	BOA0kaQben9xa\fOjQ8YZt2L9UV_FYiMKYa3n	0BL0DaCp�3weX�j7aIq4ZB�haKzr0Uwpdpm	21820865	xATcz14a�2hYJcGJb2gRs�IIQ9K2A5fOGo	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https://mail.ru/redir?u=https://mail&v=2&s=90BtYWlsLnJ1%26source%3Dmn&c=swm&r=https://click.mail.ru/redir?u=https%3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6ts%3D158192YWlsLnJ1%26source%3Dmn&c=swm&r=https%3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6userId%3D27457%26ts%3D158192YWlsLnJ1%26source%3Dmn&c=swm&r=http&o=mail.ru/redir?u=http&o=mail.ru/redir?u=http	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=77256160&userId=274634&email=bWFpbC5ydQ%3D%3D&source=mn	1855925025
+2020-02-17	2020-02-17 01:38:01	EMC	uNpwiDVW�2_p6�HEMl�4YIHw0z_vOpEqPyir	napvpdbMVeSjy KJz5spT94�Yu6wGCm_xgkX	27148457	R8wZY_PR�klNk�xX0D�n6uA{sGHEPaMT3kc8		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=3C3274695&userId=2745528194401908193467BD9189&ts=1581926&ts=1581939772B67B6AE4CD2DBD910B392YW5haWwucnU	2249892028
+2020-02-17	2020-02-17 23:32:22	EMC	uKrr0IYx�5Kt1\'QPtv@toS9�nQ4ZcrvAD0l8	4dWFNMPK50HO0z6ZPSoX6g0�2Ceu35kLzjtB	25390020	CXq72Vsb�olj2.WJUo�a8Uo�TZlw2wHUHxWX	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=http&o=mail%3DYWlsLnJ1%26source%3Dmn&c=swm&r=https%3A%2F%2Fwwwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6email&v=2&s=3A%2F%2Fwwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D67972	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=BE304F9CDBC6&ts=1581934&email=YW5haWwuY29t&source=mn	3489237198
+2020-02-17	2020-02-17 22:36:42	EMC	hQkEQSur�ISp_�uy9O�Rd05yjXxg7QByQH9	FURnnJ51�xl2NiQHcN\r6hpI�kfYa17e26Mt8	25551409	NBIMBrwi mVoY�YuxJq_8Df0CM8CZkrZezFE		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=158194463465&userId=25616C3D&source=mn	2913713687
+2020-02-17	2020-02-17 20:19:16	EMC	Xahxt0V3`JlMo+tyHm8MxJF�T9NEywdtAZeD	YV0_62kSEM9X0GW7�eHjX�xF95KfKoarfR	27982291	iuEEgZrU�YV04*QSEQg_uWbYo8cewVWPriGO		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=DFD92Y29t&source=mn	1136043092
+2020-02-17	2020-02-17 20:05:44	EMC	gAraBOGN
zDEd�kz8q�xSB6�8sQXrgyTFpX6	KC9GBYXC�c2sL;fDL4#6DMT%2ELzdoqHtjMp	27849879	hmvHfcpQ7YcXy�k6byq3rSq�EUfDuei6M62h		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=71&source=mn	795793408
+2020-02-17	2020-02-17 01:03:41	EMC	Sujwqk4k�hZF0�dd3dRrn0z�sOYRdkd9ynAR	NaKyF0O3rR3r�0wXN�Lf_3KSANWona7Ftry	30475228	calDaEsa�910bkYOkR\bbKjQ�eUvAOoR7CQJH	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=http&o=mail.ru/redir?u=https%3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6source%3Dmn&c=swm&r=http&o=mail%3DYWlsLnJ1%26source%3Dmn&c=swm&r=https://click.mail&v	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=B1581918970&userId=274634&userId=27457&email=YW5hce=mn	3165615449
+2020-02-17	2020-02-17 18:40:14	EMC	zBHwSrKI�MfAF�FJxuvWtIs�Z2DA_nAX98Aa	NQiRPz123luXCH0fvH?59tb�jHdGAlVgiHhc	26251034	0KbAptV7�u8SZ?V4N7�WHvx[0vY5D_saLU2G	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https://click.mail%3DYWlsLnJ1%26source%3Dmn&c=swm&r=https://click.mail&v=2&s=9158192Ymail&v=2&s=936source%3Dmn&c=swm&r=http&o=mail%3DbW9%26userId%3D2746158192YWlsLnJ1%26source	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=5&userId=274552746887BF4LnJ1&source=mn	115808044
+2020-02-17	2020-02-17 22:03:35	EMC	dvrVwEyo�cDVpJ3YEo�zYAH1XuC51DQcby59	wcCGF71w�VXRj�0pz2h131p\\1c7jnQYzejlC	24153343	mHlXjDve�l9Ab�8kZxtR6ELtFOSq7X1Q8_7	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=http&o=mail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D65	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=A63468874692&email=ce=pcap	1125992029
+2020-02-17	2020-02-17 13:11:31	EMC	ELifUEth�NP3xZTZ5RZd2tIgUWpctZJNKkdp	3qxE4qStGm1jI�RPc9,JCHKxNAJqeRRgVaJ5	32523658	opEAQkCl*DGc0�VozrkcYlaUmWvfN0IAYa	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=http&o=mail%3DYWlsLnJ1%26source%3Dmn&c=swm&r=https://click.mail&v=2&s=3%26userId%3D274570%26userId%3D274615819383%26userId%3D2745746193%26userId%3D2745745776source%3Dmn&c=swm&r=http&o=mail&v=2&s=34Fww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6source%3Dmn&c=swm&r=https%3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6126source%3Dmn&c=swm&r=https://click.mail%3DbWFpbC5ydQ%253D%253D	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=616&userId=27465&email=YWlsLmNvbQ%3D%3D&source=pcap	660510007
+2020-02-17	2020-02-17 23:08:20	EMC	Ww8aun1EXuPT3stlfv12_A4nVx4RfDfGaK	Gj6ixLDK{FkfiIQ0ZqHK9IsJMWys80zowLQy	28692974	nB20Riv63uMc]E9s0�Bv7x1XcxILAKq2zwc		pcap										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=3D&source=pcap	458549627
+2020-02-17	2020-02-17 15:45:11	EMC	_nkKyF0NV20B2�AAxI@oOj8�uSGus_lOrqBS	0vXB63kV�oAv97D8cU�CQSR\bmet_h2UhiQ4X	31753811	zA7KKses�YWPr�I1KD"FFmd�emMIJ3AcxwZr	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https://click.mail%3DZG1haWwucnU%253D%26source%3Dmn&c=swm&r=http&o=mail&v=2&s=9%26ts%3D1581944577457EC6source%3Dmn&c=swm&r=https://click.mail.ru/redir?u=https%3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D67D492YWlsLnJ1%26source%3Dmn&c=swm&r=https%3A%2F%2Fwwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D685764%26email&v=2&s=9%26email.ru/redir?u=https%3A%2F%2Fwwww.wildberries.ru%2Fconfirm	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=DA8189&email=a2DC9serId=274634&userId=25695&userId=274688&userId=27465	4039013194
+2020-02-17	2020-02-17 14:54:17	EMC	wHTfVQKq�lnQp!PXzWG80q�c0oWU1pvtKeL	EaNVlRXu�zZ33\fLB0Fmjet2�9f3xaJplvaiX	30992376	vJnOh3mp�Ob0h!xPw1slS0e[Ehj1AIx920Jo	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https://click.mail.ru/redir?u=https://click.mail&v=2&s=3c	pcap										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=7462B60&email=cnU%3D&source=mn	2569826295
+2020-02-17	2020-02-17 22:51:48	EMC	C1a6zgETp1v0mSMHJJgMxp8w7uJ7zN14J9Z	5cn50cUip60T6�nRihVzsS�RarErtllzgH6	26715992	0D4M2m041DlURNuFG�9QJt�IcjHVkp5_TEZ	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https://click.mail&v=2&s=9446158194419447	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=DEBE34&email=Z21haWwucnU%3D&source=mn	1504060974
+2020-02-17	2020-02-17 13:53:54	EMC	5a2rniMp�duiE�UzYFsgZjKHqgoZWbDYts	ziqQkEkD�3tYg�LGhe&QwROPlb3hkAIZW_UI	32696355	0rSWK9Um�GDPu�RvPwXqs3m�K9cF0balOqbS	http&o=mail.ru/redir?u=http&o=mail.ru/redir?u=https%3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6email	pcap										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=77274552E&ts=158192158192YXNvbQ%3D%3D&source=pcap	1675780127
+2020-02-17	2020-02-17 20:08:49	EMC	0IBozTjy/SMIk�Y5tY\\_GAd\'0VIzFgFK2VSY	V0fFnEjk�PzK3.y_ng3Nfs1oJtBLCWPq35z	33102601	NQhlj1k9�bLy234WJi\rKtOd�2239rJDYJSKh	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=http&o=mail.ru/redir?u=http&o=mail.ru/redir?u=https%3A%2F%2Fwwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D61haWwucnU%253D%253D%26source%3Dmn&c=swm&r=https://click.mail%3DYWlsLnJ1%26source%3Dmn&c=swm&r=http&o=mail.ru/redir?u=https://click.mail.ru/redir?u=https%3A%2F%2Fwwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69%26userId%3D2746193A%2F%2Fwwwwwww.wildberries.ru%2Fconfirm%2Fvipstatus	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=D8BD7B082&ts=15819392YW5haWwuY29t&source=mn	832387252
+2020-02-17	2020-02-17 21:39:56	EMC	3aiRtPZF[Sjelo0e2EVnauT�wl0tsW9SRi7f	i8OkH4au\rfMiB�V0E4�frxZ�u9WJpRG5bhp4	29554136	gzPJTobN�MGgG�1kUY�Ve9s�OwfpHmKiF1p0	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=http&o=mail%3DbW97B365	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=AE&ts=158194466&ts=15819213CD274688&ts=1581934688&email=Ymn	74272104
+2020-02-17	2020-02-17 17:21:43	EMC	yqM2AqiU�S2Lm�o0S5�KNYR-RKGxDrME0w5U	T1Q2fyBqs8qhiWB_J�BE3K�o_Ro0VD79pT0	18406884	mVoGr7MM�zfVX�0HtU�KAYe�CVX68_NOaiQY	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https%3A%2F%2Fwwwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6email%3DbWFpbC5ydQ%253D%253D%26source%3Dmn&c=swm&r=http&o=mail&v=2&s=9158192YWlsLnJ1%26source%3Dmn&c=swm&r=https	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=5819392&userId=27465D8B446887AF95&email=bWFpbC5ydQ%3D%3D&source=pcap	1422459332
+2020-02-17	2020-02-17 15:06:14	EMC	NXf_EKmdwAkZZIKr0N�0w4HuVTpNqZ30Ldk0	PpN_0d7y�wY45�Y_Wm#L2S0�1pN5klv0Wbsa	21385681	JH3MQSLt�n6gUIe92ilJPT.EFDSucrCuGHZ		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=A4627465D819397D96AE65&email=dG1haWwuY29t&source=pcap	959722117
+2020-02-17	2020-02-17 19:52:47	EMC	OY90Zcjc�8ebLcK_uULQHDNvpwO1hN3MwIat	ASdCffii�d6DweXoVj�mpL8~5DHOzhFnPRuW	33422761	4T1zxu2F4evId\rETTryZBYs�CBkTxsXLeuC9		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=E97265D8A70F&ts=15819403193634&userId=2746887&email=ce=mn	997133189
+2020-02-17	2020-02-17 17:55:20	EMC	ManEZ_qt�zd0v\bdxdB�07LpVWqoZUfpDzZW	MnNRUoUE2ASWa�g8Cl�KK5l�0yPRafHb0gsw	29428950	clyoRyW5\00F6wDd5P1�88B6�k_OdKCbUFQqT		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=52666799692&email=dGF7470E634&email=ce=mn	3015790084
+2020-02-17	2020-02-17 20:07:34	EMC	pr7D2hoMay0b�wJjvH6UaP�LZpeDXtq0Njs	bDlkkbc3OnwzdcNZHD�cBkR|V0ApitIVgPOv	22120090	TC2aVmxXNfOhS�L0y2�oqzb=drQElfeo0_OQ		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=581943D&source=mn	2666495492
+2020-02-17	2020-02-17 18:20:19	EMC	a_EpwBGW�RXNS�TY_F�yAhRB25SpX8Gw8JaW	0XDllB5u!HDENidqN/CRrkxwnjTXLeEKIyb	32607759	OKByJYVH�3pLb:BWhW�wvLn�npw0urPjP04i	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https%3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6userId%3D27461203Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6email&v=2&s=3D158192YWlsLnJ1%26source%3Dpc&c=swm&r=https	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=BC66&ts=158193926&userId=256BFC35FD2AEAZ21haWwucnU%3D&source=mn	590291581
+2020-02-17	2020-02-17 10:27:11	EMC	bn8x3TAHKlMOs�dlmzybH0S*iDc0hYull7Zt	7PY8hsBE�rbST�npoU7iPxD8CFjIMxK8eN8F	30559777	lgByHEG2ffme�xKgJXv5r�C6CW78KV4Kia		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=6667&email=bWFpbC5ydQ%3D%3D&source=mn	1898373016
+2020-02-17	2020-02-17 14:09:12	EMC	jYFR0DCE�2eTk�EFCuAIyAHVYLUy9ENh9Cqc	dhe0lvmT�Dchj�8XhgtsZ3mCPzv5hTGPM8Q	24813187	Y7MmPfU2�YhtaY85K"2xIfqNraWWqBp5QvK	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https%3A%2F%2Fwwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6source%3Dmn&c=swm&r=http&o=mail&v=2&s=96email%3DZG1haWwucnU%253D%26source%3Dmn&c=swm&r=http&o=mail&v=2&s=944576userId%3D27461819443Fkey%3D646192YWlsLnJ1%26source%3Dmn&c=swm&r=https://click.mail.ru/redir?u=https%3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6userId%3D274615819360	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=AD9397158192&userId=25616AEBD63465D88&email=bmRleC5ydQ%3D%3D&source=mn	1726149098
+2020-02-17	2020-02-17 18:27:00	EMC	VnzEdMoo�2lhW?Yr6AQyNVwyFnUj_l0pBiG	EnQ5CpnJ�t3LM1hiu�C7ls�qJChM0i4JMZ0	28169388	zBJ8p4pS
lPo_�bkk4\t2ObECW2ReYnHoWrQH	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https%3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6userId%3D27461203Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6email&v=2&s=3D158192YWlsLnJ1%26source%3Dmn&c=swm&r=https	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=397729t&source=mn	1102313974
+2020-02-17	2020-02-17 19:49:07	EMC	fxpHcK20ouDlc�jYFR�hFT4DMtbMML0qb8ds	hBF63Vf_�qZrp�sfD4�EOfd�6YGUBvflIJqF	22835697	TkBqOH2Jjp0wH�35uDnjAxqgyb2P4HqtznK	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https%3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6userId%3D2746158192YWlsLnJ1%26source%3Dmn&c=swm&r=https://click.mail&v=2&s=9158192YWlsLnJ1%26source%3Dmn&c=swm&r=https://click.mail&v=2&s=92YWlsLnJ1%26source%3Dmn&c=swm&r=https://click.mail%3DbWFpbC5ydQ%253D%253D%26source%3Dpc&c=swm&r=http&o=mail%3DZG1haWwucnU%253D%26source%3Dmn&c=swm&r=https://click.mail&v=2&s=9%26source%3Dpc&c	pcap										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=ADDABAA48E95&userId=2746346887D9420%3D&source=mn	3497823576
+2020-02-17	2020-02-17 13:28:56	EMC	hZbZqj2vRV2F�pCXlvxIEy�qdeQrwceKg35	8fpYiVLE�9gfSIzfP_$dzlj1H8tMzVbcg59	20129530	8bY0GHO5KMpHW9cIAk�e4yQQ94oQfvoHBxY		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=C5ydQ%3D%3D&source=mn	3102735519
+2020-02-17	2020-02-17 21:37:19	EMC	TAlSFJzIsXzTWbuEymOTU3�20iUcNXmLT5j	J2Vf06b6�CMLF�AvwH�_wsU�NOv4O9nnCy20	17587521	nEczrivz�drNaBYYvdnQRUT2pBBo8hu3kCvI		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=66&email=YWlsLnJ1&source=mn	1613554206
+2020-02-17	2020-02-17 21:58:32	EMC	XPXxpWq0�wHcK2FB4$JaOa�9xIGhEaajLeS	l59HKsiwgQuXg�AweGZRZrE�kN0Pqk_plZ4c	24296347	3s8mV_3W&hisrcwCm6QXCg_�0zgLyr6nnqPB	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https%3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6userId%3D274616	pcap										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=467105695&email=a2YWlsLnJ1&source=pcap	2033387311
+2020-02-17	2020-02-17 16:54:16	EMC	2beVLC_ZYMmZuZbJZ�1mjVLQeoXV0BsZKq	J4Y22CvJx36nq6KND�MXBg!H5lUx1Z7BoFf	31434491	dabaBlCY�351p�XUW0�HdEr�DTW_XGkqwmta	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https%3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6userId%3D27461203Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6Fconfirm%2Fvipstatus%3Fkey%3D661076userId%3D27457CD15819444613	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=76031938874634&userId=2564BFA93626&userId=256&email=ce=mn	1613641575
+2020-02-17	2020-02-17 00:27:36	EMC	Au3PRMFC�W0Ij8iL0Z�tsrryAvKJowqXJ8k	ntqK0ayP�8ltR0twuGraE`Vwgcmux6jFwg	21307924	ruixkHLK�jmENcmpbvvbrrLU8FDLIpWOFtU	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https://click.mail.ru/redir?u=https%3A%2F%2Fwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6email.ru/redir?u=https%3A%2F%2Fwwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D6CD2745785dQ	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=A41E772FFC5ydQ%3D%3D&source=pcap	2395792653
+2020-02-17	2020-02-17 22:03:46	EMC	Tc9EmNpb5Xy0edNdTC�Lg7X�2eLmGuTdEkIt	YsCMp8oiUBPy�hHBs�DmEZs0xvixA2L6atY	23431515	Bh67PWI6�jKZ6�Q102�nTeN(YP6e8kLoqiVA	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https://mail&v=2&s=96userId%3D274616	pcap										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=5D819446&userId=27465D8191897C7462F400&userId=27463465D8192&email=Z21haWwuY29t&source=pcap	2211382541
+2020-02-17	2020-02-17 06:34:15	EMC	S13SmJGaHVUBXqd3bs\tRpB3�G_hXZBJjxOt6	FP_GWMlf�w2REi3Up5Cqu0�qWcA99YlCduy	27883082	xJ1eD6ix�our7M0K_�l6Ziv6awoxMpnzsaW	https://click.mail%3DYmail%3Dberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D69444%26email.ru/redir?u=https://mail%3Dby%3D6%26ts%3D158193DBDB3A%2F%2Fwwww.wildberries.ru%2Fconfirm%2Fvipstatus%3Fkey%3D63158192YWlsLnJ1%26source%3Dmn&c=swm&r=https://click.mail&v=2&s=92YWlsLnJ1%26source	mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=15819362hQG1haWwuY29t&source=pcap	2045856196
+2020-02-17	2020-02-17 00:36:06	EMC	2mjWiSh70SN_�1xOB�st7vbYq1z5EL26Z0i	7e5aIak4�Pca2�woYG�ZIJi!P_gRQdnPpHDP	28784173	b5hw0Lly�fMCy�YstJ9XbB6Pq7VN5hO9kIy		mn										1	Ru	ru	7.0.32	/confirm/vipstatus	?key=15819387AE3634&email=cnU%3D&source=mn	1255281910

From 5548c9d1ad8b752cac133193d774e6a84a03321c Mon Sep 17 00:00:00 2001
From: Alexey Milovidov <milovidov@yandex-team.ru>
Date: Fri, 29 May 2020 01:53:15 +0300
Subject: [PATCH 114/122] Repeat test on curl timeout

---
 tests/clickhouse-test | 1 +
 1 file changed, 1 insertion(+)

diff --git a/tests/clickhouse-test b/tests/clickhouse-test
index 1a2735296e8..86a9ecebd96 100755
--- a/tests/clickhouse-test
+++ b/tests/clickhouse-test
@@ -30,6 +30,7 @@ from contextlib import closing
 MESSAGES_TO_RETRY = [
     "DB::Exception: ZooKeeper session has been expired",
     "Coordination::Exception: Connection loss",
+    "Operation timed out",
 ]
 
 

From 7df5280060cd05942b255710acb0815136d67887 Mon Sep 17 00:00:00 2001
From: Alexey Milovidov <milovidov@yandex-team.ru>
Date: Fri, 29 May 2020 02:09:11 +0300
Subject: [PATCH 115/122] Fix potentially flacky test

---
 .../0_stateless/00731_long_merge_tree_select_opened_files.sh  | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/tests/queries/0_stateless/00731_long_merge_tree_select_opened_files.sh b/tests/queries/0_stateless/00731_long_merge_tree_select_opened_files.sh
index 5177cb6087c..5ae9dcf67ec 100755
--- a/tests/queries/0_stateless/00731_long_merge_tree_select_opened_files.sh
+++ b/tests/queries/0_stateless/00731_long_merge_tree_select_opened_files.sh
@@ -14,9 +14,11 @@ $CLICKHOUSE_CLIENT $settings -n -q "
 DROP TABLE IF EXISTS merge_tree_table;
 CREATE TABLE merge_tree_table (id UInt64, date Date, uid UInt32) ENGINE = MergeTree(date, id, 8192);"
 
+
 $CLICKHOUSE_CLIENT $settings -q "INSERT INTO merge_tree_table SELECT (intHash64(number)) % 10000, toDate('2018-08-01'), rand() FROM system.numbers LIMIT 10000000;"
 
-$CLICKHOUSE_CLIENT $settings -q "OPTIMIZE TABLE merge_tree_table FINAL;"
+# If merge is already happening, OPTIMIZE will be noop. But we have to ensure that the data is merged.
+for i in {1..100}; do $CLICKHOUSE_CLIENT $settings --optimize_throw_if_noop=1 -q "OPTIMIZE TABLE merge_tree_table FINAL;" && break; sleep 1; done
 
 # The query may open more files if query log will be flushed during the query.
 # To lower this chance, we also flush logs before the query.

From e8f52337aacb261f55d72716a09bc85490464286 Mon Sep 17 00:00:00 2001
From: Alexey Milovidov <milovidov@yandex-team.ru>
Date: Fri, 29 May 2020 03:09:12 +0300
Subject: [PATCH 116/122] Fix incorrect usage of fmtlib

---
 base/common/logger_useful.h    | 18 +++++++++++++++---
 src/Common/tests/gtest_log.cpp | 19 +++++++++++++++++++
 2 files changed, 34 insertions(+), 3 deletions(-)
 create mode 100644 src/Common/tests/gtest_log.cpp

diff --git a/base/common/logger_useful.h b/base/common/logger_useful.h
index c89ec63ef22..ad56ef7e857 100644
--- a/base/common/logger_useful.h
+++ b/base/common/logger_useful.h
@@ -15,15 +15,27 @@ using Poco::Message;
 using DB::LogsLevel;
 using DB::CurrentThread;
 
-/// Logs a message to a specified logger with that level.
 
-#define LOG_IMPL(logger, priority, PRIORITY, ...) do                    \
+namespace
+{
+    template <typename... Ts> constexpr size_t numArgs(Ts &&...) { return sizeof...(Ts); }
+    template <typename T, typename... Ts> constexpr auto firstArg(T && x, Ts &&...) { return std::forward<T>(x); }
+}
+
+
+/// Logs a message to a specified logger with that level.
+/// If more than one argument is provided,
+///  the first argument is interpreted as template with {}-substitutions
+///  and the latter arguments treat as values to substitute.
+/// If only one argument is provided, it is threat as message without substitutions.
+
+#define LOG_IMPL(logger, priority, PRIORITY, ...) do                              \
 {                                                                                 \
     const bool is_clients_log = (CurrentThread::getGroup() != nullptr) &&         \
             (CurrentThread::getGroup()->client_logs_level >= (priority));         \
     if ((logger)->is((PRIORITY)) || is_clients_log)                               \
     {                                                                             \
-        std::string formatted_message = fmt::format(__VA_ARGS__);                 \
+        std::string formatted_message = numArgs(__VA_ARGS__) > 1 ? fmt::format(__VA_ARGS__) : firstArg(__VA_ARGS__); \
         if (auto channel = (logger)->getChannel())                                \
         {                                                                         \
             std::string file_function;                                            \
diff --git a/src/Common/tests/gtest_log.cpp b/src/Common/tests/gtest_log.cpp
new file mode 100644
index 00000000000..a1e532f92e6
--- /dev/null
+++ b/src/Common/tests/gtest_log.cpp
@@ -0,0 +1,19 @@
+#include <string>
+#include <vector>
+#include <common/logger_useful.h>
+#include <gtest/gtest.h>
+
+#include <Poco/Logger.h>
+#include <Poco/AutoPtr.h>
+#include <Poco/NullChannel.h>
+
+
+TEST(Logger, Log)
+{
+    Poco::Logger::root().setLevel("none");
+    Poco::Logger::root().setChannel(Poco::AutoPtr<Poco::NullChannel>(new Poco::NullChannel()));
+    Logger * log = &Logger::get("Log");
+
+    /// This test checks that we don't pass this string to fmtlib, because it is the only argument.
+    EXPECT_NO_THROW(LOG_INFO(log, "Hello {} World"));
+}

From 263bc6e097a271df52af743caa6339db2a95abfb Mon Sep 17 00:00:00 2001
From: Alexey Milovidov <milovidov@yandex-team.ru>
Date: Fri, 29 May 2020 03:26:56 +0300
Subject: [PATCH 117/122] Added missing file

---
 tests/queries/0_stateless/00342_escape_sequences.reference | 1 +
 1 file changed, 1 insertion(+)
 create mode 100644 tests/queries/0_stateless/00342_escape_sequences.reference

diff --git a/tests/queries/0_stateless/00342_escape_sequences.reference b/tests/queries/0_stateless/00342_escape_sequences.reference
new file mode 100644
index 00000000000..50e0756d16f
--- /dev/null
+++ b/tests/queries/0_stateless/00342_escape_sequences.reference
@@ -0,0 +1 @@
+07080C0A0D090B5C27225C3FAA

From 13b0acdc9f44e474a0c02fa186dc2372d33130d6 Mon Sep 17 00:00:00 2001
From: Alexey Milovidov <milovidov@yandex-team.ru>
Date: Fri, 29 May 2020 03:42:25 +0300
Subject: [PATCH 118/122] Fix strange code

---
 docker/test/test_runner.sh          | 76 ++++++++++++++---------------
 utils/ci/docker-multiarch/update.sh | 60 +++++++++++------------
 2 files changed, 68 insertions(+), 68 deletions(-)

diff --git a/docker/test/test_runner.sh b/docker/test/test_runner.sh
index 76e142e61f9..561117492b0 100755
--- a/docker/test/test_runner.sh
+++ b/docker/test/test_runner.sh
@@ -12,7 +12,7 @@ readonly CLICKHOUSE_PACKAGES_ARG="${2}"
 CLICKHOUSE_SERVER_IMAGE="${3}"
 
 if [ ${CLICKHOUSE_PACKAGES_ARG} != ${NO_REBUILD_FLAG} ]; then
-	readonly CLICKHOUSE_PACKAGES_DIR="$(realpath ${2})" # or --no-rebuild
+    readonly CLICKHOUSE_PACKAGES_DIR="$(realpath ${2})" # or --no-rebuild
 fi
 
 
@@ -26,19 +26,19 @@ fi
 # TODO: optionally mount most recent clickhouse-test and queries directory from local machine
 
 if [ ${CLICKHOUSE_PACKAGES_ARG} != ${NO_REBUILD_FLAG} ]; then
-	docker build \
-		-f "${CLICKHOUSE_DOCKER_DIR}/test/stateless/clickhouse-statelest-test-runner.Dockerfile" \
-		--target clickhouse-test-runner-base \
-		-t clickhouse-test-runner-base:preinstall \
-		"${CLICKHOUSE_DOCKER_DIR}/test/stateless"
+    docker build \
+        -f "${CLICKHOUSE_DOCKER_DIR}/test/stateless/clickhouse-statelest-test-runner.Dockerfile" \
+        --target clickhouse-test-runner-base \
+        -t clickhouse-test-runner-base:preinstall \
+        "${CLICKHOUSE_DOCKER_DIR}/test/stateless"
 
-	docker rm -f clickhouse-test-runner-installing-packages || true
-	docker run \
-		-v "${CLICKHOUSE_PACKAGES_DIR}:/packages" \
-		--name clickhouse-test-runner-installing-packages \
-		clickhouse-test-runner-base:preinstall
-	docker commit clickhouse-test-runner-installing-packages clickhouse-statelest-test-runner:local
-	docker rm -f clickhouse-test-runner-installing-packages || true
+    docker rm -f clickhouse-test-runner-installing-packages || true
+    docker run \
+        -v "${CLICKHOUSE_PACKAGES_DIR}:/packages" \
+        --name clickhouse-test-runner-installing-packages \
+        clickhouse-test-runner-base:preinstall
+    docker commit clickhouse-test-runner-installing-packages clickhouse-statelest-test-runner:local
+    docker rm -f clickhouse-test-runner-installing-packages || true
 fi
 
 # # Create a bind-volume to the clickhouse-test script file
@@ -47,38 +47,38 @@ fi
 
 # Build server image (optional) from local packages
 if [ -z "${CLICKHOUSE_SERVER_IMAGE}" ]; then
-	CLICKHOUSE_SERVER_IMAGE="yandex/clickhouse-server:local"
+    CLICKHOUSE_SERVER_IMAGE="yandex/clickhouse-server:local"
 
-	if [ ${CLICKHOUSE_PACKAGES_ARG} != ${NO_REBUILD_FLAG} ]; then
-		docker build \
-			-f "${CLICKHOUSE_DOCKER_DIR}/server/local.Dockerfile" \
-			--target clickhouse-server-base \
-			-t clickhouse-server-base:preinstall \
-			"${CLICKHOUSE_DOCKER_DIR}/server"
+    if [ ${CLICKHOUSE_PACKAGES_ARG} != ${NO_REBUILD_FLAG} ]; then
+        docker build \
+            -f "${CLICKHOUSE_DOCKER_DIR}/server/local.Dockerfile" \
+            --target clickhouse-server-base \
+            -t clickhouse-server-base:preinstall \
+            "${CLICKHOUSE_DOCKER_DIR}/server"
 
-		docker rm -f clickhouse_server_base_installing_server || true
-		docker run -v "${CLICKHOUSE_PACKAGES_DIR}:/packages" \
-			--name clickhouse_server_base_installing_server \
-			clickhouse-server-base:preinstall
-		docker commit clickhouse_server_base_installing_server clickhouse-server-base:postinstall
+        docker rm -f clickhouse_server_base_installing_server || true
+        docker run -v "${CLICKHOUSE_PACKAGES_DIR}:/packages" \
+            --name clickhouse_server_base_installing_server \
+            clickhouse-server-base:preinstall
+        docker commit clickhouse_server_base_installing_server clickhouse-server-base:postinstall
 
-		docker build \
-			-f "${CLICKHOUSE_DOCKER_DIR}/server/local.Dockerfile" \
-			--target clickhouse-server \
-			-t "${CLICKHOUSE_SERVER_IMAGE}" \
-			"${CLICKHOUSE_DOCKER_DIR}/server"
-	fi
+        docker build \
+            -f "${CLICKHOUSE_DOCKER_DIR}/server/local.Dockerfile" \
+            --target clickhouse-server \
+            -t "${CLICKHOUSE_SERVER_IMAGE}" \
+            "${CLICKHOUSE_DOCKER_DIR}/server"
+    fi
 fi
 
 docker rm -f test-runner || true
 docker-compose down
 CLICKHOUSE_SERVER_IMAGE="${CLICKHOUSE_SERVER_IMAGE}" \
-	docker-compose -f "${CLICKHOUSE_DOCKER_DIR}/test/test_runner_docker_compose.yaml" \
-	create \
-	--build --force-recreate
+    docker-compose -f "${CLICKHOUSE_DOCKER_DIR}/test/test_runner_docker_compose.yaml" \
+    create \
+    --build --force-recreate
 
 CLICKHOUSE_SERVER_IMAGE="${CLICKHOUSE_SERVER_IMAGE}" \
-	docker-compose -f "${CLICKHOUSE_DOCKER_DIR}/test/test_runner_docker_compose.yaml" \
-	run \
-	--name test-runner \
-	test-runner
+    docker-compose -f "${CLICKHOUSE_DOCKER_DIR}/test/test_runner_docker_compose.yaml" \
+    run \
+    --name test-runner \
+    test-runner
diff --git a/utils/ci/docker-multiarch/update.sh b/utils/ci/docker-multiarch/update.sh
index 6e3e18047de..6abcf339607 100755
--- a/utils/ci/docker-multiarch/update.sh
+++ b/utils/ci/docker-multiarch/update.sh
@@ -37,27 +37,27 @@ fi
 
 # get the image
 if \
-	wget -q --spider "$baseUrl/current" \
-	&& wget -q --spider "$baseUrl/current/$thisTar" \
-	; then
-		baseUrl+='/current'
+    wget -q --spider "$baseUrl/current" \
+    && wget -q --spider "$baseUrl/current/$thisTar" \
+    ; then
+        baseUrl+='/current'
 fi
 wget -qN "$baseUrl/"{{MD5,SHA{1,256}}SUMS{,.gpg},"$thisTarBase.manifest",'unpacked/build-info.txt'} || true
 wget -N "$baseUrl/$thisTar"
 
 # check checksum
 if [ -f SHA256SUMS ]; then
-	sha256sum="$(sha256sum "$thisTar" | cut -d' ' -f1)"
-	if ! grep -q "$sha256sum" SHA256SUMS; then
-		echo >&2 "error: '$thisTar' has invalid SHA256"
-		exit 1
-	fi
+    sha256sum="$(sha256sum "$thisTar" | cut -d' ' -f1)"
+    if ! grep -q "$sha256sum" SHA256SUMS; then
+        echo >&2 "error: '$thisTar' has invalid SHA256"
+        exit 1
+    fi
 fi
 
 cat > Dockerfile <<-EOF
-	FROM scratch
-	ADD $thisTar /
-	ENV ARCH=${ARCH} UBUNTU_SUITE=${VERSION} DOCKER_REPO=${DOCKER_REPO}
+    FROM scratch
+    ADD $thisTar /
+    ENV ARCH=${ARCH} UBUNTU_SUITE=${VERSION} DOCKER_REPO=${DOCKER_REPO}
 EOF
 
 # add qemu-user-static binary
@@ -70,26 +70,26 @@ EOF
 fi
 
 cat >> Dockerfile <<-EOF
-	# a few minor docker-specific tweaks
-	# see https://github.com/docker/docker/blob/master/contrib/mkimage/debootstrap
-	RUN echo '#!/bin/sh' > /usr/sbin/policy-rc.d \\
-		&& echo 'exit 101' >> /usr/sbin/policy-rc.d \\
-		&& chmod +x /usr/sbin/policy-rc.d \\
-		&& dpkg-divert --local --rename --add /sbin/initctl \\
-		&& cp -a /usr/sbin/policy-rc.d /sbin/initctl \\
-		&& sed -i 's/^exit.*/exit 0/' /sbin/initctl \\
-		&& echo 'force-unsafe-io' > /etc/dpkg/dpkg.cfg.d/docker-apt-speedup \\
-		&& echo 'DPkg::Post-Invoke { "rm -f /var/cache/apt/archives/*.deb /var/cache/apt/archives/partial/*.deb /var/cache/apt/*.bin || true"; };' > /etc/apt/apt.conf.d/docker-clean \\
-		&& echo 'APT::Update::Post-Invoke { "rm -f /var/cache/apt/archives/*.deb /var/cache/apt/archives/partial/*.deb /var/cache/apt/*.bin || true"; };' >> /etc/apt/apt.conf.d/docker-clean \\
-		&& echo 'Dir::Cache::pkgcache ""; Dir::Cache::srcpkgcache "";' >> /etc/apt/apt.conf.d/docker-clean \\
-		&& echo 'Acquire::Languages "none";' > /etc/apt/apt.conf.d/docker-no-languages \\
-		&& echo 'Acquire::GzipIndexes "true"; Acquire::CompressionTypes::Order:: "gz";' > /etc/apt/apt.conf.d/docker-gzip-indexes
+    # a few minor docker-specific tweaks
+    # see https://github.com/docker/docker/blob/master/contrib/mkimage/debootstrap
+    RUN echo '#!/bin/sh' > /usr/sbin/policy-rc.d \\
+        && echo 'exit 101' >> /usr/sbin/policy-rc.d \\
+        && chmod +x /usr/sbin/policy-rc.d \\
+        && dpkg-divert --local --rename --add /sbin/initctl \\
+        && cp -a /usr/sbin/policy-rc.d /sbin/initctl \\
+        && sed -i 's/^exit.*/exit 0/' /sbin/initctl \\
+        && echo 'force-unsafe-io' > /etc/dpkg/dpkg.cfg.d/docker-apt-speedup \\
+        && echo 'DPkg::Post-Invoke { "rm -f /var/cache/apt/archives/*.deb /var/cache/apt/archives/partial/*.deb /var/cache/apt/*.bin || true"; };' > /etc/apt/apt.conf.d/docker-clean \\
+        && echo 'APT::Update::Post-Invoke { "rm -f /var/cache/apt/archives/*.deb /var/cache/apt/archives/partial/*.deb /var/cache/apt/*.bin || true"; };' >> /etc/apt/apt.conf.d/docker-clean \\
+        && echo 'Dir::Cache::pkgcache ""; Dir::Cache::srcpkgcache "";' >> /etc/apt/apt.conf.d/docker-clean \\
+        && echo 'Acquire::Languages "none";' > /etc/apt/apt.conf.d/docker-no-languages \\
+        && echo 'Acquire::GzipIndexes "true"; Acquire::CompressionTypes::Order:: "gz";' > /etc/apt/apt.conf.d/docker-gzip-indexes
 
-	# enable the universe
-	RUN sed -i 's/^#\s*\(deb.*universe\)$/\1/g' /etc/apt/sources.list
+    # enable the universe
+    RUN sed -i 's/^#\s*\(deb.*universe\)$/\1/g' /etc/apt/sources.list
 
-	# overwrite this with 'CMD []' in a dependent Dockerfile
-	CMD ["/bin/bash"]
+    # overwrite this with 'CMD []' in a dependent Dockerfile
+    CMD ["/bin/bash"]
 EOF
 
 docker build -t "${DOCKER_REPO}:${TAG_ARCH}-${VERSION}" .

From 025c1a8830ff3fcd7a2b42557ca5942fb38fdd49 Mon Sep 17 00:00:00 2001
From: Alexey Milovidov <milovidov@yandex-team.ru>
Date: Fri, 29 May 2020 03:46:42 +0300
Subject: [PATCH 119/122] Added another check #11170

---
 tests/clickhouse-test | 1 +
 1 file changed, 1 insertion(+)

diff --git a/tests/clickhouse-test b/tests/clickhouse-test
index 86a9ecebd96..2d155313b06 100755
--- a/tests/clickhouse-test
+++ b/tests/clickhouse-test
@@ -31,6 +31,7 @@ MESSAGES_TO_RETRY = [
     "DB::Exception: ZooKeeper session has been expired",
     "Coordination::Exception: Connection loss",
     "Operation timed out",
+    "ConnectionPoolWithFailover: Connection failed at try",
 ]
 
 

From 2d12b4d3ac606296675496c90fafcd62d7b17d30 Mon Sep 17 00:00:00 2001
From: Vitaly Baranov <vitbar@yandex-team.ru>
Date: Fri, 29 May 2020 09:54:27 +0300
Subject: [PATCH 120/122] Fix crash when SET DEFAULT ROLE is called with wrong
 arguments.

---
 src/Access/ExtendedRoleSet.cpp                | 24 ++++--
 src/Interpreters/InterpreterSetRoleQuery.cpp  |  2 +-
 src/Parsers/ASTExtendedRoleSet.h              |  5 +-
 src/Parsers/ParserCreateUserQuery.cpp         |  1 +
 src/Parsers/ParserSetRoleQuery.cpp            |  2 +
 .../integration/test_default_role/__init__.py |  0
 tests/integration/test_default_role/test.py   | 77 +++++++++++++++++++
 7 files changed, 103 insertions(+), 8 deletions(-)
 create mode 100644 tests/integration/test_default_role/__init__.py
 create mode 100644 tests/integration/test_default_role/test.py

diff --git a/src/Access/ExtendedRoleSet.cpp b/src/Access/ExtendedRoleSet.cpp
index a29ee40380c..a8e674b3722 100644
--- a/src/Access/ExtendedRoleSet.cpp
+++ b/src/Access/ExtendedRoleSet.cpp
@@ -68,15 +68,27 @@ void ExtendedRoleSet::init(const ASTExtendedRoleSet & ast, const AccessControlMa
 {
     all = ast.all;
 
-    auto name_to_id = [id_mode{ast.id_mode}, manager](const String & name) -> UUID
+    auto name_to_id = [&ast, manager](const String & name) -> UUID
     {
-        if (id_mode)
+        if (ast.id_mode)
             return parse<UUID>(name);
         assert(manager);
-        auto id = manager->find<User>(name);
-        if (id)
-            return *id;
-        return manager->getID<Role>(name);
+        if (ast.can_contain_users && ast.can_contain_roles)
+        {
+            auto id = manager->find<User>(name);
+            if (id)
+                return *id;
+            return manager->getID<Role>(name);
+        }
+        else if (ast.can_contain_users)
+        {
+            return manager->getID<User>(name);
+        }
+        else
+        {
+            assert(ast.can_contain_roles);
+            return manager->getID<Role>(name);
+        }
     };
 
     if (!ast.names.empty() && !all)
diff --git a/src/Interpreters/InterpreterSetRoleQuery.cpp b/src/Interpreters/InterpreterSetRoleQuery.cpp
index f8e0167d748..c627061dd51 100644
--- a/src/Interpreters/InterpreterSetRoleQuery.cpp
+++ b/src/Interpreters/InterpreterSetRoleQuery.cpp
@@ -62,7 +62,7 @@ void InterpreterSetRoleQuery::setRole(const ASTSetRoleQuery & query)
 
 void InterpreterSetRoleQuery::setDefaultRole(const ASTSetRoleQuery & query)
 {
-    context.checkAccess(AccessType::CREATE_USER | AccessType::DROP_USER);
+    context.checkAccess(AccessType::ALTER_USER);
 
     auto & access_control = context.getAccessControlManager();
     std::vector<UUID> to_users = ExtendedRoleSet{*query.to_users, access_control, context.getUserID()}.getMatchingIDs(access_control);
diff --git a/src/Parsers/ASTExtendedRoleSet.h b/src/Parsers/ASTExtendedRoleSet.h
index 8d619e5d6a0..656f563bd9a 100644
--- a/src/Parsers/ASTExtendedRoleSet.h
+++ b/src/Parsers/ASTExtendedRoleSet.h
@@ -15,7 +15,10 @@ public:
     bool all = false;
     Strings except_names;
     bool except_current_user = false;
-    bool id_mode = false;  /// If true then `names` and `except_names` keeps UUIDs, not names.
+
+    bool id_mode = false;          /// true if `names` and `except_names` keep UUIDs, not names.
+    bool can_contain_roles = true; /// true if this set can contain names of roles.
+    bool can_contain_users = true; /// true if this set can contain names of users.
 
     bool empty() const { return names.empty() && !current_user && !all; }
     void replaceCurrentUserTagWithName(const String & current_user_name);
diff --git a/src/Parsers/ParserCreateUserQuery.cpp b/src/Parsers/ParserCreateUserQuery.cpp
index 76a06a0282f..a6a007855e2 100644
--- a/src/Parsers/ParserCreateUserQuery.cpp
+++ b/src/Parsers/ParserCreateUserQuery.cpp
@@ -227,6 +227,7 @@ namespace
                 return false;
 
             default_roles = typeid_cast<std::shared_ptr<ASTExtendedRoleSet>>(ast);
+            default_roles->can_contain_users = false;
             return true;
         });
     }
diff --git a/src/Parsers/ParserSetRoleQuery.cpp b/src/Parsers/ParserSetRoleQuery.cpp
index e6ff7893891..a69480f89eb 100644
--- a/src/Parsers/ParserSetRoleQuery.cpp
+++ b/src/Parsers/ParserSetRoleQuery.cpp
@@ -18,6 +18,7 @@ namespace
                 return false;
 
             roles = typeid_cast<std::shared_ptr<ASTExtendedRoleSet>>(ast);
+            roles->can_contain_users = false;
             return true;
         });
     }
@@ -34,6 +35,7 @@ namespace
                 return false;
 
             to_users = typeid_cast<std::shared_ptr<ASTExtendedRoleSet>>(ast);
+            to_users->can_contain_roles = false;
             return true;
         });
     }
diff --git a/tests/integration/test_default_role/__init__.py b/tests/integration/test_default_role/__init__.py
new file mode 100644
index 00000000000..e69de29bb2d
diff --git a/tests/integration/test_default_role/test.py b/tests/integration/test_default_role/test.py
new file mode 100644
index 00000000000..2b6b4698b20
--- /dev/null
+++ b/tests/integration/test_default_role/test.py
@@ -0,0 +1,77 @@
+import pytest
+from helpers.cluster import ClickHouseCluster
+from helpers.test_tools import TSV
+import re
+
+cluster = ClickHouseCluster(__file__)
+instance = cluster.add_instance('instance')
+
+
+@pytest.fixture(scope="module", autouse=True)
+def started_cluster():
+    try:
+        cluster.start()
+        
+        instance.query("CREATE USER john")
+        instance.query("CREATE ROLE rx")
+        instance.query("CREATE ROLE ry")
+
+        yield cluster
+
+    finally:
+        cluster.shutdown()
+
+
+@pytest.fixture(autouse=True)
+def reset_users_and_roles():
+    instance.query("CREATE USER OR REPLACE john")
+    yield
+
+
+def test_set_default_roles():
+    assert instance.query("SHOW CURRENT ROLES", user="john") == ""
+
+    instance.query("GRANT rx, ry TO john")
+    assert instance.query("SHOW CURRENT ROLES", user="john") == TSV( [['rx', 0, 1], ['ry', 0, 1]] )
+
+    instance.query("SET DEFAULT ROLE NONE TO john")
+    assert instance.query("SHOW CURRENT ROLES", user="john") == ""
+
+    instance.query("SET DEFAULT ROLE rx TO john")
+    assert instance.query("SHOW CURRENT ROLES", user="john") == TSV( [['rx', 0, 1]] )
+
+    instance.query("SET DEFAULT ROLE ry TO john")
+    assert instance.query("SHOW CURRENT ROLES", user="john") == TSV( [['ry', 0, 1]] )
+
+    instance.query("SET DEFAULT ROLE ALL TO john")
+    assert instance.query("SHOW CURRENT ROLES", user="john") == TSV( [['rx', 0, 1], ['ry', 0, 1]] )
+
+    instance.query("SET DEFAULT ROLE ALL EXCEPT rx TO john")
+    assert instance.query("SHOW CURRENT ROLES", user="john") == TSV( [['ry', 0, 1]] )
+
+
+def test_alter_user():
+    assert instance.query("SHOW CURRENT ROLES", user="john") == ""
+
+    instance.query("GRANT rx, ry TO john")
+    assert instance.query("SHOW CURRENT ROLES", user="john") == TSV( [['rx', 0, 1], ['ry', 0, 1]] )
+
+    instance.query("ALTER USER john DEFAULT ROLE NONE")
+    assert instance.query("SHOW CURRENT ROLES", user="john") == ""
+
+    instance.query("ALTER USER john DEFAULT ROLE rx")
+    assert instance.query("SHOW CURRENT ROLES", user="john") == TSV( [['rx', 0, 1]] )
+
+    instance.query("ALTER USER john DEFAULT ROLE ALL")
+    assert instance.query("SHOW CURRENT ROLES", user="john") == TSV( [['rx', 0, 1], ['ry', 0, 1]] )
+
+    instance.query("ALTER USER john DEFAULT ROLE ALL EXCEPT rx")
+    assert instance.query("SHOW CURRENT ROLES", user="john") == TSV( [['ry', 0, 1]] )
+
+
+def test_wrong_set_default_role():
+    assert "There is no user `rx`" in instance.query_and_get_error("SET DEFAULT ROLE NONE TO rx")
+    assert "There is no user `ry`" in instance.query_and_get_error("SET DEFAULT ROLE rx TO ry")
+    assert "There is no role `john`" in instance.query_and_get_error("SET DEFAULT ROLE john TO john")
+    assert "There is no role `john`" in instance.query_and_get_error("ALTER USER john DEFAULT ROLE john")
+    assert "There is no role `john`" in instance.query_and_get_error("ALTER USER john DEFAULT ROLE ALL EXCEPT john")

From 09369c259c1cd343ef6a0324b0b93e110170cc10 Mon Sep 17 00:00:00 2001
From: BayoNet <da-daos@yandex.ru>
Date: Fri, 29 May 2020 12:43:49 +0300
Subject: [PATCH 121/122] DOCS-527: system.query_log (#11197)

* DOCSUP-824: query log (#115)

* Draft changes.

* Next step.

* Structure was changed. Missed information was added.

* More edits.

* Update docs/en/operations/system-tables.md

Co-authored-by: BayoNet <da-daos@yandex.ru>

* Update docs/en/operations/system-tables.md

Co-authored-by: BayoNet <da-daos@yandex.ru>

* Update docs/en/operations/system-tables.md

Co-authored-by: BayoNet <da-daos@yandex.ru>

* Comments editing was done.

* Translated to russian.

Co-authored-by: Anastasia Uvarova <ana-uvarova@yandex-team.ru>
Co-authored-by: AnaUvarova <64017504+AnaUvarova@users.noreply.github.com>

* CLICKHOUSEDOCS-527: query_log description.

* CLICKHOUSEDOCS-527: Updated structure.

* CLICKHOUSEDOCS-527: Fixed links.

* Update docs/en/operations/system-tables.md

Co-authored-by: Ivan Blinkov <github@blinkov.ru>

* Update docs/en/operations/system-tables.md

Co-authored-by: Ivan Blinkov <github@blinkov.ru>

* Update docs/en/operations/system-tables.md

Co-authored-by: Ivan Blinkov <github@blinkov.ru>

* Update docs/en/operations/system-tables.md

Co-authored-by: Ivan Blinkov <github@blinkov.ru>

* Update docs/en/operations/system-tables.md

Co-authored-by: Ivan Blinkov <github@blinkov.ru>

* Updated by comments.

* CLICKHOUSEDOCS-527: Updated by comments
.

* CLICKHOUSEDOCS-527: Syncronized EN->RU

Co-authored-by: Anastasia Uvarova <ana-uvarova@yandex-team.ru>
Co-authored-by: AnaUvarova <64017504+AnaUvarova@users.noreply.github.com>
Co-authored-by: Sergei Shtykov <bayonet@yandex-team.ru>
Co-authored-by: Ivan Blinkov <github@blinkov.ru>
---
 .../settings.md                               |   4 +-
 docs/en/operations/settings/settings.md       |   2 +-
 docs/en/operations/system-tables.md           | 180 ++++++++----
 .../settings.md                               |   4 +-
 docs/ru/operations/settings/settings.md       |   2 +-
 docs/ru/operations/system-tables.md           | 271 ++++++++----------
 6 files changed, 248 insertions(+), 215 deletions(-)

diff --git a/docs/en/operations/server-configuration-parameters/settings.md b/docs/en/operations/server-configuration-parameters/settings.md
index 93acf3cae7a..5961c701283 100644
--- a/docs/en/operations/server-configuration-parameters/settings.md
+++ b/docs/en/operations/server-configuration-parameters/settings.md
@@ -586,11 +586,11 @@ If the table doesn’t exist, ClickHouse will create it. If the structure of the
 </query_log>
 ```
 
-## query\_thread\_log {#server_configuration_parameters-query-thread-log}
+## query\_thread\_log {#server_configuration_parameters-query_thread_log}
 
 Setting for logging threads of queries received with the [log\_query\_threads=1](../settings/settings.md#settings-log-query-threads) setting.
 
-Queries are logged in the [system.query\_thread\_log](../../operations/system-tables.md#system_tables-query-thread-log) table, not in a separate file. You can change the name of the table in the `table` parameter (see below).
+Queries are logged in the [system.query\_thread\_log](../../operations/system-tables.md#system_tables-query_thread_log) table, not in a separate file. You can change the name of the table in the `table` parameter (see below).
 
 Use the following parameters to configure logging:
 
diff --git a/docs/en/operations/settings/settings.md b/docs/en/operations/settings/settings.md
index 7103819e499..880f0ffedb1 100644
--- a/docs/en/operations/settings/settings.md
+++ b/docs/en/operations/settings/settings.md
@@ -598,7 +598,7 @@ log_queries_min_type='EXCEPTION_WHILE_PROCESSING'
 
 Setting up query threads logging.
 
-Queries’ threads runned by ClickHouse with this setup are logged according to the rules in the [query\_thread\_log](../server-configuration-parameters/settings.md#server_configuration_parameters-query-thread-log) server configuration parameter.
+Queries’ threads runned by ClickHouse with this setup are logged according to the rules in the [query\_thread\_log](../server-configuration-parameters/settings.md#server_configuration_parameters-query_thread_log) server configuration parameter.
 
 Example:
 
diff --git a/docs/en/operations/system-tables.md b/docs/en/operations/system-tables.md
index cc5456f5324..f364d4e3068 100644
--- a/docs/en/operations/system-tables.md
+++ b/docs/en/operations/system-tables.md
@@ -5,7 +5,7 @@ toc_title: System Tables
 
 # System Tables {#system-tables}
 
-## Introduction
+## Introduction {#system-tables-introduction}
 
 System tables provide information about:
 
@@ -18,9 +18,12 @@ System tables:
 - Available only for reading data.
 - Can't be dropped or altered, but can be detached.
 
-The `metric_log`, `query_log`, `query_thread_log`, `trace_log` system tables store data in a storage filesystem. Other system tables store their data in RAM. ClickHouse server creates such system tables at the start.
+Most of system tables store their data in RAM. ClickHouse server creates such system tables at the start.
 
-### Sources of System Metrics
+The [metric_log](#system_tables-metric_log), [query_log](#system_tables-query_log), [query_thread_log](#system_tables-query_thread_log), [trace_log](#system_tables-trace_log) system tables store data in a storage filesystem. You can alter them or remove from a disk manually. If you remove one of that tables from a disk, the ClickHouse server creates the table again at the time of the next recording. A storage period for these tables is not limited, and ClickHouse server doesn't delete their data automatically. You need to organize removing of outdated logs by yourself. For example, you can use [TTL](../sql-reference/statements/alter.md#manipulations-with-table-ttl) settings for removing outdated log records. 
+
+
+### Sources of System Metrics {#system-tables-sources-of-system-metrics}
 
 For collecting system metrics ClickHouse server uses:
 
@@ -587,97 +590,150 @@ Columns:
 -   `source_file` (LowCardinality(String)) — Source file from which the logging was done.
 -   `source_line` (UInt64) — Source line from which the logging was done.
 
-## system.query\_log {#system_tables-query_log}
+## system.query_log {#system_tables-query_log}
 
-Contains information about execution of queries. For each query, you can see processing start time, duration of processing, error messages and other information.
+Contains information about executed queries, for example, start time, duration of processing, error messages.
 
 !!! note "Note"
     The table doesn’t contain input data for `INSERT` queries.
 
-ClickHouse creates this table only if the [query\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query-log) server parameter is specified. This parameter sets the logging rules, such as the logging interval or the name of the table the queries will be logged in.
+You can change settings of queries logging in the [query_log](server-configuration-parameters/settings.md#server_configuration_parameters-query-log) section of the server configuration.
 
-To enable query logging, set the [log\_queries](settings/settings.md#settings-log-queries) parameter to 1. For details, see the [Settings](settings/settings.md) section.
+You can disable queries logging by setting [log_queries = 0](settings/settings.md#settings-log-queries). We don't recommend to turn off logging because information in this table is important for solving issues.
+
+The flushing period of logs is set in `flush_interval_milliseconds` parameter of the [query_log](server-configuration-parameters/settings.md#server_configuration_parameters-query-log) server settings section. To force flushing logs, use the [SYSTEM FLUSH LOGS](../sql-reference/statements/system.md#query_language-system-flush_logs) query.
+
+ClickHouse doesn't delete logs from the table automatically. See [Introduction](#system-tables-introduction) for more details.
 
 The `system.query_log` table registers two kinds of queries:
 
 1.  Initial queries that were run directly by the client.
 2.  Child queries that were initiated by other queries (for distributed query execution). For these types of queries, information about the parent queries is shown in the `initial_*` columns.
 
+Each query creates one or two rows in the `query_log` table, depending on the status (see the `type` column) of the query:
+
+1.  If the query execution was successful, two rows with the `QueryStart` and `QueryFinish` types are created .
+2.  If an error occurred during query processing, two events with the `QueryStart` and `ExceptionWhileProcessing` types are created .
+3.  If an error occurred before launching the query, a single event with the `ExceptionBeforeStart` type is created.
+
 Columns:
 
--   `type` (`Enum8`) — Type of event that occurred when executing the query. Values:
+-   `type` ([Enum8](../sql-reference/data-types/enum.md)) — Type of an event that occurred when executing the query. Values:
     -   `'QueryStart' = 1` — Successful start of query execution.
     -   `'QueryFinish' = 2` — Successful end of query execution.
     -   `'ExceptionBeforeStart' = 3` — Exception before the start of query execution.
     -   `'ExceptionWhileProcessing' = 4` — Exception during the query execution.
--   `event_date` (Date) — Query starting date.
--   `event_time` (DateTime) — Query starting time.
--   `query_start_time` (DateTime) — Start time of query execution.
--   `query_duration_ms` (UInt64) — Duration of query execution.
--   `read_rows` (UInt64) — Number of read rows.
--   `read_bytes` (UInt64) — Number of read bytes.
--   `written_rows` (UInt64) — For `INSERT` queries, the number of written rows. For other queries, the column value is 0.
--   `written_bytes` (UInt64) — For `INSERT` queries, the number of written bytes. For other queries, the column value is 0.
--   `result_rows` (UInt64) — Number of rows in the result.
--   `result_bytes` (UInt64) — Number of bytes in the result.
--   `memory_usage` (UInt64) — Memory consumption by the query.
--   `query` (String) — Query string.
--   `exception` (String) — Exception message.
--   `stack_trace` (String) — Stack trace (a list of methods called before the error occurred). An empty string, if the query is completed successfully.
--   `is_initial_query` (UInt8) — Query type. Possible values:
+-   `event_date` ([Date](../sql-reference/data-types/date.md)) — Query starting date.
+-   `event_time` ([DateTime](../sql-reference/data-types/datetime.md)) — Query starting time.
+-   `query_start_time` ([DateTime](../sql-reference/data-types/datetime.md)) — Start time of query execution.
+-   `query_duration_ms` ([UInt64](../sql-reference/data-types/int-uint.md#uint-ranges)) — Duration of query execution in milliseconds.
+-   `read_rows` ([UInt64](../sql-reference/data-types/int-uint.md#uint-ranges)) — Total number or rows read from all tables and table functions participated in query. It includes usual subqueries, subqueries for `IN` and `JOIN`. For distributed queries `read_rows` includes the total number of rows read at all replicas. Each replica sends it's `read_rows` value, and the server-initiator of the query summarize all received and local values. The cache volumes doesn't affect this value.
+-   `read_bytes` ([UInt64](../sql-reference/data-types/int-uint.md#uint-ranges)) — Total number or bytes read from all tables and table functions participated in query. It includes usual subqueries, subqueries for `IN` and `JOIN`. For distributed queries `read_bytes` includes the total number of rows read at all replicas. Each replica sends it's `read_bytes` value, and the server-initiator of the query summarize all received and local values. The cache volumes doesn't affect this value.
+-   `written_rows` ([UInt64](../sql-reference/data-types/int-uint.md#uint-ranges)) — For `INSERT` queries, the number of written rows. For other queries, the column value is 0.
+-   `written_bytes` ([UInt64](../sql-reference/data-types/int-uint.md#uint-ranges)) — For `INSERT` queries, the number of written bytes. For other queries, the column value is 0.
+-   `result_rows` ([UInt64](../sql-reference/data-types/int-uint.md#uint-ranges)) — Number of rows in a result of the `SELECT` query, or a number of rows in the `INSERT` query.
+-   `result_bytes` ([UInt64](../sql-reference/data-types/int-uint.md#uint-ranges)) — RAM volume in bytes used to store a query result.
+-   `memory_usage` ([UInt64](../sql-reference/data-types/int-uint.md#uint-ranges)) — Memory consumption by the query.
+-   `query` ([String](../sql-reference/data-types/string.md)) — Query string.
+-   `exception` ([String](../sql-reference/data-types/string.md)) — Exception message.
+-   `exception_code` ([Int32](../sql-reference/data-types/int-uint.md)) — Code of an exception. 
+-   `stack_trace` ([String](../sql-reference/data-types/string.md)) — [Stack trace](https://en.wikipedia.org/wiki/Stack_trace). An empty string, if the query was completed successfully.
+-   `is_initial_query` ([UInt8](../sql-reference/data-types/int-uint.md)) — Query type. Possible values:
     -   1 — Query was initiated by the client.
-    -   0 — Query was initiated by another query for distributed query execution.
--   `user` (String) — Name of the user who initiated the current query.
--   `query_id` (String) — ID of the query.
--   `address` (IPv6) — IP address that was used to make the query.
--   `port` (UInt16) — The client port that was used to make the query.
--   `initial_user` (String) — Name of the user who ran the initial query (for distributed query execution).
--   `initial_query_id` (String) — ID of the initial query (for distributed query execution).
--   `initial_address` (IPv6) — IP address that the parent query was launched from.
--   `initial_port` (UInt16) — The client port that was used to make the parent query.
--   `interface` (UInt8) — Interface that the query was initiated from. Possible values:
+    -   0 — Query was initiated by another query as part of distributed query execution.
+-   `user` ([String](../sql-reference/data-types/string.md)) — Name of the user who initiated the current query.
+-   `query_id` ([String](../sql-reference/data-types/string.md)) — ID of the query.
+-   `address` ([IPv6](../sql-reference/data-types/domains/ipv6.md)) — IP address that was used to make the query.
+-   `port` ([UInt16](../sql-reference/data-types/int-uint.md)) — The client port that was used to make the query.
+-   `initial_user` ([String](../sql-reference/data-types/string.md)) — Name of the user who ran the initial query (for distributed query execution).
+-   `initial_query_id` ([String](../sql-reference/data-types/string.md)) — ID of the initial query (for distributed query execution).
+-   `initial_address` ([IPv6](../sql-reference/data-types/domains/ipv6.md)) — IP address that the parent query was launched from.
+-   `initial_port` ([UInt16](../sql-reference/data-types/int-uint.md)) — The client port that was used to make the parent query.
+-   `interface` ([UInt8](../sql-reference/data-types/int-uint.md)) — Interface that the query was initiated from. Possible values:
     -   1 — TCP.
     -   2 — HTTP.
--   `os_user` (String) — OS’s username who runs [clickhouse-client](../interfaces/cli.md).
--   `client_hostname` (String) — Hostname of the client machine where the [clickhouse-client](../interfaces/cli.md) or another TCP client is run.
--   `client_name` (String) — The [clickhouse-client](../interfaces/cli.md) or another TCP client name.
--   `client_revision` (UInt32) — Revision of the [clickhouse-client](../interfaces/cli.md) or another TCP client.
--   `client_version_major` (UInt32) — Major version of the [clickhouse-client](../interfaces/cli.md) or another TCP client.
--   `client_version_minor` (UInt32) — Minor version of the [clickhouse-client](../interfaces/cli.md) or another TCP client.
--   `client_version_patch` (UInt32) — Patch component of the [clickhouse-client](../interfaces/cli.md) or another TCP client version.
+-   `os_user` ([String](../sql-reference/data-types/string.md)) — Operating system username who runs [clickhouse-client](../interfaces/cli.md).
+-   `client_hostname` ([String](../sql-reference/data-types/string.md)) — Hostname of the client machine where the [clickhouse-client](../interfaces/cli.md) or another TCP client is run.
+-   `client_name` ([String](../sql-reference/data-types/string.md)) — The [clickhouse-client](../interfaces/cli.md) or another TCP client name.
+-   `client_revision` ([UInt32](../sql-reference/data-types/int-uint.md)) — Revision of the [clickhouse-client](../interfaces/cli.md) or another TCP client.
+-   `client_version_major` ([UInt32](../sql-reference/data-types/int-uint.md)) — Major version of the [clickhouse-client](../interfaces/cli.md) or another TCP client.
+-   `client_version_minor` ([UInt32](../sql-reference/data-types/int-uint.md)) — Minor version of the [clickhouse-client](../interfaces/cli.md) or another TCP client.
+-   `client_version_patch` ([UInt32](../sql-reference/data-types/int-uint.md)) — Patch component of the [clickhouse-client](../interfaces/cli.md) or another TCP client version.
 -   `http_method` (UInt8) — HTTP method that initiated the query. Possible values:
     -   0 — The query was launched from the TCP interface.
     -   1 — `GET` method was used.
     -   2 — `POST` method was used.
--   `http_user_agent` (String) — The `UserAgent` header passed in the HTTP request.
--   `quota_key` (String) — The “quota key” specified in the [quotas](quotas.md) setting (see `keyed`).
--   `revision` (UInt32) — ClickHouse revision.
--   `thread_numbers` (Array(UInt32)) — Number of threads that are participating in query execution.
--   `ProfileEvents.Names` (Array(String)) — Counters that measure different metrics. The description of them could be found in the table [system.events](#system_tables-events)
--   `ProfileEvents.Values` (Array(UInt64)) — Values of metrics that are listed in the `ProfileEvents.Names` column.
--   `Settings.Names` (Array(String)) — Names of settings that were changed when the client ran the query. To enable logging changes to settings, set the `log_query_settings` parameter to 1.
--   `Settings.Values` (Array(String)) — Values of settings that are listed in the `Settings.Names` column.
+-   `http_user_agent` ([String](../sql-reference/data-types/string.md)) — The `UserAgent` header passed in the HTTP request.
+-   `quota_key` ([String](../sql-reference/data-types/string.md)) — The “quota key” specified in the [quotas](quotas.md) setting (see `keyed`).
+-   `revision` ([UInt32](../sql-reference/data-types/int-uint.md)) — ClickHouse revision.
+-   `thread_numbers` ([Array(UInt32)](../sql-reference/data-types/array.md)) — Number of threads that are participating in query execution.
+-   `ProfileEvents.Names` ([Array(String)](../sql-reference/data-types/array.md)) — Counters that measure different metrics. The description of them could be found in the table [system.events](#system_tables-events)
+-   `ProfileEvents.Values` ([Array(UInt64)](../sql-reference/data-types/array.md)) — Values of metrics that are listed in the `ProfileEvents.Names` column.
+-   `Settings.Names` ([Array(String)](../sql-reference/data-types/array.md)) — Names of settings that were changed when the client ran the query. To enable logging changes to settings, set the `log_query_settings` parameter to 1.
+-   `Settings.Values` ([Array(String)](../sql-reference/data-types/array.md)) — Values of settings that are listed in the `Settings.Names` column.
 
-Each query creates one or two rows in the `query_log` table, depending on the status of the query:
+**Example**
 
-1.  If the query execution is successful, two events with types 1 and 2 are created (see the `type` column).
-2.  If an error occurred during query processing, two events with types 1 and 4 are created.
-3.  If an error occurred before launching the query, a single event with type 3 is created.
+``` sql
+SELECT * FROM system.query_log LIMIT 1 FORMAT Vertical;
+```
 
-By default, logs are added to the table at intervals of 7.5 seconds. You can set this interval in the [query\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query-log) server setting (see the `flush_interval_milliseconds` parameter). To flush the logs forcibly from the memory buffer into the table, use the `SYSTEM FLUSH LOGS` query.
+``` text
+Row 1:
+──────
+type:                 QueryStart
+event_date:           2020-05-13
+event_time:           2020-05-13 14:02:28
+query_start_time:     2020-05-13 14:02:28
+query_duration_ms:    0
+read_rows:            0
+read_bytes:           0
+written_rows:         0
+written_bytes:        0
+result_rows:          0
+result_bytes:         0
+memory_usage:         0
+query:                SELECT 1
+exception_code:       0
+exception:
+stack_trace:
+is_initial_query:     1
+user:                 default
+query_id:             5e834082-6f6d-4e34-b47b-cd1934f4002a
+address:              ::ffff:127.0.0.1
+port:                 57720
+initial_user:         default
+initial_query_id:     5e834082-6f6d-4e34-b47b-cd1934f4002a
+initial_address:      ::ffff:127.0.0.1
+initial_port:         57720
+interface:            1
+os_user:              bayonet
+client_hostname:      clickhouse.ru-central1.internal
+client_name:          ClickHouse client
+client_revision:      54434
+client_version_major: 20
+client_version_minor: 4
+client_version_patch: 1
+http_method:          0
+http_user_agent:
+quota_key:
+revision:             54434
+thread_ids:           []
+ProfileEvents.Names:  []
+ProfileEvents.Values: []
+Settings.Names:       ['use_uncompressed_cache','load_balancing','log_queries','max_memory_usage']
+Settings.Values:      ['0','random','1','10000000000']
 
-When the table is deleted manually, it will be automatically created on the fly. Note that all the previous logs will be deleted.
+```
+**See Also**
 
-!!! note "Note"
-    The storage period for logs is unlimited. Logs aren’t automatically deleted from the table. You need to organize the removal of outdated logs yourself.
+-   [system.query_thread_log](#system_tables-query_thread_log) — This table contains information about each query execution thread.
 
-You can specify an arbitrary partitioning key for the `system.query_log` table in the [query\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query-log) server setting (see the `partition_by` parameter).
-
-## system.query\_thread\_log {#system_tables-query-thread-log}
+## system.query_thread_log {#system_tables-query_thread_log}
 
 The table contains information about each query execution thread.
 
-ClickHouse creates this table only if the [query\_thread\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query-thread-log) server parameter is specified. This parameter sets the logging rules, such as the logging interval or the name of the table the queries will be logged in.
+ClickHouse creates this table only if the [query\_thread\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query_thread_log) server parameter is specified. This parameter sets the logging rules, such as the logging interval or the name of the table the queries will be logged in.
 
 To enable query logging, set the [log\_query\_threads](settings/settings.md#settings-log-query-threads) parameter to 1. For details, see the [Settings](settings/settings.md) section.
 
@@ -729,14 +785,14 @@ Columns:
 -   `ProfileEvents.Names` (Array(String)) — Counters that measure different metrics for this thread. The description of them could be found in the table [system.events](#system_tables-events)
 -   `ProfileEvents.Values` (Array(UInt64)) — Values of metrics for this thread that are listed in the `ProfileEvents.Names` column.
 
-By default, logs are added to the table at intervals of 7.5 seconds. You can set this interval in the [query\_thread\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query-thread-log) server setting (see the `flush_interval_milliseconds` parameter). To flush the logs forcibly from the memory buffer into the table, use the `SYSTEM FLUSH LOGS` query.
+By default, logs are added to the table at intervals of 7.5 seconds. You can set this interval in the [query\_thread\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query_thread_log) server setting (see the `flush_interval_milliseconds` parameter). To flush the logs forcibly from the memory buffer into the table, use the `SYSTEM FLUSH LOGS` query.
 
 When the table is deleted manually, it will be automatically created on the fly. Note that all the previous logs will be deleted.
 
 !!! note "Note"
     The storage period for logs is unlimited. Logs aren’t automatically deleted from the table. You need to organize the removal of outdated logs yourself.
 
-You can specify an arbitrary partitioning key for the `system.query_thread_log` table in the [query\_thread\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query-thread-log) server setting (see the `partition_by` parameter).
+You can specify an arbitrary partitioning key for the `system.query_thread_log` table in the [query\_thread\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query_thread_log) server setting (see the `partition_by` parameter).
 
 ## system.trace\_log {#system_tables-trace_log}
 
diff --git a/docs/ru/operations/server-configuration-parameters/settings.md b/docs/ru/operations/server-configuration-parameters/settings.md
index c664580c659..ccde7a945ac 100644
--- a/docs/ru/operations/server-configuration-parameters/settings.md
+++ b/docs/ru/operations/server-configuration-parameters/settings.md
@@ -574,11 +574,11 @@ ClickHouse проверит условия `min_part_size` и `min_part_size_rat
 </query_log>
 ```
 
-## query\_thread\_log {#server_configuration_parameters-query-thread-log}
+## query\_thread\_log {#server_configuration_parameters-query_thread_log}
 
 Настройка логирования потоков выполнения запросов, принятых с настройкой [log\_query\_threads=1](../settings/settings.md#settings-log-query-threads).
 
-Запросы логируются не в отдельный файл, а в системную таблицу [system.query\_thread\_log](../../operations/server-configuration-parameters/settings.md#system_tables-query-thread-log). Вы можете изменить название этой таблицы в параметре `table` (см. ниже).
+Запросы логируются не в отдельный файл, а в системную таблицу [system.query\_thread\_log](../../operations/server-configuration-parameters/settings.md#system_tables-query_thread_log). Вы можете изменить название этой таблицы в параметре `table` (см. ниже).
 
 При настройке логирования используются следующие параметры:
 
diff --git a/docs/ru/operations/settings/settings.md b/docs/ru/operations/settings/settings.md
index 56c3042bfa3..4dd43e9607b 100644
--- a/docs/ru/operations/settings/settings.md
+++ b/docs/ru/operations/settings/settings.md
@@ -536,7 +536,7 @@ log_queries=1
 
 Установка логирования информации о потоках выполнения запроса.
 
-Лог информации о потоках выполнения запросов, переданных в ClickHouse с этой установкой, записывается согласно правилам конфигурационного параметра сервера [query\_thread\_log](../server-configuration-parameters/settings.md#server_configuration_parameters-query-thread-log).
+Лог информации о потоках выполнения запросов, переданных в ClickHouse с этой установкой, записывается согласно правилам конфигурационного параметра сервера [query\_thread\_log](../server-configuration-parameters/settings.md#server_configuration_parameters-query_thread_log).
 
 Пример:
 
diff --git a/docs/ru/operations/system-tables.md b/docs/ru/operations/system-tables.md
index ae0e67a4515..8954a484560 100644
--- a/docs/ru/operations/system-tables.md
+++ b/docs/ru/operations/system-tables.md
@@ -1,4 +1,7 @@
-# Системные таблицы {#sistemnye-tablitsy}
+# Системные таблицы {#system-tables}
+
+
+## Введение {#system-tables-introduction}
 
 Системные таблицы используются для реализации части функциональности системы, а также предоставляют доступ к информации о работе системы.
 Вы не можете удалить системную таблицу (хотя можете сделать DETACH).
@@ -544,182 +547,156 @@ CurrentMetric_ReplicatedChecks:                             0
 -   `source_file` (LowCardinality(String)) — Исходный файл, из которого была сделана запись.
 -   `source_line` (UInt64) — Исходная строка, из которой была сделана запись.
 
-## system.query\_log {#system_tables-query_log}
+## system.query_log {#system_tables-query_log}
 
-Содержит информацию о выполнении запросов. Для каждого запроса вы можете увидеть время начала обработки, продолжительность обработки, сообщения об ошибках и другую информацию.
+Содержит информацию о выполняемых запросах, например, время начала обработки, продолжительность обработки, сообщения об ошибках.
 
 !!! note "Внимание"
     Таблица не содержит входных данных для запросов `INSERT`.
 
-ClickHouse создаёт таблицу только в том случае, когда установлен конфигурационный параметр сервера [query\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query-log). Параметр задаёт правила ведения лога, такие как интервал логирования или имя таблицы, в которую будут логгироваться запросы.
+Настойки логгирования можно изменить в секции серверной конфигурации [query_log](server-configuration-parameters/settings.md#server_configuration_parameters-query-log).
 
-Чтобы включить логирование, задайте значение параметра [log\_queries](settings/settings.md#settings-log-queries) равным 1. Подробности смотрите в разделе [Настройки](settings/settings.md#settings).
+Можно отключить логгирование настройкой [log_queries = 0](settings/settings.md#settings-log-queries). По-возможности, не отключайте логгирование, поскольку информация из таблицы важна при решении проблем.
+
+Период сброса логов в таблицу задаётся параметром `flush_interval_milliseconds` в конфигурационной секции [query_log](server-configuration-parameters/settings.md#server_configuration_parameters-query-log). Чтобы принудительно записать логи из буффера памяти в таблицу, используйте запрос [SYSTEM FLUSH LOGS](../sql-reference/statements/system.md#query_language-system-flush_logs).
+
+ClickHouse не удаляет логи из таблица автоматически. Смотрите [Введение](#system-tables-introduction).
+
+Можно указать произвольный ключ партиционирования для таблицы `system.query_log` в конфигурации [query\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query-log) (параметр `partition_by`).
+
+
+
+Если таблицу удалить вручную, она создается заново автоматически «на лету». При этом все логи на момент удаления таблицы будут убраны.
 
 Таблица `system.query_log` содержит информацию о двух видах запросов:
 
 1.  Первоначальные запросы, которые были выполнены непосредственно клиентом.
 2.  Дочерние запросы, инициированные другими запросами (для выполнения распределенных запросов). Для дочерних запросов информация о первоначальном запросе содержится в столбцах `initial_*`.
 
+В зависимости от статуса (столбец `type`) каждый запрос создаёт одну или две строки в таблице `query_log`:
+
+1.  Если запрос выполнен успешно, создаются два события типа `QueryStart` и `QueryFinish`.
+2.  Если во время обработки запроса возникла ошибка, создаются два события с типами `QueryStart` и `ExceptionWhileProcessing`.
+3.  Если ошибка произошла ещё до запуска запроса, создается одно событие с типом `ExceptionBeforeStart`.
+
 Столбцы:
 
--   `type` (`Enum8`) — тип события, произошедшего при выполнении запроса. Значения:
+-   `type` ([Enum8](../sql-reference/data-types/enum.md)) — тип события, произошедшего при выполнении запроса. Значения:
     -   `'QueryStart' = 1` — успешное начало выполнения запроса.
     -   `'QueryFinish' = 2` — успешное завершение выполнения запроса.
     -   `'ExceptionBeforeStart' = 3` — исключение перед началом обработки запроса.
     -   `'ExceptionWhileProcessing' = 4` — исключение во время обработки запроса.
--   `event_date` (Date) — дата начала запроса.
--   `event_time` (DateTime) — время начала запроса.
--   `query_start_time` (DateTime) — время начала обработки запроса.
--   `query_duration_ms` (UInt64) — длительность обработки запроса.
--   `read_rows` (UInt64) — количество прочитанных строк.
--   `read_bytes` (UInt64) — количество прочитанных байтов.
--   `written_rows` (UInt64) — количество записанных строк для запросов `INSERT`. Для других запросов, значение столбца 0.
--   `written_bytes` (UInt64) — объём записанных данных в байтах для запросов `INSERT`. Для других запросов, значение столбца 0.
--   `result_rows` (UInt64) — количество строк в результате.
--   `result_bytes` (UInt64) — объём результата в байтах.
--   `memory_usage` (UInt64) — потребление RAM запросом.
--   `query` (String) — текст запроса.
--   `exception` (String) — сообщение исключения, если запрос завершился по исключению.
--   `stack_trace` (String) — трассировка (список функций, последовательно вызванных перед ошибкой). Пустая строка, если запрос успешно завершен.
--   `is_initial_query` (UInt8) — вид запроса. Возможные значения:
+-   `event_date` ([Date](../sql-reference/data-types/date.md)) — дата начала запроса.
+-   `event_time` ([DateTime](../sql-reference/data-types/datetime.md)) — время начала запроса.
+-   `query_start_time` ([DateTime](../sql-reference/data-types/datetime.md)) — время начала обработки запроса.
+-   `query_duration_ms` ([UInt64](../sql-reference/data-types/int-uint.md#uint-ranges)) — длительность выполнения запроса в миллисекундах.
+-   `read_rows` ([UInt64](../sql-reference/data-types/int-uint.md#uint-ranges)) — Общее количество строк, считанных из всех таблиц и табличных функций, участвующих в запросе. Включает в себя обычные подзапросы, подзапросы для `IN` и `JOIN`. Для распределенных запросов `read_rows` включает в себя общее количество строк, прочитанных на всех репликах. Каждая реплика передает собственное значение `read_rows`, а сервер-инициатор запроса суммирует все полученные и локальные значения. Объемы кэша не учитываюся.
+-   `read_bytes` ([UInt64](../sql-reference/data-types/int-uint.md#uint-ranges)) — Общее количество байтов, считанных из всех таблиц и табличных функций, участвующих в запросе. Включает в себя обычные подзапросы, подзапросы для `IN` и `JOIN`. Для распределенных запросов `read_bytes` включает в себя общее количество байтов, прочитанных на всех репликах. Каждая реплика передает собственное значение `read_bytes`, а сервер-инициатор запроса суммирует все полученные и локальные значения. Объемы кэша не учитываюся.
+-   `written_rows` ([UInt64](../sql-reference/data-types/int-uint.md#uint-ranges)) — количество записанных строк для запросов `INSERT`. Для других запросов, значение столбца 0.
+-   `written_bytes` ([UInt64](../sql-reference/data-types/int-uint.md#uint-ranges)) — объём записанных данных в байтах для запросов `INSERT`. Для других запросов, значение столбца 0.
+-   `result_rows` ([UInt64](../sql-reference/data-types/int-uint.md#uint-ranges)) — количество строк в результате запроса `SELECT` или количество строк в запросе `INSERT`.
+-   `result_bytes` ([UInt64](../sql-reference/data-types/int-uint.md#uint-ranges)) — объём RAM в байтах, использованный для хранения результата запроса.
+-   `memory_usage` ([UInt64](../sql-reference/data-types/int-uint.md#uint-ranges)) — потребление RAM запросом.
+-   `query` ([String](../sql-reference/data-types/string.md)) — текст запроса.
+-   `exception` ([String](../sql-reference/data-types/string.md)) — сообщение исключения, если запрос завершился по исключению.
+-   `exception_code` ([Int32](../sql-reference/data-types/int-uint.md)) — код исключения. 
+-   `stack_trace` ([String](../sql-reference/data-types/string.md)) — [stack trace](https://en.wikipedia.org/wiki/Stack_trace). Пустая строка, если запрос успешно завершен.
+-   `is_initial_query` ([UInt8](../sql-reference/data-types/int-uint.md)) — вид запроса. Возможные значения:
     -   1 — запрос был инициирован клиентом.
-    -   0 — запрос был инициирован другим запросом при распределенном запросе.
--   `user` (String) — пользователь, запустивший текущий запрос.
--   `query_id` (String) — ID запроса.
--   `address` (IPv6) — IP адрес, с которого пришел запрос.
--   `port` (UInt16) — порт, с которого клиент сделал запрос
--   `initial_user` (String) — пользователь, запустивший первоначальный запрос (для распределенных запросов).
--   `initial_query_id` (String) — ID родительского запроса.
--   `initial_address` (IPv6) — IP адрес, с которого пришел родительский запрос.
--   `initial_port` (UInt16) — порт, с которого клиент сделал родительский запрос.
--   `interface` (UInt8) — интерфейс, с которого ушёл запрос. Возможные значения:
+    -   0 — запрос был инициирован другим запросом при выполнении распределенного запроса.
+-   `user` ([String](../sql-reference/data-types/string.md)) — пользователь, запустивший текущий запрос.
+-   `query_id` ([String](../sql-reference/data-types/string.md)) — ID запроса.
+-   `address` ([IPv6](../sql-reference/data-types/domains/ipv6.md)) — IP адрес, с которого пришел запрос.
+-   `port` ([UInt16](../sql-reference/data-types/int-uint.md)) — порт, с которого клиент сделал запрос
+-   `initial_user` ([String](../sql-reference/data-types/string.md)) — пользователь, запустивший первоначальный запрос (для распределенных запросов).
+-   `initial_query_id` ([String](../sql-reference/data-types/string.md)) — ID родительского запроса.
+-   `initial_address` ([IPv6](../sql-reference/data-types/domains/ipv6.md)) — IP адрес, с которого пришел родительский запрос.
+-   `initial_port` ([UInt16](../sql-reference/data-types/int-uint.md)) — порт, с которого клиент сделал родительский запрос.
+-   `interface` ([UInt8](../sql-reference/data-types/int-uint.md)) — интерфейс, с которого ушёл запрос. Возможные значения:
     -   1 — TCP.
     -   2 — HTTP.
--   `os_user` (String) — имя пользователя в OS, который запустил [clickhouse-client](../interfaces/cli.md).
--   `client_hostname` (String) — имя сервера, с которого присоединился [clickhouse-client](../interfaces/cli.md) или другой TCP клиент.
--   `client_name` (String) — [clickhouse-client](../interfaces/cli.md) или другой TCP клиент.
--   `client_revision` (UInt32) — ревизия [clickhouse-client](../interfaces/cli.md) или другого TCP клиента.
--   `client_version_major` (UInt32) — старшая версия [clickhouse-client](../interfaces/cli.md) или другого TCP клиента.
--   `client_version_minor` (UInt32) — младшая версия [clickhouse-client](../interfaces/cli.md) или другого TCP клиента.
--   `client_version_patch` (UInt32) — патч [clickhouse-client](../interfaces/cli.md) или другого TCP клиента.
--   `http_method` (UInt8) — HTTP метод, инициировавший запрос. Возможные значения:
+-   `os_user` ([String](../sql-reference/data-types/string.md)) — имя пользователя операционной системы, который запустил [clickhouse-client](../interfaces/cli.md).
+-   `client_hostname` ([String](../sql-reference/data-types/string.md)) — имя сервера, с которого присоединился [clickhouse-client](../interfaces/cli.md) или другой TCP клиент.
+-   `client_name` ([String](../sql-reference/data-types/string.md)) — [clickhouse-client](../interfaces/cli.md) или другой TCP клиент.
+-   `client_revision` ([UInt32](../sql-reference/data-types/int-uint.md)) — ревизия [clickhouse-client](../interfaces/cli.md) или другого TCP клиента.
+-   `client_version_major` ([UInt32](../sql-reference/data-types/int-uint.md)) — старшая версия [clickhouse-client](../interfaces/cli.md) или другого TCP клиента.
+-   `client_version_minor` ([UInt32](../sql-reference/data-types/int-uint.md)) — младшая версия [clickhouse-client](../interfaces/cli.md) или другого TCP клиента.
+-   `client_version_patch` ([UInt32](../sql-reference/data-types/int-uint.md)) — патч [clickhouse-client](../interfaces/cli.md) или другого TCP клиента.
+-   `http_method` ([UInt8](../sql-reference/data-types/int-uint.md)) — HTTP метод, инициировавший запрос. Возможные значения:
     -   0 — запрос запущен с интерфейса TCP.
     -   1 — `GET`.
     -   2 — `POST`.
--   `http_user_agent` (String) — HTTP заголовок `UserAgent`.
--   `quota_key` (String) — «ключ квоты» из настроек [квот](quotas.md) (см. `keyed`).
--   `revision` (UInt32) — ревизия ClickHouse.
--   `thread_numbers` (Array(UInt32)) — количество потоков, участвующих в обработке запросов.
--   `ProfileEvents.Names` (Array(String)) — Счетчики для изменения различных метрик. Описание метрик можно получить из таблицы [system.events](#system_tables-events)(\#system\_tables-events
--   `ProfileEvents.Values` (Array(UInt64)) — метрики, перечисленные в столбце `ProfileEvents.Names`.
--   `Settings.Names` (Array(String)) — имена настроек, которые меняются, когда клиент выполняет запрос. Чтобы разрешить логирование изменений настроек, установите параметр `log_query_settings` равным 1.
--   `Settings.Values` (Array(String)) — Значения настроек, которые перечислены в столбце `Settings.Names`.
+-   `http_user_agent` ([String](../sql-reference/data-types/string.md)) — HTTP заголовок `UserAgent`.
+-   `quota_key` ([String](../sql-reference/data-types/string.md)) — «ключ квоты» из настроек [квот](quotas.md) (см. `keyed`).
+-   `revision` ([UInt32](../sql-reference/data-types/int-uint.md)) — ревизия ClickHouse.
+-   `thread_numbers` ([Array(UInt32)](../sql-reference/data-types/array.md)) — количество потоков, участвующих в обработке запросов.
+-   `ProfileEvents.Names` ([Array(String)](../sql-reference/data-types/array.md)) — Счетчики для изменения различных метрик. Описание метрик можно получить из таблицы [system.events](#system_tables-events)(\#system\_tables-events
+-   `ProfileEvents.Values` ([Array(UInt64)](../sql-reference/data-types/array.md)) — метрики, перечисленные в столбце `ProfileEvents.Names`.
+-   `Settings.Names` ([Array(String)](../sql-reference/data-types/array.md)) — имена настроек, которые меняются, когда клиент выполняет запрос. Чтобы разрешить логирование изменений настроек, установите параметр `log_query_settings` равным 1.
+-   `Settings.Values` ([Array(String)](../sql-reference/data-types/array.md)) — Значения настроек, которые перечислены в столбце `Settings.Names`.
 
-Каждый запрос создаёт одну или две строки в таблице `query_log`, в зависимости от статуса запроса:
+**Пример**
 
-1.  Если запрос выполнен успешно, создаются два события типа 1 и 2 (смотрите столбец `type`).
-2.  Если во время обработки запроса произошла ошибка, создаются два события с типами 1 и 4.
-3.  Если ошибка произошла до запуска запроса, создается одно событие с типом 3.
+``` sql
+SELECT * FROM system.query_log LIMIT 1 FORMAT Vertical;
+```
 
-По умолчанию, строки добавляются в таблицу логирования с интервалом в 7,5 секунд. Можно задать интервал в конфигурационном параметре сервера [query\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query-log) (смотрите параметр `flush_interval_milliseconds`). Чтобы принудительно записать логи из буффера памяти в таблицу, используйте запрос `SYSTEM FLUSH LOGS`.
+``` text
+Row 1:
+──────
+type:                 QueryStart
+event_date:           2020-05-13
+event_time:           2020-05-13 14:02:28
+query_start_time:     2020-05-13 14:02:28
+query_duration_ms:    0
+read_rows:            0
+read_bytes:           0
+written_rows:         0
+written_bytes:        0
+result_rows:          0
+result_bytes:         0
+memory_usage:         0
+query:                SELECT 1
+exception_code:       0
+exception:
+stack_trace:
+is_initial_query:     1
+user:                 default
+query_id:             5e834082-6f6d-4e34-b47b-cd1934f4002a
+address:              ::ffff:127.0.0.1
+port:                 57720
+initial_user:         default
+initial_query_id:     5e834082-6f6d-4e34-b47b-cd1934f4002a
+initial_address:      ::ffff:127.0.0.1
+initial_port:         57720
+interface:            1
+os_user:              bayonet
+client_hostname:      clickhouse.ru-central1.internal
+client_name:          ClickHouse client
+client_revision:      54434
+client_version_major: 20
+client_version_minor: 4
+client_version_patch: 1
+http_method:          0
+http_user_agent:
+quota_key:
+revision:             54434
+thread_ids:           []
+ProfileEvents.Names:  []
+ProfileEvents.Values: []
+Settings.Names:       ['use_uncompressed_cache','load_balancing','log_queries','max_memory_usage']
+Settings.Values:      ['0','random','1','10000000000']
 
-Если таблицу удалить вручную, она пересоздастся автоматически «на лету». При этом все логи на момент удаления таблицы будут удалены.
+```
+**Смотрите также**
 
-!!! note "Примечание"
-    Срок хранения логов не ограничен. Логи не удаляются из таблицы автоматически. Вам необходимо самостоятельно организовать удаление устаревших логов.
+-   [system.query_thread_log](#system_tables-query_thread_log) — в этой таблице содержится информация о цепочке каждого выполненного запроса.
 
-Можно указать произвольный ключ партиционирования для таблицы `system.query_log` в конфигурации [query\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query-log) (параметр `partition_by`).
-
-## system.query\_log {#system_tables-query_log}
-
-Contains information about execution of queries. For each query, you can see processing start time, duration of processing, error messages and other information.
-
-!!! note "Note"
-    The table doesn’t contain input data for `INSERT` queries.
-
-ClickHouse creates this table only if the [query\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query-log) server parameter is specified. This parameter sets the logging rules, such as the logging interval or the name of the table the queries will be logged in.
-
-To enable query logging, set the [log\_queries](settings/settings.md#settings-log-queries) parameter to 1. For details, see the [Settings](settings/settings.md) section.
-
-The `system.query_log` table registers two kinds of queries:
-
-1.  Initial queries that were run directly by the client.
-2.  Child queries that were initiated by other queries (for distributed query execution). For these types of queries, information about the parent queries is shown in the `initial_*` columns.
-
-Columns:
-
--   `type` (`Enum8`) — Type of event that occurred when executing the query. Values:
-    -   `'QueryStart' = 1` — Successful start of query execution.
-    -   `'QueryFinish' = 2` — Successful end of query execution.
-    -   `'ExceptionBeforeStart' = 3` — Exception before the start of query execution.
-    -   `'ExceptionWhileProcessing' = 4` — Exception during the query execution.
--   `event_date` (Date) — Query starting date.
--   `event_time` (DateTime) — Query starting time.
--   `query_start_time` (DateTime) — Start time of query execution.
--   `query_duration_ms` (UInt64) — Duration of query execution.
--   `read_rows` (UInt64) — Number of read rows.
--   `read_bytes` (UInt64) — Number of read bytes.
--   `written_rows` (UInt64) — For `INSERT` queries, the number of written rows. For other queries, the column value is 0.
--   `written_bytes` (UInt64) — For `INSERT` queries, the number of written bytes. For other queries, the column value is 0.
--   `result_rows` (UInt64) — Number of rows in the result.
--   `result_bytes` (UInt64) — Number of bytes in the result.
--   `memory_usage` (UInt64) — Memory consumption by the query.
--   `query` (String) — Query string.
--   `exception` (String) — Exception message.
--   `stack_trace` (String) — Stack trace (a list of methods called before the error occurred). An empty string, if the query is completed successfully.
--   `is_initial_query` (UInt8) — Query type. Possible values:
-    -   1 — Query was initiated by the client.
-    -   0 — Query was initiated by another query for distributed query execution.
--   `user` (String) — Name of the user who initiated the current query.
--   `query_id` (String) — ID of the query.
--   `address` (IPv6) — IP address that was used to make the query.
--   `port` (UInt16) — The client port that was used to make the query.
--   `initial_user` (String) — Name of the user who ran the initial query (for distributed query execution).
--   `initial_query_id` (String) — ID of the initial query (for distributed query execution).
--   `initial_address` (IPv6) — IP address that the parent query was launched from.
--   `initial_port` (UInt16) — The client port that was used to make the parent query.
--   `interface` (UInt8) — Interface that the query was initiated from. Possible values:
-    -   1 — TCP.
-    -   2 — HTTP.
--   `os_user` (String) — OS’s username who runs [clickhouse-client](../interfaces/cli.md).
--   `client_hostname` (String) — Hostname of the client machine where the [clickhouse-client](../interfaces/cli.md) or another TCP client is run.
--   `client_name` (String) — The [clickhouse-client](../interfaces/cli.md) or another TCP client name.
--   `client_revision` (UInt32) — Revision of the [clickhouse-client](../interfaces/cli.md) or another TCP client.
--   `client_version_major` (UInt32) — Major version of the [clickhouse-client](../interfaces/cli.md) or another TCP client.
--   `client_version_minor` (UInt32) — Minor version of the [clickhouse-client](../interfaces/cli.md) or another TCP client.
--   `client_version_patch` (UInt32) — Patch component of the [clickhouse-client](../interfaces/cli.md) or another TCP client version.
--   `http_method` (UInt8) — HTTP method that initiated the query. Possible values:
-    -   0 — The query was launched from the TCP interface.
-    -   1 — `GET` method was used.
-    -   2 — `POST` method was used.
--   `http_user_agent` (String) — The `UserAgent` header passed in the HTTP request.
--   `quota_key` (String) — The «quota key» specified in the [quotas](quotas.md) setting (see `keyed`).
--   `revision` (UInt32) — ClickHouse revision.
--   `thread_numbers` (Array(UInt32)) — Number of threads that are participating in query execution.
--   `ProfileEvents.Names` (Array(String)) — Counters that measure different metrics. The description of them could be found in the table [system.events](#system_tables-events)
--   `ProfileEvents.Values` (Array(UInt64)) — Values of metrics that are listed in the `ProfileEvents.Names` column.
--   `Settings.Names` (Array(String)) — Names of settings that were changed when the client ran the query. To enable logging changes to settings, set the `log_query_settings` parameter to 1.
--   `Settings.Values` (Array(String)) — Values of settings that are listed in the `Settings.Names` column.
-
-Each query creates one or two rows in the `query_log` table, depending on the status of the query:
-
-1.  If the query execution is successful, two events with types 1 and 2 are created (see the `type` column).
-2.  If an error occurred during query processing, two events with types 1 and 4 are created.
-3.  If an error occurred before launching the query, a single event with type 3 is created.
-
-By default, logs are added to the table at intervals of 7.5 seconds. You can set this interval in the [query\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query-log) server setting (see the `flush_interval_milliseconds` parameter). To flush the logs forcibly from the memory buffer into the table, use the `SYSTEM FLUSH LOGS` query.
-
-When the table is deleted manually, it will be automatically created on the fly. Note that all the previous logs will be deleted.
-
-!!! note "Note"
-    The storage period for logs is unlimited. Logs aren’t automatically deleted from the table. You need to organize the removal of outdated logs yourself.
-
-You can specify an arbitrary partitioning key for the `system.query_log` table in the [query\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query-log) server setting (see the `partition_by` parameter).
-\#\# system.query\_thread\_log {\#system\_tables-query-thread-log}
+## system.query_thread_log {#system_tables-query_thread_log}
 
 Содержит информацию о каждом потоке выполняемых запросов.
 
-ClickHouse создаёт таблицу только в том случае, когда установлен конфигурационный параметр сервера [query\_thread\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query-thread-log). Параметр задаёт правила ведения лога, такие как интервал логирования или имя таблицы, в которую будут логгироваться запросы.
+ClickHouse создаёт таблицу только в том случае, когда установлен конфигурационный параметр сервера [query\_thread\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query_thread_log). Параметр задаёт правила ведения лога, такие как интервал логирования или имя таблицы, в которую будут логгироваться запросы.
 
 Чтобы включить логирование, задайте значение параметра [log\_query\_threads](settings/settings.md#settings-log-query-threads) равным 1. Подробности смотрите в разделе [Настройки](settings/settings.md#settings).
 
@@ -770,16 +747,16 @@ ClickHouse создаёт таблицу только в том случае, к
 -   `ProfileEvents.Names` (Array(String)) — Счетчики для изменения различных метрик для данного потока. Описание метрик можно получить из таблицы [system.events](#system_tables-events)(\#system\_tables-events
 -   `ProfileEvents.Values` (Array(UInt64)) — метрики для данного потока, перечисленные в столбце `ProfileEvents.Names`.
 
-По умолчанию, строки добавляются в таблицу логирования с интервалом в 7,5 секунд. Можно задать интервал в конфигурационном параметре сервера [query\_thread\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query-thread-log) (смотрите параметр `flush_interval_milliseconds`). Чтобы принудительно записать логи из буффера памяти в таблицу, используйте запрос `SYSTEM FLUSH LOGS`.
+По умолчанию, строки добавляются в таблицу логирования с интервалом в 7,5 секунд. Можно задать интервал в конфигурационном параметре сервера [query\_thread\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query_thread_log) (смотрите параметр `flush_interval_milliseconds`). Чтобы принудительно записать логи из буффера памяти в таблицу, используйте запрос `SYSTEM FLUSH LOGS`.
 
 Если таблицу удалить вручную, она пересоздастся автоматически «на лету». При этом все логи на момент удаления таблицы будут удалены.
 
 !!! note "Примечание"
     Срок хранения логов не ограничен. Логи не удаляются из таблицы автоматически. Вам необходимо самостоятельно организовать удаление устаревших логов.
 
-Можно указать произвольный ключ партиционирования для таблицы `system.query_log` в конфигурации [query\_thread\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query-thread-log) (параметр `partition_by`).
+Можно указать произвольный ключ партиционирования для таблицы `system.query_log` в конфигурации [query\_thread\_log](server-configuration-parameters/settings.md#server_configuration_parameters-query_thread_log) (параметр `partition_by`).
 
-## system.query_thread_log {#system_tables-query-thread-log}
+## system.query_thread_log {#system_tables-query_thread_log}
 
 Содержит информацию о каждом потоке исполнения запроса.
 

From 3f53b5b335c9993c092809253f736af406d8ade1 Mon Sep 17 00:00:00 2001
From: Alexander Kazakov <Akazz@users.noreply.github.com>
Date: Fri, 29 May 2020 12:45:55 +0300
Subject: [PATCH 122/122] Fixed improper logging of failures in destructors of
 certain metrics (#11264)

---
 src/Common/MemoryStatisticsOS.cpp    | 15 ++++++++++++++-
 src/Common/ProcfsMetricsProvider.cpp | 22 +++++++++++++++++++---
 2 files changed, 33 insertions(+), 4 deletions(-)

diff --git a/src/Common/MemoryStatisticsOS.cpp b/src/Common/MemoryStatisticsOS.cpp
index adc5bf5d904..6082d23cbd0 100644
--- a/src/Common/MemoryStatisticsOS.cpp
+++ b/src/Common/MemoryStatisticsOS.cpp
@@ -9,6 +9,7 @@
 #include <Common/Exception.h>
 #include <IO/ReadBufferFromMemory.h>
 #include <IO/ReadHelpers.h>
+#include <common/logger_useful.h>
 
 
 namespace DB
@@ -19,6 +20,7 @@ namespace ErrorCodes
     extern const int FILE_DOESNT_EXIST;
     extern const int CANNOT_OPEN_FILE;
     extern const int CANNOT_READ_FROM_FILE_DESCRIPTOR;
+    extern const int CANNOT_CLOSE_FILE;
 }
 
 static constexpr auto filename = "/proc/self/statm";
@@ -35,7 +37,18 @@ MemoryStatisticsOS::MemoryStatisticsOS()
 MemoryStatisticsOS::~MemoryStatisticsOS()
 {
     if (0 != ::close(fd))
-        tryLogCurrentException(__PRETTY_FUNCTION__);
+    {
+        try
+        {
+            throwFromErrno(
+                    "File descriptor for \"" + std::string(filename) + "\" could not be closed. "
+                    "Something seems to have gone wrong. Inspect errno.", ErrorCodes::CANNOT_CLOSE_FILE);
+        }
+        catch (const ErrnoException &)
+        {
+            DB::tryLogCurrentException(__PRETTY_FUNCTION__);
+        }
+    }
 }
 
 MemoryStatisticsOS::Data MemoryStatisticsOS::get() const
diff --git a/src/Common/ProcfsMetricsProvider.cpp b/src/Common/ProcfsMetricsProvider.cpp
index 633558a7b67..fcc4124aa85 100644
--- a/src/Common/ProcfsMetricsProvider.cpp
+++ b/src/Common/ProcfsMetricsProvider.cpp
@@ -7,6 +7,7 @@
 #include <IO/ReadHelpers.h>
 
 #include <common/find_symbols.h>
+#include <common/logger_useful.h>
 
 #include <cassert>
 #include <sys/types.h>
@@ -22,6 +23,7 @@ namespace ErrorCodes
 {
     extern const int FILE_DOESNT_EXIST;
     extern const int CANNOT_OPEN_FILE;
+    extern const int CANNOT_CLOSE_FILE;
     extern const int CANNOT_READ_FROM_FILE_DESCRIPTOR;
 }
 
@@ -39,6 +41,20 @@ namespace
             errno == ENOENT ? ErrorCodes::FILE_DOESNT_EXIST : ErrorCodes::CANNOT_OPEN_FILE);
 }
 
+inline void emitErrorMsgWithFailedToCloseFile(const std::string & filename)
+{
+    try
+    {
+        throwFromErrno(
+                "File descriptor for \"" + filename + "\" could not be closed. "
+                "Something seems to have gone wrong. Inspect errno.", ErrorCodes::CANNOT_CLOSE_FILE);
+    }
+    catch (const ErrnoException &)
+    {
+        DB::tryLogCurrentException(__PRETTY_FUNCTION__);
+    }
+}
+
 ssize_t readFromFD(const int fd, const char * filename, char * buf, size_t buf_size)
 {
     ssize_t res = 0;
@@ -100,11 +116,11 @@ ProcfsMetricsProvider::ProcfsMetricsProvider(const pid_t /*tid*/)
 ProcfsMetricsProvider::~ProcfsMetricsProvider()
 {
     if (stats_version >= 3 && 0 != ::close(thread_io_fd))
-        tryLogCurrentException(__PRETTY_FUNCTION__);
+        emitErrorMsgWithFailedToCloseFile(thread_io);
     if (0 != ::close(thread_stat_fd))
-        tryLogCurrentException(__PRETTY_FUNCTION__);
+        emitErrorMsgWithFailedToCloseFile(thread_stat);
     if (0 != ::close(thread_schedstat_fd))
-        tryLogCurrentException(__PRETTY_FUNCTION__);
+        emitErrorMsgWithFailedToCloseFile(thread_schedstat);
 }