Merge branch 'master' into revert-65361-revert-65164-ecs

.github/PULL_REQUEST_TEMPLATE.md

@@ -48,9 +48,7 @@ At a minimum, the following information should be added (but add more as needed)
 - [ ] <!---ci_include_stateful--> Allow: Stateful tests
 - [ ] <!---ci_include_integration--> Allow: Integration Tests
 - [ ] <!---ci_include_performance--> Allow: Performance tests
-- [ ] <!---ci_set_normal_builds--> Allow: Normal Builds
-- [ ] <!---ci_set_special_builds--> Allow: Special Builds
-- [ ] <!---ci_set_non_required--> Allow: All NOT Required Checks
+- [ ] <!---ci_set_builds--> Allow: All Builds
 - [ ] <!---batch_0_1--> Allow: batch 1, 2 for multi-batch jobs
 - [ ] <!---batch_2_3--> Allow: batch 3, 4, 5, 6 for multi-batch jobs
 ---
@@ -61,6 +59,7 @@ At a minimum, the following information should be added (but add more as needed)
 - [ ] <!---ci_exclude_aarch64|release|debug--> Exclude: All with aarch64, release, debug
 ---
 - [ ] <!---do_not_test--> Do not test
+- [ ] <!---woolen_wolfdog--> Woolen Wolfdog
 - [ ] <!---upload_all--> Upload binaries for special builds
 - [ ] <!---no_merge_commit--> Disable merge-commit
 - [ ] <!---no_ci_cache--> Disable CI cache

.github/workflows/backport_branches.yml

@@ -70,7 +70,7 @@ jobs:
     if: ${{ !failure() && !cancelled() }}
     uses: ./.github/workflows/reusable_test.yml
     with:
-      test_name: Compatibility check (amd64)
+      test_name: Compatibility check (release)
       runner_type: style-checker
       data: ${{ needs.RunConfig.outputs.data }}
   CompatibilityCheckAarch64:
@@ -194,7 +194,7 @@ jobs:
     if: ${{ !failure() && !cancelled() }}
     uses: ./.github/workflows/reusable_test.yml
     with:
-      test_name: Install packages (amd64)
+      test_name: Install packages (release)
       runner_type: style-checker
       data: ${{ needs.RunConfig.outputs.data }}
       run_command: |
@@ -204,7 +204,7 @@ jobs:
     if: ${{ !failure() && !cancelled() }}
     uses: ./.github/workflows/reusable_test.yml
     with:
-      test_name: Install packages (arm64)
+      test_name: Install packages (aarch64)
       runner_type: style-checker-aarch64
       data: ${{ needs.RunConfig.outputs.data }}
       run_command: |

.github/workflows/master.yml

@@ -104,10 +104,9 @@ jobs:
     with:
       stage: Tests_2
       data: ${{ needs.RunConfig.outputs.data }}
-  # stage for jobs that do not prohibit merge
   Tests_3:
     # Test_3 should not wait for Test_1/Test_2 and should not be blocked by them on master branch since all jobs need to run there.
-    needs: [RunConfig, Builds_1, Builds_2]
+    needs: [RunConfig, Builds_1]
     if: ${{ !failure() && !cancelled() && contains(fromJson(needs.RunConfig.outputs.data).stages_data.stages_to_do, 'Tests_3') }}
     uses: ./.github/workflows/reusable_test_stage.yml
     with:
@@ -115,25 +114,16 @@ jobs:
       data: ${{ needs.RunConfig.outputs.data }}
 
   ################################# Reports #################################
-  # Reports should be run even if Builds_1/2 failed - put them separately in wf (not in Tests_1/2)
-  Builds_1_Report:
+  # Reports should run even if Builds_1/2 fail - run them separately, not in Tests_1/2/3
+  Builds_Report:
     # run report check for failed builds to indicate the CI error
     if: ${{ !cancelled() && needs.RunConfig.result == 'success' && contains(fromJson(needs.RunConfig.outputs.data).jobs_data.jobs_to_do, 'ClickHouse build check') }}
-    needs: [RunConfig, Builds_1]
+    needs: [RunConfig, Builds_1, Builds_2]
     uses: ./.github/workflows/reusable_test.yml
     with:
       test_name: ClickHouse build check
       runner_type: style-checker-aarch64
       data: ${{ needs.RunConfig.outputs.data }}
-  Builds_2_Report:
-    # run report check for failed builds to indicate the CI error
-    if: ${{ !cancelled() && needs.RunConfig.result == 'success' && contains(fromJson(needs.RunConfig.outputs.data).jobs_data.jobs_to_do, 'ClickHouse special build check') }}
-    needs: [RunConfig, Builds_2]
-    uses: ./.github/workflows/reusable_test.yml
-    with:
-      test_name: ClickHouse special build check
-      runner_type: style-checker-aarch64
-      data: ${{ needs.RunConfig.outputs.data }}
 
   MarkReleaseReady:
     if: ${{ !failure() && !cancelled() }}
@@ -165,7 +155,7 @@ jobs:
 
   FinishCheck:
     if: ${{ !cancelled() }}
-    needs: [RunConfig, Builds_1, Builds_2, Builds_1_Report, Builds_2_Report, Tests_1, Tests_2, Tests_3]
+    needs: [RunConfig, Builds_1, Builds_2, Builds_Report, Tests_1, Tests_2, Tests_3]
     runs-on: [self-hosted, style-checker-aarch64]
     steps:
       - name: Check out repository code

.github/workflows/pull_request.yml

@@ -126,16 +126,16 @@ jobs:
     with:
       stage: Builds_2
       data: ${{ needs.RunConfig.outputs.data }}
+  # stage for running non-required checks without being blocked by required checks (Test_1) if corresponding settings is selected
   Tests_2:
-    needs: [RunConfig, Builds_2]
+    needs: [RunConfig, Builds_1]
     if: ${{ !failure() && !cancelled() && contains(fromJson(needs.RunConfig.outputs.data).stages_data.stages_to_do, 'Tests_2') }}
     uses: ./.github/workflows/reusable_test_stage.yml
     with:
       stage: Tests_2
       data: ${{ needs.RunConfig.outputs.data }}
-  # stage for jobs that do not prohibit merge
   Tests_3:
-    needs: [RunConfig, Builds_1, Tests_1, Builds_2, Tests_2]
+    needs: [RunConfig, Builds_1, Tests_1]
     if: ${{ !failure() && !cancelled() && contains(fromJson(needs.RunConfig.outputs.data).stages_data.stages_to_do, 'Tests_3') }}
     uses: ./.github/workflows/reusable_test_stage.yml
     with:
@@ -143,29 +143,21 @@ jobs:
       data: ${{ needs.RunConfig.outputs.data }}
 
   ################################# Reports #################################
-  # Reports should by run even if Builds_1/2 fail, so put them separately in wf (not in Tests_1/2)
-  Builds_1_Report:
+  # Reports should run even if Builds_1/2 fail - run them separately (not in Tests_1/2/3)
+  Builds_Report:
     # run report check for failed builds to indicate the CI error
-    if: ${{ !cancelled() && needs.StyleCheck.result == 'success' && contains(fromJson(needs.RunConfig.outputs.data).jobs_data.jobs_to_do, 'ClickHouse build check') }}
-    needs: [RunConfig, StyleCheck, Builds_1]
+    if: ${{ !cancelled() && needs.RunConfig.result == 'success' && contains(fromJson(needs.RunConfig.outputs.data).jobs_data.jobs_to_do, 'ClickHouse build check') }}
+    needs: [RunConfig, StyleCheck, Builds_1, Builds_2]
     uses: ./.github/workflows/reusable_test.yml
     with:
       test_name: ClickHouse build check
       runner_type: style-checker-aarch64
       data: ${{ needs.RunConfig.outputs.data }}
-  Builds_2_Report:
-    # run report check for failed builds to indicate the CI error
-    if: ${{ !cancelled() && needs.StyleCheck.result == 'success' && contains(fromJson(needs.RunConfig.outputs.data).jobs_data.jobs_to_do, 'ClickHouse special build check') }}
-    needs: [RunConfig, StyleCheck, Builds_2]
-    uses: ./.github/workflows/reusable_test.yml
-    with:
-      test_name: ClickHouse special build check
-      runner_type: style-checker-aarch64
-      data: ${{ needs.RunConfig.outputs.data }}
 
   CheckReadyForMerge:
     if: ${{ !cancelled() && needs.StyleCheck.result == 'success' }}
-    needs: [RunConfig, BuildDockers, StyleCheck, FastTest, Builds_1, Builds_2, Builds_1_Report, Builds_2_Report, Tests_1, Tests_2]
+    # Test_2 or Test_3 must not have jobs required for Mergeable check
+    needs: [RunConfig, BuildDockers, StyleCheck, FastTest, Builds_1, Builds_2, Builds_Report, Tests_1]
     runs-on: [self-hosted, style-checker-aarch64]
     steps:
       - name: Check out repository code
@@ -181,7 +173,7 @@ jobs:
   #
   FinishCheck:
     if: ${{ !cancelled() }}
-    needs: [RunConfig, BuildDockers, StyleCheck, FastTest, Builds_1, Builds_2, Builds_1_Report, Builds_2_Report, Tests_1, Tests_2, Tests_3]
+    needs: [RunConfig, BuildDockers, StyleCheck, FastTest, Builds_1, Builds_2, Builds_Report, Tests_1, Tests_2, Tests_3]
     runs-on: [self-hosted, style-checker-aarch64]
     steps:
       - name: Check out repository code
@@ -204,8 +196,7 @@ jobs:
     concurrency:
       group: jepsen
     if: ${{ !failure() && !cancelled() && contains(fromJson(needs.RunConfig.outputs.data).jobs_data.jobs_to_do, 'ClickHouse Keeper Jepsen') }}
-    # jepsen needs binary_release build which is in Builds_2
-    needs: [RunConfig, Builds_2]
+    needs: [RunConfig, Builds_1]
     uses: ./.github/workflows/reusable_test.yml
     with:
       test_name: ClickHouse Keeper Jepsen

.github/workflows/release_branches.yml

@@ -65,7 +65,7 @@ jobs:
     if: ${{ !failure() && !cancelled() }}
     uses: ./.github/workflows/reusable_test.yml
     with:
-      test_name: Compatibility check (amd64)
+      test_name: Compatibility check (release)
       runner_type: style-checker
       data: ${{ needs.RunConfig.outputs.data }}
   CompatibilityCheckAarch64:
@@ -244,7 +244,7 @@ jobs:
     if: ${{ !failure() && !cancelled() }}
     uses: ./.github/workflows/reusable_test.yml
     with:
-      test_name: Install packages (amd64)
+      test_name: Install packages (release)
       runner_type: style-checker
       data: ${{ needs.RunConfig.outputs.data }}
       run_command: |
@@ -254,7 +254,7 @@ jobs:
     if: ${{ !failure() && !cancelled() }}
     uses: ./.github/workflows/reusable_test.yml
     with:
-      test_name: Install packages (arm64)
+      test_name: Install packages (aarch64)
       runner_type: style-checker-aarch64
       data: ${{ needs.RunConfig.outputs.data }}
       run_command: |

base/poco/Foundation/CMakeLists.txt

@@ -213,6 +213,7 @@ target_compile_definitions (_poco_foundation
 )
 
 target_include_directories (_poco_foundation SYSTEM PUBLIC "include")
+target_link_libraries (_poco_foundation PRIVATE clickhouse_common_io)
 
 target_link_libraries (_poco_foundation
     PRIVATE

base/poco/Foundation/include/Poco/ThreadPool.h

@@ -48,7 +48,13 @@ class Foundation_API ThreadPool
 /// from the pool.
 {
 public:
-    ThreadPool(int minCapacity = 2, int maxCapacity = 16, int idleTime = 60, int stackSize = POCO_THREAD_STACK_SIZE);
+    explicit ThreadPool(
+        int minCapacity = 2,
+        int maxCapacity = 16,
+        int idleTime = 60,
+        int stackSize = POCO_THREAD_STACK_SIZE,
+        size_t global_profiler_real_time_period_ns_ = 0,
+        size_t global_profiler_cpu_time_period_ns_ = 0);
     /// Creates a thread pool with minCapacity threads.
     /// If required, up to maxCapacity threads are created
     /// a NoThreadAvailableException exception is thrown.
@@ -56,8 +62,14 @@ public:
     /// and more than minCapacity threads are running, the thread
     /// is killed. Threads are created with given stack size.
 
-    ThreadPool(
-        const std::string & name, int minCapacity = 2, int maxCapacity = 16, int idleTime = 60, int stackSize = POCO_THREAD_STACK_SIZE);
+    explicit ThreadPool(
+        const std::string & name,
+        int minCapacity = 2,
+        int maxCapacity = 16,
+        int idleTime = 60,
+        int stackSize = POCO_THREAD_STACK_SIZE,
+        size_t global_profiler_real_time_period_ns_ = 0,
+        size_t global_profiler_cpu_time_period_ns_ = 0);
     /// Creates a thread pool with the given name and minCapacity threads.
     /// If required, up to maxCapacity threads are created
     /// a NoThreadAvailableException exception is thrown.
@@ -171,6 +183,8 @@ private:
     int _serial;
     int _age;
     int _stackSize;
+    size_t _globalProfilerRealTimePeriodNs;
+    size_t _globalProfilerCPUTimePeriodNs;
     ThreadVec _threads;
     mutable FastMutex _mutex;
 };

base/poco/Foundation/src/ThreadPool.cpp

@@ -20,6 +20,7 @@
 #include "Poco/ErrorHandler.h"
 #include <sstream>
 #include <ctime>
+#include <Common/ThreadPool.h>
 
 
 namespace Poco {
@@ -28,7 +29,11 @@ namespace Poco {
 class PooledThread: public Runnable
 {
 public:
-	PooledThread(const std::string& name, int stackSize = POCO_THREAD_STACK_SIZE);
+	explicit PooledThread(
+		const std::string& name,
+		int stackSize = POCO_THREAD_STACK_SIZE,
+        size_t globalProfilerRealTimePeriodNs_ = 0,
+        size_t globalProfilerCPUTimePeriodNs_ = 0);
 	~PooledThread();
 
 	void start();
@@ -51,16 +56,24 @@ private:
 	Event                _targetCompleted;
 	Event                _started;
 	FastMutex            _mutex;
+	size_t _globalProfilerRealTimePeriodNs;
+	size_t _globalProfilerCPUTimePeriodNs;
 };
 
 
-PooledThread::PooledThread(const std::string& name, int stackSize): 
-	_idle(true), 
-	_idleTime(0), 
-	_pTarget(0), 
-	_name(name), 
+PooledThread::PooledThread(
+	const std::string& name,
+	int stackSize,
+	size_t globalProfilerRealTimePeriodNs_,
+	size_t globalProfilerCPUTimePeriodNs_) :
+	_idle(true),
+	_idleTime(0),
+	_pTarget(0),
+	_name(name),
 	_thread(name),
-	_targetCompleted(false)
+	_targetCompleted(false),
+	_globalProfilerRealTimePeriodNs(globalProfilerRealTimePeriodNs_),
+	_globalProfilerCPUTimePeriodNs(globalProfilerCPUTimePeriodNs_)
 {
 	poco_assert_dbg (stackSize >= 0);
 	_thread.setStackSize(stackSize);
@@ -83,7 +96,7 @@ void PooledThread::start()
 void PooledThread::start(Thread::Priority priority, Runnable& target)
 {
 	FastMutex::ScopedLock lock(_mutex);
-	
+
 	poco_assert (_pTarget == 0);
 
 	_pTarget = &target;
@@ -109,7 +122,7 @@ void PooledThread::start(Thread::Priority priority, Runnable& target, const std:
 	}
 	_thread.setName(fullName);
 	_thread.setPriority(priority);
-	
+
 	poco_assert (_pTarget == 0);
 
 	_pTarget = &target;
@@ -145,7 +158,7 @@ void PooledThread::join()
 void PooledThread::activate()
 {
 	FastMutex::ScopedLock lock(_mutex);
-	
+
 	poco_assert (_idle);
 	_idle = false;
 	_targetCompleted.reset();
@@ -155,7 +168,7 @@ void PooledThread::activate()
 void PooledThread::release()
 {
 	const long JOIN_TIMEOUT = 10000;
-	
+
 	_mutex.lock();
 	_pTarget = 0;
 	_mutex.unlock();
@@ -174,6 +187,10 @@ void PooledThread::release()
 
 void PooledThread::run()
 {
+    DB::ThreadStatus thread_status;
+	if (unlikely(_globalProfilerRealTimePeriodNs != 0 || _globalProfilerCPUTimePeriodNs != 0))
+		thread_status.initGlobalProfiler(_globalProfilerRealTimePeriodNs, _globalProfilerCPUTimePeriodNs);
+
 	_started.set();
 	for (;;)
 	{
@@ -220,13 +237,17 @@ void PooledThread::run()
 ThreadPool::ThreadPool(int minCapacity,
 	int maxCapacity,
 	int idleTime,
-	int stackSize): 
-	_minCapacity(minCapacity), 
-	_maxCapacity(maxCapacity), 
+	int stackSize,
+	size_t globalProfilerRealTimePeriodNs_,
+	size_t globalProfilerCPUTimePeriodNs_) :
+	_minCapacity(minCapacity),
+	_maxCapacity(maxCapacity),
 	_idleTime(idleTime),
 	_serial(0),
 	_age(0),
-	_stackSize(stackSize)
+	_stackSize(stackSize),
+    _globalProfilerRealTimePeriodNs(globalProfilerRealTimePeriodNs_),
+    _globalProfilerCPUTimePeriodNs(globalProfilerCPUTimePeriodNs_)
 {
 	poco_assert (minCapacity >= 1 && maxCapacity >= minCapacity && idleTime > 0);
 
@@ -243,14 +264,18 @@ ThreadPool::ThreadPool(const std::string& name,
 	int minCapacity,
 	int maxCapacity,
 	int idleTime,
-	int stackSize):
+	int stackSize,
+	size_t globalProfilerRealTimePeriodNs_,
+	size_t globalProfilerCPUTimePeriodNs_) :
 	_name(name),
-	_minCapacity(minCapacity), 
-	_maxCapacity(maxCapacity), 
+	_minCapacity(minCapacity),
+	_maxCapacity(maxCapacity),
 	_idleTime(idleTime),
 	_serial(0),
 	_age(0),
-	_stackSize(stackSize)
+	_stackSize(stackSize),
+    _globalProfilerRealTimePeriodNs(globalProfilerRealTimePeriodNs_),
+    _globalProfilerCPUTimePeriodNs(globalProfilerCPUTimePeriodNs_)
 {
 	poco_assert (minCapacity >= 1 && maxCapacity >= minCapacity && idleTime > 0);
 
@@ -393,15 +418,15 @@ void ThreadPool::housekeep()
 	ThreadVec activeThreads;
 	idleThreads.reserve(_threads.size());
 	activeThreads.reserve(_threads.size());
-	
+
 	for (ThreadVec::iterator it = _threads.begin(); it != _threads.end(); ++it)
 	{
 		if ((*it)->idle())
 		{
 			if ((*it)->idleTime() < _idleTime)
 				idleThreads.push_back(*it);
-			else 
-				expiredThreads.push_back(*it);	
+			else
+				expiredThreads.push_back(*it);
 		}
 		else activeThreads.push_back(*it);
 	}
@@ -463,7 +488,7 @@ PooledThread* ThreadPool::createThread()
 {
 	std::ostringstream name;
 	name << _name << "[#" << ++_serial << "]";
-	return new PooledThread(name.str(), _stackSize);
+	return new PooledThread(name.str(), _stackSize, _globalProfilerRealTimePeriodNs, _globalProfilerCPUTimePeriodNs);
 }
 
 
@@ -481,7 +506,7 @@ public:
 	ThreadPool* pool()
 	{
 		FastMutex::ScopedLock lock(_mutex);
-		
+
 		if (!_pPool)
 		{
 			_pPool = new ThreadPool("default");
@@ -490,7 +515,7 @@ public:
 		}
 		return _pPool;
 	}
-	
+
 private:
 	ThreadPool* _pPool;
 	FastMutex   _mutex;

base/poco/Net/include/Poco/Net/HTTPBasicStreamBuf.h

@@ -26,7 +26,7 @@ namespace Poco
 {
 namespace Net
 {
-    constexpr size_t HTTP_DEFAULT_BUFFER_SIZE = 8 * 1024;
+    constexpr size_t HTTP_DEFAULT_BUFFER_SIZE = 1024 * 1024;
 
     typedef Poco::BasicBufferedStreamBuf<char, std::char_traits<char>> HTTPBasicStreamBuf;
 

base/poco/NetSSL_OpenSSL/src/SSLManager.cpp

@@ -330,27 +330,26 @@ void SSLManager::initDefaultContext(bool server)
 	else
 		_ptrDefaultClientContext->disableProtocols(disabledProtocols);
 
-    /// Temporarily disabled during the transition from boringssl to OpenSSL due to tsan issues.
-	/// bool cacheSessions = config.getBool(prefix + CFG_CACHE_SESSIONS, false);
-	/// if (server)
-	/// {
-	/// 	std::string sessionIdContext = config.getString(prefix + CFG_SESSION_ID_CONTEXT, config.getString("application.name", ""));
-	/// 	_ptrDefaultServerContext->enableSessionCache(cacheSessions, sessionIdContext);
-	/// 	if (config.hasProperty(prefix + CFG_SESSION_CACHE_SIZE))
-	/// 	{
-	/// 		int cacheSize = config.getInt(prefix + CFG_SESSION_CACHE_SIZE);
-	/// 		_ptrDefaultServerContext->setSessionCacheSize(cacheSize);
-	/// 	}
-	/// 	if (config.hasProperty(prefix + CFG_SESSION_TIMEOUT))
-	/// 	{
-	/// 		int timeout = config.getInt(prefix + CFG_SESSION_TIMEOUT);
-	/// 		_ptrDefaultServerContext->setSessionTimeout(timeout);
-	/// 	}
-	/// }
-	/// else
-	/// {
-	/// 	_ptrDefaultClientContext->enableSessionCache(cacheSessions);
-	/// }
+	bool cacheSessions = config.getBool(prefix + CFG_CACHE_SESSIONS, false);
+	if (server)
+	{
+		std::string sessionIdContext = config.getString(prefix + CFG_SESSION_ID_CONTEXT, config.getString("application.name", ""));
+		_ptrDefaultServerContext->enableSessionCache(cacheSessions, sessionIdContext);
+		if (config.hasProperty(prefix + CFG_SESSION_CACHE_SIZE))
+		{
+			int cacheSize = config.getInt(prefix + CFG_SESSION_CACHE_SIZE);
+			_ptrDefaultServerContext->setSessionCacheSize(cacheSize);
+		}
+		if (config.hasProperty(prefix + CFG_SESSION_TIMEOUT))
+		{
+			int timeout = config.getInt(prefix + CFG_SESSION_TIMEOUT);
+			_ptrDefaultServerContext->setSessionTimeout(timeout);
+		}
+	}
+	else
+	{
+		_ptrDefaultClientContext->enableSessionCache(cacheSessions);
+	}
 	bool extendedVerification = config.getBool(prefix + CFG_EXTENDED_VERIFICATION, false);
 	if (server)
 		_ptrDefaultServerContext->enableExtendedCertificateVerification(extendedVerification);

contrib/openssl

@@ -1 +1 @@
-Subproject commit 67c0b63e578e4c751ac9edf490f5a96124fff8dc
+Subproject commit e0d6ae2bf93cf6dc26bb86aa39992bc6a410869a

docker/test/stateless/run.sh

@@ -254,7 +254,7 @@ function run_tests()
 
     set +e
     clickhouse-test --testname --shard --zookeeper --check-zookeeper-session --hung-check --print-time \
-        --test-runs "$NUM_TRIES" "${ADDITIONAL_OPTIONS[@]}" 2>&1 \
+         --no-drop-if-fail --test-runs "$NUM_TRIES" "${ADDITIONAL_OPTIONS[@]}" 2>&1 \
     | ts '%Y-%m-%d %H:%M:%S' \
     | tee -a test_output/test_result.txt
     set -e
@@ -379,6 +379,10 @@ fi
 
 tar -chf /test_output/coordination.tar /var/lib/clickhouse/coordination ||:
 
+rm -rf /var/lib/clickhouse/data/system/*/
+tar -chf /test_output/store.tar /var/lib/clickhouse/store ||:
+tar -chf /test_output/metadata.tar /var/lib/clickhouse/metadata/*.sql ||:
+
 if [[ -n "$USE_DATABASE_REPLICATED" ]] && [[ "$USE_DATABASE_REPLICATED" -eq 1 ]]; then
     rg -Fa "<Fatal>" /var/log/clickhouse-server/clickhouse-server1.log ||:
     rg -Fa "<Fatal>" /var/log/clickhouse-server/clickhouse-server2.log ||:

docs/en/engines/table-engines/integrations/iceberg.md

@@ -37,7 +37,7 @@ Using named collections:
     <named_collections>
         <iceberg_conf>
             <url>http://test.s3.amazonaws.com/clickhouse-bucket/</url>
-            <access_key_id>test<access_key_id>
+            <access_key_id>test</access_key_id>
             <secret_access_key>test</secret_access_key>
         </iceberg_conf>
     </named_collections>

docs/en/engines/table-engines/integrations/s3queue.md

@@ -13,7 +13,7 @@ This engine provides integration with [Amazon S3](https://aws.amazon.com/s3/) ec
 CREATE TABLE s3_queue_engine_table (name String, value UInt32)
     ENGINE = S3Queue(path, [NOSIGN, | aws_access_key_id, aws_secret_access_key,] format, [compression])
     [SETTINGS]
-    [mode = 'unordered',]
+    [mode = '',]
     [after_processing = 'keep',]
     [keeper_path = '',]
     [s3queue_loading_retries = 0,]

docs/en/operations/allocation-profiling.md

@@ -59,10 +59,10 @@ For that, we need to use `jemalloc`'s tool called [jeprof](https://github.com/je
 If that’s the case, we recommend installing an [alternative implementation](https://github.com/gimli-rs/addr2line) of the tool.
 
 ```
-git clone https://github.com/gimli-rs/addr2line
+git clone https://github.com/gimli-rs/addr2line.git --depth=1 --branch=0.23.0
 cd addr2line
-cargo b --examples -r
-cp ./target/release/examples/addr2line path/to/current/addr2line
+cargo build --features bin --release
+cp ./target/release/addr2line path/to/current/addr2line
 ```
 :::
 

docs/en/operations/server-configuration-parameters/settings.md

@@ -591,6 +591,22 @@ Default value: 100000
 <max_part_num_to_warn>400</max_part_num_to_warn>
 ```
 
+## max\_table\_num\_to\_throw {#max-table-num-to-throw}
+If number of tables is greater than this value, server will throw an exception. 0 means no limitation. View, remote tables, dictionary, system tables are not counted. Only count table in Atomic/Ordinary/Replicated/Lazy database engine.Default value: 0
+
+**Example**
+```xml
+<max_table_num_to_throw>400</max_table_num_to_throw>
+```
+
+## max\_database\_num\_to\_throw {#max-table-num-to-throw}
+If number of _database is greater than this value, server will throw an exception. 0 means no limitation.
+Default value: 0
+
+**Example**
+```xml
+<max_database_num_to_throw>400</max_database_num_to_throw>
+```
 
 ## max_temporary_data_on_disk_size
 
@@ -3084,3 +3100,21 @@ This setting is only necessary for the migration period and will become obsolete
 Type: Bool
 
 Default: 1
+
+## merge_workload {#merge_workload}
+
+Used to regulate how resources are utilized and shared between merges and other workloads. Specified value is used as `workload` setting value for all background merges. Can be overridden by a merge tree setting.
+
+Default value: "default"
+
+**See Also**
+- [Workload Scheduling](/docs/en/operations/workload-scheduling.md)
+
+## mutation_workload {#mutation_workload}
+
+Used to regulate how resources are utilized and shared between mutations and other workloads. Specified value is used as `workload` setting value for all background mutations. Can be overridden by a merge tree setting.
+
+Default value: "default"
+
+**See Also**
+- [Workload Scheduling](/docs/en/operations/workload-scheduling.md)

docs/en/operations/settings/merge-tree-settings.md

@@ -974,6 +974,24 @@ Default value: false
 
 - [exclude_deleted_rows_for_part_size_in_merge](#exclude_deleted_rows_for_part_size_in_merge) setting
 
+## merge_workload
+
+Used to regulate how resources are utilized and shared between merges and other workloads. Specified value is used as `workload` setting value for background merges of this table. If not specified (empty string), then server setting `merge_workload` is used instead.
+
+Default value: an empty string
+
+**See Also**
+- [Workload Scheduling](/docs/en/operations/workload-scheduling.md)
+
+## mutation_workload
+
+Used to regulate how resources are utilized and shared between mutations and other workloads. Specified value is used as `workload` setting value for background mutations of this table. If not specified (empty string), then server setting `mutation_workload` is used instead.
+
+Default value: an empty string
+
+**See Also**
+- [Workload Scheduling](/docs/en/operations/workload-scheduling.md)
+
 ### optimize_row_order
 
 Controls if the row order should be optimized during inserts to improve the compressability of the newly inserted table part.

docs/en/operations/settings/settings.md

@@ -1592,19 +1592,19 @@ Default value: `default`.
 
 ## parallel_replicas_custom_key_range_lower {#parallel_replicas_custom_key_range_lower}
 
-Allows the filter type `range` to split the work evenly between replicas based on the custom range `[parallel_replicas_custom_key_range_lower, INT_MAX]`. 
+Allows the filter type `range` to split the work evenly between replicas based on the custom range `[parallel_replicas_custom_key_range_lower, INT_MAX]`.
 
-When used in conjuction with [parallel_replicas_custom_key_range_upper](#parallel_replicas_custom_key_range_upper), it lets the filter evenly split the work over replicas for the range `[parallel_replicas_custom_key_range_lower, parallel_replicas_custom_key_range_upper]`. 
+When used in conjuction with [parallel_replicas_custom_key_range_upper](#parallel_replicas_custom_key_range_upper), it lets the filter evenly split the work over replicas for the range `[parallel_replicas_custom_key_range_lower, parallel_replicas_custom_key_range_upper]`.
 
-Note: This setting will not cause any additional data to be filtered during query processing, rather it changes the points at which the range filter breaks up the range `[0, INT_MAX]` for parallel processing. 
+Note: This setting will not cause any additional data to be filtered during query processing, rather it changes the points at which the range filter breaks up the range `[0, INT_MAX]` for parallel processing.
 
 ## parallel_replicas_custom_key_range_upper {#parallel_replicas_custom_key_range_upper}
 
 Allows the filter type `range` to split the work evenly between replicas based on the custom range `[0, parallel_replicas_custom_key_range_upper]`. A value of 0 disables the upper bound, setting it the max value of the custom key expression.
 
-When used in conjuction with [parallel_replicas_custom_key_range_lower](#parallel_replicas_custom_key_range_lower), it lets the filter evenly split the work over replicas for the range `[parallel_replicas_custom_key_range_lower, parallel_replicas_custom_key_range_upper]`. 
+When used in conjuction with [parallel_replicas_custom_key_range_lower](#parallel_replicas_custom_key_range_lower), it lets the filter evenly split the work over replicas for the range `[parallel_replicas_custom_key_range_lower, parallel_replicas_custom_key_range_upper]`.
 
-Note: This setting will not cause any additional data to be filtered during query processing, rather it changes the points at which the range filter breaks up the range `[0, INT_MAX]` for parallel processing. 
+Note: This setting will not cause any additional data to be filtered during query processing, rather it changes the points at which the range filter breaks up the range `[0, INT_MAX]` for parallel processing.
 
 ## allow_experimental_parallel_reading_from_replicas
 
@@ -3188,7 +3188,7 @@ Default value: `0`.
 
 ## lightweight_deletes_sync {#lightweight_deletes_sync}
 
-The same as 'mutation_sync', but controls only execution of lightweight deletes. 
+The same as 'mutation_sync', but controls only execution of lightweight deletes.
 
 Possible values:
 
@@ -5150,7 +5150,7 @@ Allows using statistic to optimize the order of [prewhere conditions](../../sql-
 
 ## analyze_index_with_space_filling_curves
 
-If a table has a space-filling curve in its index, e.g. `ORDER BY mortonEncode(x, y)`, and the query has conditions on its arguments, e.g. `x >= 10 AND x <= 20 AND y >= 20 AND y <= 30`, use the space-filling curve for index analysis.
+If a table has a space-filling curve in its index, e.g. `ORDER BY mortonEncode(x, y)` or `ORDER BY hilbertEncode(x, y)`, and the query has conditions on its arguments, e.g. `x >= 10 AND x <= 20 AND y >= 20 AND y <= 30`, use the space-filling curve for index analysis.
 
 ## query_plan_enable_optimizations {#query_plan_enable_optimizations}
 
@@ -5418,6 +5418,12 @@ When set to `false` than all attempts are made with identical timeouts.
 
 Default value: `true`.
 
+## uniform_snowflake_conversion_functions {#uniform_snowflake_conversion_functions}
+
+If set to `true`, then functions `snowflakeIDToDateTime`, `snowflakeIDToDateTime64`, `dateTimeToSnowflakeID`, and `dateTime64ToSnowflakeID` are enabled, and functions `snowflakeToDateTime`, `snowflakeToDateTime64`, `dateTimeToSnowflake`, and `dateTime64ToSnowflake` are disabled (and vice versa if set to `false`).
+
+Default value: `true`
+
 ## allow_experimental_variant_type {#allow_experimental_variant_type}
 
 Allows creation of experimental [Variant](../../sql-reference/data-types/variant.md).

docs/en/operations/system-tables/query_log.md

@@ -113,6 +113,8 @@ Columns:
 - `used_functions` ([Array(String)](../../sql-reference/data-types/array.md)) — Canonical names of `functions`, which were used during query execution.
 - `used_storages` ([Array(String)](../../sql-reference/data-types/array.md)) — Canonical names of `storages`, which were used during query execution.
 - `used_table_functions` ([Array(String)](../../sql-reference/data-types/array.md)) — Canonical names of `table functions`, which were used during query execution.
+- `used_privileges` ([Array(String)](../../sql-reference/data-types/array.md)) - Privileges which were successfully checked during query execution.
+- `missing_privileges` ([Array(String)](../../sql-reference/data-types/array.md)) - Privileges that are missing during query execution.
 - `query_cache_usage` ([Enum8](../../sql-reference/data-types/enum.md)) — Usage of the [query cache](../query-cache.md) during query execution. Values:
     - `'Unknown'` = Status unknown.
     - `'None'` = The query result was neither written into nor read from the query cache.
@@ -194,6 +196,8 @@ used_formats:                          []
 used_functions:                        []
 used_storages:                         []
 used_table_functions:                  []
+used_privileges:                       []
+missing_privileges:                    []
 query_cache_usage:                     None
 ```
 

docs/en/operations/workload-scheduling.md

@@ -47,6 +47,8 @@ Example:
 
 Queries can be marked with setting `workload` to distinguish different workloads. If `workload` is not set, than value "default" is used. Note that you are able to specify the other value using settings profiles. Setting constraints can be used to make `workload` constant if you want all queries from the user to be marked with fixed value of `workload` setting.
 
+It is possible to assign a `workload` setting for background activities. Merges and mutations are using `merge_workload` and `mutation_workload` server settings correspondingly. These values can also be overridden for specific tables using `merge_workload` and `mutation_workload` merge tree settings
+
 Let's consider an example of a system with two different workloads: "production" and "development".
 
 ```sql
@@ -151,6 +153,9 @@ Example:
 </clickhouse>
 ```
 
-
 ## See also
  - [system.scheduler](/docs/en/operations/system-tables/scheduler.md)
+ - [merge_workload](/docs/en/operations/settings/merge-tree-settings.md#merge_workload) merge tree setting
+ - [merge_workload](/docs/en/operations/server-configuration-parameters/settings.md#merge_workload) global server setting
+ - [mutation_workload](/docs/en/operations/settings/merge-tree-settings.md#mutation_workload) merge tree setting
+ - [mutation_workload](/docs/en/operations/server-configuration-parameters/settings.md#mutation_workload) global server setting

docs/en/sql-reference/aggregate-functions/reference/groupconcat.md

@@ -0,0 +1,90 @@
+---
+slug: /en/sql-reference/aggregate-functions/reference/groupconcat
+sidebar_position: 363
+sidebar_label: groupConcat
+title: groupConcat
+---
+
+Calculates a concatenated string from a group of strings, optionally separated by a delimiter, and optionally limited by a maximum number of elements.
+
+**Syntax**
+
+``` sql
+groupConcat(expression [, delimiter] [, limit]);
+```
+
+**Arguments**
+
+- `expression` — The expression or column name that outputs strings to be concatenated..
+- `delimiter` — A [string](../../../sql-reference/data-types/string.md) that will be used to separate concatenated values. This parameter is optional and defaults to an empty string if not specified.
+- `limit` — A positive [integer](../../../sql-reference/data-types/int-uint.md) specifying the maximum number of elements to concatenate. If more elements are present, excess elements are ignored. This parameter is optional.
+
+:::note
+If delimiter is specified without limit, it must be the first parameter following the expression. If both delimiter and limit are specified, delimiter must precede limit.
+:::
+
+**Returned value**
+
+- Returns a [string](../../../sql-reference/data-types/string.md) consisting of the concatenated values of the column or expression. If the group has no elements or only null elements, and the function does not specify a handling for only null values, the result is a nullable string with a null value.
+
+**Examples**
+
+Input table:
+
+``` text
+┌─id─┬─name─┐
+│ 1  │  John│
+│ 2  │  Jane│
+│ 3  │   Bob│
+└────┴──────┘
+```
+
+1.	Basic usage without a delimiter:
+
+Query:
+
+``` sql
+SELECT groupConcat(Name) FROM Employees;
+```
+
+Result:
+
+``` text
+JohnJaneBob
+```
+
+This concatenates all names into one continuous string without any separator.
+
+
+2. Using comma as a delimiter:
+
+Query:
+
+``` sql
+SELECT groupConcat(Name, ', ', 2) FROM Employees;
+```
+
+Result:
+
+``` text
+John, Jane, Bob
+```
+
+This output shows the names separated by a comma followed by a space.
+
+
+3. Limiting the number of concatenated elements
+
+Query:
+
+``` sql
+SELECT groupConcat(Name, ', ', 2) FROM Employees;
+```
+
+Result:
+
+``` text
+John, Jane
+```
+
+This query limits the output to the first two names, even though there are more names in the table.

docs/en/sql-reference/data-types/datetime.md

@@ -137,7 +137,7 @@ If the time transition (due to daylight saving time or for other reasons) was pe
 
 Non-monotonic calendar dates. For example, in Happy Valley - Goose Bay, the time was transitioned one hour backwards at 00:01:00 7 Nov 2010 (one minute after midnight). So after 6th Nov has ended, people observed a whole one minute of 7th Nov, then time was changed back to 23:01 6th Nov and after another 59 minutes the 7th Nov started again. ClickHouse does not (yet) support this kind of fun. During these days the results of time processing functions may be slightly incorrect.
 
-Similar issue exists for Casey Antarctic station in year 2010. They changed time three hours back at 5 Mar, 02:00. If you are working in antarctic station, please don't afraid to use ClickHouse. Just make sure you set timezone to UTC or be aware of inaccuracies.
+Similar issue exists for Casey Antarctic station in year 2010. They changed time three hours back at 5 Mar, 02:00. If you are working in antarctic station, please don't be afraid to use ClickHouse. Just make sure you set timezone to UTC or be aware of inaccuracies.
 
 Time shifts for multiple days. Some pacific islands changed their timezone offset from UTC+14 to UTC-12. That's alright but some inaccuracies may present if you do calculations with their timezone for historical time points at the days of conversion.
 

docs/en/sql-reference/functions/string-functions.md

@@ -2178,6 +2178,32 @@ Result:
 
 Alias: levenshteinDistance
 
+## editDistanceUTF8
+
+Calculates the [edit distance](https://en.wikipedia.org/wiki/Edit_distance) between two UTF8 strings.
+
+**Syntax**
+
+```sql
+editDistanceUTF8(string1, string2)
+```
+
+**Examples**
+
+``` sql
+SELECT editDistanceUTF8('我是谁', '我是我');
+```
+
+Result:
+
+``` text
+┌─editDistanceUTF8('我是谁', '我是我')──┐
+│                                   1 │
+└─────────────────────────────────────┘
+```
+
+Alias: levenshteinDistanceUTF8
+
 ## damerauLevenshteinDistance
 
 Calculates the [Damerau-Levenshtein distance](https://en.wikipedia.org/wiki/Damerau%E2%80%93Levenshtein_distance) between two byte strings.

docs/en/sql-reference/functions/uuid-functions.md

@@ -543,12 +543,17 @@ serverUUID()
 
 Generates a [Snowflake ID](https://en.wikipedia.org/wiki/Snowflake_ID).
 
-The generated Snowflake ID contains the current Unix timestamp in milliseconds 41 (+ 1 top zero bit) bits, followed by machine id (10 bits), a counter (12 bits) to distinguish IDs within a millisecond.
+The generated Snowflake ID contains the current Unix timestamp in milliseconds (41 + 1 top zero bits), followed by a machine id (10 bits), and a counter (12 bits) to distinguish IDs within a millisecond.
 For any given timestamp (unix_ts_ms), the counter starts at 0 and is incremented by 1 for each new Snowflake ID until the timestamp changes.
 In case the counter overflows, the timestamp field is incremented by 1 and the counter is reset to 0.
 
 Function `generateSnowflakeID` guarantees that the counter field within a timestamp increments monotonically across all function invocations in concurrently running threads and queries.
 
+:::note
+The generated Snowflake IDs are based on the UNIX epoch 1970-01-01.
+While no standard or recommendation exists for the epoch of Snowflake IDs, implementations in other systems may use a different epoch, e.g. Twitter/X (2010-11-04) or Mastodon (2015-01-01).
+:::
+
 ```
  0                   1                   2                   3
  0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
@@ -605,6 +610,11 @@ SELECT generateSnowflakeID(1), generateSnowflakeID(2);
 
 ## snowflakeToDateTime
 
+:::warning
+This function is deprecated and can only be used if setting [uniform_snowflake_conversion_functions](../../operations/settings/settings.md#uniform_snowflake_conversion_functions) is disabled.
+The function will be removed at some point in future.
+:::
+
 Extracts the timestamp component of a [Snowflake ID](https://en.wikipedia.org/wiki/Snowflake_ID) in [DateTime](../data-types/datetime.md) format.
 
 **Syntax**
@@ -641,6 +651,11 @@ Result:
 
 ## snowflakeToDateTime64
 
+:::warning
+This function is deprecated and can only be used if setting [uniform_snowflake_conversion_functions](../../operations/settings/settings.md#uniform_snowflake_conversion_functions) is disabled.
+The function will be removed at some point in future.
+:::
+
 Extracts the timestamp component of a [Snowflake ID](https://en.wikipedia.org/wiki/Snowflake_ID) in [DateTime64](../data-types/datetime64.md) format.
 
 **Syntax**
@@ -677,6 +692,11 @@ Result:
 
 ## dateTimeToSnowflake
 
+:::warning
+This function is deprecated and can only be used if setting [uniform_snowflake_conversion_functions](../../operations/settings/settings.md#uniform_snowflake_conversion_functions) is disabled.
+The function will be removed at some point in future.
+:::
+
 Converts a [DateTime](../data-types/datetime.md) value to the first [Snowflake ID](https://en.wikipedia.org/wiki/Snowflake_ID) at the giving time.
 
 **Syntax**
@@ -711,6 +731,11 @@ Result:
 
 ## dateTime64ToSnowflake
 
+:::warning
+This function is deprecated and can only be used if setting [uniform_snowflake_conversion_functions](../../operations/settings/settings.md#uniform_snowflake_conversion_functions) is disabled.
+The function will be removed at some point in future.
+:::
+
 Convert a [DateTime64](../data-types/datetime64.md) to the first [Snowflake ID](https://en.wikipedia.org/wiki/Snowflake_ID) at the giving time.
 
 **Syntax**
@@ -743,6 +768,148 @@ Result:
 └─────────────────────────────┘
 ```
 
+## snowflakeIDToDateTime
+
+Returns the timestamp component of a [Snowflake ID](https://en.wikipedia.org/wiki/Snowflake_ID) as a value of type [DateTime](../data-types/datetime.md).
+
+**Syntax**
+
+``` sql
+snowflakeIDToDateTime(value[, epoch[, time_zone]])
+```
+
+**Arguments**
+
+- `value` — Snowflake ID. [UInt64](../data-types/int-uint.md).
+- `epoch` - Epoch of the Snowflake ID in milliseconds since 1970-01-01. Defaults to 0 (1970-01-01). For the Twitter/X epoch (2015-01-01), provide 1288834974657. Optional. [UInt*](../data-types/int-uint.md).
+- `time_zone` — [Timezone](/docs/en/operations/server-configuration-parameters/settings.md/#server_configuration_parameters-timezone). The function parses `time_string` according to the timezone. Optional. [String](../data-types/string.md).
+
+**Returned value**
+
+- The timestamp component of `value` as a [DateTime](../data-types/datetime.md) value.
+
+**Example**
+
+Query:
+
+```sql
+SELECT snowflakeIDToDateTime(7204436857747984384) AS res
+```
+
+Result:
+
+```
+┌─────────────────res─┐
+│ 2024-06-06 10:59:58 │
+└─────────────────────┘
+```
+
+## snowflakeIDToDateTime64
+
+Returns the timestamp component of a [Snowflake ID](https://en.wikipedia.org/wiki/Snowflake_ID) as a value of type [DateTime64](../data-types/datetime64.md).
+
+**Syntax**
+
+``` sql
+snowflakeIDToDateTime64(value[, epoch[, time_zone]])
+```
+
+**Arguments**
+
+- `value` — Snowflake ID. [UInt64](../data-types/int-uint.md).
+- `epoch` - Epoch of the Snowflake ID in milliseconds since 1970-01-01. Defaults to 0 (1970-01-01). For the Twitter/X epoch (2015-01-01), provide 1288834974657. Optional. [UInt*](../data-types/int-uint.md).
+- `time_zone` — [Timezone](/docs/en/operations/server-configuration-parameters/settings.md/#server_configuration_parameters-timezone). The function parses `time_string` according to the timezone. Optional. [String](../data-types/string.md).
+
+**Returned value**
+
+- The timestamp component of `value` as a [DateTime64](../data-types/datetime64.md) with scale = 3, i.e. millisecond precision.
+
+**Example**
+
+Query:
+
+```sql
+SELECT snowflakeIDToDateTime64(7204436857747984384) AS res
+```
+
+Result:
+
+```
+┌─────────────────res─┐
+│ 2024-06-06 10:59:58 │
+└─────────────────────┘
+```
+
+## dateTimeToSnowflakeID
+
+Converts a [DateTime](../data-types/datetime.md) value to the first [Snowflake ID](https://en.wikipedia.org/wiki/Snowflake_ID) at the giving time.
+
+**Syntax**
+
+``` sql
+dateTimeToSnowflakeID(value[, epoch])
+```
+
+**Arguments**
+
+- `value` — Date with time. [DateTime](../data-types/datetime.md).
+- `epoch` - Epoch of the Snowflake ID in milliseconds since 1970-01-01. Defaults to 0 (1970-01-01). For the Twitter/X epoch (2015-01-01), provide 1288834974657. Optional. [UInt*](../data-types/int-uint.md).
+
+**Returned value**
+
+- Input value converted to [UInt64](../data-types/int-uint.md) as the first Snowflake ID at that time.
+
+**Example**
+
+Query:
+
+```sql
+SELECT toDateTime('2021-08-15 18:57:56', 'Asia/Shanghai') AS dt, dateTimeToSnowflakeID(dt) AS res;
+```
+
+Result:
+
+```
+┌──────────────────dt─┬─────────────────res─┐
+│ 2021-08-15 18:57:56 │ 6832626392367104000 │
+└─────────────────────┴─────────────────────┘
+```
+
+## dateTime64ToSnowflakeID
+
+Convert a [DateTime64](../data-types/datetime64.md) to the first [Snowflake ID](https://en.wikipedia.org/wiki/Snowflake_ID) at the giving time.
+
+**Syntax**
+
+``` sql
+dateTime64ToSnowflakeID(value[, epoch])
+```
+
+**Arguments**
+
+- `value` — Date with time. [DateTime64](../data-types/datetime64.md).
+- `epoch` - Epoch of the Snowflake ID in milliseconds since 1970-01-01. Defaults to 0 (1970-01-01). For the Twitter/X epoch (2015-01-01), provide 1288834974657. Optional. [UInt*](../data-types/int-uint.md).
+
+**Returned value**
+
+- Input value converted to [UInt64](../data-types/int-uint.md) as the first Snowflake ID at that time.
+
+**Example**
+
+Query:
+
+```sql
+SELECT toDateTime('2021-08-15 18:57:56.493', 3, 'Asia/Shanghai') AS dt, dateTime64ToSnowflakeID(dt) AS res;
+```
+
+Result:
+
+```
+┌──────────────────────dt─┬─────────────────res─┐
+│ 2021-08-15 18:57:56.493 │ 6832626394434895872 │
+└─────────────────────────┴─────────────────────┘
+```
+
 ## See also
 
 - [dictGetUUID](../functions/ext-dict-functions.md#ext_dict_functions-other)

programs/server/Server.cpp

@@ -10,6 +10,7 @@
 #include <Poco/Net/NetException.h>
 #include <Poco/Util/HelpFormatter.h>
 #include <Poco/Environment.h>
+#include <Poco/Config.h>
 #include <Common/scope_guard_safe.h>
 #include <Common/logger_useful.h>
 #include <base/phdr_cache.h>
@@ -721,11 +722,6 @@ try
     CurrentMetrics::set(CurrentMetrics::Revision, ClickHouseRevision::getVersionRevision());
     CurrentMetrics::set(CurrentMetrics::VersionInteger, ClickHouseRevision::getVersionInteger());
 
-    Poco::ThreadPool server_pool(3, server_settings.max_connections);
-    std::mutex servers_lock;
-    std::vector<ProtocolServerAdapter> servers;
-    std::vector<ProtocolServerAdapter> servers_to_start_before_tables;
-
     /** Context contains all that query execution is dependent:
       *  settings, available functions, data types, aggregate functions, databases, ...
       */
@@ -823,6 +819,18 @@ try
             total_memory_tracker.setSampleMaxAllocationSize(server_settings.total_memory_profiler_sample_max_allocation_size);
     }
 
+    Poco::ThreadPool server_pool(
+        /* minCapacity */3,
+        /* maxCapacity */server_settings.max_connections,
+        /* idleTime */60,
+        /* stackSize */POCO_THREAD_STACK_SIZE,
+        server_settings.global_profiler_real_time_period_ns,
+        server_settings.global_profiler_cpu_time_period_ns);
+
+    std::mutex servers_lock;
+    std::vector<ProtocolServerAdapter> servers;
+    std::vector<ProtocolServerAdapter> servers_to_start_before_tables;
+
     /// Wait for all threads to avoid possible use-after-free (for example logging objects can be already destroyed).
     SCOPE_EXIT({
         Stopwatch watch;
@@ -1609,6 +1617,10 @@ try
                 0, // We don't need any threads one all the parts will be deleted
                 new_server_settings.max_parts_cleaning_thread_pool_size);
 
+
+            global_context->setMergeWorkload(new_server_settings.merge_workload);
+            global_context->setMutationWorkload(new_server_settings.mutation_workload);
+
             if (config->has("resources"))
             {
                 global_context->getResourceManager()->updateConfiguration(*config);

programs/server/config.xml

@@ -371,7 +371,7 @@
     <!-- Enables asynchronous loading of databases and tables to speedup server startup.
          Queries to not yet loaded entity will be blocked until load is finished.
       -->
-    <!-- <async_load_databases>true</async_load_databases> -->
+    <async_load_databases>true</async_load_databases>
 
     <!-- On memory constrained environments you may have to set this to value larger than 1.
       -->
@@ -1396,6 +1396,14 @@
         <!-- <host_name>replica</host_name> -->
     </distributed_ddl>
 
+    <!-- Used to regulate how resources are utilized and shared between merges, mutations and other workloads.
+         Specified value is used as `workload` setting value for background merge or mutation.
+    -->
+    <!--
+    <merge_workload>merges_and_mutations</merge_workload>
+    <mutation_workload>merges_and_mutations</mutation_workload>
+    -->
+
     <!-- Settings to fine-tune MergeTree tables. See documentation in source code, in MergeTreeSettings.h -->
     <!--
     <merge_tree>

src/Access/CachedAccessChecking.cpp

@@ -4,12 +4,12 @@
 
 namespace DB
 {
-CachedAccessChecking::CachedAccessChecking(const std::shared_ptr<const ContextAccess> & access_, AccessFlags access_flags_)
+CachedAccessChecking::CachedAccessChecking(const std::shared_ptr<const ContextAccessWrapper> & access_, AccessFlags access_flags_)
     : CachedAccessChecking(access_, AccessRightsElement{access_flags_})
 {
 }
 
-CachedAccessChecking::CachedAccessChecking(const std::shared_ptr<const ContextAccess> & access_, const AccessRightsElement & element_)
+CachedAccessChecking::CachedAccessChecking(const std::shared_ptr<const ContextAccessWrapper> & access_, const AccessRightsElement & element_)
     : access(access_), element(element_)
 {
 }

src/Access/CachedAccessChecking.h

@@ -1,6 +1,7 @@
 #pragma once
 
 #include <Access/Common/AccessRightsElement.h>
+#include <Access/ContextAccess.h>
 #include <memory>
 
 
@@ -13,14 +14,14 @@ class ContextAccess;
 class CachedAccessChecking
 {
 public:
-    CachedAccessChecking(const std::shared_ptr<const ContextAccess> & access_, AccessFlags access_flags_);
-    CachedAccessChecking(const std::shared_ptr<const ContextAccess> & access_, const AccessRightsElement & element_);
+    CachedAccessChecking(const std::shared_ptr<const ContextAccessWrapper> & access_, AccessFlags access_flags_);
+    CachedAccessChecking(const std::shared_ptr<const ContextAccessWrapper> & access_, const AccessRightsElement & element_);
     ~CachedAccessChecking();
 
     bool checkAccess(bool throw_if_denied = true);
 
 private:
-    const std::shared_ptr<const ContextAccess> access;
+    const std::shared_ptr<const ContextAccessWrapper> access;
     const AccessRightsElement element;
     bool checked = false;
     bool result = false;

src/Access/ContextAccess.cpp

@@ -20,6 +20,7 @@
 #include <boost/algorithm/string/join.hpp>
 #include <boost/range/algorithm/set_algorithm.hpp>
 #include <cassert>
+#include <unordered_set>
 
 
 namespace DB
@@ -271,7 +272,7 @@ namespace
 
 std::shared_ptr<const ContextAccess> ContextAccess::fromContext(const ContextPtr & context)
 {
-    return context->getAccess();
+    return ContextAccessWrapper::fromContext(context)->getAccess();
 }
 
 
@@ -560,7 +561,7 @@ std::shared_ptr<const AccessRights> ContextAccess::getAccessRightsWithImplicit()
 
 
 template <bool throw_if_denied, bool grant_option, typename... Args>
-bool ContextAccess::checkAccessImplHelper(AccessFlags flags, const Args &... args) const
+bool ContextAccess::checkAccessImplHelper(const ContextPtr & context, AccessFlags flags, const Args &... args) const
 {
     if (user_was_dropped)
     {
@@ -573,8 +574,10 @@ bool ContextAccess::checkAccessImplHelper(AccessFlags flags, const Args &... arg
     if (params.full_access)
         return true;
 
-    auto access_granted = []
+    auto access_granted = [&]
     {
+        if constexpr (throw_if_denied)
+            context->addQueryPrivilegesInfo(AccessRightsElement{flags, args...}.toStringWithoutOptions(), true);
         return true;
     };
 
@@ -583,7 +586,10 @@ bool ContextAccess::checkAccessImplHelper(AccessFlags flags, const Args &... arg
                                                FmtArgs && ...fmt_args [[maybe_unused]])
     {
         if constexpr (throw_if_denied)
+        {
+            context->addQueryPrivilegesInfo(AccessRightsElement{flags, args...}.toStringWithoutOptions(), false);
             throw Exception(error_code, std::move(fmt_string), getUserName(), std::forward<FmtArgs>(fmt_args)...);
+        }
         return false;
     };
 
@@ -686,102 +692,102 @@ bool ContextAccess::checkAccessImplHelper(AccessFlags flags, const Args &... arg
 }
 
 template <bool throw_if_denied, bool grant_option>
-bool ContextAccess::checkAccessImpl(const AccessFlags & flags) const
+bool ContextAccess::checkAccessImpl(const ContextPtr & context, const AccessFlags & flags) const
 {
-    return checkAccessImplHelper<throw_if_denied, grant_option>(flags);
+    return checkAccessImplHelper<throw_if_denied, grant_option>(context, flags);
 }
 
 template <bool throw_if_denied, bool grant_option, typename... Args>
-bool ContextAccess::checkAccessImpl(const AccessFlags & flags, std::string_view database, const Args &... args) const
+bool ContextAccess::checkAccessImpl(const ContextPtr & context, const AccessFlags & flags, std::string_view database, const Args &... args) const
 {
-    return checkAccessImplHelper<throw_if_denied, grant_option>(flags, database.empty() ? params.current_database : database, args...);
+    return checkAccessImplHelper<throw_if_denied, grant_option>(context, flags, database.empty() ? params.current_database : database, args...);
 }
 
 template <bool throw_if_denied, bool grant_option>
-bool ContextAccess::checkAccessImplHelper(const AccessRightsElement & element) const
+bool ContextAccess::checkAccessImplHelper(const ContextPtr & context, const AccessRightsElement & element) const
 {
     assert(!element.grant_option || grant_option);
     if (element.isGlobalWithParameter())
     {
         if (element.any_parameter)
-            return checkAccessImpl<throw_if_denied, grant_option>(element.access_flags);
+            return checkAccessImpl<throw_if_denied, grant_option>(context, element.access_flags);
         else
-            return checkAccessImpl<throw_if_denied, grant_option>(element.access_flags, element.parameter);
+            return checkAccessImpl<throw_if_denied, grant_option>(context, element.access_flags, element.parameter);
     }
     else if (element.any_database)
-        return checkAccessImpl<throw_if_denied, grant_option>(element.access_flags);
+        return checkAccessImpl<throw_if_denied, grant_option>(context, element.access_flags);
     else if (element.any_table)
-        return checkAccessImpl<throw_if_denied, grant_option>(element.access_flags, element.database);
+        return checkAccessImpl<throw_if_denied, grant_option>(context, element.access_flags, element.database);
     else if (element.any_column)
-        return checkAccessImpl<throw_if_denied, grant_option>(element.access_flags, element.database, element.table);
+        return checkAccessImpl<throw_if_denied, grant_option>(context, element.access_flags, element.database, element.table);
     else
-        return checkAccessImpl<throw_if_denied, grant_option>(element.access_flags, element.database, element.table, element.columns);
+        return checkAccessImpl<throw_if_denied, grant_option>(context, element.access_flags, element.database, element.table, element.columns);
 }
 
 template <bool throw_if_denied, bool grant_option>
-bool ContextAccess::checkAccessImpl(const AccessRightsElement & element) const
+bool ContextAccess::checkAccessImpl(const ContextPtr & context, const AccessRightsElement & element) const
 {
     if constexpr (grant_option)
     {
-        return checkAccessImplHelper<throw_if_denied, true>(element);
+        return checkAccessImplHelper<throw_if_denied, true>(context, element);
     }
     else
     {
         if (element.grant_option)
-            return checkAccessImplHelper<throw_if_denied, true>(element);
+            return checkAccessImplHelper<throw_if_denied, true>(context, element);
         else
-            return checkAccessImplHelper<throw_if_denied, false>(element);
+            return checkAccessImplHelper<throw_if_denied, false>(context, element);
     }
 }
 
 template <bool throw_if_denied, bool grant_option>
-bool ContextAccess::checkAccessImpl(const AccessRightsElements & elements) const
+bool ContextAccess::checkAccessImpl(const ContextPtr & context, const AccessRightsElements & elements) const
 {
     for (const auto & element : elements)
-        if (!checkAccessImpl<throw_if_denied, grant_option>(element))
+        if (!checkAccessImpl<throw_if_denied, grant_option>(context, element))
             return false;
     return true;
 }
 
-bool ContextAccess::isGranted(const AccessFlags & flags) const { return checkAccessImpl<false, false>(flags); }
-bool ContextAccess::isGranted(const AccessFlags & flags, std::string_view database) const { return checkAccessImpl<false, false>(flags, database); }
-bool ContextAccess::isGranted(const AccessFlags & flags, std::string_view database, std::string_view table) const { return checkAccessImpl<false, false>(flags, database, table); }
-bool ContextAccess::isGranted(const AccessFlags & flags, std::string_view database, std::string_view table, std::string_view column) const { return checkAccessImpl<false, false>(flags, database, table, column); }
-bool ContextAccess::isGranted(const AccessFlags & flags, std::string_view database, std::string_view table, const std::vector<std::string_view> & columns) const { return checkAccessImpl<false, false>(flags, database, table, columns); }
-bool ContextAccess::isGranted(const AccessFlags & flags, std::string_view database, std::string_view table, const Strings & columns) const { return checkAccessImpl<false, false>(flags, database, table, columns); }
-bool ContextAccess::isGranted(const AccessRightsElement & element) const { return checkAccessImpl<false, false>(element); }
-bool ContextAccess::isGranted(const AccessRightsElements & elements) const { return checkAccessImpl<false, false>(elements); }
+bool ContextAccess::isGranted(const ContextPtr & context, const AccessFlags & flags) const { return checkAccessImpl<false, false>(context, flags); }
+bool ContextAccess::isGranted(const ContextPtr & context, const AccessFlags & flags, std::string_view database) const { return checkAccessImpl<false, false>(context, flags, database); }
+bool ContextAccess::isGranted(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table) const { return checkAccessImpl<false, false>(context, flags, database, table); }
+bool ContextAccess::isGranted(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, std::string_view column) const { return checkAccessImpl<false, false>(context, flags, database, table, column); }
+bool ContextAccess::isGranted(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, const std::vector<std::string_view> & columns) const { return checkAccessImpl<false, false>(context, flags, database, table, columns); }
+bool ContextAccess::isGranted(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, const Strings & columns) const { return checkAccessImpl<false, false>(context, flags, database, table, columns); }
+bool ContextAccess::isGranted(const ContextPtr & context, const AccessRightsElement & element) const { return checkAccessImpl<false, false>(context, element); }
+bool ContextAccess::isGranted(const ContextPtr & context, const AccessRightsElements & elements) const { return checkAccessImpl<false, false>(context, elements); }
 
-bool ContextAccess::hasGrantOption(const AccessFlags & flags) const { return checkAccessImpl<false, true>(flags); }
-bool ContextAccess::hasGrantOption(const AccessFlags & flags, std::string_view database) const { return checkAccessImpl<false, true>(flags, database); }
-bool ContextAccess::hasGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table) const { return checkAccessImpl<false, true>(flags, database, table); }
-bool ContextAccess::hasGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table, std::string_view column) const { return checkAccessImpl<false, true>(flags, database, table, column); }
-bool ContextAccess::hasGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table, const std::vector<std::string_view> & columns) const { return checkAccessImpl<false, true>(flags, database, table, columns); }
-bool ContextAccess::hasGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table, const Strings & columns) const { return checkAccessImpl<false, true>(flags, database, table, columns); }
-bool ContextAccess::hasGrantOption(const AccessRightsElement & element) const { return checkAccessImpl<false, true>(element); }
-bool ContextAccess::hasGrantOption(const AccessRightsElements & elements) const { return checkAccessImpl<false, true>(elements); }
+bool ContextAccess::hasGrantOption(const ContextPtr & context, const AccessFlags & flags) const { return checkAccessImpl<false, true>(context, flags); }
+bool ContextAccess::hasGrantOption(const ContextPtr & context, const AccessFlags & flags, std::string_view database) const { return checkAccessImpl<false, true>(context, flags, database); }
+bool ContextAccess::hasGrantOption(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table) const { return checkAccessImpl<false, true>(context, flags, database, table); }
+bool ContextAccess::hasGrantOption(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, std::string_view column) const { return checkAccessImpl<false, true>(context, flags, database, table, column); }
+bool ContextAccess::hasGrantOption(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, const std::vector<std::string_view> & columns) const { return checkAccessImpl<false, true>(context, flags, database, table, columns); }
+bool ContextAccess::hasGrantOption(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, const Strings & columns) const { return checkAccessImpl<false, true>(context, flags, database, table, columns); }
+bool ContextAccess::hasGrantOption(const ContextPtr & context, const AccessRightsElement & element) const { return checkAccessImpl<false, true>(context, element); }
+bool ContextAccess::hasGrantOption(const ContextPtr & context, const AccessRightsElements & elements) const { return checkAccessImpl<false, true>(context, elements); }
 
-void ContextAccess::checkAccess(const AccessFlags & flags) const { checkAccessImpl<true, false>(flags); }
-void ContextAccess::checkAccess(const AccessFlags & flags, std::string_view database) const { checkAccessImpl<true, false>(flags, database); }
-void ContextAccess::checkAccess(const AccessFlags & flags, std::string_view database, std::string_view table) const { checkAccessImpl<true, false>(flags, database, table); }
-void ContextAccess::checkAccess(const AccessFlags & flags, std::string_view database, std::string_view table, std::string_view column) const { checkAccessImpl<true, false>(flags, database, table, column); }
-void ContextAccess::checkAccess(const AccessFlags & flags, std::string_view database, std::string_view table, const std::vector<std::string_view> & columns) const { checkAccessImpl<true, false>(flags, database, table, columns); }
-void ContextAccess::checkAccess(const AccessFlags & flags, std::string_view database, std::string_view table, const Strings & columns) const { checkAccessImpl<true, false>(flags, database, table, columns); }
-void ContextAccess::checkAccess(const AccessRightsElement & element) const { checkAccessImpl<true, false>(element); }
-void ContextAccess::checkAccess(const AccessRightsElements & elements) const { checkAccessImpl<true, false>(elements); }
+void ContextAccess::checkAccess(const ContextPtr & context, const AccessFlags & flags) const { checkAccessImpl<true, false>(context, flags); }
+void ContextAccess::checkAccess(const ContextPtr & context, const AccessFlags & flags, std::string_view database) const { checkAccessImpl<true, false>(context, flags, database); }
+void ContextAccess::checkAccess(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table) const { checkAccessImpl<true, false>(context, flags, database, table); }
+void ContextAccess::checkAccess(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, std::string_view column) const { checkAccessImpl<true, false>(context, flags, database, table, column); }
+void ContextAccess::checkAccess(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, const std::vector<std::string_view> & columns) const { checkAccessImpl<true, false>(context, flags, database, table, columns); }
+void ContextAccess::checkAccess(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, const Strings & columns) const { checkAccessImpl<true, false>(context, flags, database, table, columns); }
+void ContextAccess::checkAccess(const ContextPtr & context, const AccessRightsElement & element) const { checkAccessImpl<true, false>(context, element); }
+void ContextAccess::checkAccess(const ContextPtr & context, const AccessRightsElements & elements) const { checkAccessImpl<true, false>(context, elements); }
 
-void ContextAccess::checkGrantOption(const AccessFlags & flags) const { checkAccessImpl<true, true>(flags); }
-void ContextAccess::checkGrantOption(const AccessFlags & flags, std::string_view database) const { checkAccessImpl<true, true>(flags, database); }
-void ContextAccess::checkGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table) const { checkAccessImpl<true, true>(flags, database, table); }
-void ContextAccess::checkGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table, std::string_view column) const { checkAccessImpl<true, true>(flags, database, table, column); }
-void ContextAccess::checkGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table, const std::vector<std::string_view> & columns) const { checkAccessImpl<true, true>(flags, database, table, columns); }
-void ContextAccess::checkGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table, const Strings & columns) const { checkAccessImpl<true, true>(flags, database, table, columns); }
-void ContextAccess::checkGrantOption(const AccessRightsElement & element) const { checkAccessImpl<true, true>(element); }
-void ContextAccess::checkGrantOption(const AccessRightsElements & elements) const { checkAccessImpl<true, true>(elements); }
+void ContextAccess::checkGrantOption(const ContextPtr & context, const AccessFlags & flags) const { checkAccessImpl<true, true>(context, flags); }
+void ContextAccess::checkGrantOption(const ContextPtr & context, const AccessFlags & flags, std::string_view database) const { checkAccessImpl<true, true>(context, flags, database); }
+void ContextAccess::checkGrantOption(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table) const { checkAccessImpl<true, true>(context, flags, database, table); }
+void ContextAccess::checkGrantOption(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, std::string_view column) const { checkAccessImpl<true, true>(context, flags, database, table, column); }
+void ContextAccess::checkGrantOption(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, const std::vector<std::string_view> & columns) const { checkAccessImpl<true, true>(context, flags, database, table, columns); }
+void ContextAccess::checkGrantOption(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, const Strings & columns) const { checkAccessImpl<true, true>(context, flags, database, table, columns); }
+void ContextAccess::checkGrantOption(const ContextPtr & context, const AccessRightsElement & element) const { checkAccessImpl<true, true>(context, element); }
+void ContextAccess::checkGrantOption(const ContextPtr & context, const AccessRightsElements & elements) const { checkAccessImpl<true, true>(context, elements); }
 
 
 template <bool throw_if_denied, typename Container, typename GetNameFunction>
-bool ContextAccess::checkAdminOptionImplHelper(const Container & role_ids, const GetNameFunction & get_name_function) const
+bool ContextAccess::checkAdminOptionImplHelper(const ContextPtr & context, const Container & role_ids, const GetNameFunction & get_name_function) const
 {
     auto show_error = []<typename... FmtArgs>(int error_code [[maybe_unused]],
                                                   FormatStringHelper<FmtArgs...> fmt_string [[maybe_unused]],
@@ -804,7 +810,7 @@ bool ContextAccess::checkAdminOptionImplHelper(const Container & role_ids, const
     if (!std::size(role_ids))
         return true;
 
-    if (isGranted(AccessType::ROLE_ADMIN))
+    if (isGranted(context, AccessType::ROLE_ADMIN))
         return true;
 
     auto info = getRolesInfo();
@@ -840,54 +846,54 @@ bool ContextAccess::checkAdminOptionImplHelper(const Container & role_ids, const
 }
 
 template <bool throw_if_denied>
-bool ContextAccess::checkAdminOptionImpl(const UUID & role_id) const
+bool ContextAccess::checkAdminOptionImpl(const ContextPtr & context, const UUID & role_id) const
 {
-    return checkAdminOptionImplHelper<throw_if_denied>(to_array(role_id), [this](const UUID & id, size_t) { return access_control->tryReadName(id); });
+    return checkAdminOptionImplHelper<throw_if_denied>(context, to_array(role_id), [this](const UUID & id, size_t) { return access_control->tryReadName(id); });
 }
 
 template <bool throw_if_denied>
-bool ContextAccess::checkAdminOptionImpl(const UUID & role_id, const String & role_name) const
+bool ContextAccess::checkAdminOptionImpl(const ContextPtr & context, const UUID & role_id, const String & role_name) const
 {
-    return checkAdminOptionImplHelper<throw_if_denied>(to_array(role_id), [&role_name](const UUID &, size_t) { return std::optional<String>{role_name}; });
+    return checkAdminOptionImplHelper<throw_if_denied>(context, to_array(role_id), [&role_name](const UUID &, size_t) { return std::optional<String>{role_name}; });
 }
 
 template <bool throw_if_denied>
-bool ContextAccess::checkAdminOptionImpl(const UUID & role_id, const std::unordered_map<UUID, String> & names_of_roles) const
+bool ContextAccess::checkAdminOptionImpl(const ContextPtr & context, const UUID & role_id, const std::unordered_map<UUID, String> & names_of_roles) const
 {
-    return checkAdminOptionImplHelper<throw_if_denied>(to_array(role_id), [&names_of_roles](const UUID & id, size_t) { auto it = names_of_roles.find(id); return (it != names_of_roles.end()) ? it->second : std::optional<String>{}; });
+    return checkAdminOptionImplHelper<throw_if_denied>(context, to_array(role_id), [&names_of_roles](const UUID & id, size_t) { auto it = names_of_roles.find(id); return (it != names_of_roles.end()) ? it->second : std::optional<String>{}; });
 }
 
 template <bool throw_if_denied>
-bool ContextAccess::checkAdminOptionImpl(const std::vector<UUID> & role_ids) const
+bool ContextAccess::checkAdminOptionImpl(const ContextPtr & context, const std::vector<UUID> & role_ids) const
 {
-    return checkAdminOptionImplHelper<throw_if_denied>(role_ids, [this](const UUID & id, size_t) { return access_control->tryReadName(id); });
+    return checkAdminOptionImplHelper<throw_if_denied>(context, role_ids, [this](const UUID & id, size_t) { return access_control->tryReadName(id); });
 }
 
 template <bool throw_if_denied>
-bool ContextAccess::checkAdminOptionImpl(const std::vector<UUID> & role_ids, const Strings & names_of_roles) const
+bool ContextAccess::checkAdminOptionImpl(const ContextPtr & context, const std::vector<UUID> & role_ids, const Strings & names_of_roles) const
 {
-    return checkAdminOptionImplHelper<throw_if_denied>(role_ids, [&names_of_roles](const UUID &, size_t i) { return std::optional<String>{names_of_roles[i]}; });
+    return checkAdminOptionImplHelper<throw_if_denied>(context, role_ids, [&names_of_roles](const UUID &, size_t i) { return std::optional<String>{names_of_roles[i]}; });
 }
 
 template <bool throw_if_denied>
-bool ContextAccess::checkAdminOptionImpl(const std::vector<UUID> & role_ids, const std::unordered_map<UUID, String> & names_of_roles) const
+bool ContextAccess::checkAdminOptionImpl(const ContextPtr & context, const std::vector<UUID> & role_ids, const std::unordered_map<UUID, String> & names_of_roles) const
 {
-    return checkAdminOptionImplHelper<throw_if_denied>(role_ids, [&names_of_roles](const UUID & id, size_t) { auto it = names_of_roles.find(id); return (it != names_of_roles.end()) ? it->second : std::optional<String>{}; });
+    return checkAdminOptionImplHelper<throw_if_denied>(context, role_ids, [&names_of_roles](const UUID & id, size_t) { auto it = names_of_roles.find(id); return (it != names_of_roles.end()) ? it->second : std::optional<String>{}; });
 }
 
-bool ContextAccess::hasAdminOption(const UUID & role_id) const { return checkAdminOptionImpl<false>(role_id); }
-bool ContextAccess::hasAdminOption(const UUID & role_id, const String & role_name) const { return checkAdminOptionImpl<false>(role_id, role_name); }
-bool ContextAccess::hasAdminOption(const UUID & role_id, const std::unordered_map<UUID, String> & names_of_roles) const { return checkAdminOptionImpl<false>(role_id, names_of_roles); }
-bool ContextAccess::hasAdminOption(const std::vector<UUID> & role_ids) const { return checkAdminOptionImpl<false>(role_ids); }
-bool ContextAccess::hasAdminOption(const std::vector<UUID> & role_ids, const Strings & names_of_roles) const { return checkAdminOptionImpl<false>(role_ids, names_of_roles); }
-bool ContextAccess::hasAdminOption(const std::vector<UUID> & role_ids, const std::unordered_map<UUID, String> & names_of_roles) const { return checkAdminOptionImpl<false>(role_ids, names_of_roles); }
+bool ContextAccess::hasAdminOption(const ContextPtr & context, const UUID & role_id) const { return checkAdminOptionImpl<false>(context, role_id); }
+bool ContextAccess::hasAdminOption(const ContextPtr & context, const UUID & role_id, const String & role_name) const { return checkAdminOptionImpl<false>(context, role_id, role_name); }
+bool ContextAccess::hasAdminOption(const ContextPtr & context, const UUID & role_id, const std::unordered_map<UUID, String> & names_of_roles) const { return checkAdminOptionImpl<false>(context, role_id, names_of_roles); }
+bool ContextAccess::hasAdminOption(const ContextPtr & context, const std::vector<UUID> & role_ids) const { return checkAdminOptionImpl<false>(context, role_ids); }
+bool ContextAccess::hasAdminOption(const ContextPtr & context, const std::vector<UUID> & role_ids, const Strings & names_of_roles) const { return checkAdminOptionImpl<false>(context, role_ids, names_of_roles); }
+bool ContextAccess::hasAdminOption(const ContextPtr & context, const std::vector<UUID> & role_ids, const std::unordered_map<UUID, String> & names_of_roles) const { return checkAdminOptionImpl<false>(context, role_ids, names_of_roles); }
 
-void ContextAccess::checkAdminOption(const UUID & role_id) const { checkAdminOptionImpl<true>(role_id); }
-void ContextAccess::checkAdminOption(const UUID & role_id, const String & role_name) const { checkAdminOptionImpl<true>(role_id, role_name); }
-void ContextAccess::checkAdminOption(const UUID & role_id, const std::unordered_map<UUID, String> & names_of_roles) const { checkAdminOptionImpl<true>(role_id, names_of_roles); }
-void ContextAccess::checkAdminOption(const std::vector<UUID> & role_ids) const { checkAdminOptionImpl<true>(role_ids); }
-void ContextAccess::checkAdminOption(const std::vector<UUID> & role_ids, const Strings & names_of_roles) const { checkAdminOptionImpl<true>(role_ids, names_of_roles); }
-void ContextAccess::checkAdminOption(const std::vector<UUID> & role_ids, const std::unordered_map<UUID, String> & names_of_roles) const { checkAdminOptionImpl<true>(role_ids, names_of_roles); }
+void ContextAccess::checkAdminOption(const ContextPtr & context, const UUID & role_id) const { checkAdminOptionImpl<true>(context, role_id); }
+void ContextAccess::checkAdminOption(const ContextPtr & context, const UUID & role_id, const String & role_name) const { checkAdminOptionImpl<true>(context, role_id, role_name); }
+void ContextAccess::checkAdminOption(const ContextPtr & context, const UUID & role_id, const std::unordered_map<UUID, String> & names_of_roles) const { checkAdminOptionImpl<true>(context, role_id, names_of_roles); }
+void ContextAccess::checkAdminOption(const ContextPtr & context, const std::vector<UUID> & role_ids) const { checkAdminOptionImpl<true>(context, role_ids); }
+void ContextAccess::checkAdminOption(const ContextPtr & context, const std::vector<UUID> & role_ids, const Strings & names_of_roles) const { checkAdminOptionImpl<true>(context, role_ids, names_of_roles); }
+void ContextAccess::checkAdminOption(const ContextPtr & context, const std::vector<UUID> & role_ids, const std::unordered_map<UUID, String> & names_of_roles) const { checkAdminOptionImpl<true>(context, role_ids, names_of_roles); }
 
 
 void ContextAccess::checkGranteeIsAllowed(const UUID & grantee_id, const IAccessEntity & grantee) const
@@ -919,4 +925,10 @@ void ContextAccess::checkGranteesAreAllowed(const std::vector<UUID> & grantee_id
     }
 }
 
+std::shared_ptr<const ContextAccessWrapper> ContextAccessWrapper::fromContext(const ContextPtr & context)
+{
+    return context->getAccess();
+}
+
+
 }

src/Access/ContextAccess.h

@@ -4,9 +4,12 @@
 #include <Access/ContextAccessParams.h>
 #include <Access/EnabledRowPolicies.h>
 #include <Interpreters/ClientInfo.h>
+#include <Access/QuotaUsage.h>
+#include <Common/SettingsChanges.h>
 #include <Core/UUID.h>
 #include <base/scope_guard.h>
 #include <boost/container/flat_set.hpp>
+#include <memory>
 #include <mutex>
 #include <optional>
 #include <unordered_map>
@@ -71,59 +74,59 @@ public:
 
     /// Checks if a specified access is granted, and throws an exception if not.
     /// Empty database means the current database.
-    void checkAccess(const AccessFlags & flags) const;
-    void checkAccess(const AccessFlags & flags, std::string_view database) const;
-    void checkAccess(const AccessFlags & flags, std::string_view database, std::string_view table) const;
-    void checkAccess(const AccessFlags & flags, std::string_view database, std::string_view table, std::string_view column) const;
-    void checkAccess(const AccessFlags & flags, std::string_view database, std::string_view table, const std::vector<std::string_view> & columns) const;
-    void checkAccess(const AccessFlags & flags, std::string_view database, std::string_view table, const Strings & columns) const;
-    void checkAccess(const AccessRightsElement & element) const;
-    void checkAccess(const AccessRightsElements & elements) const;
+    void checkAccess(const ContextPtr & context, const AccessFlags & flags) const;
+    void checkAccess(const ContextPtr & context, const AccessFlags & flags, std::string_view database) const;
+    void checkAccess(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table) const;
+    void checkAccess(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, std::string_view column) const;
+    void checkAccess(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, const std::vector<std::string_view> & columns) const;
+    void checkAccess(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, const Strings & columns) const;
+    void checkAccess(const ContextPtr & context, const AccessRightsElement & element) const;
+    void checkAccess(const ContextPtr & context, const AccessRightsElements & elements) const;
 
-    void checkGrantOption(const AccessFlags & flags) const;
-    void checkGrantOption(const AccessFlags & flags, std::string_view database) const;
-    void checkGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table) const;
-    void checkGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table, std::string_view column) const;
-    void checkGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table, const std::vector<std::string_view> & columns) const;
-    void checkGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table, const Strings & columns) const;
-    void checkGrantOption(const AccessRightsElement & element) const;
-    void checkGrantOption(const AccessRightsElements & elements) const;
+    void checkGrantOption(const ContextPtr & context, const AccessFlags & flags) const;
+    void checkGrantOption(const ContextPtr & context, const AccessFlags & flags, std::string_view database) const;
+    void checkGrantOption(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table) const;
+    void checkGrantOption(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, std::string_view column) const;
+    void checkGrantOption(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, const std::vector<std::string_view> & columns) const;
+    void checkGrantOption(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, const Strings & columns) const;
+    void checkGrantOption(const ContextPtr & context, const AccessRightsElement & element) const;
+    void checkGrantOption(const ContextPtr & context, const AccessRightsElements & elements) const;
 
     /// Checks if a specified access is granted, and returns false if not.
     /// Empty database means the current database.
-    bool isGranted(const AccessFlags & flags) const;
-    bool isGranted(const AccessFlags & flags, std::string_view database) const;
-    bool isGranted(const AccessFlags & flags, std::string_view database, std::string_view table) const;
-    bool isGranted(const AccessFlags & flags, std::string_view database, std::string_view table, std::string_view column) const;
-    bool isGranted(const AccessFlags & flags, std::string_view database, std::string_view table, const std::vector<std::string_view> & columns) const;
-    bool isGranted(const AccessFlags & flags, std::string_view database, std::string_view table, const Strings & columns) const;
-    bool isGranted(const AccessRightsElement & element) const;
-    bool isGranted(const AccessRightsElements & elements) const;
+    bool isGranted(const ContextPtr & context, const AccessFlags & flags) const;
+    bool isGranted(const ContextPtr & context, const AccessFlags & flags, std::string_view database) const;
+    bool isGranted(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table) const;
+    bool isGranted(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, std::string_view column) const;
+    bool isGranted(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, const std::vector<std::string_view> & columns) const;
+    bool isGranted(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, const Strings & columns) const;
+    bool isGranted(const ContextPtr & context, const AccessRightsElement & element) const;
+    bool isGranted(const ContextPtr & context, const AccessRightsElements & elements) const;
 
-    bool hasGrantOption(const AccessFlags & flags) const;
-    bool hasGrantOption(const AccessFlags & flags, std::string_view database) const;
-    bool hasGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table) const;
-    bool hasGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table, std::string_view column) const;
-    bool hasGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table, const std::vector<std::string_view> & columns) const;
-    bool hasGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table, const Strings & columns) const;
-    bool hasGrantOption(const AccessRightsElement & element) const;
-    bool hasGrantOption(const AccessRightsElements & elements) const;
+    bool hasGrantOption(const ContextPtr & context, const AccessFlags & flags) const;
+    bool hasGrantOption(const ContextPtr & context, const AccessFlags & flags, std::string_view database) const;
+    bool hasGrantOption(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table) const;
+    bool hasGrantOption(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, std::string_view column) const;
+    bool hasGrantOption(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, const std::vector<std::string_view> & columns) const;
+    bool hasGrantOption(const ContextPtr & context, const AccessFlags & flags, std::string_view database, std::string_view table, const Strings & columns) const;
+    bool hasGrantOption(const ContextPtr & context, const AccessRightsElement & element) const;
+    bool hasGrantOption(const ContextPtr & context, const AccessRightsElements & elements) const;
 
     /// Checks if a specified role is granted with admin option, and throws an exception if not.
-    void checkAdminOption(const UUID & role_id) const;
-    void checkAdminOption(const UUID & role_id, const String & role_name) const;
-    void checkAdminOption(const UUID & role_id, const std::unordered_map<UUID, String> & names_of_roles) const;
-    void checkAdminOption(const std::vector<UUID> & role_ids) const;
-    void checkAdminOption(const std::vector<UUID> & role_ids, const Strings & names_of_roles) const;
-    void checkAdminOption(const std::vector<UUID> & role_ids, const std::unordered_map<UUID, String> & names_of_roles) const;
+    void checkAdminOption(const ContextPtr & context, const UUID & role_id) const;
+    void checkAdminOption(const ContextPtr & context, const UUID & role_id, const String & role_name) const;
+    void checkAdminOption(const ContextPtr & context, const UUID & role_id, const std::unordered_map<UUID, String> & names_of_roles) const;
+    void checkAdminOption(const ContextPtr & context, const std::vector<UUID> & role_ids) const;
+    void checkAdminOption(const ContextPtr & context, const std::vector<UUID> & role_ids, const Strings & names_of_roles) const;
+    void checkAdminOption(const ContextPtr & context, const std::vector<UUID> & role_ids, const std::unordered_map<UUID, String> & names_of_roles) const;
 
     /// Checks if a specified role is granted with admin option, and returns false if not.
-    bool hasAdminOption(const UUID & role_id) const;
-    bool hasAdminOption(const UUID & role_id, const String & role_name) const;
-    bool hasAdminOption(const UUID & role_id, const std::unordered_map<UUID, String> & names_of_roles) const;
-    bool hasAdminOption(const std::vector<UUID> & role_ids) const;
-    bool hasAdminOption(const std::vector<UUID> & role_ids, const Strings & names_of_roles) const;
-    bool hasAdminOption(const std::vector<UUID> & role_ids, const std::unordered_map<UUID, String> & names_of_roles) const;
+    bool hasAdminOption(const ContextPtr & context, const UUID & role_id) const;
+    bool hasAdminOption(const ContextPtr & context, const UUID & role_id, const String & role_name) const;
+    bool hasAdminOption(const ContextPtr & context, const UUID & role_id, const std::unordered_map<UUID, String> & names_of_roles) const;
+    bool hasAdminOption(const ContextPtr & context, const std::vector<UUID> & role_ids) const;
+    bool hasAdminOption(const ContextPtr & context, const std::vector<UUID> & role_ids, const Strings & names_of_roles) const;
+    bool hasAdminOption(const ContextPtr & context, const std::vector<UUID> & role_ids, const std::unordered_map<UUID, String> & names_of_roles) const;
 
     /// Checks if a grantee is allowed for the current user, throws an exception if not.
     void checkGranteeIsAllowed(const UUID & grantee_id, const IAccessEntity & grantee) const;
@@ -142,43 +145,43 @@ private:
     void calculateAccessRights() const TSA_REQUIRES(mutex);
 
     template <bool throw_if_denied, bool grant_option>
-    bool checkAccessImpl(const AccessFlags & flags) const;
+    bool checkAccessImpl(const ContextPtr & context, const AccessFlags & flags) const;
 
     template <bool throw_if_denied, bool grant_option, typename... Args>
-    bool checkAccessImpl(const AccessFlags & flags, std::string_view database, const Args &... args) const;
+    bool checkAccessImpl(const ContextPtr & context, const AccessFlags & flags, std::string_view database, const Args &... args) const;
 
     template <bool throw_if_denied, bool grant_option>
-    bool checkAccessImpl(const AccessRightsElement & element) const;
+    bool checkAccessImpl(const ContextPtr & context, const AccessRightsElement & element) const;
 
     template <bool throw_if_denied, bool grant_option>
-    bool checkAccessImpl(const AccessRightsElements & elements) const;
+    bool checkAccessImpl(const ContextPtr & context, const AccessRightsElements & elements) const;
 
     template <bool throw_if_denied, bool grant_option, typename... Args>
-    bool checkAccessImplHelper(AccessFlags flags, const Args &... args) const;
+    bool checkAccessImplHelper(const ContextPtr & context, AccessFlags flags, const Args &... args) const;
 
     template <bool throw_if_denied, bool grant_option>
-    bool checkAccessImplHelper(const AccessRightsElement & element) const;
+    bool checkAccessImplHelper(const ContextPtr & context, const AccessRightsElement & element) const;
 
     template <bool throw_if_denied>
-    bool checkAdminOptionImpl(const UUID & role_id) const;
+    bool checkAdminOptionImpl(const ContextPtr & context, const UUID & role_id) const;
 
     template <bool throw_if_denied>
-    bool checkAdminOptionImpl(const UUID & role_id, const String & role_name) const;
+    bool checkAdminOptionImpl(const ContextPtr & context, const UUID & role_id, const String & role_name) const;
 
     template <bool throw_if_denied>
-    bool checkAdminOptionImpl(const UUID & role_id, const std::unordered_map<UUID, String> & names_of_roles) const;
+    bool checkAdminOptionImpl(const ContextPtr & context, const UUID & role_id, const std::unordered_map<UUID, String> & names_of_roles) const;
 
     template <bool throw_if_denied>
-    bool checkAdminOptionImpl(const std::vector<UUID> & role_ids) const;
+    bool checkAdminOptionImpl(const ContextPtr & context, const std::vector<UUID> & role_ids) const;
 
     template <bool throw_if_denied>
-    bool checkAdminOptionImpl(const std::vector<UUID> & role_ids, const Strings & names_of_roles) const;
+    bool checkAdminOptionImpl(const ContextPtr & context, const std::vector<UUID> & role_ids, const Strings & names_of_roles) const;
 
     template <bool throw_if_denied>
-    bool checkAdminOptionImpl(const std::vector<UUID> & role_ids, const std::unordered_map<UUID, String> & names_of_roles) const;
+    bool checkAdminOptionImpl(const ContextPtr & context, const std::vector<UUID> & role_ids, const std::unordered_map<UUID, String> & names_of_roles) const;
 
     template <bool throw_if_denied, typename Container, typename GetNameFunction>
-    bool checkAdminOptionImplHelper(const Container & role_ids, const GetNameFunction & get_name_function) const;
+    bool checkAdminOptionImplHelper(const ContextPtr & context, const Container & role_ids, const GetNameFunction & get_name_function) const;
 
     const AccessControl * access_control = nullptr;
     const Params params;
@@ -203,4 +206,115 @@ private:
     mutable std::shared_ptr<const EnabledSettings> enabled_settings TSA_GUARDED_BY(mutex);
 };
 
+/// This wrapper was added to be able to pass the current context to the access
+/// without the need to change the signature and all calls to the ContextAccess itself.
+/// Right now a context is used to store privileges that are checked for a query,
+/// and might be useful for something else in the future as well.
+class ContextAccessWrapper : public std::enable_shared_from_this<ContextAccessWrapper>
+{
+public:
+    using ContextAccessPtr = std::shared_ptr<const ContextAccess>;
+
+    ContextAccessWrapper(const ContextAccessPtr & access_, const ContextPtr & context_): access(access_), context(context_) {}
+    ~ContextAccessWrapper() = default;
+
+    static std::shared_ptr<const ContextAccessWrapper> fromContext(const ContextPtr & context);
+
+    const ContextAccess::Params & getParams() const { return access->getParams(); }
+
+    const ContextAccessPtr & getAccess() const { return access; }
+
+    /// Returns the current user. Throws if user is nullptr.
+    ALWAYS_INLINE UserPtr getUser() const { return access->getUser(); }
+    /// Same as above, but can return nullptr.
+    ALWAYS_INLINE UserPtr tryGetUser() const { return access->tryGetUser(); }
+    ALWAYS_INLINE String getUserName() const { return access->getUserName(); }
+    ALWAYS_INLINE std::optional<UUID> getUserID() const { return access->getUserID(); }
+
+    /// Returns information about current and enabled roles.
+    ALWAYS_INLINE std::shared_ptr<const EnabledRolesInfo> getRolesInfo() const { return access->getRolesInfo(); }
+
+    /// Returns the row policy filter for a specified table.
+    /// The function returns nullptr if there is no filter to apply.
+    ALWAYS_INLINE RowPolicyFilterPtr getRowPolicyFilter(const String & database, const String & table_name, RowPolicyFilterType filter_type) const { return access->getRowPolicyFilter(database, table_name, filter_type); }
+
+    /// Returns the quota to track resource consumption.
+    ALWAYS_INLINE std::shared_ptr<const EnabledQuota> getQuota() const { return access->getQuota(); }
+    ALWAYS_INLINE std::optional<QuotaUsage> getQuotaUsage() const { return access->getQuotaUsage(); }
+
+    /// Returns the default settings, i.e. the settings which should be applied on user's login.
+    ALWAYS_INLINE SettingsChanges getDefaultSettings() const { return access->getDefaultSettings(); }
+    ALWAYS_INLINE std::shared_ptr<const SettingsProfilesInfo> getDefaultProfileInfo() const { return access->getDefaultProfileInfo(); }
+
+    /// Returns the current access rights.
+    ALWAYS_INLINE std::shared_ptr<const AccessRights> getAccessRights() const { return access->getAccessRights(); }
+    ALWAYS_INLINE std::shared_ptr<const AccessRights> getAccessRightsWithImplicit() const { return access->getAccessRightsWithImplicit(); }
+
+    /// Checks if a specified access is granted, and throws an exception if not.
+    /// Empty database means the current database.
+    ALWAYS_INLINE void checkAccess(const AccessFlags & flags) const { access->checkAccess(context, flags); }
+    ALWAYS_INLINE void checkAccess(const AccessFlags & flags, std::string_view database) const { access->checkAccess(context, flags, database); }
+    ALWAYS_INLINE void checkAccess(const AccessFlags & flags, std::string_view database, std::string_view table) const { access->checkAccess(context, flags, database, table); }
+    ALWAYS_INLINE void checkAccess(const AccessFlags & flags, std::string_view database, std::string_view table, std::string_view column) const { access->checkAccess(context, flags, database, table, column); }
+    ALWAYS_INLINE void checkAccess(const AccessFlags & flags, std::string_view database, std::string_view table, const std::vector<std::string_view> & columns) const { access->checkAccess(context, flags, database, table, columns); }
+    ALWAYS_INLINE void checkAccess(const AccessFlags & flags, std::string_view database, std::string_view table, const Strings & columns) const { access->checkAccess(context, flags, database, table, columns); }
+    ALWAYS_INLINE void checkAccess(const AccessRightsElement & element) const { access->checkAccess(context, element); }
+    ALWAYS_INLINE void checkAccess(const AccessRightsElements & elements) const { access->checkAccess(context, elements); }
+
+    ALWAYS_INLINE void checkGrantOption(const AccessFlags & flags) const { access->checkGrantOption(context, flags); }
+    ALWAYS_INLINE void checkGrantOption(const AccessFlags & flags, std::string_view database) const { access->checkGrantOption(context, flags, database); }
+    ALWAYS_INLINE void checkGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table) const { access->checkGrantOption(context, flags, database, table); }
+    ALWAYS_INLINE void checkGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table, std::string_view column) const { access->checkGrantOption(context, flags, database, table, column); }
+    ALWAYS_INLINE void checkGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table, const std::vector<std::string_view> & columns) const { access->checkGrantOption(context, flags, database, table, columns); }
+    ALWAYS_INLINE void checkGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table, const Strings & columns) const { access->checkGrantOption(context, flags, database, table, columns); }
+    ALWAYS_INLINE void checkGrantOption(const AccessRightsElement & element) const { access->checkGrantOption(context, element); }
+    ALWAYS_INLINE void checkGrantOption(const AccessRightsElements & elements) const { access->checkGrantOption(context, elements); }
+
+    /// Checks if a specified access is granted, and returns false if not.
+    /// Empty database means the current database.
+    ALWAYS_INLINE bool isGranted(const AccessFlags & flags) const { return access->isGranted(context, flags); }
+    ALWAYS_INLINE bool isGranted(const AccessFlags & flags, std::string_view database) const { return access->isGranted(context, flags, database); }
+    ALWAYS_INLINE bool isGranted(const AccessFlags & flags, std::string_view database, std::string_view table) const { return access->isGranted(context, flags, database, table); }
+    ALWAYS_INLINE bool isGranted(const AccessFlags & flags, std::string_view database, std::string_view table, std::string_view column) const { return access->isGranted(context, flags, database, table, column); }
+    ALWAYS_INLINE bool isGranted(const AccessFlags & flags, std::string_view database, std::string_view table, const std::vector<std::string_view> & columns) const { return access->isGranted(context, flags, database, table, columns); }
+    ALWAYS_INLINE bool isGranted(const AccessFlags & flags, std::string_view database, std::string_view table, const Strings & columns) const { return access->isGranted(context, flags, database, table, columns); }
+    ALWAYS_INLINE bool isGranted(const AccessRightsElement & element) const { return access->isGranted(context, element); }
+    ALWAYS_INLINE bool isGranted(const AccessRightsElements & elements) const { return access->isGranted(context, elements); }
+
+    ALWAYS_INLINE bool hasGrantOption(const AccessFlags & flags) const { return access->hasGrantOption(context, flags); }
+    ALWAYS_INLINE bool hasGrantOption(const AccessFlags & flags, std::string_view database) const { return access->hasGrantOption(context, flags, database); }
+    ALWAYS_INLINE bool hasGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table) const { return access->hasGrantOption(context, flags, database, table); }
+    ALWAYS_INLINE bool hasGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table, std::string_view column) const { return access->hasGrantOption(context, flags, database, table, column); }
+    ALWAYS_INLINE bool hasGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table, const std::vector<std::string_view> & columns) const { return access->hasGrantOption(context, flags, database, table, columns); }
+    ALWAYS_INLINE bool hasGrantOption(const AccessFlags & flags, std::string_view database, std::string_view table, const Strings & columns) const { return access->hasGrantOption(context, flags, database, table, columns); }
+    ALWAYS_INLINE bool hasGrantOption(const AccessRightsElement & element) const { return access->hasGrantOption(context, element); }
+    ALWAYS_INLINE bool hasGrantOption(const AccessRightsElements & elements) const { return access->hasGrantOption(context, elements); }
+
+    /// Checks if a specified role is granted with admin option, and throws an exception if not.
+    ALWAYS_INLINE void checkAdminOption(const UUID & role_id) const { access->checkAdminOption(context, role_id); }
+    ALWAYS_INLINE void checkAdminOption(const UUID & role_id, const String & role_name) const { access->checkAdminOption(context, role_id, role_name); }
+    ALWAYS_INLINE void checkAdminOption(const UUID & role_id, const std::unordered_map<UUID, String> & names_of_roles) const { access->checkAdminOption(context, role_id, names_of_roles); }
+    ALWAYS_INLINE void checkAdminOption(const std::vector<UUID> & role_ids) const { access->checkAdminOption(context, role_ids); }
+    ALWAYS_INLINE void checkAdminOption(const std::vector<UUID> & role_ids, const Strings & names_of_roles) const { access->checkAdminOption(context, role_ids, names_of_roles); }
+    ALWAYS_INLINE void checkAdminOption(const std::vector<UUID> & role_ids, const std::unordered_map<UUID, String> & names_of_roles) const { access->checkAdminOption(context, role_ids, names_of_roles); }
+
+    /// Checks if a specified role is granted with admin option, and returns false if not.
+    ALWAYS_INLINE bool hasAdminOption(const UUID & role_id) const { return access->hasAdminOption(context, role_id); }
+    ALWAYS_INLINE bool hasAdminOption(const UUID & role_id, const String & role_name) const { return access->hasAdminOption(context, role_id, role_name); }
+    ALWAYS_INLINE bool hasAdminOption(const UUID & role_id, const std::unordered_map<UUID, String> & names_of_roles) const { return access->hasAdminOption(context, role_id, names_of_roles); }
+    ALWAYS_INLINE bool hasAdminOption(const std::vector<UUID> & role_ids) const { return access->hasAdminOption(context, role_ids); }
+    ALWAYS_INLINE bool hasAdminOption(const std::vector<UUID> & role_ids, const Strings & names_of_roles) const { return access->hasAdminOption(context, role_ids, names_of_roles); }
+    ALWAYS_INLINE bool hasAdminOption(const std::vector<UUID> & role_ids, const std::unordered_map<UUID, String> & names_of_roles) const { return access->hasAdminOption(context, role_ids, names_of_roles); }
+
+    /// Checks if a grantee is allowed for the current user, throws an exception if not.
+    ALWAYS_INLINE void checkGranteeIsAllowed(const UUID & grantee_id, const IAccessEntity & grantee) const { access->checkGranteeIsAllowed(grantee_id, grantee); }
+    /// Checks if grantees are allowed for the current user, throws an exception if not.
+    ALWAYS_INLINE void checkGranteesAreAllowed(const std::vector<UUID> & grantee_ids) const { access->checkGranteesAreAllowed(grantee_ids); }
+
+private:
+    ContextAccessPtr access;
+    ContextPtr context;
+};
+
+
 }

src/Access/SettingsProfilesInfo.cpp

@@ -15,22 +15,8 @@ namespace ErrorCodes
 
 bool operator==(const SettingsProfilesInfo & lhs, const SettingsProfilesInfo & rhs)
 {
-    if (lhs.settings != rhs.settings)
-        return false;
-
-    if (lhs.constraints != rhs.constraints)
-        return false;
-
-    if (lhs.profiles != rhs.profiles)
-        return false;
-
-    if (lhs.profiles_with_implicit != rhs.profiles_with_implicit)
-        return false;
-
-    if (lhs.names_of_profiles != rhs.names_of_profiles)
-        return false;
-
-    return true;
+    return std::tie(lhs.settings, lhs.constraints, lhs.profiles, lhs.profiles_with_implicit, lhs.names_of_profiles)
+        == std::tie(rhs.settings, rhs.constraints, rhs.profiles, rhs.profiles_with_implicit, rhs.names_of_profiles);
 }
 
 std::shared_ptr<const SettingsConstraintsAndProfileIDs>
@@ -66,18 +52,20 @@ Strings SettingsProfilesInfo::getProfileNames() const
 {
     Strings result;
     result.reserve(profiles.size());
-    for (const auto & profile_id : profiles)
+    for (const UUID & profile_uuid : profiles)
     {
-        const auto p = names_of_profiles.find(profile_id);
-        if (p != names_of_profiles.end())
-            result.push_back(p->second);
+        const auto names_it = names_of_profiles.find(profile_uuid);
+        if (names_it != names_of_profiles.end())
+        {
+            result.push_back(names_it->second);
+        }
         else
         {
-            if (const auto name = access_control.tryReadName(profile_id))
+            if (const auto name = access_control.tryReadName(profile_uuid))
                 // We could've updated cache here, but it is a very rare case, so don't bother.
                 result.push_back(*name);
             else
-                throw Exception(ErrorCodes::LOGICAL_ERROR, "Unable to get profile name for {}", toString(profile_id));
+                throw Exception(ErrorCodes::LOGICAL_ERROR, "Unable to get profile name for {}", toString(profile_uuid));
         }
     }
 

src/Access/SettingsProfilesInfo.h

@@ -29,7 +29,11 @@ struct SettingsProfilesInfo
     /// Names of all the profiles in `profiles`.
     std::unordered_map<UUID, String> names_of_profiles;
 
-    explicit SettingsProfilesInfo(const AccessControl & access_control_) : constraints(access_control_), access_control(access_control_) {}
+    explicit SettingsProfilesInfo(const AccessControl & access_control_)
+        : constraints(access_control_), access_control(access_control_)
+    {
+    }
+
     std::shared_ptr<const SettingsConstraintsAndProfileIDs> getConstraintsAndProfileIDs(
         const std::shared_ptr<const SettingsConstraintsAndProfileIDs> & previous = nullptr) const;
 

src/AggregateFunctions/Combinators/AggregateFunctionDistinct.h

@@ -228,6 +228,11 @@ public:
         return prefix_size + nested_func->sizeOfData();
     }
 
+    size_t alignOfData() const override
+    {
+        return std::max(alignof(Data), nested_func->alignOfData());
+    }
+
     void create(AggregateDataPtr __restrict place) const override
     {
         new (place) Data;

src/Analyzer/Resolve/QueryAnalyzer.cpp

@@ -985,18 +985,18 @@ std::string QueryAnalyzer::rewriteAggregateFunctionNameIfNeeded(
     {
         result_aggregate_function_name = settings.count_distinct_implementation;
     }
-    else if (aggregate_function_name_lowercase == "countdistinctif" || aggregate_function_name_lowercase == "countifdistinct")
+    else if (aggregate_function_name_lowercase == "countifdistinct" ||
+        (settings.rewrite_count_distinct_if_with_count_distinct_implementation && aggregate_function_name_lowercase == "countdistinctif"))
     {
         result_aggregate_function_name = settings.count_distinct_implementation;
         result_aggregate_function_name += "If";
     }
-
-    /// Replace aggregateFunctionIfDistinct into aggregateFunctionDistinctIf to make execution more optimal
-    if (result_aggregate_function_name.ends_with("ifdistinct"))
+    else if (aggregate_function_name_lowercase.ends_with("ifdistinct"))
     {
+        /// Replace aggregateFunctionIfDistinct into aggregateFunctionDistinctIf to make execution more optimal
         size_t prefix_length = result_aggregate_function_name.size() - strlen("ifdistinct");
         result_aggregate_function_name = result_aggregate_function_name.substr(0, prefix_length) + "DistinctIf";
-   }
+    }
 
     bool need_add_or_null = settings.aggregate_functions_null_for_empty && !result_aggregate_function_name.ends_with("OrNull");
     if (need_add_or_null)

src/Analyzer/SetUtils.cpp

@@ -9,6 +9,8 @@
 #include <Interpreters/convertFieldToType.h>
 #include <Interpreters/Set.h>
 
+#include <Common/assert_cast.h>
+
 namespace DB
 {
 
@@ -54,8 +56,9 @@ size_t getCompoundTypeDepth(const IDataType & type)
 }
 
 template <typename Collection>
-Block createBlockFromCollection(const Collection & collection, const DataTypes & block_types, bool transform_null_in)
+Block createBlockFromCollection(const Collection & collection, const DataTypes& value_types, const DataTypes & block_types, bool transform_null_in)
 {
+    assert(collection.size() == value_types.size());
     size_t columns_size = block_types.size();
     MutableColumns columns(columns_size);
     for (size_t i = 0; i < columns_size; ++i)
@@ -66,13 +69,17 @@ Block createBlockFromCollection(const Collection & collection, const DataTypes &
 
     Row tuple_values;
 
-    for (const auto & value : collection)
+    for (size_t collection_index = 0; collection_index < collection.size(); ++collection_index)
     {
+        const auto & value = collection[collection_index];
         if (columns_size == 1)
         {
-            auto field = convertFieldToTypeStrict(value, *block_types[0]);
+            const DataTypePtr & data_type = value_types[collection_index];
+            auto field = convertFieldToTypeStrict(value, *data_type, *block_types[0]);
             if (!field)
+            {
                 continue;
+            }
 
             bool need_insert_null = transform_null_in && block_types[0]->isNullable();
             if (!field->isNull() || need_insert_null)
@@ -87,6 +94,9 @@ Block createBlockFromCollection(const Collection & collection, const DataTypes &
                 value.getTypeName());
 
         const auto & tuple = value.template get<const Tuple &>();
+        const DataTypePtr & value_type = value_types[collection_index];
+        const DataTypes & tuple_value_type = typeid_cast<const DataTypeTuple *>(value_type.get())->getElements();
+
         size_t tuple_size = tuple.size();
 
         if (tuple_size != columns_size)
@@ -101,7 +111,7 @@ Block createBlockFromCollection(const Collection & collection, const DataTypes &
         size_t i = 0;
         for (; i < tuple_size; ++i)
         {
-            auto converted_field = convertFieldToTypeStrict(tuple[i], *block_types[i]);
+            auto converted_field = convertFieldToTypeStrict(tuple[i], *tuple_value_type[i], *block_types[i]);
             if (!converted_field)
                 break;
             tuple_values[i] = std::move(*converted_field);
@@ -147,20 +157,28 @@ Block getSetElementsForConstantValue(const DataTypePtr & expression_type, const
     if (lhs_type_depth == rhs_type_depth)
     {
         /// 1 in 1; (1, 2) in (1, 2); identity(tuple(tuple(tuple(1)))) in tuple(tuple(tuple(1))); etc.
-
         Array array{value};
-        result_block = createBlockFromCollection(array, set_element_types, transform_null_in);
+        DataTypes value_types{value_type};
+        result_block = createBlockFromCollection(array, value_types, set_element_types, transform_null_in);
     }
     else if (lhs_type_depth + 1 == rhs_type_depth)
     {
         /// 1 in (1, 2); (1, 2) in ((1, 2), (3, 4))
-
         WhichDataType rhs_which_type(value_type);
 
         if (rhs_which_type.isArray())
-            result_block = createBlockFromCollection(value.get<const Array &>(), set_element_types, transform_null_in);
+        {
+            const DataTypeArray * value_array_type = assert_cast<const DataTypeArray *>(value_type.get());
+            size_t value_array_size = value.get<const Array &>().size();
+            DataTypes value_types(value_array_size, value_array_type->getNestedType());
+            result_block = createBlockFromCollection(value.get<const Array &>(), value_types, set_element_types, transform_null_in);
+        }
         else if (rhs_which_type.isTuple())
-            result_block = createBlockFromCollection(value.get<const Tuple &>(), set_element_types, transform_null_in);
+        {
+            const DataTypeTuple * value_tuple_type = assert_cast<const DataTypeTuple *>(value_type.get());
+            const DataTypes & value_types = value_tuple_type->getElements();
+            result_block = createBlockFromCollection(value.get<const Tuple &>(), value_types, set_element_types, transform_null_in);
+        }
         else
             throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT,
                 "Unsupported type at the right-side of IN. Expected Array or Tuple. Actual {}",

src/Client/ClientBase.cpp

@@ -44,13 +44,12 @@
 #include <Parsers/ASTIdentifier.h>
 #include <Parsers/ASTColumnDeclaration.h>
 #include <Parsers/ASTFunction.h>
-#include <Parsers/Kusto/ParserKQLStatement.h>
 #include <Parsers/PRQL/ParserPRQLQuery.h>
+#include <Parsers/Kusto/ParserKQLStatement.h>
 #include <Parsers/Kusto/parseKQLQuery.h>
 
 #include <Processors/Formats/Impl/NullFormat.h>
 #include <Processors/Formats/IInputFormat.h>
-#include <Processors/Formats/IOutputFormat.h>
 #include <Processors/QueryPlan/QueryPlan.h>
 #include <Processors/QueryPlan/BuildQueryPipelineSettings.h>
 #include <Processors/QueryPlan/Optimizations/QueryPlanOptimizationSettings.h>

src/Client/HedgedConnections.cpp

@@ -255,6 +255,17 @@ void HedgedConnections::sendCancel()
     if (!sent_query || cancelled)
         throw Exception(ErrorCodes::LOGICAL_ERROR, "Cannot cancel. Either no query sent or already cancelled.");
 
+    /// All hedged connections should be stopped, since otherwise before the
+    /// HedgedConnectionsFactory will be destroyed (that will happen from
+    /// QueryPipeline dtor) they could still do some work.
+    /// And not only this does not make sense, but it also could lead to
+    /// use-after-free of the current_thread, since the thread from which they
+    /// had been created differs from the thread where the dtor of
+    /// QueryPipeline will be called and the initial thread could be already
+    /// destroyed (especially when the system is under pressure).
+    if (hedged_connections_factory.hasEventsInProcess())
+        hedged_connections_factory.stopChoosingReplicas();
+
     cancelled = true;
 
     for (auto & offset_status : offset_states)

src/Common/ErrorCodes.cpp

@@ -602,6 +602,8 @@
     M(721, DEPRECATED_FUNCTION) \
     M(722, ASYNC_LOAD_WAIT_FAILED) \
     M(723, PARQUET_EXCEPTION) \
+    M(724, TOO_MANY_TABLES) \
+    M(725, TOO_MANY_DATABASES) \
     \
     M(900, DISTRIBUTED_CACHE_ERROR) \
     M(901, CANNOT_USE_DISTRIBUTED_CACHE) \

src/Common/HilbertUtils.h

@@ -0,0 +1,161 @@
+#pragma once
+
+#include <Core/Types.h>
+#include <Common/BitHelpers.h>
+#include "base/types.h"
+#include <Functions/hilbertDecode2DLUT.h>
+#include <base/defines.h>
+#include <array>
+#include <set>
+
+
+namespace HilbertDetails
+{
+
+    struct Segment // represents [begin; end], all bounds are included
+    {
+        UInt64 begin;
+        UInt64 end;
+    };
+
+}
+
+/*
+    Given the range of values of hilbert code - and this function will return segments of the Hilbert curve
+    such that each of them lies in a whole domain (aka square)
+           0                 1
+    ┌────────────────────────────────┐
+    │               │                │
+    │               │                │
+  0 │    00xxx      │     11xxx      │
+    │      |        │       |        │
+    │      |        │       |        │
+    │_______________│________________│
+    │      |        │       |        │
+    │      |        │       |        │
+    │      |        │       |        │
+  1 │    01xxx______│_____10xxx      │
+    │               │                │
+    │               │                │
+    └────────────────────────────────┘
+    Imagine a square, one side of which is a x-axis, other is a y-axis.
+    First approximation of the Hilbert curve is on the picture - U curve.
+    So we divide Hilbert Code Interval on 4 parts each of which is represented by a square
+    and look where the given interval [start, finish] is located:
+    [00xxxxxx      |      01xxxxxx      |      10xxxxxx      |      11xxxxxx      ]
+    1:     [                                                ]
+           start = 0010111                                  end = 10111110
+    2:     [       ]                    [                   ]
+    If it contains a whole sector (that represents a domain=square),
+    then we take this range. In the example above - it is a sector [01000000, 01111111]
+    Then we dig into the recursion and check the remaining ranges.
+    Note that after the first call all other ranges in the recursion will have either start or finish on the end of a range,
+    so the complexity of the algorithm will be O(logN), where N is the maximum of hilbert code.
+*/
+template <typename F>
+void segmentBinaryPartition(UInt64 start, UInt64 finish, UInt8 current_bits, F && callback)
+{
+    if (current_bits == 0)
+        return;
+
+    const auto next_bits = current_bits - 2;
+    const auto history = current_bits == 64 ? 0 : (start >> current_bits) << current_bits;
+
+    const auto chunk_mask = 0b11;
+    const auto start_chunk = (start >> next_bits) & chunk_mask;
+    const auto finish_chunk = (finish >> next_bits) & chunk_mask;
+
+    auto construct_range = [next_bits, history](UInt64 chunk)
+    {
+        return HilbertDetails::Segment{
+            .begin = history + (chunk << next_bits),
+            .end = history + ((chunk + 1) << next_bits) - 1
+        };
+    };
+
+    if (start_chunk == finish_chunk)
+    {
+        if ((finish - start + 1) == (1 << next_bits)) // it means that [begin, end] is a range
+        {
+            callback(HilbertDetails::Segment{.begin = start, .end = finish});
+            return;
+        }
+        segmentBinaryPartition(start, finish, next_bits, callback);
+        return;
+    }
+
+    for (auto range_chunk = start_chunk + 1; range_chunk < finish_chunk; ++range_chunk)
+    {
+        callback(construct_range(range_chunk));
+    }
+
+    const auto start_range = construct_range(start_chunk);
+    if (start == start_range.begin)
+    {
+        callback(start_range);
+    }
+    else
+    {
+        segmentBinaryPartition(start, start_range.end, next_bits, callback);
+    }
+
+    const auto finish_range = construct_range(finish_chunk);
+    if (finish == finish_range.end)
+    {
+        callback(finish_range);
+    }
+    else
+    {
+        segmentBinaryPartition(finish_range.begin, finish, next_bits, callback);
+    }
+}
+
+// Given 2 points representing ends of the range of Hilbert Curve that lies in a whole domain.
+// The are neighbour corners of some square - and the function returns ranges of both sides of this square
+inline std::array<std::pair<UInt64, UInt64>, 2> createRangeFromCorners(UInt64 x1, UInt64 y1, UInt64 x2, UInt64 y2)
+{
+    UInt64 dist_x = x1 > x2 ? x1 - x2 : x2 - x1;
+    UInt64 dist_y = y1 > y2 ? y1 - y2 : y2 - y1;
+    UInt64 range_size = std::max(dist_x, dist_y);
+    bool contains_minimum_vertice = x1 % (range_size + 1) == 0;
+    if (contains_minimum_vertice)
+    {
+        UInt64 x_min = std::min(x1, x2);
+        UInt64 y_min = std::min(y1, y2);
+        return {
+            std::pair<UInt64, UInt64>{x_min, x_min + range_size},
+            std::pair<UInt64, UInt64>{y_min, y_min + range_size}
+        };
+    }
+    else
+    {
+        UInt64 x_max = std::max(x1, x2);
+        UInt64 y_max = std::max(y1, y2);
+        chassert(x_max >= range_size);
+        chassert(y_max >= range_size);
+        return {
+            std::pair<UInt64, UInt64>{x_max - range_size, x_max},
+            std::pair<UInt64, UInt64>{y_max - range_size, y_max}
+        };
+    }
+}
+
+/** Unpack an interval of Hilbert curve to hyperrectangles covered by it across N dimensions.
+  */
+template <typename F>
+void hilbertIntervalToHyperrectangles2D(UInt64 first, UInt64 last, F && callback)
+{
+    const auto equal_bits_count = getLeadingZeroBits(last | first);
+    const auto even_equal_bits_count = equal_bits_count - equal_bits_count % 2;
+    segmentBinaryPartition(first, last, 64 - even_equal_bits_count, [&](HilbertDetails::Segment range)
+    {
+        auto interval1 = DB::FunctionHilbertDecode2DWIthLookupTableImpl<3>::decode(range.begin);
+        auto interval2 = DB::FunctionHilbertDecode2DWIthLookupTableImpl<3>::decode(range.end);
+
+        std::array<std::pair<UInt64, UInt64>, 2> unpacked_range = createRangeFromCorners(
+            std::get<0>(interval1), std::get<1>(interval1),
+            std::get<0>(interval2), std::get<1>(interval2));
+
+        callback(unpacked_range);
+    });
+}

src/Common/Scheduler/ISchedulerNode.h

@@ -11,10 +11,10 @@
 #include <Poco/Util/XMLConfiguration.h>
 
 #include <boost/noncopyable.hpp>
+#include <boost/intrusive/list.hpp>
 
 #include <chrono>
 #include <deque>
-#include <queue>
 #include <algorithm>
 #include <functional>
 #include <memory>
@@ -30,6 +30,8 @@ namespace ErrorCodes
 }
 
 class ISchedulerNode;
+class EventQueue;
+using EventId = UInt64;
 
 inline const Poco::Util::AbstractConfiguration & emptyConfig()
 {
@@ -82,6 +84,115 @@ struct SchedulerNodeInfo
     }
 };
 
+
+/*
+ * Node of hierarchy for scheduling requests for resource. Base class for all
+ * kinds of scheduling elements (queues, policies, constraints and schedulers).
+ *
+ * Root node is a scheduler, which has it's thread to dequeue requests,
+ * execute requests (see ResourceRequest) and process events in a thread-safe manner.
+ * Immediate children of the scheduler represent independent resources.
+ * Each resource has it's own hierarchy to achieve required scheduling policies.
+ * Non-leaf nodes do not hold requests, but keep scheduling state
+ * (e.g. consumption history, amount of in-flight requests, etc).
+ * Leafs of hierarchy are queues capable of holding pending requests.
+ *
+ *        scheduler         (SchedulerRoot)
+ *         /     \
+ *  constraint  constraint  (SemaphoreConstraint)
+ *      |           |
+ *   policy      policy     (PriorityPolicy)
+ *   /    \      /    \
+ *  q1    q2    q3    q4    (FifoQueue)
+ *
+ * Dequeueing request from an inner node will dequeue request from one of active leaf-queues in its subtree.
+ * Node is considered to be active iff:
+ *  - it has at least one pending request in one of leaves of it's subtree;
+ *  - and enforced constraints, if any, are satisfied
+ *    (e.g. amount of concurrent requests is not greater than some number).
+ *
+ * All methods must be called only from scheduler thread for thread-safety.
+ */
+class ISchedulerNode : public boost::intrusive::list_base_hook<>, private boost::noncopyable
+{
+public:
+    explicit ISchedulerNode(EventQueue * event_queue_, const Poco::Util::AbstractConfiguration & config = emptyConfig(), const String & config_prefix = {})
+        : event_queue(event_queue_)
+        , info(config, config_prefix)
+    {}
+
+    virtual ~ISchedulerNode() = default;
+
+    /// Checks if two nodes configuration is equal
+    virtual bool equals(ISchedulerNode * other)
+    {
+        return info.equals(other->info);
+    }
+
+    /// Attach new child
+    virtual void attachChild(const std::shared_ptr<ISchedulerNode> & child) = 0;
+
+    /// Detach and destroy child
+    virtual void removeChild(ISchedulerNode * child) = 0;
+
+    /// Get attached child by name
+    virtual ISchedulerNode * getChild(const String & child_name) = 0;
+
+    /// Activation of child due to the first pending request
+    /// Should be called on leaf node (i.e. queue) to propagate activation signal through chain to the root
+    virtual void activateChild(ISchedulerNode * child) = 0;
+
+    /// Returns true iff node is active
+    virtual bool isActive() = 0;
+
+    /// Returns number of active children
+    virtual size_t activeChildren() = 0;
+
+    /// Returns the first request to be executed as the first component of resulting pair.
+    /// The second pair component is `true` iff node is still active after dequeueing.
+    virtual std::pair<ResourceRequest *, bool> dequeueRequest() = 0;
+
+    /// Returns full path string using names of every parent
+    String getPath()
+    {
+        String result;
+        ISchedulerNode * ptr = this;
+        while (ptr->parent)
+        {
+            result = "/" + ptr->basename + result;
+            ptr = ptr->parent;
+        }
+        return result.empty() ? "/" : result;
+    }
+
+    /// Attach to a parent (used by attachChild)
+    virtual void setParent(ISchedulerNode * parent_)
+    {
+        parent = parent_;
+    }
+
+protected:
+    /// Notify parents about the first pending request or constraint becoming satisfied.
+    /// Postponed to be handled in scheduler thread, so it is intended to be called from outside.
+    void scheduleActivation();
+
+public:
+    EventQueue * const event_queue;
+    String basename;
+    SchedulerNodeInfo info;
+    ISchedulerNode * parent = nullptr;
+    EventId activation_event_id = 0; // Valid for `ISchedulerNode` placed in EventQueue::activations
+
+    /// Introspection
+    std::atomic<UInt64> dequeued_requests{0};
+    std::atomic<UInt64> canceled_requests{0};
+    std::atomic<ResourceCost> dequeued_cost{0};
+    std::atomic<ResourceCost> canceled_cost{0};
+    std::atomic<UInt64> busy_periods{0};
+};
+
+using SchedulerNodePtr = std::shared_ptr<ISchedulerNode>;
+
 /*
  * Simple waitable thread-safe FIFO task queue.
  * Intended to hold postponed events for later handling (usually by scheduler thread).
@@ -89,57 +200,70 @@ struct SchedulerNodeInfo
 class EventQueue
 {
 public:
-    using Event = std::function<void()>;
+    using Task = std::function<void()>;
+
+    static constexpr EventId not_postponed = 0;
+
     using TimePoint = std::chrono::system_clock::time_point;
     using Duration = std::chrono::system_clock::duration;
-    static constexpr UInt64 not_postponed = 0;
+
+    struct Event
+    {
+        const EventId event_id;
+        Task task;
+
+        Event(EventId event_id_, Task && task_)
+            : event_id(event_id_)
+            , task(std::move(task_))
+        {}
+    };
 
     struct Postponed
     {
         TimePoint key;
-        UInt64 id; // for canceling
-        std::unique_ptr<Event> event;
+        EventId event_id; // for canceling
+        std::unique_ptr<Task> task;
 
-        Postponed(TimePoint key_, UInt64 id_, Event && event_)
+        Postponed(TimePoint key_, EventId event_id_, Task && task_)
             : key(key_)
-            , id(id_)
-            , event(std::make_unique<Event>(std::move(event_)))
+            , event_id(event_id_)
+            , task(std::make_unique<Task>(std::move(task_)))
         {}
 
         bool operator<(const Postponed & rhs) const
         {
-            return std::tie(key, id) > std::tie(rhs.key, rhs.id); // reversed for min-heap
+            return std::tie(key, event_id) > std::tie(rhs.key, rhs.event_id); // reversed for min-heap
         }
     };
 
     /// Add an `event` to be processed after `until` time point.
-    /// Returns a unique id for canceling.
-    [[nodiscard]] UInt64 postpone(TimePoint until, Event && event)
+    /// Returns a unique event id for canceling.
+    [[nodiscard]] EventId postpone(TimePoint until, Task && task)
     {
         std::unique_lock lock{mutex};
         if (postponed.empty() || until < postponed.front().key)
             pending.notify_one();
-        auto id = ++last_id;
-        postponed.emplace_back(until, id, std::move(event));
+        auto event_id = ++last_event_id;
+        postponed.emplace_back(until, event_id, std::move(task));
         std::push_heap(postponed.begin(), postponed.end());
-        return id;
+        return event_id;
     }
 
     /// Cancel a postponed event using its unique id.
     /// NOTE: Only postponed events can be canceled.
     /// NOTE: If you need to cancel enqueued event, consider doing your actions inside another enqueued
     /// NOTE: event instead. This ensures that all previous events are processed.
-    bool cancelPostponed(UInt64 postponed_id)
+    bool cancelPostponed(EventId postponed_event_id)
     {
-        if (postponed_id == not_postponed)
+        if (postponed_event_id == not_postponed)
             return false;
         std::unique_lock lock{mutex};
         for (auto i = postponed.begin(), e = postponed.end(); i != e; ++i)
         {
-            if (i->id == postponed_id)
+            if (i->event_id == postponed_event_id)
             {
                 postponed.erase(i);
-                // It is O(n), but we do not expect either big heaps or frequent cancels. So it is fine.
+                // It is O(n), but we do not expect neither big heaps nor frequent cancels. So it is fine.
                 std::make_heap(postponed.begin(), postponed.end());
                 return true;
             }
@@ -148,11 +272,23 @@ public:
     }
 
     /// Add an `event` for immediate processing
-    void enqueue(Event && event)
+    void enqueue(Task && task)
     {
         std::unique_lock lock{mutex};
-        bool was_empty = queue.empty();
-        queue.emplace_back(event);
+        bool was_empty = events.empty() && activations.empty();
+        auto event_id = ++last_event_id;
+        events.emplace_back(event_id, std::move(task));
+        if (was_empty)
+            pending.notify_one();
+    }
+
+    /// Add an activation `event` for immediate processing. Activations use a separate queue for performance reasons.
+    void enqueueActivation(ISchedulerNode * node)
+    {
+        std::unique_lock lock{mutex};
+        bool was_empty = events.empty() && activations.empty();
+        node->activation_event_id = ++last_event_id;
+        activations.push_back(*node);
         if (was_empty)
             pending.notify_one();
     }
@@ -163,7 +299,7 @@ public:
     bool forceProcess()
     {
         std::unique_lock lock{mutex};
-        if (!queue.empty())
+        if (!events.empty() || !activations.empty())
         {
             processQueue(std::move(lock));
             return true;
@@ -181,7 +317,7 @@ public:
     bool tryProcess()
     {
         std::unique_lock lock{mutex};
-        if (!queue.empty())
+        if (!events.empty() || !activations.empty())
         {
             processQueue(std::move(lock));
             return true;
@@ -205,7 +341,7 @@ public:
         std::unique_lock lock{mutex};
         while (true)
         {
-            if (!queue.empty())
+            if (!events.empty() || !activations.empty())
             {
                 processQueue(std::move(lock));
                 return;
@@ -269,141 +405,69 @@ private:
 
     void processQueue(std::unique_lock<std::mutex> && lock)
     {
-        Event event = std::move(queue.front());
-        queue.pop_front();
+        if (events.empty())
+        {
+            processActivation(std::move(lock));
+            return;
+        }
+        if (activations.empty())
+        {
+            processEvent(std::move(lock));
+            return;
+        }
+        if (activations.front().activation_event_id < events.front().event_id)
+            processActivation(std::move(lock));
+        else
+            processEvent(std::move(lock));
+    }
+
+    void processActivation(std::unique_lock<std::mutex> && lock)
+    {
+        ISchedulerNode * node = &activations.front();
+        activations.pop_front();
+        node->activation_event_id = 0;
         lock.unlock(); // do not hold queue mutex while processing events
-        event();
+        node->parent->activateChild(node);
+    }
+
+    void processEvent(std::unique_lock<std::mutex> && lock)
+    {
+        Task task = std::move(events.front().task);
+        events.pop_front();
+        lock.unlock(); // do not hold queue mutex while processing events
+        task();
     }
 
     void processPostponed(std::unique_lock<std::mutex> && lock)
     {
-        Event event = std::move(*postponed.front().event);
+        Task task = std::move(*postponed.front().task);
         std::pop_heap(postponed.begin(), postponed.end());
         postponed.pop_back();
         lock.unlock(); // do not hold queue mutex while processing events
-        event();
+        task();
     }
 
     std::mutex mutex;
     std::condition_variable pending;
-    std::deque<Event> queue;
+
+    // `events` and `activations` logically represent one ordered queue. To preserve the common order we use `EventId`
+    // Activations are stored in a separate queue for performance reasons (mostly to avoid any allocations)
+    std::deque<Event> events;
+    boost::intrusive::list<ISchedulerNode> activations;
+
     std::vector<Postponed> postponed;
-    UInt64 last_id = 0;
+    EventId last_event_id = 0;
 
     std::atomic<TimePoint> manual_time{TimePoint()}; // for tests only
 };
 
-/*
- * Node of hierarchy for scheduling requests for resource. Base class for all
- * kinds of scheduling elements (queues, policies, constraints and schedulers).
- *
- * Root node is a scheduler, which has it's thread to dequeue requests,
- * execute requests (see ResourceRequest) and process events in a thread-safe manner.
- * Immediate children of the scheduler represent independent resources.
- * Each resource has it's own hierarchy to achieve required scheduling policies.
- * Non-leaf nodes do not hold requests, but keep scheduling state
- * (e.g. consumption history, amount of in-flight requests, etc).
- * Leafs of hierarchy are queues capable of holding pending requests.
- *
- *        scheduler         (SchedulerRoot)
- *         /     \
- *  constraint  constraint  (SemaphoreConstraint)
- *      |           |
- *   policy      policy     (PriorityPolicy)
- *   /    \      /    \
- *  q1    q2    q3    q4    (FifoQueue)
- *
- * Dequeueing request from an inner node will dequeue request from one of active leaf-queues in its subtree.
- * Node is considered to be active iff:
- *  - it has at least one pending request in one of leaves of it's subtree;
- *  - and enforced constraints, if any, are satisfied
- *    (e.g. amount of concurrent requests is not greater than some number).
- *
- * All methods must be called only from scheduler thread for thread-safety.
- */
-class ISchedulerNode : private boost::noncopyable
+inline void ISchedulerNode::scheduleActivation()
 {
-public:
-    explicit ISchedulerNode(EventQueue * event_queue_, const Poco::Util::AbstractConfiguration & config = emptyConfig(), const String & config_prefix = {})
-        : event_queue(event_queue_)
-        , info(config, config_prefix)
-    {}
-
-    virtual ~ISchedulerNode() = default;
-
-    /// Checks if two nodes configuration is equal
-    virtual bool equals(ISchedulerNode * other)
+    if (likely(parent))
     {
-        return info.equals(other->info);
+        // The same as `enqueue([this] { parent->activateChild(this); });` but faster
+        event_queue->enqueueActivation(this);
     }
-
-    /// Attach new child
-    virtual void attachChild(const std::shared_ptr<ISchedulerNode> & child) = 0;
-
-    /// Detach and destroy child
-    virtual void removeChild(ISchedulerNode * child) = 0;
-
-    /// Get attached child by name
-    virtual ISchedulerNode * getChild(const String & child_name) = 0;
-
-    /// Activation of child due to the first pending request
-    /// Should be called on leaf node (i.e. queue) to propagate activation signal through chain to the root
-    virtual void activateChild(ISchedulerNode * child) = 0;
-
-    /// Returns true iff node is active
-    virtual bool isActive() = 0;
-
-    /// Returns number of active children
-    virtual size_t activeChildren() = 0;
-
-    /// Returns the first request to be executed as the first component of resulting pair.
-    /// The second pair component is `true` iff node is still active after dequeueing.
-    virtual std::pair<ResourceRequest *, bool> dequeueRequest() = 0;
-
-    /// Returns full path string using names of every parent
-    String getPath()
-    {
-        String result;
-        ISchedulerNode * ptr = this;
-        while (ptr->parent)
-        {
-            result = "/" + ptr->basename + result;
-            ptr = ptr->parent;
-        }
-        return result.empty() ? "/" : result;
-    }
-
-    /// Attach to a parent (used by attachChild)
-    virtual void setParent(ISchedulerNode * parent_)
-    {
-        parent = parent_;
-    }
-
-protected:
-    /// Notify parents about the first pending request or constraint becoming satisfied.
-    /// Postponed to be handled in scheduler thread, so it is intended to be called from outside.
-    void scheduleActivation()
-    {
-        if (likely(parent))
-        {
-            event_queue->enqueue([this] { parent->activateChild(this); });
-        }
-    }
-
-public:
-    EventQueue * const event_queue;
-    String basename;
-    SchedulerNodeInfo info;
-    ISchedulerNode * parent = nullptr;
-
-    /// Introspection
-    std::atomic<UInt64> dequeued_requests{0};
-    std::atomic<UInt64> canceled_requests{0};
-    std::atomic<ResourceCost> dequeued_cost{0};
-    std::atomic<ResourceCost> canceled_cost{0};
-    std::atomic<UInt64> busy_periods{0};
-};
-
-using SchedulerNodePtr = std::shared_ptr<ISchedulerNode>;
+}
 
 }

src/Common/Scheduler/Nodes/tests/gtest_event_queue.cpp

@@ -0,0 +1,143 @@
+#include <chrono>
+#include <gtest/gtest.h>
+
+#include <Common/Scheduler/ISchedulerNode.h>
+
+using namespace DB;
+
+class FakeSchedulerNode : public ISchedulerNode
+{
+public:
+    explicit FakeSchedulerNode(String & log_, EventQueue * event_queue_, const Poco::Util::AbstractConfiguration & config = emptyConfig(), const String & config_prefix = {})
+        : ISchedulerNode(event_queue_, config, config_prefix)
+        , log(log_)
+    {}
+
+    void attachChild(const SchedulerNodePtr & child) override
+    {
+        log += " +" + child->basename;
+    }
+
+    void removeChild(ISchedulerNode * child) override
+    {
+        log += " -" + child->basename;
+    }
+
+    ISchedulerNode * getChild(const String & /* child_name */) override
+    {
+        return nullptr;
+    }
+
+    void activateChild(ISchedulerNode * child) override
+    {
+        log += " A" + child->basename;
+    }
+
+    bool isActive() override
+    {
+        return false;
+    }
+
+    size_t activeChildren() override
+    {
+        return 0;
+    }
+
+    std::pair<ResourceRequest *, bool> dequeueRequest() override
+    {
+        log += " D";
+        return {nullptr, false};
+    }
+
+private:
+    String & log;
+};
+
+struct QueueTest {
+    String log;
+    EventQueue event_queue;
+    FakeSchedulerNode root_node;
+
+    QueueTest()
+        : root_node(log, &event_queue)
+    {}
+
+    SchedulerNodePtr makeNode(const String & name)
+    {
+        auto node = std::make_shared<FakeSchedulerNode>(log, &event_queue);
+        node->basename = name;
+        node->setParent(&root_node);
+        return std::static_pointer_cast<ISchedulerNode>(node);
+    }
+
+    void process(EventQueue::TimePoint now, const String & expected_log, size_t limit = size_t(-1))
+    {
+        event_queue.setManualTime(now);
+        for (;limit > 0; limit--)
+        {
+            if (!event_queue.tryProcess())
+                break;
+        }
+        EXPECT_EQ(log, expected_log);
+        log.clear();
+    }
+
+    void activate(const SchedulerNodePtr & node)
+    {
+        event_queue.enqueueActivation(node.get());
+    }
+
+    void event(const String & text)
+    {
+        event_queue.enqueue([this, text] { log += " " + text; });
+    }
+
+    EventId postpone(EventQueue::TimePoint until, const String & text)
+    {
+        return event_queue.postpone(until, [this, text] { log += " " + text; });
+    }
+
+    void cancel(EventId event_id)
+    {
+        event_queue.cancelPostponed(event_id);
+    }
+};
+
+TEST(SchedulerEventQueue, Smoke)
+{
+    QueueTest t;
+
+    using namespace std::chrono_literals;
+
+    EventQueue::TimePoint start = std::chrono::system_clock::now();
+    t.process(start, "", 0);
+
+    // Activations
+    auto node1 = t.makeNode("1");
+    auto node2 = t.makeNode("2");
+    t.activate(node2);
+    t.activate(node1);
+    t.process(start + 42s, " A2 A1");
+
+    // Events
+    t.event("E1");
+    t.event("E2");
+    t.process(start + 100s, " E1 E2");
+
+    // Postponed events
+    t.postpone(start + 200s, "P200");
+    auto p190 = t.postpone(start + 200s, "P190");
+    t.postpone(start + 150s, "P150");
+    t.postpone(start + 175s, "P175");
+    t.process(start + 180s, " P150 P175");
+    t.event("E3");
+    t.cancel(p190);
+    t.process(start + 300s, " E3 P200");
+
+    // Ordering of events and activations
+    t.event("E1");
+    t.activate(node1);
+    t.event("E2");
+    t.activate(node2);
+    t.process(start + 300s, " E1 A1 E2 A2");
+}

src/Common/Scheduler/Nodes/tests/gtest_throttler_constraint.cpp

@@ -5,8 +5,6 @@
 
 #include <Common/Scheduler/Nodes/FairPolicy.h>
 #include <Common/Scheduler/Nodes/ThrottlerConstraint.h>
-#include "Common/Scheduler/ISchedulerNode.h"
-#include "Common/Scheduler/ResourceRequest.h"
 
 using namespace DB;
 

src/Coordination/KeeperStorage.cpp

@@ -609,7 +609,10 @@ void KeeperStorage::UncommittedState::commit(int64_t commit_zxid)
             uncommitted_auth.pop_front();
             if (uncommitted_auth.empty())
                 session_and_auth.erase(add_auth->session_id);
-
+        }
+        else if (auto * close_session = std::get_if<CloseSessionDelta>(&front_delta.operation))
+        {
+            closed_sessions.erase(close_session->session_id);
         }
 
         deltas.pop_front();
@@ -682,6 +685,10 @@ void KeeperStorage::UncommittedState::rollback(int64_t rollback_zxid)
                     session_and_auth.erase(add_auth->session_id);
             }
         }
+        else if (auto * close_session = std::get_if<CloseSessionDelta>(&delta_it->operation))
+        {
+           closed_sessions.erase(close_session->session_id);
+        }
     }
 
     if (delta_it == deltas.rend())
@@ -878,6 +885,10 @@ Coordination::Error KeeperStorage::commit(int64_t commit_zxid)
                     session_and_auth[operation.session_id].emplace_back(std::move(operation.auth_id));
                     return Coordination::Error::ZOK;
                 }
+                else if constexpr (std::same_as<DeltaType, KeeperStorage::CloseSessionDelta>)
+                {
+                    return Coordination::Error::ZOK;
+                }
                 else
                 {
                     // shouldn't be called in any process functions
@@ -2366,12 +2377,15 @@ void KeeperStorage::preprocessRequest(
 
             ephemerals.erase(session_ephemerals);
         }
+        new_deltas.emplace_back(transaction.zxid, CloseSessionDelta{session_id});
+        uncommitted_state.closed_sessions.insert(session_id);
 
         new_digest = calculateNodesDigest(new_digest, new_deltas);
         return;
     }
 
-    if (check_acl && !request_processor->checkAuth(*this, session_id, false))
+    if ((check_acl && !request_processor->checkAuth(*this, session_id, false)) ||
+        uncommitted_state.closed_sessions.contains(session_id))  // Is session closed but not committed yet
     {
         uncommitted_state.deltas.emplace_back(new_last_zxid, Coordination::Error::ZNOAUTH);
         return;

src/Coordination/KeeperStorage.h

@@ -314,8 +314,13 @@ public:
         AuthID auth_id;
     };
 
+    struct CloseSessionDelta
+    {
+        int64_t session_id;
+    };
+
     using Operation = std::
-        variant<CreateNodeDelta, RemoveNodeDelta, UpdateNodeDelta, SetACLDelta, AddAuthDelta, ErrorDelta, SubDeltaEnd, FailedMultiDelta>;
+        variant<CreateNodeDelta, RemoveNodeDelta, UpdateNodeDelta, SetACLDelta, AddAuthDelta, ErrorDelta, SubDeltaEnd, FailedMultiDelta, CloseSessionDelta>;
 
     struct Delta
     {
@@ -351,6 +356,7 @@ public:
         std::shared_ptr<Node> tryGetNodeFromStorage(StringRef path) const;
 
         std::unordered_map<int64_t, std::list<const AuthID *>> session_and_auth;
+        std::unordered_set<int64_t> closed_sessions;
 
         struct UncommittedNode
         {

src/Coordination/tests/gtest_coordination.cpp

@@ -2019,6 +2019,67 @@ TEST_P(CoordinationTest, TestCreateNodeWithAuthSchemeForAclWhenAuthIsPrecommitte
     EXPECT_EQ(acls[0].permissions, 31);
 }
 
+TEST_P(CoordinationTest, TestPreprocessWhenCloseSessionIsPrecommitted)
+{
+    using namespace Coordination;
+    using namespace DB;
+
+    ChangelogDirTest snapshots("./snapshots");
+    setSnapshotDirectory("./snapshots");
+    ResponsesQueue queue(std::numeric_limits<size_t>::max());
+    SnapshotsQueue snapshots_queue{1};
+    int64_t session_id = 1;
+    size_t term = 0;
+
+    auto state_machine = std::make_shared<KeeperStateMachine>(queue, snapshots_queue, keeper_context, nullptr);
+    state_machine->init();
+
+    auto & storage = state_machine->getStorageUnsafe();    
+    const auto & uncommitted_state = storage.uncommitted_state;    
+
+    // Create first node for the session
+    String node_path_1 = "/node_1";
+    std::shared_ptr<ZooKeeperCreateRequest> create_req_1 = std::make_shared<ZooKeeperCreateRequest>();
+    create_req_1->path = node_path_1;
+    auto create_entry_1 = getLogEntryFromZKRequest(term, session_id, state_machine->getNextZxid(), create_req_1);
+
+    state_machine->pre_commit(1, create_entry_1->get_buf());    
+    EXPECT_TRUE(uncommitted_state.nodes.contains(node_path_1));
+
+    state_machine->commit(1, create_entry_1->get_buf());
+    EXPECT_TRUE(storage.container.contains(node_path_1));
+
+    // Close session
+    std::shared_ptr<ZooKeeperCloseRequest> close_req = std::make_shared<ZooKeeperCloseRequest>();
+    auto close_entry = getLogEntryFromZKRequest(term, session_id, state_machine->getNextZxid(), close_req);
+    // Pre-commit close session
+    state_machine->pre_commit(2, close_entry->get_buf());
+
+    // Try to create second node after close session is pre-committed
+    String node_path_2 = "/node_2";
+    std::shared_ptr<ZooKeeperCreateRequest> create_req_2 = std::make_shared<ZooKeeperCreateRequest>();
+    create_req_2->path = node_path_2;
+    auto create_entry_2 = getLogEntryFromZKRequest(term, session_id, state_machine->getNextZxid(), create_req_2);
+    
+    // Pre-commit creating second node
+    state_machine->pre_commit(3, create_entry_2->get_buf());
+    // Second node wasn't created
+    EXPECT_FALSE(uncommitted_state.nodes.contains(node_path_2));
+
+    // Rollback pre-committed closing session
+    state_machine->rollback(3, create_entry_2->get_buf());
+    state_machine->rollback(2, close_entry->get_buf());
+
+    // Pre-commit creating second node
+    state_machine->pre_commit(2, create_entry_2->get_buf());
+    // Now second node was created
+    EXPECT_TRUE(uncommitted_state.nodes.contains(node_path_2));
+
+    state_machine->commit(2, create_entry_2->get_buf());
+    EXPECT_TRUE(storage.container.contains(node_path_1));
+    EXPECT_TRUE(storage.container.contains(node_path_2));
+}
+
 TEST_P(CoordinationTest, TestSetACLWithAuthSchemeForAclWhenAuthIsPrecommitted)
 {
     using namespace Coordination;

src/Core/ServerSettings.h

@@ -102,6 +102,8 @@ namespace DB
     M(UInt64, max_dictionary_num_to_warn, 1000lu, "If the number of dictionaries is greater than this value, the server will create a warning that will displayed to user.", 0) \
     M(UInt64, max_database_num_to_warn, 1000lu, "If the number of databases is greater than this value, the server will create a warning that will displayed to user.", 0) \
     M(UInt64, max_part_num_to_warn, 100000lu, "If the number of parts is greater than this value, the server will create a warning that will displayed to user.", 0) \
+    M(UInt64, max_table_num_to_throw, 0lu, "If number of tables is greater than this value, server will throw an exception. 0 means no limitation. View, remote tables, dictionary, system tables are not counted. Only count table in Atomic/Ordinary/Replicated/Lazy database engine.", 0) \
+    M(UInt64, max_database_num_to_throw, 0lu, "If number of databases is greater than this value, server will throw an exception. 0 means no limitation.", 0) \
     M(UInt64, concurrent_threads_soft_limit_num, 0, "Sets how many concurrent thread can be allocated before applying CPU pressure. Zero means unlimited.", 0) \
     M(UInt64, concurrent_threads_soft_limit_ratio_to_cores, 0, "Same as concurrent_threads_soft_limit_num, but with ratio to cores.", 0) \
     \
@@ -146,6 +148,8 @@ namespace DB
     M(UInt64, global_profiler_real_time_period_ns, 0, "Period for real clock timer of global profiler (in nanoseconds). Set 0 value to turn off the real clock global profiler. Recommended value is at least 10000000 (100 times a second) for single queries or 1000000000 (once a second) for cluster-wide profiling.", 0) \
     M(UInt64, global_profiler_cpu_time_period_ns, 0, "Period for CPU clock timer of global profiler (in nanoseconds). Set 0 value to turn off the CPU clock global profiler. Recommended value is at least 10000000 (100 times a second) for single queries or 1000000000 (once a second) for cluster-wide profiling.", 0) \
     M(Bool, enable_azure_sdk_logging, false, "Enables logging from Azure sdk", 0) \
+    M(String, merge_workload, "default", "Name of workload to be used to access resources for all merges (may be overridden by a merge tree setting)", 0) \
+    M(String, mutation_workload, "default", "Name of workload to be used to access resources for all mutations (may be overridden by a merge tree setting)", 0) \
     M(Double, gwp_asan_force_sample_probability, 0, "Probability that an allocation from specific places will be sampled by GWP Asan (i.e. PODArray allocations)", 0) \
 
 /// If you add a setting which can be updated at runtime, please update 'changeable_settings' map in StorageSystemServerSettings.cpp

src/Core/Settings.h

@@ -933,6 +933,7 @@ class IColumn;
     M(Int64, prefer_warmed_unmerged_parts_seconds, 0, "Only available in ClickHouse Cloud. If a merged part is less than this many seconds old and is not pre-warmed (see cache_populated_by_fetch), but all its source parts are available and pre-warmed, SELECT queries will read from those parts instead. Only for ReplicatedMergeTree. Note that this only checks whether CacheWarmer processed the part; if the part was fetched into cache by something else, it'll still be considered cold until CacheWarmer gets to it; if it was warmed, then evicted from cache, it'll still be considered warm.", 0) \
     M(Bool, iceberg_engine_ignore_schema_evolution, false, "Ignore schema evolution in Iceberg table engine and read all data using latest schema saved on table creation. Note that it can lead to incorrect result", 0) \
     M(Bool, allow_deprecated_error_prone_window_functions, false, "Allow usage of deprecated error prone window functions (neighbor, runningAccumulate, runningDifferenceStartingWithFirstValue, runningDifference)", 0) \
+    M(Bool, uniform_snowflake_conversion_functions, true, "Enables functions snowflakeIDToDateTime[64] and dateTime[64]ToSnowflakeID while disabling functions snowflakeToDateTime[64] and dateTime[64]ToSnowflake.", 0) \
 
 // End of COMMON_SETTINGS
 // Please add settings related to formats into the FORMAT_FACTORY_SETTINGS, move obsolete settings to OBSOLETE_SETTINGS and obsolete format settings to OBSOLETE_FORMAT_SETTINGS.

src/Core/SettingsChangesHistory.h

@@ -101,6 +101,7 @@ static const std::map<ClickHouseVersion, SettingsChangesHistory::SettingsChanges
               {"input_format_parquet_max_block_size", 8192, DEFAULT_BLOCK_SIZE, "Increase block size for parquet reader."},
               {"input_format_parquet_prefer_block_bytes", 0, DEFAULT_BLOCK_SIZE * 256, "Average block bytes output by parquet reader."},
               {"enable_blob_storage_log", true, true, "Write information about blob storage operations to system.blob_storage_log table"},
+              {"uniform_snowflake_conversion_functions", false, true, "Enable functions snowflakeIDToDateTime[64] and dateTime[64]ToSnowflakeID."},
               {"allow_statistic_optimize", false, false, "Old setting which popped up here being renamed."},
               {"allow_experimental_statistic", false, false, "Old setting which popped up here being renamed."},
               {"allow_statistics_optimize", false, false, "The setting was renamed. The previous name is `allow_statistic_optimize`."},

src/Databases/DatabaseLazy.cpp

@@ -186,6 +186,7 @@ void DatabaseLazy::attachTable(ContextPtr /* context_ */, const String & table_n
         throw Exception(ErrorCodes::TABLE_ALREADY_EXISTS, "Table {}.{} already exists.", backQuote(database_name), backQuote(table_name));
 
     it->second.expiration_iterator = cache_expiration_queue.emplace(cache_expiration_queue.end(), current_time, table_name);
+
     CurrentMetrics::add(CurrentMetrics::AttachedTable, 1);
 }
 
@@ -202,6 +203,7 @@ StoragePtr DatabaseLazy::detachTable(ContextPtr /* context */, const String & ta
         if (it->second.expiration_iterator != cache_expiration_queue.end())
             cache_expiration_queue.erase(it->second.expiration_iterator);
         tables_cache.erase(it);
+
         CurrentMetrics::sub(CurrentMetrics::AttachedTable, 1);
     }
     return res;

src/Databases/DatabasesCommon.cpp

@@ -260,7 +260,9 @@ StoragePtr DatabaseWithOwnTablesBase::detachTableUnlocked(const String & table_n
     res = it->second;
     tables.erase(it);
     res->is_detached = true;
-    CurrentMetrics::sub(getAttachedCounterForStorage(res), 1);
+
+    if (res->isSystemStorage() == false)
+        CurrentMetrics::sub(getAttachedCounterForStorage(res), 1);
 
     auto table_id = res->getStorageID();
     if (table_id.hasUUID())
@@ -301,7 +303,9 @@ void DatabaseWithOwnTablesBase::attachTableUnlocked(const String & table_name, c
     /// It is important to reset is_detached here since in case of RENAME in
     /// non-Atomic database the is_detached is set to true before RENAME.
     table->is_detached = false;
-    CurrentMetrics::add(getAttachedCounterForStorage(table), 1);
+
+    if (table->isSystemStorage() == false && table_id.database_name != DatabaseCatalog::SYSTEM_DATABASE)
+        CurrentMetrics::add(getAttachedCounterForStorage(table), 1);
 }
 
 void DatabaseWithOwnTablesBase::shutdown()

src/Functions/FunctionsStringDistance.cpp

@@ -113,6 +113,36 @@ struct ByteHammingDistanceImpl
     }
 };
 
+void parseUTF8String(const char * __restrict data, size_t size, std::function<void(UInt32)> utf8_consumer, std::function<void(unsigned char)> ascii_consumer = nullptr)
+{
+    const char * end = data + size;
+    while (data < end)
+    {
+        size_t len = UTF8::seqLength(*data);
+        if (len == 1)
+        {
+            if (ascii_consumer)
+                ascii_consumer(static_cast<unsigned char>(*data));
+            else
+                utf8_consumer(static_cast<UInt32>(*data));
+            ++data;
+        }
+        else
+        {
+            auto code_point = UTF8::convertUTF8ToCodePoint(data, end - data);
+            if (code_point.has_value())
+            {
+                utf8_consumer(code_point.value());
+                data += len;
+            }
+            else
+            {
+                throw Exception(ErrorCodes::BAD_ARGUMENTS, "Illegal UTF-8 sequence, while processing '{}'", StringRef(data, end - data));
+            }
+        }
+    }
+}
+
 template <bool is_utf8>
 struct ByteJaccardIndexImpl
 {
@@ -138,57 +168,28 @@ struct ByteJaccardIndexImpl
         haystack_set.fill(0);
         needle_set.fill(0);
 
-        while (haystack < haystack_end)
+        if constexpr (is_utf8)
         {
-            size_t len = 1;
-            if constexpr (is_utf8)
-                len = UTF8::seqLength(*haystack);
-
-            if (len == 1)
+            parseUTF8String(
+                haystack,
+                haystack_size,
+                [&](UInt32 data) { haystack_utf8_set.insert(data); },
+                [&](unsigned char data) { haystack_set[data] = 1; });
+            parseUTF8String(
+                needle, needle_size, [&](UInt32 data) { needle_utf8_set.insert(data); }, [&](unsigned char data) { needle_set[data] = 1; });
+        }
+        else
+        {
+            while (haystack < haystack_end)
             {
                 haystack_set[static_cast<unsigned char>(*haystack)] = 1;
                 ++haystack;
             }
-            else
-            {
-                auto code_point = UTF8::convertUTF8ToCodePoint(haystack, haystack_end - haystack);
-                if (code_point.has_value())
-                {
-                    haystack_utf8_set.insert(code_point.value());
-                    haystack += len;
-                }
-                else
-                {
-                    throw Exception(ErrorCodes::BAD_ARGUMENTS, "Illegal UTF-8 sequence, while processing '{}'", StringRef(haystack, haystack_end - haystack));
-                }
-            }
-        }
-
-        while (needle < needle_end)
-        {
-
-            size_t len = 1;
-            if constexpr (is_utf8)
-                len = UTF8::seqLength(*needle);
-
-            if (len == 1)
+            while (needle < needle_end)
             {
                 needle_set[static_cast<unsigned char>(*needle)] = 1;
                 ++needle;
             }
-            else
-            {
-                auto code_point = UTF8::convertUTF8ToCodePoint(needle, needle_end - needle);
-                if (code_point.has_value())
-                {
-                    needle_utf8_set.insert(code_point.value());
-                    needle += len;
-                }
-                else
-                {
-                    throw Exception(ErrorCodes::BAD_ARGUMENTS, "Illegal UTF-8 sequence, while processing '{}'", StringRef(needle, needle_end - needle));
-                }
-            }
         }
 
         UInt8 intersection = 0;
@@ -226,6 +227,7 @@ struct ByteJaccardIndexImpl
 
 static constexpr size_t max_string_size = 1u << 16;
 
+template<bool is_utf8>
 struct ByteEditDistanceImpl
 {
     using ResultType = UInt64;
@@ -242,6 +244,16 @@ struct ByteEditDistanceImpl
                 ErrorCodes::TOO_LARGE_STRING_SIZE,
                 "The string size is too big for function editDistance, should be at most {}", max_string_size);
 
+        PaddedPODArray<UInt32> haystack_utf8;
+        PaddedPODArray<UInt32> needle_utf8;
+        if constexpr (is_utf8)
+        {
+            parseUTF8String(haystack, haystack_size, [&](UInt32 data) { haystack_utf8.push_back(data); });
+            parseUTF8String(needle, needle_size, [&](UInt32 data) { needle_utf8.push_back(data); });
+            haystack_size = haystack_utf8.size();
+            needle_size = needle_utf8.size();
+        }
+
         PaddedPODArray<ResultType> distances0(haystack_size + 1, 0);
         PaddedPODArray<ResultType> distances1(haystack_size + 1, 0);
 
@@ -261,9 +273,16 @@ struct ByteEditDistanceImpl
                 insertion = distances1[pos_haystack] + 1;
                 substitution = distances0[pos_haystack];
 
-                if (*(needle + pos_needle) != *(haystack + pos_haystack))
-                    substitution += 1;
-
+                if constexpr (is_utf8)
+                {
+                    if (needle_utf8[pos_needle] != haystack_utf8[pos_haystack])
+                        substitution += 1;
+                }
+                else
+                {
+                    if (*(needle + pos_needle) != *(haystack + pos_haystack))
+                        substitution += 1;
+                }
                 distances1[pos_haystack + 1] = std::min(deletion, std::min(substitution, insertion));
             }
             distances0.swap(distances1);
@@ -457,7 +476,12 @@ struct NameEditDistance
 {
     static constexpr auto name = "editDistance";
 };
-using FunctionEditDistance = FunctionsStringSimilarity<FunctionStringDistanceImpl<ByteEditDistanceImpl>, NameEditDistance>;
+using FunctionEditDistance = FunctionsStringSimilarity<FunctionStringDistanceImpl<ByteEditDistanceImpl<false>>, NameEditDistance>;
+struct NameEditDistanceUTF8
+{
+    static constexpr auto name = "editDistanceUTF8";
+};
+using FunctionEditDistanceUTF8 = FunctionsStringSimilarity<FunctionStringDistanceImpl<ByteEditDistanceImpl<true>>, NameEditDistanceUTF8>;
 
 struct NameDamerauLevenshteinDistance
 {
@@ -499,6 +523,10 @@ REGISTER_FUNCTION(StringDistance)
         FunctionDocumentation{.description = R"(Calculates the edit distance between two byte-strings.)"});
     factory.registerAlias("levenshteinDistance", NameEditDistance::name);
 
+    factory.registerFunction<FunctionEditDistanceUTF8>(
+        FunctionDocumentation{.description = R"(Calculates the edit distance between two UTF8 strings.)"});
+    factory.registerAlias("levenshteinDistanceUTF8", NameEditDistanceUTF8::name);
+
     factory.registerFunction<FunctionDamerauLevenshteinDistance>(
         FunctionDocumentation{.description = R"(Calculates the Damerau-Levenshtein distance two between two byte-string.)"});
 

src/Functions/dateTimeToSnowflakeID.cpp

@@ -0,0 +1,181 @@
+#include <Functions/FunctionFactory.h>
+#include <Functions/IFunction.h>
+#include <Functions/FunctionHelpers.h>
+#include <DataTypes/DataTypeDateTime64.h>
+#include <DataTypes/DataTypesDecimal.h>
+#include <DataTypes/DataTypesNumber.h>
+#include <Columns/ColumnConst.h>
+#include <Columns/ColumnsDateTime.h>
+#include <Columns/ColumnsNumber.h>
+#include <Core/DecimalFunctions.h>
+#include <Interpreters/Context.h>
+
+
+namespace DB
+{
+
+namespace ErrorCodes
+{
+    extern const int UNKNOWN_FUNCTION;
+}
+
+namespace
+{
+
+/// See generateSnowflakeID.cpp
+constexpr int time_shift = 22;
+
+}
+
+class FunctionDateTimeToSnowflakeID : public IFunction
+{
+private:
+    const bool uniform_snowflake_conversion_functions;
+
+public:
+    static constexpr auto name = "dateTimeToSnowflakeID";
+
+    static FunctionPtr create(ContextPtr context) { return std::make_shared<FunctionDateTimeToSnowflakeID>(context); }
+    explicit FunctionDateTimeToSnowflakeID(ContextPtr context)
+        : uniform_snowflake_conversion_functions(context->getSettingsRef().uniform_snowflake_conversion_functions)
+    {}
+
+    String getName() const override { return name; }
+    size_t getNumberOfArguments() const override { return 0; }
+    bool isVariadic() const override { return true; }
+    bool useDefaultImplementationForConstants() const override { return true; }
+    bool isSuitableForShortCircuitArgumentsExecution(const DataTypesWithConstInfo & /*arguments*/) const override { return true; }
+
+    DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override
+    {
+        FunctionArgumentDescriptors args{
+            {"value", static_cast<FunctionArgumentDescriptor::TypeValidator>(&isDateTime), nullptr, "DateTime"}
+        };
+        FunctionArgumentDescriptors optional_args{
+            {"epoch", static_cast<FunctionArgumentDescriptor::TypeValidator>(&isNativeUInt), isColumnConst, "UInt*"}
+        };
+        validateFunctionArgumentTypes(*this, arguments, args, optional_args);
+
+        return std::make_shared<DataTypeUInt64>();
+    }
+
+    ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override
+    {
+        if (!uniform_snowflake_conversion_functions)
+            throw Exception(ErrorCodes::UNKNOWN_FUNCTION, "To use function {}, setting 'uniform_snowflake_conversion_functions' must be enabled", getName());
+
+        const auto & col_src = *arguments[0].column;
+
+        size_t epoch = 0;
+        if (arguments.size() == 2 && input_rows_count != 0)
+        {
+            const auto & col_epoch = *arguments[1].column;
+            epoch = col_epoch.getUInt(0);
+        }
+
+        auto col_res = ColumnUInt64::create(input_rows_count);
+        auto & res_data = col_res->getData();
+
+        const auto & src_data = typeid_cast<const ColumnDateTime &>(col_src).getData();
+        for (size_t i = 0; i < input_rows_count; ++i)
+            res_data[i] = (static_cast<UInt64>(src_data[i]) * 1000 - epoch) << time_shift;
+        return col_res;
+    }
+};
+
+
+class FunctionDateTime64ToSnowflakeID : public IFunction
+{
+private:
+    const bool uniform_snowflake_conversion_functions;
+
+public:
+    static constexpr auto name = "dateTime64ToSnowflakeID";
+
+    static FunctionPtr create(ContextPtr context) { return std::make_shared<FunctionDateTime64ToSnowflakeID>(context); }
+    explicit FunctionDateTime64ToSnowflakeID(ContextPtr context)
+        : uniform_snowflake_conversion_functions(context->getSettingsRef().uniform_snowflake_conversion_functions)
+    {}
+
+    String getName() const override { return name; }
+    size_t getNumberOfArguments() const override { return 0; }
+    bool isVariadic() const override { return true; }
+    bool useDefaultImplementationForConstants() const override { return true; }
+    bool isSuitableForShortCircuitArgumentsExecution(const DataTypesWithConstInfo & /*arguments*/) const override { return true; }
+
+    DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override
+    {
+        FunctionArgumentDescriptors args{
+            {"value", static_cast<FunctionArgumentDescriptor::TypeValidator>(&isDateTime64), nullptr, "DateTime64"}
+        };
+        FunctionArgumentDescriptors optional_args{
+            {"epoch", static_cast<FunctionArgumentDescriptor::TypeValidator>(&isNativeUInt), isColumnConst, "UInt*"}
+        };
+        validateFunctionArgumentTypes(*this, arguments, args, optional_args);
+
+        return std::make_shared<DataTypeUInt64>();
+    }
+
+    ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override
+    {
+        if (!uniform_snowflake_conversion_functions)
+            throw Exception(ErrorCodes::UNKNOWN_FUNCTION, "To use function {}, setting 'uniform_snowflake_conversion_functions' must be enabled", getName());
+
+        const auto & col_src = *arguments[0].column;
+        const auto & src_data = typeid_cast<const ColumnDateTime64 &>(col_src).getData();
+
+        size_t epoch = 0;
+        if (arguments.size() == 2 && input_rows_count != 0)
+        {
+            const auto & col_epoch = *arguments[1].column;
+            epoch = col_epoch.getUInt(0);
+        }
+
+        auto col_res = ColumnUInt64::create(input_rows_count);
+        auto & res_data = col_res->getData();
+
+        /// timestamps in snowflake-ids are millisecond-based, convert input to milliseconds
+        UInt32 src_scale = getDecimalScale(*arguments[0].type);
+        Int64 multiplier_msec = DecimalUtils::scaleMultiplier<DateTime64>(3);
+        Int64 multiplier_src = DecimalUtils::scaleMultiplier<DateTime64>(src_scale);
+        auto factor = multiplier_msec / static_cast<double>(multiplier_src);
+
+        for (size_t i = 0; i < input_rows_count; ++i)
+            res_data[i] = static_cast<UInt64>(src_data[i] * factor - epoch) << time_shift;
+
+        return col_res;
+    }
+};
+
+REGISTER_FUNCTION(DateTimeToSnowflakeID)
+{
+    {
+        FunctionDocumentation::Description description = R"(Converts a [DateTime](../data-types/datetime.md) value to the first [Snowflake ID](https://en.wikipedia.org/wiki/Snowflake_ID) at the giving time.)";
+        FunctionDocumentation::Syntax syntax = "dateTimeToSnowflakeID(value[, epoch])";
+        FunctionDocumentation::Arguments arguments = {
+            {"value", "Date with time. [DateTime](../data-types/datetime.md)."},
+            {"epoch", "Epoch of the Snowflake ID in milliseconds since 1970-01-01. Defaults to 0 (1970-01-01). For the Twitter/X epoch (2015-01-01), provide 1288834974657. Optional. [UInt*](../data-types/int-uint.md)"}
+        };
+        FunctionDocumentation::ReturnedValue returned_value = "Input value converted to [UInt64](../data-types/int-uint.md) as the first Snowflake ID at that time.";
+        FunctionDocumentation::Examples examples = {{"simple", "SELECT dateTimeToSnowflakeID(toDateTime('2021-08-15 18:57:56', 'Asia/Shanghai'))", "6832626392367104000"}};
+        FunctionDocumentation::Categories categories = {"Snowflake ID"};
+
+        factory.registerFunction<FunctionDateTimeToSnowflakeID>({description, syntax, arguments, returned_value, examples, categories});
+    }
+
+    {
+        FunctionDocumentation::Description description = R"(Converts a [DateTime64](../data-types/datetime64.md) value to the first [Snowflake ID](https://en.wikipedia.org/wiki/Snowflake_ID) at the giving time.)";
+        FunctionDocumentation::Syntax syntax = "dateTime64ToSnowflakeID(value[, epoch])";
+        FunctionDocumentation::Arguments arguments = {
+            {"value", "Date with time. [DateTime64](../data-types/datetime.md)."},
+            {"epoch", "Epoch of the Snowflake ID in milliseconds since 1970-01-01. Defaults to 0 (1970-01-01). For the Twitter/X epoch (2015-01-01), provide 1288834974657. Optional. [UInt*](../data-types/int-uint.md)"}
+        };
+        FunctionDocumentation::ReturnedValue returned_value = "Input value converted to [UInt64](../data-types/int-uint.md) as the first Snowflake ID at that time.";
+        FunctionDocumentation::Examples examples = {{"simple", "SELECT dateTime64ToSnowflakeID(toDateTime64('2021-08-15 18:57:56', 3, 'Asia/Shanghai'))", "6832626394434895872"}};
+        FunctionDocumentation::Categories categories = {"Snowflake ID"};
+
+        factory.registerFunction<FunctionDateTime64ToSnowflakeID>({description, syntax, arguments, returned_value, examples, categories});
+    }
+}
+
+}

src/Functions/generateSnowflakeID.cpp

@@ -207,7 +207,7 @@ public:
 
 REGISTER_FUNCTION(GenerateSnowflakeID)
 {
-    FunctionDocumentation::Description description = R"(Generates a Snowflake ID. The generated Snowflake ID contains the current Unix timestamp in milliseconds 41 (+ 1 top zero bit) bits, followed by machine id (10 bits), a counter (12 bits) to distinguish IDs within a millisecond. For any given timestamp (unix_ts_ms), the counter starts at 0 and is incremented by 1 for each new Snowflake ID until the timestamp changes. In case the counter overflows, the timestamp field is incremented by 1 and the counter is reset to 0. Function generateSnowflakeID guarantees that the counter field within a timestamp increments monotonically across all function invocations in concurrently running threads and queries.)";
+    FunctionDocumentation::Description description = R"(Generates a Snowflake ID. The generated Snowflake ID contains the current Unix timestamp in milliseconds (41 + 1 top zero bits), followed by a machine id (10 bits), and a counter (12 bits) to distinguish IDs within a millisecond. For any given timestamp (unix_ts_ms), the counter starts at 0 and is incremented by 1 for each new Snowflake ID until the timestamp changes. In case the counter overflows, the timestamp field is incremented by 1 and the counter is reset to 0. Function generateSnowflakeID guarantees that the counter field within a timestamp increments monotonically across all function invocations in concurrently running threads and queries.)";
     FunctionDocumentation::Syntax syntax = "generateSnowflakeID([expression])";
     FunctionDocumentation::Arguments arguments = {{"expression", "The expression is used to bypass common subexpression elimination if the function is called multiple times in a query but otherwise ignored. Optional."}};
     FunctionDocumentation::ReturnedValue returned_value = "A value of type UInt64";

src/Functions/snowflake.cpp

@@ -11,11 +11,17 @@
 #include <Interpreters/Context.h>
 
 
+/// ------------------------------------------------------------------------------------------------------------------------------
+/// The functions in this file are deprecated and should be removed in favor of functions 'snowflakeIDToDateTime[64]' and
+/// 'dateTime[64]ToSnowflakeID' by summer 2025. Please also mark setting `uniform_snowflake_conversion_functions` as obsolete then.
+/// ------------------------------------------------------------------------------------------------------------------------------
+
 namespace DB
 {
 
 namespace ErrorCodes
 {
+    extern const int DEPRECATED_FUNCTION;
     extern const int ILLEGAL_TYPE_OF_ARGUMENT;
 }
 
@@ -34,10 +40,19 @@ constexpr int time_shift = 22;
 class FunctionDateTimeToSnowflake : public IFunction
 {
 private:
-    const char * name;
+    const bool uniform_snowflake_conversion_functions;
 
 public:
-    explicit FunctionDateTimeToSnowflake(const char * name_) : name(name_) { }
+    static constexpr auto name = "dateTimeToSnowflake";
+
+    static FunctionPtr create(ContextPtr context)
+    {
+        return std::make_shared<FunctionDateTimeToSnowflake>(context);
+    }
+
+    explicit FunctionDateTimeToSnowflake(ContextPtr context)
+        : uniform_snowflake_conversion_functions(context->getSettingsRef().uniform_snowflake_conversion_functions)
+    {}
 
     String getName() const override { return name; }
     size_t getNumberOfArguments() const override { return 1; }
@@ -56,6 +71,9 @@ public:
 
     ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override
     {
+        if (uniform_snowflake_conversion_functions)
+            throw Exception(ErrorCodes::DEPRECATED_FUNCTION, "Function {} is deprecated, to enable it disable setting 'uniform_snowflake_conversion_functions'", getName());
+
         const auto & src = arguments[0];
         const auto & src_column = *src.column;
 
@@ -73,13 +91,20 @@ public:
 class FunctionSnowflakeToDateTime : public IFunction
 {
 private:
-    const char * name;
     const bool allow_nonconst_timezone_arguments;
+    const bool uniform_snowflake_conversion_functions;
 
 public:
-    explicit FunctionSnowflakeToDateTime(const char * name_, ContextPtr context)
-        : name(name_)
-        , allow_nonconst_timezone_arguments(context->getSettings().allow_nonconst_timezone_arguments)
+    static constexpr auto name = "snowflakeToDateTime";
+
+    static FunctionPtr create(ContextPtr context)
+    {
+        return std::make_shared<FunctionSnowflakeToDateTime>(context);
+    }
+
+    explicit FunctionSnowflakeToDateTime(ContextPtr context)
+        : allow_nonconst_timezone_arguments(context->getSettingsRef().allow_nonconst_timezone_arguments)
+        , uniform_snowflake_conversion_functions(context->getSettingsRef().uniform_snowflake_conversion_functions)
     {}
 
     String getName() const override { return name; }
@@ -107,6 +132,9 @@ public:
 
     ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override
     {
+        if (uniform_snowflake_conversion_functions)
+            throw Exception(ErrorCodes::DEPRECATED_FUNCTION, "Function {} is deprecated, to enable it disable setting 'uniform_snowflake_conversion_functions'", getName());
+
         const auto & src = arguments[0];
         const auto & src_column = *src.column;
 
@@ -138,10 +166,19 @@ public:
 class FunctionDateTime64ToSnowflake : public IFunction
 {
 private:
-    const char * name;
+    const bool uniform_snowflake_conversion_functions;
 
 public:
-    explicit FunctionDateTime64ToSnowflake(const char * name_) : name(name_) { }
+    static constexpr auto name = "dateTime64ToSnowflake";
+
+    static FunctionPtr create(ContextPtr context)
+    {
+        return std::make_shared<FunctionDateTime64ToSnowflake>(context);
+    }
+
+    explicit FunctionDateTime64ToSnowflake(ContextPtr context)
+        : uniform_snowflake_conversion_functions(context->getSettingsRef().uniform_snowflake_conversion_functions)
+    {}
 
     String getName() const override { return name; }
     size_t getNumberOfArguments() const override { return 1; }
@@ -160,6 +197,9 @@ public:
 
     ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override
     {
+        if (uniform_snowflake_conversion_functions)
+            throw Exception(ErrorCodes::DEPRECATED_FUNCTION, "Function {} is deprecated, to enable it disable setting 'uniform_snowflake_conversion_functions'", getName());
+
         const auto & src = arguments[0];
 
         const auto & src_column = *src.column;
@@ -185,13 +225,20 @@ public:
 class FunctionSnowflakeToDateTime64 : public IFunction
 {
 private:
-    const char * name;
     const bool allow_nonconst_timezone_arguments;
+    const bool uniform_snowflake_conversion_functions;
 
 public:
-    explicit FunctionSnowflakeToDateTime64(const char * name_, ContextPtr context)
-        : name(name_)
-        , allow_nonconst_timezone_arguments(context->getSettings().allow_nonconst_timezone_arguments)
+    static constexpr auto name = "snowflakeToDateTime64";
+
+    static FunctionPtr create(ContextPtr context)
+    {
+        return std::make_shared<FunctionSnowflakeToDateTime64>(context);
+    }
+
+    explicit FunctionSnowflakeToDateTime64(ContextPtr context)
+        : allow_nonconst_timezone_arguments(context->getSettingsRef().allow_nonconst_timezone_arguments)
+        , uniform_snowflake_conversion_functions(context->getSettingsRef().uniform_snowflake_conversion_functions)
     {}
 
     String getName() const override { return name; }
@@ -219,6 +266,9 @@ public:
 
     ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override
     {
+        if (uniform_snowflake_conversion_functions)
+            throw Exception(ErrorCodes::DEPRECATED_FUNCTION, "Function {} is deprecated, to enable it disable setting 'uniform_snowflake_conversion_functions'", getName());
+
         const auto & src = arguments[0];
         const auto & src_column = *src.column;
 
@@ -246,27 +296,12 @@ public:
 
 }
 
-REGISTER_FUNCTION(DateTimeToSnowflake)
+REGISTER_FUNCTION(LegacySnowflakeConversion)
 {
-    factory.registerFunction("dateTimeToSnowflake",
-        [](ContextPtr){ return std::make_shared<FunctionDateTimeToSnowflake>("dateTimeToSnowflake"); });
-}
-
-REGISTER_FUNCTION(DateTime64ToSnowflake)
-{
-    factory.registerFunction("dateTime64ToSnowflake",
-        [](ContextPtr){ return std::make_shared<FunctionDateTime64ToSnowflake>("dateTime64ToSnowflake"); });
-}
-
-REGISTER_FUNCTION(SnowflakeToDateTime)
-{
-    factory.registerFunction("snowflakeToDateTime",
-        [](ContextPtr context){ return std::make_shared<FunctionSnowflakeToDateTime>("snowflakeToDateTime", context); });
-}
-REGISTER_FUNCTION(SnowflakeToDateTime64)
-{
-    factory.registerFunction("snowflakeToDateTime64",
-        [](ContextPtr context){ return std::make_shared<FunctionSnowflakeToDateTime64>("snowflakeToDateTime64", context); });
+    factory.registerFunction<FunctionSnowflakeToDateTime>();
+    factory.registerFunction<FunctionSnowflakeToDateTime64>();
+    factory.registerFunction<FunctionDateTimeToSnowflake>();
+    factory.registerFunction<FunctionDateTime64ToSnowflake>();
 }
 
 }

src/Functions/snowflakeIDToDateTime.cpp

@@ -0,0 +1,217 @@
+#include <Functions/FunctionFactory.h>
+#include <Functions/extractTimeZoneFromFunctionArguments.h>
+#include <Functions/IFunction.h>
+#include <Functions/FunctionHelpers.h>
+#include <DataTypes/DataTypeDateTime64.h>
+#include <DataTypes/DataTypesDecimal.h>
+#include <DataTypes/DataTypesNumber.h>
+#include <Columns/ColumnConst.h>
+#include <Columns/ColumnsDateTime.h>
+#include <Columns/ColumnsNumber.h>
+#include <Core/DecimalFunctions.h>
+#include <Interpreters/Context.h>
+
+
+namespace DB
+{
+
+namespace ErrorCodes
+{
+    extern const int ILLEGAL_TYPE_OF_ARGUMENT;
+    extern const int UNKNOWN_FUNCTION;
+}
+
+namespace
+{
+
+/// See generateSnowflakeID.cpp
+constexpr int time_shift = 22;
+
+}
+
+class FunctionSnowflakeIDToDateTime : public IFunction
+{
+private:
+    const bool uniform_snowflake_conversion_functions;
+    const bool allow_nonconst_timezone_arguments;
+
+public:
+    static constexpr auto name = "snowflakeIDToDateTime";
+
+    static FunctionPtr create(ContextPtr context) { return std::make_shared<FunctionSnowflakeIDToDateTime>(context); }
+    explicit FunctionSnowflakeIDToDateTime(ContextPtr context)
+        : uniform_snowflake_conversion_functions(context->getSettingsRef().uniform_snowflake_conversion_functions)
+        , allow_nonconst_timezone_arguments(context->getSettings().allow_nonconst_timezone_arguments)
+    {}
+
+    String getName() const override { return name; }
+    size_t getNumberOfArguments() const override { return 0; }
+    bool isVariadic() const override { return true; }
+    bool useDefaultImplementationForConstants() const override { return true; }
+    bool isSuitableForShortCircuitArgumentsExecution(const DataTypesWithConstInfo & /*arguments*/) const override { return true; }
+
+    DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override
+    {
+        FunctionArgumentDescriptors args{
+            {"value", static_cast<FunctionArgumentDescriptor::TypeValidator>(&isUInt64), nullptr, "UInt64"}
+        };
+        FunctionArgumentDescriptors optional_args{
+            {"epoch", static_cast<FunctionArgumentDescriptor::TypeValidator>(&isNativeUInt), isColumnConst, "UInt*"},
+            {"time_zone", static_cast<FunctionArgumentDescriptor::TypeValidator>(&isString), nullptr, "String"}
+        };
+        validateFunctionArgumentTypes(*this, arguments, args, optional_args);
+
+        String timezone;
+        if (arguments.size() == 3)
+            timezone = extractTimeZoneNameFromFunctionArguments(arguments, 2, 0, allow_nonconst_timezone_arguments);
+
+        return std::make_shared<DataTypeDateTime>(timezone);
+    }
+
+    ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override
+    {
+        if (!uniform_snowflake_conversion_functions)
+            throw Exception(ErrorCodes::UNKNOWN_FUNCTION, "To use function {}, setting 'uniform_snowflake_conversion_functions' must be enabled", getName());
+
+        const auto & col_src = *arguments[0].column;
+
+        size_t epoch = 0;
+        if (arguments.size() >= 2 && input_rows_count != 0)
+        {
+            const auto & col_epoch = *arguments[1].column;
+            epoch = col_epoch.getUInt(0);
+        }
+
+        auto col_res = ColumnDateTime::create(input_rows_count);
+        auto & res_data = col_res->getData();
+
+        if (const auto * col_src_non_const = typeid_cast<const ColumnUInt64 *>(&col_src))
+        {
+            const auto & src_data = col_src_non_const->getData();
+            for (size_t i = 0; i < input_rows_count; ++i)
+                res_data[i] = static_cast<UInt32>(((src_data[i] >> time_shift) + epoch) / 1000);
+        }
+        else if (const auto * col_src_const = typeid_cast<const ColumnConst *>(&col_src))
+        {
+            UInt64 src_val = col_src_const->getValue<UInt64>();
+            for (size_t i = 0; i < input_rows_count; ++i)
+                res_data[i] = static_cast<UInt32>(((src_val >> time_shift) + epoch) / 1000);
+        }
+        else
+            throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Illegal argument for function {}", name);
+
+        return col_res;
+    }
+};
+
+
+class FunctionSnowflakeIDToDateTime64 : public IFunction
+{
+private:
+    const bool uniform_snowflake_conversion_functions;
+    const bool allow_nonconst_timezone_arguments;
+
+public:
+    static constexpr auto name = "snowflakeIDToDateTime64";
+
+    static FunctionPtr create(ContextPtr context) { return std::make_shared<FunctionSnowflakeIDToDateTime64>(context); }
+    explicit FunctionSnowflakeIDToDateTime64(ContextPtr context)
+        : uniform_snowflake_conversion_functions(context->getSettingsRef().uniform_snowflake_conversion_functions)
+        , allow_nonconst_timezone_arguments(context->getSettings().allow_nonconst_timezone_arguments)
+    {}
+
+    String getName() const override { return name; }
+    size_t getNumberOfArguments() const override { return 0; }
+    bool isVariadic() const override { return true; }
+    bool useDefaultImplementationForConstants() const override { return true; }
+    bool isSuitableForShortCircuitArgumentsExecution(const DataTypesWithConstInfo & /*arguments*/) const override { return true; }
+
+    DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override
+    {
+        FunctionArgumentDescriptors args{
+            {"value", static_cast<FunctionArgumentDescriptor::TypeValidator>(&isUInt64), nullptr, "UInt64"}
+        };
+        FunctionArgumentDescriptors optional_args{
+            {"epoch", static_cast<FunctionArgumentDescriptor::TypeValidator>(&isNativeUInt), isColumnConst, "UInt*"},
+            {"time_zone", static_cast<FunctionArgumentDescriptor::TypeValidator>(&isString), nullptr, "String"}
+        };
+        validateFunctionArgumentTypes(*this, arguments, args, optional_args);
+
+        String timezone;
+        if (arguments.size() == 3)
+            timezone = extractTimeZoneNameFromFunctionArguments(arguments, 2, 0, allow_nonconst_timezone_arguments);
+
+        return std::make_shared<DataTypeDateTime64>(3, timezone);
+    }
+
+    ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override
+    {
+        if (!uniform_snowflake_conversion_functions)
+            throw Exception(ErrorCodes::UNKNOWN_FUNCTION, "To use function {}, setting 'uniform_snowflake_conversion_functions' must be enabled", getName());
+
+        const auto & col_src = *arguments[0].column;
+
+        size_t epoch = 0;
+        if (arguments.size() >= 2 && input_rows_count != 0)
+        {
+            const auto & col_epoch = *arguments[1].column;
+            epoch = col_epoch.getUInt(0);
+        }
+
+        auto col_res = ColumnDateTime64::create(input_rows_count, 3);
+        auto & res_data = col_res->getData();
+
+        if (const auto * col_src_non_const = typeid_cast<const ColumnUInt64 *>(&col_src))
+        {
+            const auto & src_data = col_src_non_const->getData();
+            for (size_t i = 0; i < input_rows_count; ++i)
+                res_data[i] = (src_data[i] >> time_shift) + epoch;
+        }
+        else if (const auto * col_src_const = typeid_cast<const ColumnConst *>(&col_src))
+        {
+            UInt64 src_val = col_src_const->getValue<UInt64>();
+            for (size_t i = 0; i < input_rows_count; ++i)
+                res_data[i] = (src_val >> time_shift) + epoch;
+        }
+        else
+            throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Illegal argument for function {}", name);
+
+        return col_res;
+
+    }
+};
+
+REGISTER_FUNCTION(SnowflakeIDToDateTime)
+{
+    {
+        FunctionDocumentation::Description description = R"(Returns the timestamp component of a [Snowflake ID](https://en.wikipedia.org/wiki/Snowflake_ID) as a value of type [DateTime](../data-types/datetime.md).)";
+        FunctionDocumentation::Syntax syntax = "snowflakeIDToDateTime(value[, epoch[, time_zone]])";
+        FunctionDocumentation::Arguments arguments = {
+            {"value", "Snowflake ID. [UInt64](../data-types/int-uint.md)"},
+            {"epoch", "Epoch of the Snowflake ID in milliseconds since 1970-01-01. Defaults to 0 (1970-01-01). For the Twitter/X epoch (2015-01-01), provide 1288834974657. Optional. [UInt*](../data-types/int-uint.md)"},
+            {"time_zone", "[Timezone](/docs/en/operations/server-configuration-parameters/settings.md/#server_configuration_parameters-timezone). The function parses `time_string` according to the timezone. Optional. [String](../data-types/string.md)"}
+        };
+        FunctionDocumentation::ReturnedValue returned_value = "The timestamp component of `value` as a [DateTime](../data-types/datetime.md) value.";
+        FunctionDocumentation::Examples examples = {{"simple", "SELECT snowflakeIDToDateTime(7204436857747984384)", "2024-06-06 10:59:58"}};
+        FunctionDocumentation::Categories categories = {"Snowflake ID"};
+
+        factory.registerFunction<FunctionSnowflakeIDToDateTime>({description, syntax, arguments, returned_value, examples, categories});
+    }
+
+    {
+        FunctionDocumentation::Description description = R"(Returns the timestamp component of a [Snowflake ID](https://en.wikipedia.org/wiki/Snowflake_ID) as a value of type [DateTime64](../data-types/datetime64.md).)";
+        FunctionDocumentation::Syntax syntax = "snowflakeIDToDateTime64(value[, epoch[, time_zone]])";
+        FunctionDocumentation::Arguments arguments = {
+            {"value", "Snowflake ID. [UInt64](../data-types/int-uint.md)"},
+            {"epoch", "Epoch of the Snowflake ID in milliseconds since 1970-01-01. Defaults to 0 (1970-01-01). For the Twitter/X epoch (2015-01-01), provide 1288834974657. Optional. [UInt*](../data-types/int-uint.md)"},
+            {"time_zone", "[Timezone](/docs/en/operations/server-configuration-parameters/settings.md/#server_configuration_parameters-timezone). The function parses `time_string` according to the timezone. Optional. [String](../data-types/string.md)"}
+        };
+        FunctionDocumentation::ReturnedValue returned_value = "The timestamp component of `value` as a [DateTime64](../data-types/datetime64.md) with scale = 3, i.e. millisecond precision.";
+        FunctionDocumentation::Examples examples = {{"simple", "SELECT snowflakeIDToDateTime64(7204436857747984384)", "2024-06-06 10:59:58"}};
+        FunctionDocumentation::Categories categories = {"Snowflake ID"};
+
+        factory.registerFunction<FunctionSnowflakeIDToDateTime64>({description, syntax, arguments, returned_value, examples, categories});
+    }
+}
+
+}

src/IO/S3/PocoHTTPClient.cpp

@@ -551,7 +551,7 @@ void PocoHTTPClient::makeRequestInternalImpl(
                 const static std::string_view needle = "<Error>";
                 if (auto it = std::search(response_string.begin(), response_string.end(), std::default_searcher(needle.begin(), needle.end())); it != response_string.end())
                 {
-                    LOG_WARNING(log, "Response for request contain <Error> tag in body, settings internal server error (500 code)");
+                    LOG_WARNING(log, "Response for the request contains an <Error> tag in the body, will treat it as an internal server error (code 500)");
                     response->SetResponseCode(Aws::Http::HttpResponseCode::INTERNAL_SERVER_ERROR);
 
                     addMetric(request, S3MetricType::Errors);

src/Interpreters/Access/InterpreterGrantQuery.cpp

@@ -118,7 +118,7 @@ namespace
     /// Checks if the current user has enough access rights granted with grant option to grant or revoke specified access rights.
     void checkGrantOption(
         const AccessControl & access_control,
-        const ContextAccess & current_user_access,
+        const ContextAccessWrapper & current_user_access,
         const std::vector<UUID> & grantees_from_query,
         bool & need_check_grantees_are_allowed,
         const AccessRightsElements & elements_to_grant,
@@ -200,7 +200,7 @@ namespace
     /// Checks if the current user has enough roles granted with admin option to grant or revoke specified roles.
     void checkAdminOption(
         const AccessControl & access_control,
-        const ContextAccess & current_user_access,
+        const ContextAccessWrapper & current_user_access,
         const std::vector<UUID> & grantees_from_query,
         bool & need_check_grantees_are_allowed,
         const std::vector<UUID> & roles_to_grant,
@@ -277,7 +277,7 @@ namespace
     /// This function is less accurate than checkAdminOption() because it cannot use any information about
     /// granted roles the grantees currently have (due to those grantees are located on multiple nodes,
     /// we just don't have the full information about them).
-    void checkAdminOptionForExecutingOnCluster(const ContextAccess & current_user_access,
+    void checkAdminOptionForExecutingOnCluster(const ContextAccessWrapper & current_user_access,
                                                const std::vector<UUID> roles_to_grant,
                                                const RolesOrUsersSet & roles_to_revoke)
     {
@@ -376,7 +376,7 @@ namespace
     /// Calculates all available rights to grant with current user intersection.
     void calculateCurrentGrantRightsWithIntersection(
         AccessRights & rights,
-        std::shared_ptr<const ContextAccess> current_user_access,
+        std::shared_ptr<const ContextAccessWrapper> current_user_access,
         const AccessRightsElements & elements_to_grant)
     {
         AccessRightsElements current_user_grantable_elements;

src/Interpreters/ActionsVisitor.cpp

@@ -4,6 +4,7 @@
 #include <Common/typeid_cast.h>
 #include <Common/FieldVisitorsAccurateComparison.h>
 #include <Common/checkStackSize.h>
+#include <Common/assert_cast.h>
 
 #include <Core/ColumnNumbers.h>
 #include <Core/ColumnWithTypeAndName.h>
@@ -102,7 +103,7 @@ static size_t getTypeDepth(const DataTypePtr & type)
 /// 33.33 in the set is converted to 33.3, but it is not equal to 33.3 in the column, so the result should still be empty.
 /// We can not include values that don't represent any possible value from the type of filtered column to the set.
 template<typename Collection>
-static Block createBlockFromCollection(const Collection & collection, const DataTypes & types, bool transform_null_in)
+static Block createBlockFromCollection(const Collection & collection, const DataTypes & value_types, const DataTypes & types, bool transform_null_in)
 {
     size_t columns_num = types.size();
     MutableColumns columns(columns_num);
@@ -113,11 +114,12 @@ static Block createBlockFromCollection(const Collection & collection, const Data
     }
 
     Row tuple_values;
-    for (const auto & value : collection)
+    for (size_t collection_index = 0; collection_index < collection.size(); ++collection_index)
     {
+        const auto& value = collection[collection_index];
         if (columns_num == 1)
         {
-            auto field = convertFieldToTypeStrict(value, *types[0]);
+            auto field = convertFieldToTypeStrict(value, *value_types[collection_index], *types[0]);
             bool need_insert_null = transform_null_in && types[0]->isNullable();
             if (field && (!field->isNull() || need_insert_null))
                 columns[0]->insert(*field);
@@ -130,7 +132,6 @@ static Block createBlockFromCollection(const Collection & collection, const Data
 
             const auto & tuple = value.template get<const Tuple &>();
             size_t tuple_size = tuple.size();
-
             if (tuple_size != columns_num)
                 throw Exception(ErrorCodes::INCORRECT_ELEMENT_OF_SET, "Incorrect size of tuple in set: {} instead of {}",
                     tuple_size, columns_num);
@@ -138,10 +139,13 @@ static Block createBlockFromCollection(const Collection & collection, const Data
             if (tuple_values.empty())
                 tuple_values.resize(tuple_size);
 
+            const DataTypePtr & value_type = value_types[collection_index];
+            const DataTypes & tuple_value_type = typeid_cast<const DataTypeTuple *>(value_type.get())->getElements();
+
             size_t i = 0;
             for (; i < tuple_size; ++i)
             {
-                auto converted_field = convertFieldToTypeStrict(tuple[i], *types[i]);
+                auto converted_field = convertFieldToTypeStrict(tuple[i], *tuple_value_type[i], *types[i]);
                 if (!converted_field)
                     break;
                 tuple_values[i] = std::move(*converted_field);
@@ -317,16 +321,25 @@ Block createBlockForSet(
     if (left_type_depth == right_type_depth)
     {
         Array array{right_arg_value};
-        block = createBlockFromCollection(array, set_element_types, tranform_null_in);
+        DataTypes value_types{right_arg_type};
+        block = createBlockFromCollection(array, value_types, set_element_types, tranform_null_in);
     }
     /// 1 in (1, 2); (1, 2) in ((1, 2), (3, 4)); etc.
     else if (left_type_depth + 1 == right_type_depth)
     {
         auto type_index = right_arg_type->getTypeId();
         if (type_index == TypeIndex::Tuple)
-            block = createBlockFromCollection(right_arg_value.get<const Tuple &>(), set_element_types, tranform_null_in);
+        {
+            const DataTypes & value_types = assert_cast<const DataTypeTuple *>(right_arg_type.get())->getElements();
+            block = createBlockFromCollection(right_arg_value.get<const Tuple &>(), value_types, set_element_types, tranform_null_in);
+        }
         else if (type_index == TypeIndex::Array)
-            block = createBlockFromCollection(right_arg_value.get<const Array &>(), set_element_types, tranform_null_in);
+        {
+            const auto* right_arg_array_type =  assert_cast<const DataTypeArray *>(right_arg_type.get());
+            size_t right_arg_array_size = right_arg_value.get<const Array &>().size();
+            DataTypes value_types(right_arg_array_size, right_arg_array_type->getNestedType());
+            block = createBlockFromCollection(right_arg_value.get<const Array &>(), value_types, set_element_types, tranform_null_in);
+        }
         else
             throw_unsupported_type(right_arg_type);
     }

src/Interpreters/Context.cpp

@@ -281,6 +281,8 @@ struct ContextSharedPart : boost::noncopyable
     String default_profile_name;                                /// Default profile name used for default values.
     String system_profile_name;                                 /// Profile used by system processes
     String buffer_profile_name;                                 /// Profile used by Buffer engine for flushing to the underlying
+    String merge_workload TSA_GUARDED_BY(mutex);                /// Workload setting value that is used by all merges
+    String mutation_workload TSA_GUARDED_BY(mutex);             /// Workload setting value that is used by all mutations
     std::unique_ptr<AccessControl> access_control TSA_GUARDED_BY(mutex);
     mutable OnceFlag resource_manager_initialized;
     mutable ResourceManagerPtr resource_manager;
@@ -833,6 +835,7 @@ ContextMutablePtr Context::createGlobal(ContextSharedPart * shared_part)
     auto res = std::shared_ptr<Context>(new Context);
     res->shared = shared_part;
     res->query_access_info = std::make_shared<QueryAccessInfo>();
+    res->query_privileges_info = std::make_shared<QueryPrivilegesInfo>();
     return res;
 }
 
@@ -1425,7 +1428,7 @@ void Context::checkAccess(const AccessFlags & flags, const StorageID & table_id,
 void Context::checkAccess(const AccessRightsElement & element) const { checkAccessImpl(element); }
 void Context::checkAccess(const AccessRightsElements & elements) const { checkAccessImpl(elements); }
 
-std::shared_ptr<const ContextAccess> Context::getAccess() const
+std::shared_ptr<const ContextAccessWrapper> Context::getAccess() const
 {
     /// A helper function to collect parameters for calculating access rights, called with Context::getLocalSharedLock() acquired.
     auto get_params = [this]()
@@ -1442,14 +1445,14 @@ std::shared_ptr<const ContextAccess> Context::getAccess() const
     {
         SharedLockGuard lock(mutex);
         if (access && !need_recalculate_access)
-            return access; /// No need to recalculate access rights.
+            return std::make_shared<const ContextAccessWrapper>(access, shared_from_this()); /// No need to recalculate access rights.
 
         params.emplace(get_params());
 
         if (access && (access->getParams() == *params))
         {
             need_recalculate_access = false;
-            return access; /// No need to recalculate access rights.
+            return std::make_shared<const ContextAccessWrapper>(access, shared_from_this()); /// No need to recalculate access rights.
         }
     }
 
@@ -1469,7 +1472,7 @@ std::shared_ptr<const ContextAccess> Context::getAccess() const
         }
     }
 
-    return res;
+    return std::make_shared<const ContextAccessWrapper>(res, shared_from_this());
 }
 
 RowPolicyFilterPtr Context::getRowPolicyFilter(const String & database, const String & table_name, RowPolicyFilterType filter_type) const
@@ -1561,11 +1564,36 @@ ResourceManagerPtr Context::getResourceManager() const
 ClassifierPtr Context::getWorkloadClassifier() const
 {
     std::lock_guard lock(mutex);
+    // NOTE: Workload cannot be changed after query start, and getWorkloadClassifier() should not be called before proper `workload` is set
     if (!classifier)
         classifier = getResourceManager()->acquire(getSettingsRef().workload);
     return classifier;
 }
 
+String Context::getMergeWorkload() const
+{
+    SharedLockGuard lock(shared->mutex);
+    return shared->merge_workload;
+}
+
+void Context::setMergeWorkload(const String & value)
+{
+    std::lock_guard lock(shared->mutex);
+    shared->merge_workload = value;
+}
+
+String Context::getMutationWorkload() const
+{
+    SharedLockGuard lock(shared->mutex);
+    return shared->mutation_workload;
+}
+
+void Context::setMutationWorkload(const String & value)
+{
+    std::lock_guard lock(shared->mutex);
+    shared->mutation_workload = value;
+}
+
 
 Scalars Context::getScalars() const
 {
@@ -1830,6 +1858,15 @@ void Context::addQueryFactoriesInfo(QueryLogFactories factory_type, const String
     }
 }
 
+void Context::addQueryPrivilegesInfo(const String & privilege, bool granted) const
+{
+    std::lock_guard lock(query_privileges_info->mutex);
+    if (granted)
+        query_privileges_info->used_privileges.emplace(privilege);
+    else
+        query_privileges_info->missing_privileges.emplace(privilege);
+}
+
 static bool findIdentifier(const ASTFunction * function)
 {
     if (!function || !function->arguments)
@@ -2511,6 +2548,21 @@ void Context::makeQueryContext()
     local_read_query_throttler.reset();
     local_write_query_throttler.reset();
     backups_query_throttler.reset();
+    query_privileges_info = std::make_shared<QueryPrivilegesInfo>(*query_privileges_info);
+}
+
+void Context::makeQueryContextForMerge(const MergeTreeSettings & merge_tree_settings)
+{
+    makeQueryContext();
+    classifier.reset(); // It is assumed that there are no active queries running using this classifier, otherwise this will lead to crashes
+    settings.workload = merge_tree_settings.merge_workload.value.empty() ? getMergeWorkload() : merge_tree_settings.merge_workload;
+}
+
+void Context::makeQueryContextForMutate(const MergeTreeSettings & merge_tree_settings)
+{
+    makeQueryContext();
+    classifier.reset(); // It is assumed that there are no active queries running using this classifier, otherwise this will lead to crashes
+    settings.workload = merge_tree_settings.mutation_workload.value.empty() ? getMutationWorkload() : merge_tree_settings.mutation_workload;
 }
 
 void Context::makeSessionContext()

src/Interpreters/Context.h

@@ -50,6 +50,7 @@ class ASTSelectQuery;
 
 struct ContextSharedPart;
 class ContextAccess;
+class ContextAccessWrapper;
 struct User;
 using UserPtr = std::shared_ptr<const User>;
 struct SettingsProfilesInfo;
@@ -403,9 +404,31 @@ public:
         mutable std::mutex mutex;
     };
 
+    struct QueryPrivilegesInfo
+    {
+        QueryPrivilegesInfo() = default;
+
+        QueryPrivilegesInfo(const QueryPrivilegesInfo & rhs)
+        {
+            std::lock_guard<std::mutex> lock(rhs.mutex);
+            used_privileges = rhs.used_privileges;
+            missing_privileges = rhs.missing_privileges;
+        }
+
+        QueryPrivilegesInfo(QueryPrivilegesInfo && rhs) = delete;
+
+        std::unordered_set<std::string> used_privileges TSA_GUARDED_BY(mutex);
+        std::unordered_set<std::string> missing_privileges TSA_GUARDED_BY(mutex);
+
+        mutable std::mutex mutex;
+    };
+
+    using QueryPrivilegesInfoPtr = std::shared_ptr<QueryPrivilegesInfo>;
+
 protected:
     /// Needs to be changed while having const context in factories methods
     mutable QueryFactoriesInfo query_factories_info;
+    QueryPrivilegesInfoPtr query_privileges_info;
     /// Query metrics for reading data asynchronously with IAsynchronousReader.
     mutable std::shared_ptr<AsyncReadCounters> async_read_counters;
 
@@ -612,7 +635,7 @@ public:
     void checkAccess(const AccessRightsElement & element) const;
     void checkAccess(const AccessRightsElements & elements) const;
 
-    std::shared_ptr<const ContextAccess> getAccess() const;
+    std::shared_ptr<const ContextAccessWrapper> getAccess() const;
 
     RowPolicyFilterPtr getRowPolicyFilter(const String & database, const String & table_name, RowPolicyFilterType filter_type) const;
 
@@ -622,6 +645,10 @@ public:
     /// Resource management related
     ResourceManagerPtr getResourceManager() const;
     ClassifierPtr getWorkloadClassifier() const;
+    String getMergeWorkload() const;
+    void setMergeWorkload(const String & value);
+    String getMutationWorkload() const;
+    void setMutationWorkload(const String & value);
 
     /// We have to copy external tables inside executeQuery() to track limits. Therefore, set callback for it. Must set once.
     void setExternalTablesInitializer(ExternalTablesInitializer && initializer);
@@ -737,6 +764,10 @@ public:
     QueryFactoriesInfo getQueryFactoriesInfo() const;
     void addQueryFactoriesInfo(QueryLogFactories factory_type, const String & created_object) const;
 
+    const QueryPrivilegesInfo & getQueryPrivilegesInfo() const { return *getQueryPrivilegesInfoPtr(); }
+    QueryPrivilegesInfoPtr getQueryPrivilegesInfoPtr() const { return query_privileges_info; }
+    void addQueryPrivilegesInfo(const String & privilege, bool granted) const;
+
     /// For table functions s3/file/url/hdfs/input we can use structure from
     /// insertion table depending on select expression.
     StoragePtr executeTableFunction(const ASTPtr & table_expression, const ASTSelectQuery * select_query_hint = nullptr);
@@ -907,6 +938,8 @@ public:
     void setSessionContext(ContextMutablePtr context_) { session_context = context_; }
 
     void makeQueryContext();
+    void makeQueryContextForMerge(const MergeTreeSettings & merge_tree_settings);
+    void makeQueryContextForMutate(const MergeTreeSettings & merge_tree_settings);
     void makeSessionContext();
     void makeGlobalContext();
 

src/Interpreters/DatabaseCatalog.h

@@ -129,6 +129,7 @@ public:
     static constexpr const char * SYSTEM_DATABASE = "system";
     static constexpr const char * INFORMATION_SCHEMA = "information_schema";
     static constexpr const char * INFORMATION_SCHEMA_UPPERCASE = "INFORMATION_SCHEMA";
+    static constexpr const char * DEFAULT_DATABASE = "default";
 
     /// Returns true if a passed name is one of the predefined databases' names.
     static bool isPredefinedDatabase(std::string_view database_name);

src/Interpreters/InterpreterCreateQuery.cpp

@@ -88,6 +88,11 @@
 #include <Interpreters/ReplaceQueryParameterVisitor.h>
 #include <Parsers/QueryParameterVisitor.h>
 
+namespace CurrentMetrics
+{
+    extern const Metric AttachedTable;
+}
+
 namespace DB
 {
 
@@ -113,6 +118,8 @@ namespace ErrorCodes
     extern const int UNKNOWN_STORAGE;
     extern const int SYNTAX_ERROR;
     extern const int SUPPORT_IS_DISABLED;
+    extern const int TOO_MANY_TABLES;
+    extern const int TOO_MANY_DATABASES;
 }
 
 namespace fs = std::filesystem;
@@ -138,6 +145,31 @@ BlockIO InterpreterCreateQuery::createDatabase(ASTCreateQuery & create)
             throw Exception(ErrorCodes::DATABASE_ALREADY_EXISTS, "Database {} already exists.", database_name);
     }
 
+    auto db_num_limit = getContext()->getGlobalContext()->getServerSettings().max_database_num_to_throw;
+    if (db_num_limit > 0)
+    {
+        size_t db_count = DatabaseCatalog::instance().getDatabases().size();
+        std::vector<String> system_databases = {
+            DatabaseCatalog::TEMPORARY_DATABASE,
+            DatabaseCatalog::SYSTEM_DATABASE,
+            DatabaseCatalog::INFORMATION_SCHEMA,
+            DatabaseCatalog::INFORMATION_SCHEMA_UPPERCASE,
+            DatabaseCatalog::DEFAULT_DATABASE
+        };
+
+        for (const auto & system_database : system_databases)
+        {
+            if (db_count > 0 && DatabaseCatalog::instance().isDatabaseExist(system_database))
+                db_count--;
+        }
+
+        if (db_count >= db_num_limit)
+            throw Exception(ErrorCodes::TOO_MANY_DATABASES,
+                            "Too many databases in the Clickhouse. "
+                            "The limit (setting 'max_database_num_to_throw') is set to {}, current number of databases is {}",
+                            db_num_limit, db_count);
+    }
+
     /// Will write file with database metadata, if needed.
     String database_name_escaped = escapeForFileName(database_name);
     fs::path metadata_path = fs::weakly_canonical(getContext()->getPath());
@@ -1543,6 +1575,17 @@ bool InterpreterCreateQuery::doCreateTable(ASTCreateQuery & create,
         }
     }
 
+    UInt64 table_num_limit = getContext()->getGlobalContext()->getServerSettings().max_table_num_to_throw;
+    if (table_num_limit > 0 && create.getDatabase() != DatabaseCatalog::SYSTEM_DATABASE)
+    {
+        UInt64 table_count = CurrentMetrics::get(CurrentMetrics::AttachedTable);
+        if (table_count >= table_num_limit)
+            throw Exception(ErrorCodes::TOO_MANY_TABLES,
+                            "Too many tables in the Clickhouse. "
+                            "The limit (setting 'max_table_num_to_throw') is set to {}, current number of tables is {}",
+                            table_num_limit, table_count);
+    }
+
     database->createTable(getContext(), create.getTable(), res, query_ptr);
 
     /// Move table data to the proper place. Wo do not move data earlier to avoid situations

src/Interpreters/InterpreterInsertQuery.cpp

@@ -26,7 +26,8 @@
 #include <Processors/Transforms/CountingTransform.h>
 #include <Processors/Transforms/ExpressionTransform.h>
 #include <Processors/Transforms/MaterializingTransform.h>
-#include <Processors/Transforms/SquashingChunksTransform.h>
+#include <Processors/Transforms/SquashingTransform.h>
+#include <Processors/Transforms/PlanSquashingTransform.h>
 #include <Processors/Transforms/getSourceFromASTInsertQuery.h>
 #include <Processors/QueryPlan/QueryPlan.h>
 #include <QueryPipeline/QueryPipelineBuilder.h>
@@ -625,9 +626,15 @@ BlockIO InterpreterInsertQuery::execute()
         {
             bool table_prefers_large_blocks = table->prefersLargeBlocks();
 
+            pipeline.addTransform(std::make_shared<PlanSquashingTransform>(
+                    header,
+                    table_prefers_large_blocks ? settings.min_insert_block_size_rows : settings.max_block_size,
+                    table_prefers_large_blocks ? settings.min_insert_block_size_bytes : 0ULL,
+                    presink_chains.size()));
+
             pipeline.addSimpleTransform([&](const Block & in_header) -> ProcessorPtr
             {
-                return std::make_shared<SimpleSquashingChunksTransform>(
+                return std::make_shared<ApplySquashingTransform>(
                     in_header,
                     table_prefers_large_blocks ? settings.min_insert_block_size_rows : settings.max_block_size,
                     table_prefers_large_blocks ? settings.min_insert_block_size_bytes : 0ULL);
@@ -683,12 +690,20 @@ BlockIO InterpreterInsertQuery::execute()
         {
             bool table_prefers_large_blocks = table->prefersLargeBlocks();
 
-            auto squashing = std::make_shared<SimpleSquashingChunksTransform>(
-                chain.getInputHeader(),
-                table_prefers_large_blocks ? settings.min_insert_block_size_rows : settings.max_block_size,
-                table_prefers_large_blocks ? settings.min_insert_block_size_bytes : 0ULL);
+            auto squashing = std::make_shared<ApplySquashingTransform>(
+                    chain.getInputHeader(),
+                    table_prefers_large_blocks ? settings.min_insert_block_size_rows : settings.max_block_size,
+                    table_prefers_large_blocks ? settings.min_insert_block_size_bytes : 0ULL);
 
             chain.addSource(std::move(squashing));
+
+            auto balancing = std::make_shared<PlanSquashingTransform>(
+                    chain.getInputHeader(),
+                    table_prefers_large_blocks ? settings.min_insert_block_size_rows : settings.max_block_size,
+                    table_prefers_large_blocks ? settings.min_insert_block_size_bytes : 0ULL,
+                    presink_chains.size());
+
+            chain.addSource(std::move(balancing));
         }
 
         auto context_ptr = getContext();

src/Interpreters/InterpreterSelectQuery.cpp

@@ -1474,6 +1474,9 @@ void InterpreterSelectQuery::executeImpl(QueryPlan & query_plan, std::optional<P
     if (expressions.hasHaving() && query.group_by_with_totals && (query.group_by_with_rollup || query.group_by_with_cube))
         throw Exception(ErrorCodes::NOT_IMPLEMENTED, "WITH TOTALS and WITH ROLLUP or CUBE are not supported together in presence of HAVING");
 
+    if (query.qualify())
+        throw Exception(ErrorCodes::NOT_IMPLEMENTED, "QUALIFY clause is not supported in the old analyzer");
+
     if (options.only_analyze)
     {
         auto read_nothing = std::make_unique<ReadNothingStep>(source_header);

src/Interpreters/QueryLog.cpp

@@ -136,6 +136,9 @@ ColumnsDescription QueryLogElement::getColumnsDescription()
 
         {"used_row_policies", array_low_cardinality_string, "The list of row policies names that were used during query execution."},
 
+        {"used_privileges", array_low_cardinality_string, "Privileges which were successfully checked during query execution."},
+        {"missing_privileges", array_low_cardinality_string, "Privileges that are missing during query execution."},
+
         {"transaction_id", getTransactionIDDataType(), "The identifier of the transaction in scope of which this query was executed."},
 
         {"query_cache_usage", std::move(query_cache_usage_datatype), "Usage of the query cache during query execution. Values: 'Unknown' = Status unknown, 'None' = The query result was neither written into nor read from the query cache, 'Write' = The query result was written into the query cache, 'Read' = The query result was read from the query cache."},
@@ -267,6 +270,8 @@ void QueryLogElement::appendToBlock(MutableColumns & columns) const
         auto & column_storage_factory_objects = typeid_cast<ColumnArray &>(*columns[i++]);
         auto & column_table_function_factory_objects = typeid_cast<ColumnArray &>(*columns[i++]);
         auto & column_row_policies_names = typeid_cast<ColumnArray &>(*columns[i++]);
+        auto & column_used_privileges = typeid_cast<ColumnArray &>(*columns[i++]);
+        auto & column_missing_privileges = typeid_cast<ColumnArray &>(*columns[i++]);
 
         auto fill_column = [](const auto & data, ColumnArray & column)
         {
@@ -290,6 +295,8 @@ void QueryLogElement::appendToBlock(MutableColumns & columns) const
         fill_column(used_storages, column_storage_factory_objects);
         fill_column(used_table_functions, column_table_function_factory_objects);
         fill_column(used_row_policies, column_row_policies_names);
+        fill_column(used_privileges, column_used_privileges);
+        fill_column(missing_privileges, column_missing_privileges);
     }
 
     columns[i++]->insert(Tuple{tid.start_csn, tid.local_tid, tid.host_id});

src/Interpreters/QueryLog.h

@@ -81,6 +81,8 @@ struct QueryLogElement
     std::unordered_set<String> used_storages;
     std::unordered_set<String> used_table_functions;
     std::set<String> used_row_policies;
+    std::unordered_set<String> used_privileges;
+    std::unordered_set<String> missing_privileges;
 
     Int32 exception_code{}; // because ErrorCodes are int
     String exception;

src/Interpreters/Session.cpp

@@ -532,7 +532,7 @@ ContextMutablePtr Session::makeSessionContext()
     session_context->checkSettingsConstraints(settings_from_auth_server, SettingSource::QUERY);
     session_context->applySettingsChanges(settings_from_auth_server);
 
-    recordLoginSucess(session_context);
+    recordLoginSuccess(session_context);
 
     return session_context;
 }
@@ -596,7 +596,7 @@ ContextMutablePtr Session::makeSessionContext(const String & session_name_, std:
         { session_name_ },
         max_sessions_for_user);
 
-    recordLoginSucess(session_context);
+    recordLoginSuccess(session_context);
 
     return session_context;
 }
@@ -672,13 +672,13 @@ ContextMutablePtr Session::makeQueryContextImpl(const ClientInfo * client_info_t
         user = query_context->getUser();
 
     /// Interserver does not create session context
-    recordLoginSucess(query_context);
+    recordLoginSuccess(query_context);
 
     return query_context;
 }
 
 
-void Session::recordLoginSucess(ContextPtr login_context) const
+void Session::recordLoginSuccess(ContextPtr login_context) const
 {
     if (notified_session_log_about_login)
         return;
@@ -694,7 +694,7 @@ void Session::recordLoginSucess(ContextPtr login_context) const
         session_log->addLoginSuccess(auth_id,
                                      named_session ? named_session->key.second : "",
                                      settings,
-                                     access,
+                                     access->getAccess(),
                                      getClientInfo(),
                                      user);
     }

src/Interpreters/Session.h

@@ -102,8 +102,7 @@ public:
 private:
     std::shared_ptr<SessionLog> getSessionLog() const;
     ContextMutablePtr makeQueryContextImpl(const ClientInfo * client_info_to_copy, ClientInfo * client_info_to_move) const;
-    void recordLoginSucess(ContextPtr login_context) const;
-
+    void recordLoginSuccess(ContextPtr login_context) const;
 
     mutable bool notified_session_log_about_login = false;
     const UUID auth_id;

src/Interpreters/SessionLog.cpp

@@ -214,7 +214,7 @@ void SessionLog::addLoginSuccess(const UUID & auth_id,
                                  const ClientInfo & client_info,
                                  const UserPtr & login_user)
 {
-    DB::SessionLogElement log_entry(auth_id, SESSION_LOGIN_SUCCESS);
+    SessionLogElement log_entry(auth_id, SESSION_LOGIN_SUCCESS);
     log_entry.client_info = client_info;
 
     if (login_user)

src/Interpreters/Squashing.cpp

@@ -0,0 +1,159 @@
+#include <vector>
+#include <Interpreters/Squashing.h>
+#include <Common/CurrentThread.h>
+
+
+namespace DB
+{
+namespace ErrorCodes
+{
+    extern const int LOGICAL_ERROR;
+}
+
+Squashing::Squashing(Block header_, size_t min_block_size_rows_, size_t min_block_size_bytes_)
+    : header(header_)
+    , min_block_size_rows(min_block_size_rows_)
+    , min_block_size_bytes(min_block_size_bytes_)
+{
+}
+
+Chunk Squashing::flush()
+{
+    return convertToChunk(std::move(chunks_to_merge_vec));
+}
+
+Chunk Squashing::squash(Chunk && input_chunk)
+{
+    if (!input_chunk.hasChunkInfo())
+        return Chunk();
+
+    const auto *info = getInfoFromChunk(input_chunk);
+    return squash(info->chunks);
+}
+
+Chunk Squashing::add(Chunk && input_chunk)
+{
+    if (!input_chunk)
+        return {};
+
+    /// Just read block is already enough.
+    if (isEnoughSize(input_chunk.getNumRows(), input_chunk.bytes()))
+    {
+        /// If no accumulated data, return just read block.
+        if (chunks_to_merge_vec.empty())
+        {
+            chunks_to_merge_vec.push_back(std::move(input_chunk));
+            Chunk res_chunk = convertToChunk(std::move(chunks_to_merge_vec));
+            chunks_to_merge_vec.clear();
+            return res_chunk;
+        }
+
+        /// Return accumulated data (maybe it has small size) and place new block to accumulated data.
+        Chunk res_chunk = convertToChunk(std::move(chunks_to_merge_vec));
+        chunks_to_merge_vec.clear();
+        changeCurrentSize(input_chunk.getNumRows(), input_chunk.bytes());
+        chunks_to_merge_vec.push_back(std::move(input_chunk));
+        return res_chunk;
+    }
+
+    /// Accumulated block is already enough.
+    if (isEnoughSize(accumulated_size.rows, accumulated_size.bytes))
+    {
+        /// Return accumulated data and place new block to accumulated data.
+        Chunk res_chunk = convertToChunk(std::move(chunks_to_merge_vec));
+        chunks_to_merge_vec.clear();
+        changeCurrentSize(input_chunk.getNumRows(), input_chunk.bytes());
+        chunks_to_merge_vec.push_back(std::move(input_chunk));
+        return res_chunk;
+    }
+
+    /// Pushing data into accumulating vector
+    expandCurrentSize(input_chunk.getNumRows(), input_chunk.bytes());
+    chunks_to_merge_vec.push_back(std::move(input_chunk));
+
+    /// If accumulated data is big enough, we send it
+    if (isEnoughSize(accumulated_size.rows, accumulated_size.bytes))
+    {
+        Chunk res_chunk = convertToChunk(std::move(chunks_to_merge_vec));
+        changeCurrentSize(0, 0);
+        chunks_to_merge_vec.clear();
+        return res_chunk;
+    }
+    return {};
+}
+
+Chunk Squashing::convertToChunk(std::vector<Chunk> && chunks) const
+{
+    if (chunks.empty())
+        return {};
+
+    auto info = std::make_shared<ChunksToSquash>();
+    info->chunks = std::move(chunks);
+
+    chunks.clear();
+
+    return Chunk(header.cloneEmptyColumns(), 0, info);
+}
+
+Chunk Squashing::squash(std::vector<Chunk> & input_chunks)
+{
+    Chunk accumulated_chunk;
+    std::vector<IColumn::MutablePtr> mutable_columns = {};
+    size_t rows = 0;
+    for (const Chunk & chunk : input_chunks)
+        rows += chunk.getNumRows();
+
+    {
+        auto & first_chunk = input_chunks[0];
+        Columns columns = first_chunk.detachColumns();
+        for (auto & column : columns)
+        {
+            mutable_columns.push_back(IColumn::mutate(std::move(column)));
+            mutable_columns.back()->reserve(rows);
+        }
+    }
+
+    for (size_t i = 1; i < input_chunks.size(); ++i) // We've already processed the first chunk above
+    {
+        Columns columns = input_chunks[i].detachColumns();
+        for (size_t j = 0, size = mutable_columns.size(); j < size; ++j)
+        {
+            const auto source_column = columns[j];
+
+            mutable_columns[j]->insertRangeFrom(*source_column, 0, source_column->size());
+        }
+    }
+    accumulated_chunk.setColumns(std::move(mutable_columns), rows);
+    return accumulated_chunk;
+}
+
+const ChunksToSquash* Squashing::getInfoFromChunk(const Chunk & chunk)
+{
+    const auto& info = chunk.getChunkInfo();
+    const auto * agg_info = typeid_cast<const ChunksToSquash *>(info.get());
+
+    if (!agg_info)
+        throw Exception(ErrorCodes::LOGICAL_ERROR, "There is no ChunksToSquash in ChunkInfoPtr");
+
+    return agg_info;
+}
+
+void Squashing::expandCurrentSize(size_t rows, size_t bytes)
+{
+    accumulated_size.rows += rows;
+    accumulated_size.bytes += bytes;
+}
+
+void Squashing::changeCurrentSize(size_t rows, size_t bytes)
+{
+    accumulated_size.rows = rows;
+    accumulated_size.bytes = bytes;
+}
+
+bool Squashing::isEnoughSize(size_t rows, size_t bytes) const
+{
+    return (!min_block_size_rows && !min_block_size_bytes)
+        || (min_block_size_rows && rows >= min_block_size_rows)
+        || (min_block_size_bytes && bytes >= min_block_size_bytes);
+}
+}

src/Interpreters/Squashing.h

@@ -0,0 +1,69 @@
+#pragma once
+
+#include <vector>
+#include <Core/Block.h>
+#include <Processors/Chunk.h>
+
+
+namespace DB
+{
+
+struct ChunksToSquash : public ChunkInfo
+{
+    mutable std::vector<Chunk> chunks = {};
+};
+
+/** Merging consecutive passed blocks to specified minimum size.
+  *
+  * (But if one of input blocks has already at least specified size,
+  *  then don't merge it with neighbours, even if neighbours are small.)
+  *
+  * Used to prepare blocks to adequate size for INSERT queries,
+  *  because such storages as Memory, StripeLog, Log, TinyLog...
+  *  store or compress data in blocks exactly as passed to it,
+  *  and blocks of small size are not efficient.
+  *
+  * Order of data is kept.
+  */
+
+class Squashing
+{
+public:
+    explicit Squashing(Block header_, size_t min_block_size_rows_, size_t min_block_size_bytes_);
+    Squashing(Squashing && other) = default;
+
+    Chunk add(Chunk && input_chunk);
+    static Chunk squash(Chunk && input_chunk);
+    Chunk flush();
+
+    bool isDataLeft()
+    {
+        return !chunks_to_merge_vec.empty();
+    }
+
+    Block header;
+private:
+    struct CurrentSize
+    {
+        size_t rows = 0;
+        size_t bytes = 0;
+    };
+
+    std::vector<Chunk> chunks_to_merge_vec = {};
+    size_t min_block_size_rows;
+    size_t min_block_size_bytes;
+
+    CurrentSize accumulated_size;
+
+    static const ChunksToSquash * getInfoFromChunk(const Chunk & chunk);
+
+    static Chunk squash(std::vector<Chunk> & input_chunks);
+
+    void expandCurrentSize(size_t rows, size_t bytes);
+    void changeCurrentSize(size_t rows, size_t bytes);
+    bool isEnoughSize(size_t rows, size_t bytes) const;
+
+    Chunk convertToChunk(std::vector<Chunk> && chunks) const;
+};
+
+}

src/Interpreters/SquashingTransform.cpp

@@ -1,145 +0,0 @@
-#include <Interpreters/SquashingTransform.h>
-
-
-namespace DB
-{
-namespace ErrorCodes
-{
-    extern const int SIZES_OF_COLUMNS_DOESNT_MATCH;
-    extern const int LOGICAL_ERROR;
-}
-
-SquashingTransform::SquashingTransform(size_t min_block_size_rows_, size_t min_block_size_bytes_)
-    : min_block_size_rows(min_block_size_rows_)
-    , min_block_size_bytes(min_block_size_bytes_)
-{
-}
-
-Block SquashingTransform::add(Block && input_block)
-{
-    return addImpl<Block &&>(std::move(input_block));
-}
-
-Block SquashingTransform::add(const Block & input_block)
-{
-    return addImpl<const Block &>(input_block);
-}
-
-/*
- * To minimize copying, accept two types of argument: const reference for output
- * stream, and rvalue reference for input stream, and decide whether to copy
- * inside this function. This allows us not to copy Block unless we absolutely
- * have to.
- */
-template <typename ReferenceType>
-Block SquashingTransform::addImpl(ReferenceType input_block)
-{
-    /// End of input stream.
-    if (!input_block)
-    {
-        Block to_return;
-        std::swap(to_return, accumulated_block);
-        return to_return;
-    }
-
-    /// Just read block is already enough.
-    if (isEnoughSize(input_block))
-    {
-        /// If no accumulated data, return just read block.
-        if (!accumulated_block)
-        {
-            return std::move(input_block);
-        }
-
-        /// Return accumulated data (maybe it has small size) and place new block to accumulated data.
-        Block to_return = std::move(input_block);
-        std::swap(to_return, accumulated_block);
-        return to_return;
-    }
-
-    /// Accumulated block is already enough.
-    if (isEnoughSize(accumulated_block))
-    {
-        /// Return accumulated data and place new block to accumulated data.
-        Block to_return = std::move(input_block);
-        std::swap(to_return, accumulated_block);
-        return to_return;
-    }
-
-    append<ReferenceType>(std::move(input_block));
-    if (isEnoughSize(accumulated_block))
-    {
-        Block to_return;
-        std::swap(to_return, accumulated_block);
-        return to_return;
-    }
-
-    /// Squashed block is not ready.
-    return {};
-}
-
-
-template <typename ReferenceType>
-void SquashingTransform::append(ReferenceType input_block)
-{
-    if (!accumulated_block)
-    {
-        accumulated_block = std::move(input_block);
-        return;
-    }
-
-    assert(blocksHaveEqualStructure(input_block, accumulated_block));
-
-    try
-    {
-        for (size_t i = 0, size = accumulated_block.columns(); i < size; ++i)
-        {
-            const auto source_column = input_block.getByPosition(i).column;
-
-            auto mutable_column = IColumn::mutate(std::move(accumulated_block.getByPosition(i).column));
-            mutable_column->insertRangeFrom(*source_column, 0, source_column->size());
-            accumulated_block.getByPosition(i).column = std::move(mutable_column);
-        }
-    }
-    catch (...)
-    {
-        /// add() may be called again even after a previous add() threw an exception.
-        /// Keep accumulated_block in a valid state.
-        /// Seems ok to discard accumulated data because we're throwing an exception, which the caller will
-        /// hopefully interpret to mean "this block and all *previous* blocks are potentially lost".
-        accumulated_block.clear();
-        throw;
-    }
-}
-
-
-bool SquashingTransform::isEnoughSize(const Block & block)
-{
-    size_t rows = 0;
-    size_t bytes = 0;
-
-    for (const auto & [column, type, name] : block)
-    {
-        if (!column)
-            throw Exception(ErrorCodes::LOGICAL_ERROR, "Invalid column in block.");
-
-        if (!rows)
-            rows = column->size();
-        else if (rows != column->size())
-            throw Exception(ErrorCodes::SIZES_OF_COLUMNS_DOESNT_MATCH, "Sizes of columns doesn't match");
-
-        bytes += column->byteSize();
-    }
-
-    return isEnoughSize(rows, bytes);
-}
-
-
-bool SquashingTransform::isEnoughSize(size_t rows, size_t bytes) const
-{
-    return (!min_block_size_rows && !min_block_size_bytes)
-        || (min_block_size_rows && rows >= min_block_size_rows)
-        || (min_block_size_bytes && bytes >= min_block_size_bytes);
-}
-
-}

src/Interpreters/SquashingTransform.h

@@ -1,50 +0,0 @@
-#pragma once
-
-#include <Core/Block.h>
-
-
-namespace DB
-{
-
-
-/** Merging consecutive passed blocks to specified minimum size.
-  *
-  * (But if one of input blocks has already at least specified size,
-  *  then don't merge it with neighbours, even if neighbours are small.)
-  *
-  * Used to prepare blocks to adequate size for INSERT queries,
-  *  because such storages as Memory, StripeLog, Log, TinyLog...
-  *  store or compress data in blocks exactly as passed to it,
-  *  and blocks of small size are not efficient.
-  *
-  * Order of data is kept.
-  */
-class SquashingTransform
-{
-public:
-    /// Conditions on rows and bytes are OR-ed. If one of them is zero, then corresponding condition is ignored.
-    SquashingTransform(size_t min_block_size_rows_, size_t min_block_size_bytes_);
-
-    /** Add next block and possibly returns squashed block.
-      * At end, you need to pass empty block. As the result for last (empty) block, you will get last Result with ready = true.
-      */
-    Block add(Block && block);
-    Block add(const Block & block);
-
-private:
-    size_t min_block_size_rows;
-    size_t min_block_size_bytes;
-
-    Block accumulated_block;
-
-    template <typename ReferenceType>
-    Block addImpl(ReferenceType block);
-
-    template <typename ReferenceType>
-    void append(ReferenceType block);
-
-    bool isEnoughSize(const Block & block);
-    bool isEnoughSize(size_t rows, size_t bytes) const;
-};
-
-}

src/Interpreters/convertFieldToType.cpp

@@ -615,9 +615,9 @@ static bool decimalEqualsFloat(Field field, Float64 float_value)
     return decimal_to_float == float_value;
 }
 
-std::optional<Field> convertFieldToTypeStrict(const Field & from_value, const IDataType & to_type)
+std::optional<Field> convertFieldToTypeStrict(const Field & from_value, const IDataType & from_type, const IDataType & to_type)
 {
-    Field result_value = convertFieldToType(from_value, to_type);
+    Field result_value = convertFieldToType(from_value, to_type, &from_type);
 
     if (Field::isDecimal(from_value.getType()) && Field::isDecimal(result_value.getType()))
     {

src/Interpreters/convertFieldToType.h

@@ -22,6 +22,6 @@ Field convertFieldToTypeOrThrow(const Field & from_value, const IDataType & to_t
 
 /// Applies stricter rules than convertFieldToType, doesn't allow loss of precision converting to Decimal.
 /// Returns `Field` if the conversion was successful and the result is equal to the original value, otherwise returns nullopt.
-std::optional<Field> convertFieldToTypeStrict(const Field & from_value, const IDataType & to_type);
+std::optional<Field> convertFieldToTypeStrict(const Field & from_value, const IDataType & from_type, const IDataType & to_type);
 
 }

src/Interpreters/executeQuery.cpp

@@ -44,6 +44,7 @@
 #include <Formats/FormatFactory.h>
 #include <Storages/StorageInput.h>
 
+#include <Access/ContextAccess.h>
 #include <Access/EnabledQuota.h>
 #include <Interpreters/ApplyWithGlobalVisitor.h>
 #include <Interpreters/Context.h>
@@ -221,6 +222,17 @@ static void logException(ContextPtr context, QueryLogElement & elem, bool log_er
         LOG_INFO(getLogger("executeQuery"), message);
 }
 
+static void
+addPrivilegesInfoToQueryLogElement(QueryLogElement & element, const ContextPtr context_ptr)
+{
+    const auto & privileges_info = context_ptr->getQueryPrivilegesInfo();
+    {
+        std::lock_guard lock(privileges_info.mutex);
+        element.used_privileges = privileges_info.used_privileges;
+        element.missing_privileges = privileges_info.missing_privileges;
+    }
+}
+
 static void
 addStatusInfoToQueryLogElement(QueryLogElement & element, const QueryStatusInfo & info, const ASTPtr query_ast, const ContextPtr context_ptr)
 {
@@ -286,6 +298,7 @@ addStatusInfoToQueryLogElement(QueryLogElement & element, const QueryStatusInfo
     }
 
     element.async_read_counters = context_ptr->getAsyncReadCounters();
+    addPrivilegesInfoToQueryLogElement(element, context_ptr);
 }
 
 
@@ -601,6 +614,8 @@ void logExceptionBeforeStart(
             elem.formatted_query = queryToString(ast);
     }
 
+    addPrivilegesInfoToQueryLogElement(elem, context);
+
     // We don't calculate databases, tables and columns when the query isn't able to start
 
     elem.exception_code = getCurrentExceptionCode();

src/Parsers/ExpressionListParsers.cpp

@@ -2179,7 +2179,7 @@ public:
 
     bool parse(IParser::Pos & pos, Expected & expected, Action & /*action*/) override
     {
-        /// kql(table|project ...)
+        /// kql('table|project ...')
         /// 0. Parse the kql query
         /// 1. Parse closing token
         if (state == 0)

src/Parsers/Kusto/KQL_ReleaseNote.md

@@ -853,7 +853,7 @@ Please note that the functions listed below only take constant parameters for no
 ## KQL() function
 
  - create table  
-  `CREATE TABLE kql_table4 ENGINE = Memory AS select *, now() as new_column From kql(Customers | project LastName,Age);`  
+  `CREATE TABLE kql_table4 ENGINE = Memory AS select *, now() as new_column From kql($$Customers | project LastName,Age$$);`  
    verify the content of `kql_table`  
     `select * from kql_table`
    
@@ -867,12 +867,12 @@ Please note that the functions listed below only take constant parameters for no
         Age Nullable(UInt8)
     ) ENGINE = Memory;
     ```
-    `INSERT INTO temp select * from kql(Customers|project FirstName,LastName,Age);`  
+    `INSERT INTO temp select * from kql($$Customers|project FirstName,LastName,Age$$);`  
     verify the content of `temp`   
         `select * from temp`
 
- - Select from kql()  
-    `Select * from kql(Customers|project FirstName)`
+ - Select from kql(...)  
+    `Select * from kql($$Customers|project FirstName$$)`
 
 ## KQL operators:
  - Tabular expression statements  
@@ -993,4 +993,3 @@ Please note that the functions listed below only take constant parameters for no
  - dcount()
  - dcountif()
  - bin
- 

src/Parsers/Kusto/KustoFunctions/IParserKQLFunction.cpp

@@ -301,8 +301,8 @@ String IParserKQLFunction::kqlCallToExpression(
         });
 
     const auto kql_call = std::format("{}({})", function_name, params_str);
-    DB::Tokens call_tokens(kql_call.c_str(), kql_call.c_str() + kql_call.length());
-    DB::IParser::Pos tokens_pos(call_tokens, max_depth, max_backtracks);
+    Tokens call_tokens(kql_call.data(), kql_call.data() + kql_call.length(), 0, true);
+    IParser::Pos tokens_pos(call_tokens, max_depth, max_backtracks);
     return DB::IParserKQLFunction::getExpression(tokens_pos);
 }
 

src/Parsers/Kusto/ParserKQLDistinct.cpp

@@ -11,7 +11,7 @@ bool ParserKQLDistinct::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
 
     expr = getExprFromToken(pos);
 
-    Tokens tokens(expr.c_str(), expr.c_str() + expr.size());
+    Tokens tokens(expr.data(), expr.data() + expr.size(), 0, true);
     IParser::Pos new_pos(tokens, pos.max_depth, pos.max_backtracks);
 
     if (!ParserNotEmptyExpressionList(false).parse(new_pos, select_expression_list, expected))

src/Parsers/Kusto/ParserKQLExtend.cpp

@@ -22,7 +22,7 @@ bool ParserKQLExtend ::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
 
     String except_str;
     String new_extend_str;
-    Tokens ntokens(extend_expr.c_str(), extend_expr.c_str() + extend_expr.size());
+    Tokens ntokens(extend_expr.data(), extend_expr.data() + extend_expr.size(), 0, true);
     IParser::Pos npos(ntokens, pos.max_depth, pos.max_backtracks);
 
     String alias;
@@ -76,7 +76,7 @@ bool ParserKQLExtend ::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
     apply_alias();
 
     String expr = std::format("SELECT * {}, {} from prev", except_str, new_extend_str);
-    Tokens tokens(expr.c_str(), expr.c_str() + expr.size());
+    Tokens tokens(expr.data(), expr.data() + expr.size(), 0, true);
     IParser::Pos new_pos(tokens, pos.max_depth, pos.max_backtracks);
 
     if (!ParserSelectQuery().parse(new_pos, select_query, expected))

src/Parsers/Kusto/ParserKQLFilter.cpp

@@ -13,7 +13,7 @@ bool ParserKQLFilter::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
     String expr = getExprFromToken(pos);
     ASTPtr where_expression;
 
-    Tokens token_filter(expr.c_str(), expr.c_str() + expr.size());
+    Tokens token_filter(expr.data(), expr.data() + expr.size(), 0, true);
     IParser::Pos pos_filter(token_filter, pos.max_depth, pos.max_backtracks);
     if (!ParserExpressionWithOptionalAlias(false).parse(pos_filter, where_expression, expected))
         return false;

src/Parsers/Kusto/ParserKQLLimit.cpp

@@ -13,7 +13,7 @@ bool ParserKQLLimit::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
 
     auto expr = getExprFromToken(pos);
 
-    Tokens tokens(expr.c_str(), expr.c_str() + expr.size());
+    Tokens tokens(expr.data(), expr.data() + expr.size(), 0, true);
     IParser::Pos new_pos(tokens, pos.max_depth, pos.max_backtracks);
 
     if (!ParserExpressionWithOptionalAlias(false).parse(new_pos, limit_length, expected))

src/Parsers/Kusto/ParserKQLMVExpand.cpp

@@ -298,7 +298,7 @@ bool ParserKQLMVExpand::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
         return false;
 
     const String setting_str = "enable_unaligned_array_join = 1";
-    Tokens token_settings(setting_str.c_str(), setting_str.c_str() + setting_str.size());
+    Tokens token_settings(setting_str.data(), setting_str.data() + setting_str.size(), 0, true);
     IParser::Pos pos_settings(token_settings, pos.max_depth, pos.max_backtracks);
 
     if (!ParserSetQuery(true).parse(pos_settings, setting, expected))

src/Parsers/Kusto/ParserKQLMakeSeries.cpp

@@ -173,7 +173,7 @@ bool ParserKQLMakeSeries ::makeSeries(KQLMakeSeries & kql_make_series, ASTPtr &
 
     auto date_type_cast = [&](String & src)
     {
-        Tokens tokens(src.c_str(), src.c_str() + src.size());
+        Tokens tokens(src.data(), src.data() + src.size(), 0, true);
         IParser::Pos pos(tokens, max_depth, max_backtracks);
         String res;
         while (isValidKQLPos(pos))
@@ -200,7 +200,7 @@ bool ParserKQLMakeSeries ::makeSeries(KQLMakeSeries & kql_make_series, ASTPtr &
     auto get_group_expression_alias = [&]
     {
         std::vector<String> group_expression_tokens;
-        Tokens tokens(group_expression.c_str(), group_expression.c_str() + group_expression.size());
+        Tokens tokens(group_expression.data(), group_expression.data() + group_expression.size(), 0, true);
         IParser::Pos pos(tokens, max_depth, max_backtracks);
         while (isValidKQLPos(pos))
         {
@@ -413,7 +413,7 @@ bool ParserKQLMakeSeries ::parseImpl(Pos & pos, ASTPtr & node, Expected & expect
 
     makeSeries(kql_make_series, node, pos.max_depth, pos.max_backtracks);
 
-    Tokens token_main_query(kql_make_series.main_query.c_str(), kql_make_series.main_query.c_str() + kql_make_series.main_query.size());
+    Tokens token_main_query(kql_make_series.main_query.data(), kql_make_series.main_query.data() + kql_make_series.main_query.size(), 0, true);
     IParser::Pos pos_main_query(token_main_query, pos.max_depth, pos.max_backtracks);
 
     if (!ParserNotEmptyExpressionList(true).parse(pos_main_query, select_expression_list, expected))

src/Parsers/Kusto/ParserKQLOperators.cpp

@@ -1,20 +1,26 @@
 #include <Parsers/ASTLiteral.h>
 #include <Parsers/CommonParsers.h>
 #include <Parsers/Kusto/KustoFunctions/IParserKQLFunction.h>
-#include <Parsers/Kusto/KustoFunctions/KQLFunctionFactory.h>
 #include <Parsers/Kusto/ParserKQLOperators.h>
-#include <Parsers/Kusto/ParserKQLQuery.h>
 #include <Parsers/Kusto/ParserKQLStatement.h>
 #include <Parsers/Kusto/Utilities.h>
 #include <Parsers/ASTFunction.h>
 #include <Parsers/ASTIdentifier.h>
 #include <Parsers/formatAST.h>
-#include "KustoFunctions/IParserKQLFunction.h"
+
+
+namespace DB
+{
+
+namespace ErrorCodes
+{
+    extern const int SYNTAX_ERROR;
+}
 
 namespace
 {
 
-enum class KQLOperatorValue : uint16_t
+enum class KQLOperatorValue
 {
     none,
     between,
@@ -56,7 +62,8 @@ enum class KQLOperatorValue : uint16_t
     not_startswith_cs,
 };
 
-const std::unordered_map<String, KQLOperatorValue> KQLOperator = {
+const std::unordered_map<String, KQLOperatorValue> KQLOperator =
+{
     {"between", KQLOperatorValue::between},
     {"!between", KQLOperatorValue::not_between},
     {"contains", KQLOperatorValue::contains},
@@ -96,44 +103,37 @@ const std::unordered_map<String, KQLOperatorValue> KQLOperator = {
     {"!startswith_cs", KQLOperatorValue::not_startswith_cs},
 };
 
-void rebuildSubqueryForInOperator(DB::ASTPtr & node, bool useLowerCase)
+void rebuildSubqueryForInOperator(ASTPtr & node, bool useLowerCase)
 {
     //A sub-query for in operator in kql can have multiple columns, but only takes the first column.
     //A sub-query for in operator in ClickHouse can not have multiple columns
     //So only take the first column if there are multiple columns.
     //select * not working for subquery. (a tabular statement without project)
 
-    const auto selectColumns = node->children[0]->children[0]->as<DB::ASTSelectQuery>()->select();
+    const auto selectColumns = node->children[0]->children[0]->as<ASTSelectQuery>()->select();
     while (selectColumns->children.size() > 1)
         selectColumns->children.pop_back();
 
     if (useLowerCase)
     {
-        auto args = std::make_shared<DB::ASTExpressionList>();
+        auto args = std::make_shared<ASTExpressionList>();
         args->children.push_back(selectColumns->children[0]);
-        auto func_lower = std::make_shared<DB::ASTFunction>();
+        auto func_lower = std::make_shared<ASTFunction>();
         func_lower->name = "lower";
         func_lower->children.push_back(selectColumns->children[0]);
         func_lower->arguments = args;
-        if (selectColumns->children[0]->as<DB::ASTIdentifier>())
-            func_lower->alias = std::move(selectColumns->children[0]->as<DB::ASTIdentifier>()->alias);
-        else if (selectColumns->children[0]->as<DB::ASTFunction>())
-            func_lower->alias = std::move(selectColumns->children[0]->as<DB::ASTFunction>()->alias);
+        if (selectColumns->children[0]->as<ASTIdentifier>())
+            func_lower->alias = std::move(selectColumns->children[0]->as<ASTIdentifier>()->alias);
+        else if (selectColumns->children[0]->as<ASTFunction>())
+            func_lower->alias = std::move(selectColumns->children[0]->as<ASTFunction>()->alias);
 
-        auto funcs = std::make_shared<DB::ASTExpressionList>();
+        auto funcs = std::make_shared<ASTExpressionList>();
         funcs->children.push_back(func_lower);
         selectColumns->children[0] = std::move(funcs);
     }
 }
 
 }
-namespace DB
-{
-
-namespace ErrorCodes
-{
-    extern const int SYNTAX_ERROR;
-}
 
 String KQLOperators::genHasAnyAllOpExpr(std::vector<String> & tokens, IParser::Pos & token_pos, String kql_op, String ch_op)
 {
@@ -166,7 +166,7 @@ String KQLOperators::genHasAnyAllOpExpr(std::vector<String> & tokens, IParser::P
     return new_expr;
 }
 
-String genEqOpExprCis(std::vector<String> & tokens, DB::IParser::Pos & token_pos, const String & ch_op)
+String genEqOpExprCis(std::vector<String> & tokens, IParser::Pos & token_pos, const String & ch_op)
 {
     String tmp_arg(token_pos->begin, token_pos->end);
 
@@ -178,30 +178,30 @@ String genEqOpExprCis(std::vector<String> & tokens, DB::IParser::Pos & token_pos
     new_expr += ch_op + " ";
     ++token_pos;
 
-    if (token_pos->type == DB::TokenType::StringLiteral || token_pos->type == DB::TokenType::QuotedIdentifier)
-        new_expr += "lower('" + DB::IParserKQLFunction::escapeSingleQuotes(String(token_pos->begin + 1, token_pos->end - 1)) + "')";
+    if (token_pos->type == TokenType::StringLiteral || token_pos->type == TokenType::QuotedIdentifier)
+        new_expr += "lower('" + IParserKQLFunction::escapeSingleQuotes(String(token_pos->begin + 1, token_pos->end - 1)) + "')";
     else
-        new_expr += "lower(" + DB::IParserKQLFunction::getExpression(token_pos) + ")";
+        new_expr += "lower(" + IParserKQLFunction::getExpression(token_pos) + ")";
 
     tokens.pop_back();
     return new_expr;
 }
 
-String genInOpExprCis(std::vector<String> & tokens, DB::IParser::Pos & token_pos, const String & kql_op, const String & ch_op)
+String genInOpExprCis(std::vector<String> & tokens, IParser::Pos & token_pos, const String & kql_op, const String & ch_op)
 {
-    DB::ParserKQLTableFunction kqlfun_p;
-    DB::ParserToken s_lparen(DB::TokenType::OpeningRoundBracket);
+    ParserKQLTableFunction kqlfun_p;
+    ParserToken s_lparen(TokenType::OpeningRoundBracket);
 
-    DB::ASTPtr select;
-    DB::Expected expected;
+    ASTPtr select;
+    Expected expected;
     String new_expr;
 
     ++token_pos;
     if (!s_lparen.ignore(token_pos, expected))
-        throw DB::Exception(DB::ErrorCodes::SYNTAX_ERROR, "Syntax error near {}", kql_op);
+        throw Exception(ErrorCodes::SYNTAX_ERROR, "Syntax error near {}", kql_op);
 
     if (tokens.empty())
-        throw DB::Exception(DB::ErrorCodes::SYNTAX_ERROR, "Syntax error near {}", kql_op);
+        throw Exception(ErrorCodes::SYNTAX_ERROR, "Syntax error near {}", kql_op);
 
     new_expr = "lower(" + tokens.back() + ") ";
     tokens.pop_back();
@@ -218,39 +218,39 @@ String genInOpExprCis(std::vector<String> & tokens, DB::IParser::Pos & token_pos
     --token_pos;
 
     new_expr += ch_op;
-    while (isValidKQLPos(token_pos) && token_pos->type != DB::TokenType::PipeMark && token_pos->type != DB::TokenType::Semicolon)
+    while (isValidKQLPos(token_pos) && token_pos->type != TokenType::PipeMark && token_pos->type != TokenType::Semicolon)
     {
         auto tmp_arg = String(token_pos->begin, token_pos->end);
-        if (token_pos->type != DB::TokenType::Comma && token_pos->type != DB::TokenType::ClosingRoundBracket
-            && token_pos->type != DB::TokenType::OpeningRoundBracket && token_pos->type != DB::TokenType::OpeningSquareBracket
-            && token_pos->type != DB::TokenType::ClosingSquareBracket && tmp_arg != "~" && tmp_arg != "dynamic")
+        if (token_pos->type != TokenType::Comma && token_pos->type != TokenType::ClosingRoundBracket
+            && token_pos->type != TokenType::OpeningRoundBracket && token_pos->type != TokenType::OpeningSquareBracket
+            && token_pos->type != TokenType::ClosingSquareBracket && tmp_arg != "~" && tmp_arg != "dynamic")
         {
-            if (token_pos->type == DB::TokenType::StringLiteral || token_pos->type == DB::TokenType::QuotedIdentifier)
-                new_expr += "lower('" + DB::IParserKQLFunction::escapeSingleQuotes(String(token_pos->begin + 1, token_pos->end - 1)) + "')";
+            if (token_pos->type == TokenType::StringLiteral || token_pos->type == TokenType::QuotedIdentifier)
+                new_expr += "lower('" + IParserKQLFunction::escapeSingleQuotes(String(token_pos->begin + 1, token_pos->end - 1)) + "')";
             else
                 new_expr += "lower(" + tmp_arg + ")";
         }
         else if (tmp_arg != "~" && tmp_arg != "dynamic" && tmp_arg != "[" && tmp_arg != "]")
             new_expr += tmp_arg;
 
-        if (token_pos->type == DB::TokenType::ClosingRoundBracket)
+        if (token_pos->type == TokenType::ClosingRoundBracket)
             break;
         ++token_pos;
     }
     return new_expr;
 }
 
-std::string genInOpExpr(DB::IParser::Pos & token_pos, const std::string & kql_op, const std::string & ch_op)
+std::string genInOpExpr(IParser::Pos & token_pos, const std::string & kql_op, const std::string & ch_op)
 {
-    DB::ParserKQLTableFunction kqlfun_p;
-    DB::ParserToken s_lparen(DB::TokenType::OpeningRoundBracket);
+    ParserKQLTableFunction kqlfun_p;
+    ParserToken s_lparen(TokenType::OpeningRoundBracket);
 
-    DB::ASTPtr select;
-    DB::Expected expected;
+    ASTPtr select;
+    Expected expected;
 
     ++token_pos;
     if (!s_lparen.ignore(token_pos, expected))
-        throw DB::Exception(DB::ErrorCodes::SYNTAX_ERROR, "Syntax error near {}", kql_op);
+        throw Exception(ErrorCodes::SYNTAX_ERROR, "Syntax error near {}", kql_op);
 
     auto pos = token_pos;
     if (kqlfun_p.parse(pos, select, expected))

src/Parsers/Kusto/ParserKQLPrint.cpp

@@ -9,7 +9,7 @@ bool ParserKQLPrint::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
     ASTPtr select_expression_list;
     const String expr = getExprFromToken(pos);
 
-    Tokens tokens(expr.c_str(), expr.c_str() + expr.size());
+    Tokens tokens(expr.data(), expr.data() + expr.size(), 0, true);
     IParser::Pos new_pos(tokens, pos.max_depth, pos.max_backtracks);
 
     if (!ParserNotEmptyExpressionList(true).parse(new_pos, select_expression_list, expected))

src/Parsers/Kusto/ParserKQLProject.cpp

@@ -11,7 +11,7 @@ bool ParserKQLProject ::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
 
     expr = getExprFromToken(pos);
 
-    Tokens tokens(expr.c_str(), expr.c_str() + expr.size());
+    Tokens tokens(expr.data(), expr.data() + expr.size(), 0, true);
     IParser::Pos new_pos(tokens, pos.max_depth, pos.max_backtracks);
 
     if (!ParserNotEmptyExpressionList(false).parse(new_pos, select_expression_list, expected))

src/Parsers/Kusto/ParserKQLQuery.cpp

@@ -37,7 +37,7 @@ bool ParserKQLBase::parseByString(String expr, ASTPtr & node, uint32_t max_depth
 {
     Expected expected;
 
-    Tokens tokens(expr.c_str(), expr.c_str() + expr.size());
+    Tokens tokens(expr.data(), expr.data() + expr.size(), 0, true);
     IParser::Pos pos(tokens, max_depth, max_backtracks);
     return parse(pos, node, expected);
 }
@@ -45,7 +45,7 @@ bool ParserKQLBase::parseByString(String expr, ASTPtr & node, uint32_t max_depth
 bool ParserKQLBase::parseSQLQueryByString(ParserPtr && parser, String & query, ASTPtr & select_node, uint32_t max_depth, uint32_t max_backtracks)
 {
     Expected expected;
-    Tokens token_subquery(query.c_str(), query.c_str() + query.size());
+    Tokens token_subquery(query.data(), query.data() + query.size(), 0, true);
     IParser::Pos pos_subquery(token_subquery, max_depth, max_backtracks);
     if (!parser->parse(pos_subquery, select_node, expected))
         return false;
@@ -123,7 +123,7 @@ bool ParserKQLBase::setSubQuerySource(ASTPtr & select_query, ASTPtr & source, bo
 
 String ParserKQLBase::getExprFromToken(const String & text, uint32_t max_depth, uint32_t max_backtracks)
 {
-    Tokens tokens(text.c_str(), text.c_str() + text.size());
+    Tokens tokens(text.data(), text.data() + text.size(), 0, true);
     IParser::Pos pos(tokens, max_depth, max_backtracks);
 
     return getExprFromToken(pos);
@@ -522,7 +522,7 @@ bool ParserKQLQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
                 --last_pos;
 
             String sub_query = std::format("({})", String(operation_pos.front().second->begin, last_pos->end));
-            Tokens token_subquery(sub_query.c_str(), sub_query.c_str() + sub_query.size());
+            Tokens token_subquery(sub_query.data(), sub_query.data() + sub_query.size(), 0, true);
             IParser::Pos pos_subquery(token_subquery, pos.max_depth, pos.max_backtracks);
 
             if (!ParserKQLSubquery().parse(pos_subquery, tables, expected))
@@ -543,7 +543,7 @@ bool ParserKQLQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
             auto oprator = getOperator(op_str);
             if (oprator)
             {
-                Tokens token_clause(op_calsue.c_str(), op_calsue.c_str() + op_calsue.size());
+                Tokens token_clause(op_calsue.data(), op_calsue.data() + op_calsue.size(), 0, true);
                 IParser::Pos pos_clause(token_clause, pos.max_depth, pos.max_backtracks);
                 if (!oprator->parse(pos_clause, node, expected))
                     return false;
@@ -576,7 +576,7 @@ bool ParserKQLQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
     if (!node->as<ASTSelectQuery>()->select())
     {
         auto expr = String("*");
-        Tokens tokens(expr.c_str(), expr.c_str() + expr.size());
+        Tokens tokens(expr.data(), expr.data() + expr.size(), 0, true);
         IParser::Pos new_pos(tokens, pos.max_depth, pos.max_backtracks);
         if (!std::make_unique<ParserKQLProject>()->parse(new_pos, node, expected))
             return false;

src/Parsers/Kusto/ParserKQLSort.cpp

@@ -18,7 +18,7 @@ bool ParserKQLSort::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
 
     auto expr = getExprFromToken(pos);
 
-    Tokens tokens(expr.c_str(), expr.c_str() + expr.size());
+    Tokens tokens(expr.data(), expr.data() + expr.size(), 0, true);
     IParser::Pos new_pos(tokens, pos.max_depth, pos.max_backtracks);
 
     auto pos_backup = new_pos;

src/Parsers/Kusto/ParserKQLStatement.cpp

@@ -2,13 +2,13 @@
 #include <Parsers/ASTSelectWithUnionQuery.h>
 #include <Parsers/CommonParsers.h>
 #include <Parsers/IParserBase.h>
-#include <Parsers/Kusto/KustoFunctions/KQLFunctionFactory.h>
 #include <Parsers/Kusto/ParserKQLQuery.h>
 #include <Parsers/Kusto/ParserKQLStatement.h>
 #include <Parsers/Kusto/Utilities.h>
 #include <Parsers/ParserSetQuery.h>
 #include <Parsers/ASTLiteral.h>
 
+
 namespace DB
 {
 
@@ -63,6 +63,8 @@ bool ParserKQLWithUnionQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & exp
 
 bool ParserKQLTableFunction::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
 {
+    /// TODO: This code is idiotic, see https://github.com/ClickHouse/ClickHouse/issues/61742
+
     ParserToken lparen(TokenType::OpeningRoundBracket);
 
     ASTPtr string_literal;
@@ -101,13 +103,16 @@ bool ParserKQLTableFunction::parseImpl(Pos & pos, ASTPtr & node, Expected & expe
         ++pos;
     }
 
-    Tokens token_kql(kql_statement.data(), kql_statement.data() + kql_statement.size());
-    IParser::Pos pos_kql(token_kql, pos.max_depth, pos.max_backtracks);
+    Tokens tokens_kql(kql_statement.data(), kql_statement.data() + kql_statement.size(), 0, true);
+    IParser::Pos pos_kql(tokens_kql, pos.max_depth, pos.max_backtracks);
+
     Expected kql_expected;
     kql_expected.enable_highlighting = false;
     if (!ParserKQLWithUnionQuery().parse(pos_kql, node, kql_expected))
         return false;
+
     ++pos;
     return true;
 }
+
 }

src/Parsers/Kusto/ParserKQLStatement.h

@@ -45,7 +45,7 @@ protected:
 class ParserKQLTableFunction : public IParserBase
 {
 protected:
-    const char * getName() const override { return "KQL() function"; }
+    const char * getName() const override { return "KQL function"; }
     bool parseImpl(Pos & pos, ASTPtr & node, Expected & expected) override;
 };
 

src/Parsers/Kusto/ParserKQLSummarize.cpp

@@ -194,7 +194,7 @@ bool ParserKQLSummarize::parseImpl(Pos & pos, ASTPtr & node, Expected & expected
 
     String converted_columns = getExprFromToken(expr_columns, pos.max_depth, pos.max_backtracks);
 
-    Tokens token_converted_columns(converted_columns.c_str(), converted_columns.c_str() + converted_columns.size());
+    Tokens token_converted_columns(converted_columns.data(), converted_columns.data() + converted_columns.size(), 0, true);
     IParser::Pos pos_converted_columns(token_converted_columns, pos.max_depth, pos.max_backtracks);
 
     if (!ParserNotEmptyExpressionList(true).parse(pos_converted_columns, select_expression_list, expected))
@@ -206,7 +206,7 @@ bool ParserKQLSummarize::parseImpl(Pos & pos, ASTPtr & node, Expected & expected
     {
         String converted_groupby = getExprFromToken(expr_groupby, pos.max_depth, pos.max_backtracks);
 
-        Tokens token_converted_groupby(converted_groupby.c_str(), converted_groupby.c_str() + converted_groupby.size());
+        Tokens token_converted_groupby(converted_groupby.data(), converted_groupby.data() + converted_groupby.size(), 0, true);
         IParser::Pos postoken_converted_groupby(token_converted_groupby, pos.max_depth, pos.max_backtracks);
 
         if (!ParserNotEmptyExpressionList(false).parse(postoken_converted_groupby, group_expression_list, expected))

src/Parsers/TokenIterator.h

@@ -21,6 +21,7 @@ class Tokens
 {
 private:
     std::vector<Token> data;
+    size_t max_pos = 0;
     Lexer lexer;
     bool skip_insignificant;
 
@@ -35,10 +36,16 @@ public:
         while (true)
         {
             if (index < data.size())
+            {
+                max_pos = std::max(max_pos, index);
                 return data[index];
+            }
 
             if (!data.empty() && data.back().isEnd())
+            {
+                max_pos = data.size() - 1;
                 return data.back();
+            }
 
             Token token = lexer.nextToken();
 
@@ -51,7 +58,12 @@ public:
     {
         if (data.empty())
             return (*this)[0];
-        return data.back();
+        return data[max_pos];
+    }
+
+    void reset()
+    {
+        max_pos = 0;
     }
 };
 

src/Parsers/parseQuery.cpp

@@ -4,6 +4,7 @@
 #include <Parsers/ParserQuery.h>
 #include <Parsers/ASTInsertQuery.h>
 #include <Parsers/ASTExplainQuery.h>
+#include <Parsers/CommonParsers.h>
 #include <Parsers/Lexer.h>
 #include <Parsers/TokenIterator.h>
 #include <Common/StringUtils.h>
@@ -285,6 +286,33 @@ ASTPtr tryParseQuery(
     }
 
     Expected expected;
+
+    /** A shortcut - if Lexer found invalid tokens, fail early without full parsing.
+      * But there are certain cases when invalid tokens are permitted:
+      * 1. INSERT queries can have arbitrary data after the FORMAT clause, that is parsed by a different parser.
+      * 2. It can also be the case when there are multiple queries separated by semicolons, and the first queries are ok
+      * while subsequent queries have syntax errors.
+      *
+      * This shortcut is needed to avoid complex backtracking in case of obviously erroneous queries.
+      */
+    IParser::Pos lookahead(token_iterator);
+    if (!ParserKeyword(Keyword::INSERT_INTO).ignore(lookahead))
+    {
+        while (lookahead->type != TokenType::Semicolon && lookahead->type != TokenType::EndOfStream)
+        {
+            if (lookahead->isError())
+            {
+                out_error_message = getLexicalErrorMessage(query_begin, all_queries_end, *lookahead, hilite, query_description);
+                return nullptr;
+            }
+
+            ++lookahead;
+        }
+
+        /// We should not spoil the info about maximum parsed position in the original iterator.
+        tokens.reset();
+    }
+
     ASTPtr res;
     const bool parse_res = parser.parse(token_iterator, res, expected);
     const auto last_token = token_iterator.max();

src/Processors/QueryPlan/Optimizations/mergeExpressions.cpp

@@ -2,10 +2,25 @@
 #include <Processors/QueryPlan/FilterStep.h>
 #include <Processors/QueryPlan/ExpressionStep.h>
 #include <Interpreters/ActionsDAG.h>
+#include <Functions/FunctionsLogical.h>
+#include <Functions/IFunctionAdaptors.h>
 
 namespace DB::QueryPlanOptimizations
 {
 
+static void removeFromOutputs(ActionsDAG & dag, const ActionsDAG::Node & node)
+{
+    auto & outputs = dag.getOutputs();
+    for (size_t i = 0; i < outputs.size(); ++i)
+    {
+        if (&node == outputs[i])
+        {
+            outputs.erase(outputs.begin() + i);
+            return;
+        }
+    }
+}
+
 size_t tryMergeExpressions(QueryPlan::Node * parent_node, QueryPlan::Nodes &)
 {
     if (parent_node->children.size() != 1)
@@ -19,6 +34,7 @@ size_t tryMergeExpressions(QueryPlan::Node * parent_node, QueryPlan::Nodes &)
     auto * parent_expr = typeid_cast<ExpressionStep *>(parent.get());
     auto * parent_filter = typeid_cast<FilterStep *>(parent.get());
     auto * child_expr = typeid_cast<ExpressionStep *>(child.get());
+    auto * child_filter = typeid_cast<FilterStep *>(child.get());
 
     if (parent_expr && child_expr)
     {
@@ -60,6 +76,42 @@ size_t tryMergeExpressions(QueryPlan::Node * parent_node, QueryPlan::Nodes &)
         parent_node->children.swap(child_node->children);
         return 1;
     }
+    else if (parent_filter && child_filter)
+    {
+        const auto & child_actions = child_filter->getExpression();
+        const auto & parent_actions = parent_filter->getExpression();
+
+        if (child_actions->hasArrayJoin())
+            return 0;
+
+        auto actions = child_actions->clone();
+        const auto & child_filter_node = actions->findInOutputs(child_filter->getFilterColumnName());
+        if (child_filter->removesFilterColumn())
+            removeFromOutputs(*actions, child_filter_node);
+
+        actions->mergeInplace(std::move(*parent_actions->clone()));
+
+        const auto & parent_filter_node = actions->findInOutputs(parent_filter->getFilterColumnName());
+        if (parent_filter->removesFilterColumn())
+            removeFromOutputs(*actions, parent_filter_node);
+
+        FunctionOverloadResolverPtr func_builder_and = std::make_unique<FunctionToOverloadResolverAdaptor>(std::make_shared<FunctionAnd>());
+        const auto & condition = actions->addFunction(func_builder_and, {&child_filter_node, &parent_filter_node}, {});
+        auto & outputs = actions->getOutputs();
+        outputs.insert(outputs.begin(), &condition);
+
+        actions->removeUnusedActions(false);
+
+        auto filter = std::make_unique<FilterStep>(child_filter->getInputStreams().front(),
+                                                   actions,
+                                                   condition.result_name,
+                                                   true);
+        filter->setStepDescription("(" + parent_filter->getStepDescription() + " + " + child_filter->getStepDescription() + ")");
+
+        parent_node->step = std::move(filter);
+        parent_node->children.swap(child_node->children);
+        return 1;
+    }
 
     return 0;
 }

src/Processors/Sources/RecursiveCTESource.cpp

@@ -5,7 +5,7 @@
 #include <Processors/Sinks/SinkToStorage.h>
 #include <Processors/Transforms/ExpressionTransform.h>
 #include <Processors/Transforms/MaterializingTransform.h>
-#include <Processors/Transforms/SquashingChunksTransform.h>
+#include <Processors/Transforms/SquashingTransform.h>
 #include <Processors/Executors/PullingAsyncPipelineExecutor.h>
 
 #include <QueryPipeline/QueryPipelineBuilder.h>

src/Processors/Transforms/AggregatingTransform.cpp

@@ -9,7 +9,7 @@
 #include <Common/logger_useful.h>
 #include <Common/formatReadable.h>
 
-#include <Processors/Transforms/SquashingChunksTransform.h>
+#include <Processors/Transforms/SquashingTransform.h>
 
 
 namespace ProfileEvents
@@ -783,7 +783,7 @@ void AggregatingTransform::initGenerate()
                             {
                                 /// Just a reasonable constant, matches default value for the setting `preferred_block_size_bytes`
                                 static constexpr size_t oneMB = 1024 * 1024;
-                                return std::make_shared<SimpleSquashingChunksTransform>(header, params->params.max_block_size, oneMB);
+                                return std::make_shared<SimpleSquashingTransform>(header, params->params.max_block_size, oneMB);
                             });
                     }
                     /// AggregatingTransform::expandPipeline expects single output port.

src/Processors/Transforms/ApplySquashingTransform.h

@@ -0,0 +1,63 @@
+#pragma once
+#include <Interpreters/Squashing.h>
+#include <Processors/ISimpleTransform.h>
+#include <Processors/Sinks/SinkToStorage.h>
+
+namespace DB
+{
+
+class ApplySquashingTransform : public ExceptionKeepingTransform
+{
+public:
+    explicit ApplySquashingTransform(const Block & header, const size_t min_block_size_rows, const size_t min_block_size_bytes)
+        : ExceptionKeepingTransform(header, header, false)
+        , squashing(header, min_block_size_rows, min_block_size_bytes)
+    {
+    }
+
+    String getName() const override { return "ApplySquashingTransform"; }
+
+    void work() override
+    {
+        if (stage == Stage::Exception)
+        {
+            data.chunk.clear();
+            ready_input = false;
+            return;
+        }
+
+        ExceptionKeepingTransform::work();
+        if (finish_chunk)
+        {
+            data.chunk = std::move(finish_chunk);
+            ready_output = true;
+        }
+    }
+
+protected:
+    void onConsume(Chunk chunk) override
+    {
+        if (auto res_chunk = DB::Squashing::squash(std::move(chunk)))
+            cur_chunk.setColumns(res_chunk.getColumns(), res_chunk.getNumRows());
+    }
+
+    GenerateResult onGenerate() override
+    {
+        GenerateResult res;
+        res.chunk = std::move(cur_chunk);
+        res.is_done = true;
+        return res;
+    }
+    void onFinish() override
+    {
+        auto chunk = DB::Squashing::squash({});
+        finish_chunk.setColumns(chunk.getColumns(), chunk.getNumRows());
+    }
+
+private:
+    Squashing squashing;
+    Chunk cur_chunk;
+    Chunk finish_chunk;
+};
+
+}

src/Processors/Transforms/PlanSquashingTransform.cpp

@@ -0,0 +1,145 @@
+#include <Processors/Transforms/PlanSquashingTransform.h>
+#include <Processors/IProcessor.h>
+#include <Common/Exception.h>
+
+namespace DB
+{
+
+namespace ErrorCodes
+{
+    extern const int LOGICAL_ERROR;
+}
+
+PlanSquashingTransform::PlanSquashingTransform(const Block & header, size_t min_block_size_rows, size_t min_block_size_bytes, size_t num_ports)
+    : IProcessor(InputPorts(num_ports, header), OutputPorts(num_ports, header)), squashing(header, min_block_size_rows, min_block_size_bytes)
+{
+}
+
+IProcessor::Status PlanSquashingTransform::prepare()
+{
+    Status status = Status::Ready;
+
+    while (planning_status != PlanningStatus::FINISH)
+    {
+        switch (planning_status)
+        {
+            case INIT:
+                init();
+                break;
+            case READ_IF_CAN:
+                return prepareConsume();
+            case PUSH:
+                return sendOrFlush();
+            case FLUSH:
+                return sendOrFlush();
+            case FINISH:
+                break; /// never reached
+        }
+    }
+    if (status == Status::Ready)
+        status = finish();
+    else
+        throw Exception(ErrorCodes::LOGICAL_ERROR, "There should be a Ready status to finish the PlanSquashing");
+
+    return status;
+}
+
+void PlanSquashingTransform::work()
+{
+    prepare();
+}
+
+void PlanSquashingTransform::init()
+{
+    for (auto input: inputs)
+        if (!input.isFinished())
+            input.setNeeded();
+
+    planning_status = PlanningStatus::READ_IF_CAN;
+}
+
+IProcessor::Status PlanSquashingTransform::prepareConsume()
+{
+    bool all_finished = true;
+    for (auto & input : inputs)
+    {
+        if (!input.isFinished())
+        {
+            all_finished = false;
+            input.setNeeded();
+        }
+        else
+            continue;
+
+        if (input.hasData())
+        {
+            chunk = input.pull();
+            chunk = transform(std::move(chunk));
+
+            if (chunk.hasChunkInfo())
+            {
+                planning_status = PlanningStatus::PUSH;
+                return Status::Ready;
+            }
+        }
+    }
+
+    if (all_finished) /// If all inputs are closed, we check if we have data in balancing
+    {
+        if (squashing.isDataLeft()) /// If we have data in balancing, we process this data
+        {
+            planning_status = PlanningStatus::FLUSH;
+            chunk = flushChunk();
+            return Status::Ready;
+        }
+        planning_status = PlanningStatus::FINISH;
+        return Status::Ready;
+    }
+
+    return Status::NeedData;
+}
+
+Chunk PlanSquashingTransform::transform(Chunk && chunk_)
+{
+    return squashing.add(std::move(chunk_));
+}
+
+Chunk PlanSquashingTransform::flushChunk()
+{
+    return squashing.flush();
+}
+
+IProcessor::Status PlanSquashingTransform::sendOrFlush()
+{
+    if (!chunk)
+    {
+        planning_status = PlanningStatus::FINISH;
+        return Status::Ready;
+    }
+
+    for (auto &output : outputs)
+    {
+        if (output.canPush())
+        {
+            if (planning_status == PlanningStatus::PUSH)
+                planning_status = PlanningStatus::READ_IF_CAN;
+            else
+                planning_status = PlanningStatus::FINISH;
+
+            output.push(std::move(chunk));
+            return Status::Ready;
+        }
+    }
+    return Status::PortFull;
+}
+
+IProcessor::Status PlanSquashingTransform::finish()
+{
+    for (auto & in : inputs)
+        in.close();
+    for (auto & output : outputs)
+        output.finish();
+
+    return Status::Finished;
+}
+}