Merge branch 'master' into fix-apache-arrow

contrib/grpc

@@ -1 +1 @@
-Subproject commit f5b7fdc2dff09ada06dbf6c75df298fb40f898df
+Subproject commit 1716359d2e28d304a250f9df0e6c0ccad03de8db

docker/test/fasttest/Dockerfile

@@ -26,7 +26,10 @@ RUN apt-get update \
         zstd \
         --yes --no-install-recommends \
     && apt-get clean \
-    && rm -rf /var/lib/apt/lists/* /var/cache/debconf /tmp/*
+    && rm -rf /var/lib/apt/lists/* /var/cache/debconf /tmp/* \
+    && groupadd --system --gid 1000 clickhouse \
+    && useradd --system --gid 1000 --uid 1000 -m clickhouse
+# ^ For some reason, groupadd and useradd are needed for tests with 'expect', but I don't know, why.
 
 COPY requirements.txt /
 RUN pip3 install --no-cache-dir -r /requirements.txt

docker/test/util/Dockerfile

@@ -17,6 +17,7 @@ ENV DEBIAN_FRONTEND=noninteractive LLVM_VERSION=18
 
 RUN apt-get update \
     && apt-get install \
+        sudo \
         apt-transport-https \
         apt-utils \
         ca-certificates \

docs/en/interfaces/cli.md

@@ -185,6 +185,7 @@ You can pass parameters to `clickhouse-client` (all parameters have a default va
 - `--format, -f` – Use the specified default format to output the result.
 - `--vertical, -E` – If specified, use the [Vertical format](../interfaces/formats.md#vertical) by default to output the result. This is the same as `–format=Vertical`. In this format, each value is printed on a separate line, which is helpful when displaying wide tables.
 - `--time, -t` – If specified, print the query execution time to ‘stderr’ in non-interactive mode.
+- `--memory-usage` – If specified, print memory usage to ‘stderr’ in non-interactive mode]. Possible values: 'none' - do not print memory usage, 'default' - print number of bytes, 'readable' - print memory usage in human-readable format.
 - `--stacktrace` – If specified, also print the stack trace if an exception occurs.
 - `--config-file` – The name of the configuration file.
 - `--secure` – If specified, will connect to server over secure connection (TLS). You might need to configure your CA certificates in the [configuration file](#configuration_files). The available configuration settings are the same as for [server-side TLS configuration](../operations/server-configuration-parameters/settings.md#server_configuration_parameters-openssl).
@@ -339,7 +340,7 @@ clickhouse-client clickhouse://some_user%40some_mail.com@localhost:9000
 Connect to one of provides hosts: `192.168.1.15`, `192.168.1.25`.
 
 ``` bash
-clickhouse-client clickhouse://192.168.1.15,192.168.1.25 
+clickhouse-client clickhouse://192.168.1.15,192.168.1.25
 ```
 
 ### Configuration Files {#configuration_files}
@@ -367,7 +368,7 @@ Example of a config file:
 ```
 
 Or the same config in a YAML format:
- 
+
 ```yaml
 user: username
 password: 'password'

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

@@ -56,7 +56,6 @@ Functions:
 
 ## Related content
 
-- [Reducing ClickHouse Storage Cost with the Low Cardinality Type – Lessons from an Instana Engineer](https://altinity.com/blog/2020-5-20-reducing-clickhouse-storage-cost-with-the-low-cardinality-type-lessons-from-an-instana-engineer)
-- [String Optimization (video presentation in Russian)](https://youtu.be/rqf-ILRgBdY?list=PL0Z2YDlm0b3iwXCpEFiOOYmwXzVmjJfEt). [Slides in English](https://github.com/ClickHouse/clickhouse-presentations/raw/master/meetup19/string_optimization.pdf)
 - Blog: [Optimizing ClickHouse with Schemas and Codecs](https://clickhouse.com/blog/optimize-clickhouse-codecs-compression-schema)
 - Blog: [Working with time series data in ClickHouse](https://clickhouse.com/blog/working-with-time-series-data-and-functions-ClickHouse)
+- [String Optimization (video presentation in Russian)](https://youtu.be/rqf-ILRgBdY?list=PL0Z2YDlm0b3iwXCpEFiOOYmwXzVmjJfEt). [Slides in English](https://github.com/ClickHouse/clickhouse-presentations/raw/master/meetup19/string_optimization.pdf)

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

@@ -76,7 +76,7 @@ WHERE macro = 'test';
 └───────┴──────────────┘
 ```
 
-## FQDN
+## fqdn
 
 Returns the fully qualified domain name of the ClickHouse server.
 
@@ -86,7 +86,7 @@ Returns the fully qualified domain name of the ClickHouse server.
 fqdn();
 ```
 
-Aliases: `fullHostName`, 'FQDN'. 
+Aliases: `fullHostName`, `FQDN`. 
 
 **Returned value**
 

docs/zh/changelog/index.md

@@ -356,7 +356,7 @@ sidebar_label: "\u53D8\u66F4\u65E5\u5FD7"
 
 #### 新功能 {#new-feature-1}
 
--   添加 `deduplicate_blocks_in_dependent_materialized_views` 用于控制具有实例化视图的表中幂等插入的行为的选项。 这个新功能是由Altinity的特殊要求添加到错误修正版本中的。
+-   添加 `deduplicate_blocks_in_dependent_materialized_views` 用于控制具有实例化视图的表中幂等插入的行为的选项。
     [#9070](https://github.com/ClickHouse/ClickHouse/pull/9070) [(urykhy)](https://github.com/urykhy)
 
 ### ClickHouse版本v20.1.2.4,2020-01-22 {#clickhouse-release-v20-1-2-4-2020-01-22}

programs/local/LocalServer.cpp

@@ -423,6 +423,7 @@ void LocalServer::connect()
 {
     connection_parameters = ConnectionParameters(getClientConfiguration(), "localhost");
 
+    /// This is needed for table function input(...).
     ReadBuffer * in;
     auto table_file = getClientConfiguration().getString("table-file", "-");
     if (table_file == "-" || table_file == "stdin")

src/Client/ClientBase.cpp

@@ -2070,9 +2070,18 @@ void ClientBase::processParsedSingleQuery(const String & full_query, const Strin
         progress_indication.writeFinalProgress();
         output_stream << std::endl << std::endl;
     }
-    else if (getClientConfiguration().getBool("print-time-to-stderr", false))
+    else
     {
-        error_stream << progress_indication.elapsedSeconds() << "\n";
+        const auto & config = getClientConfiguration();
+        if (config.getBool("print-time-to-stderr", false))
+            error_stream << progress_indication.elapsedSeconds() << "\n";
+
+        const auto & print_memory_mode = config.getString("print-memory-to-stderr", "");
+        auto peak_memeory_usage = std::max<Int64>(progress_indication.getMemoryUsage().peak, 0);
+        if (print_memory_mode == "default")
+            error_stream << peak_memeory_usage << "\n";
+        else if (print_memory_mode == "readable")
+            error_stream << formatReadableSizeWithBinarySuffix(peak_memeory_usage) << "\n";
     }
 
     if (!is_interactive && getClientConfiguration().getBool("print-num-processed-rows", false))
@@ -3036,6 +3045,7 @@ void ClientBase::init(int argc, char ** argv)
         ("disable_suggestion,A", "Disable loading suggestion data. Note that suggestion data is loaded asynchronously through a second connection to ClickHouse server. Also it is reasonable to disable suggestion if you want to paste a query with TAB characters. Shorthand option -A is for those who get used to mysql client.")
         ("wait_for_suggestions_to_load", "Load suggestion data synchonously.")
         ("time,t", "print query execution time to stderr in non-interactive mode (for benchmarks)")
+        ("memory-usage", po::value<std::string>()->implicit_value("default")->default_value("none"), "print memory usage to stderr in non-interactive mode (for benchmarks). Values: 'none', 'default', 'readable'")
 
         ("echo", "in batch mode, print query before execution")
 
@@ -3121,6 +3131,14 @@ void ClientBase::init(int argc, char ** argv)
     /// Output execution time to stderr in batch mode.
     if (options.count("time"))
         getClientConfiguration().setBool("print-time-to-stderr", true);
+    if (options.count("memory-usage"))
+    {
+        const auto & memory_usage_mode = options["memory-usage"].as<std::string>();
+        if (memory_usage_mode != "none" && memory_usage_mode != "default" && memory_usage_mode != "readable")
+            throw Exception(ErrorCodes::BAD_ARGUMENTS, "Unknown memory-usage mode: {}", memory_usage_mode);
+        getClientConfiguration().setString("print-memory-to-stderr", memory_usage_mode);
+    }
+
     if (options.count("query"))
         queries = options["query"].as<std::vector<std::string>>();
     if (options.count("query_id"))

src/Client/IConnections.h

@@ -54,8 +54,6 @@ public:
 
     struct ReplicaInfo
     {
-        bool collaborate_with_initiator{false};
-        size_t all_replicas_count{0};
         size_t number_of_current_replica{0};
     };
 

src/Client/LocalConnection.h

@@ -175,4 +175,5 @@ private:
 
     ReadBuffer * in;
 };
+
 }

src/Client/MultiplexedConnections.cpp

@@ -142,13 +142,12 @@ void MultiplexedConnections::sendQuery(
             modified_settings.group_by_two_level_threshold = 0;
             modified_settings.group_by_two_level_threshold_bytes = 0;
         }
+    }
 
-        if (replica_info)
+    if (replica_info)
-        {
+    {
-            client_info.collaborate_with_initiator = true;
+        client_info.collaborate_with_initiator = true;
-            client_info.count_participating_replicas = replica_info->all_replicas_count;
+        client_info.number_of_current_replica = replica_info->number_of_current_replica;
-            client_info.number_of_current_replica = replica_info->number_of_current_replica;
-        }
     }
 
     /// FIXME: Remove once we will make `allow_experimental_analyzer` obsolete setting.

src/Common/ProgressIndication.h

@@ -72,11 +72,6 @@ public:
     /// How much seconds passed since query execution start.
     double elapsedSeconds() const { return getElapsedNanoseconds() / 1e9; }
 
-    void updateThreadEventData(HostToTimesMap & new_hosts_data);
-
-private:
-    double getCPUUsage();
-
     struct MemoryUsage
     {
         UInt64 total = 0;
@@ -86,6 +81,11 @@ private:
 
     MemoryUsage getMemoryUsage() const;
 
+    void updateThreadEventData(HostToTimesMap & new_hosts_data);
+
+private:
+    double getCPUUsage();
+
     UInt64 getElapsedNanoseconds() const;
 
     /// This flag controls whether to show the progress bar. We start showing it after

src/Interpreters/Access/InterpreterCreateUserQuery.cpp

@@ -114,6 +114,34 @@ namespace
         else if (query.grantees)
             user.grantees = *query.grantees;
     }
+
+    time_t getValidUntilFromAST(ASTPtr valid_until, ContextPtr context)
+    {
+        if (context)
+            valid_until = evaluateConstantExpressionAsLiteral(valid_until, context);
+
+        const String valid_until_str = checkAndGetLiteralArgument<String>(valid_until, "valid_until");
+
+        if (valid_until_str == "infinity")
+            return 0;
+
+        time_t time = 0;
+        ReadBufferFromString in(valid_until_str);
+
+        if (context)
+        {
+            const auto & time_zone = DateLUT::instance("");
+            const auto & utc_time_zone = DateLUT::instance("UTC");
+
+            parseDateTimeBestEffort(time, in, time_zone, utc_time_zone);
+        }
+        else
+        {
+            readDateTimeText(time, in);
+        }
+
+        return time;
+    }
 }
 
 BlockIO InterpreterCreateUserQuery::execute()
@@ -134,23 +162,7 @@ BlockIO InterpreterCreateUserQuery::execute()
 
     std::optional<time_t> valid_until;
     if (query.valid_until)
-    {
+        valid_until = getValidUntilFromAST(query.valid_until, getContext());
-        const ASTPtr valid_until_literal = evaluateConstantExpressionAsLiteral(query.valid_until, getContext());
-        const String valid_until_str = checkAndGetLiteralArgument<String>(valid_until_literal, "valid_until");
-
-        time_t time = 0;
-
-        if (valid_until_str != "infinity")
-        {
-            const auto & time_zone = DateLUT::instance("");
-            const auto & utc_time_zone = DateLUT::instance("UTC");
-
-            ReadBufferFromString in(valid_until_str);
-            parseDateTimeBestEffort(time, in, time_zone, utc_time_zone);
-        }
-
-        valid_until = time;
-    }
 
     std::optional<RolesOrUsersSet> default_roles_from_query;
     if (query.default_roles)
@@ -259,7 +271,11 @@ void InterpreterCreateUserQuery::updateUserFromQuery(User & user, const ASTCreat
     if (query.auth_data)
         auth_data = AuthenticationData::fromAST(*query.auth_data, {}, !query.attach);
 
-    updateUserFromQueryImpl(user, query, auth_data, {}, {}, {}, {}, {}, allow_no_password, allow_plaintext_password, true);
+    std::optional<time_t> valid_until;
+    if (query.valid_until)
+        valid_until = getValidUntilFromAST(query.valid_until, {});
+
+    updateUserFromQueryImpl(user, query, auth_data, {}, {}, {}, {}, valid_until, allow_no_password, allow_plaintext_password, true);
 }
 
 void registerInterpreterCreateUserQuery(InterpreterFactory & factory)

src/Interpreters/ClientInfo.cpp

@@ -95,7 +95,7 @@ void ClientInfo::write(WriteBuffer & out, UInt64 server_protocol_revision) const
     if (server_protocol_revision >= DBMS_MIN_REVISION_WITH_PARALLEL_REPLICAS)
     {
         writeVarUInt(static_cast<UInt64>(collaborate_with_initiator), out);
-        writeVarUInt(count_participating_replicas, out);
+        writeVarUInt(obsolete_count_participating_replicas, out);
         writeVarUInt(number_of_current_replica, out);
     }
 }
@@ -185,7 +185,7 @@ void ClientInfo::read(ReadBuffer & in, UInt64 client_protocol_revision)
         UInt64 value;
         readVarUInt(value, in);
         collaborate_with_initiator = static_cast<bool>(value);
-        readVarUInt(count_participating_replicas, in);
+        readVarUInt(obsolete_count_participating_replicas, in);
         readVarUInt(number_of_current_replica, in);
     }
 }

src/Interpreters/ClientInfo.h

@@ -127,7 +127,7 @@ public:
 
     /// For parallel processing on replicas
     bool collaborate_with_initiator{false};
-    UInt64 count_participating_replicas{0};
+    UInt64 obsolete_count_participating_replicas{0};
     UInt64 number_of_current_replica{0};
 
     enum class BackgroundOperationType : uint8_t

src/Interpreters/Context.cpp

@@ -5001,13 +5001,6 @@ void Context::setConnectionClientVersion(UInt64 client_version_major, UInt64 cli
     client_info.connection_tcp_protocol_version = client_tcp_protocol_version;
 }
 
-void Context::setReplicaInfo(bool collaborate_with_initiator, size_t all_replicas_count, size_t number_of_current_replica)
-{
-    client_info.collaborate_with_initiator = collaborate_with_initiator;
-    client_info.count_participating_replicas = all_replicas_count;
-    client_info.number_of_current_replica = number_of_current_replica;
-}
-
 void Context::increaseDistributedDepth()
 {
     ++client_info.distributed_depth;

src/Interpreters/Context.h

@@ -699,7 +699,6 @@ public:
     void setInitialQueryStartTime(std::chrono::time_point<std::chrono::system_clock> initial_query_start_time);
     void setQuotaClientKey(const String & quota_key);
     void setConnectionClientVersion(UInt64 client_version_major, UInt64 client_version_minor, UInt64 client_version_patch, unsigned client_tcp_protocol_version);
-    void setReplicaInfo(bool collaborate_with_initiator, size_t all_replicas_count, size_t number_of_current_replica);
     void increaseDistributedDepth();
     const OpenTelemetry::TracingContext & getClientTraceContext() const { return client_info.client_trace_context; }
     OpenTelemetry::TracingContext & getClientTraceContext() { return client_info.client_trace_context; }

src/Interpreters/Squashing.cpp

@@ -8,6 +8,7 @@
 
 namespace DB
 {
+
 namespace ErrorCodes
 {
     extern const int LOGICAL_ERROR;

src/Parsers/ASTCreateQuery.cpp

@@ -265,6 +265,16 @@ ASTPtr ASTCreateQuery::clone() const
     return res;
 }
 
+String ASTCreateQuery::getID(char delim) const
+{
+    String res = attach ? "AttachQuery" : "CreateQuery";
+    String database = getDatabase();
+    if (!database.empty())
+        res += (delim + getDatabase());
+    res += (delim + getTable());
+    return res;
+}
+
 void ASTCreateQuery::formatQueryImpl(const FormatSettings & settings, FormatState & state, FormatStateStacked frame) const
 {
     frame.need_parens = false;

src/Parsers/ASTCreateQuery.h

@@ -136,7 +136,7 @@ public:
     bool create_or_replace{false};
 
     /** Get the text that identifies this element. */
-    String getID(char delim) const override { return (attach ? "AttachQuery" : "CreateQuery") + (delim + getDatabase()) + delim + getTable(); }
+    String getID(char delim) const override;
 
     ASTPtr clone() const override;
 

src/Processors/QueryPlan/ReadFromRemote.cpp

@@ -412,8 +412,8 @@ void ReadFromParallelRemoteReplicasStep::initializePipeline(QueryPipelineBuilder
     auto timeouts = ConnectionTimeouts::getTCPTimeoutsWithFailover(current_settings);
 
     const auto & shard = cluster->getShardsInfo().at(0);
-    size_t all_replicas_count = current_settings.max_parallel_replicas;
+    size_t max_replicas_to_use = current_settings.max_parallel_replicas;
-    if (all_replicas_count > shard.getAllNodeCount())
+    if (max_replicas_to_use > shard.getAllNodeCount())
     {
         LOG_INFO(
             getLogger("ReadFromParallelRemoteReplicasStep"),
@@ -421,14 +421,14 @@ void ReadFromParallelRemoteReplicasStep::initializePipeline(QueryPipelineBuilder
             "Will use the latter number to execute the query.",
             current_settings.max_parallel_replicas,
             shard.getAllNodeCount());
-        all_replicas_count = shard.getAllNodeCount();
+        max_replicas_to_use = shard.getAllNodeCount();
     }
 
     std::vector<ConnectionPoolWithFailover::Base::ShuffledPool> shuffled_pool;
-    if (all_replicas_count < shard.getAllNodeCount())
+    if (max_replicas_to_use < shard.getAllNodeCount())
     {
         shuffled_pool = shard.pool->getShuffledPools(current_settings);
-        shuffled_pool.resize(all_replicas_count);
+        shuffled_pool.resize(max_replicas_to_use);
     }
     else
     {
@@ -438,11 +438,10 @@ void ReadFromParallelRemoteReplicasStep::initializePipeline(QueryPipelineBuilder
         shuffled_pool = shard.pool->getShuffledPools(current_settings, priority_func);
     }
 
-    for (size_t i=0; i < all_replicas_count; ++i)
+    for (size_t i=0; i < max_replicas_to_use; ++i)
     {
         IConnections::ReplicaInfo replica_info
         {
-            .all_replicas_count = all_replicas_count,
             /// we should use this number specifically because efficiency of data distribution by consistent hash depends on it.
             .number_of_current_replica = i,
         };

src/Processors/QueryPlan/ReadFromSystemNumbersStep.cpp

@@ -36,18 +36,33 @@ inline void iotaWithStepOptimized(T * begin, size_t count, T first_value, T step
         iotaWithStep(begin, count, first_value, step);
 }
 
+/// The range is defined as [start, end)
+UInt64 itemCountInRange(UInt64 start, UInt64 end, UInt64 step)
+{
+    const auto range_count = end - start;
+    if (step == 1)
+        return range_count;
+
+    return (range_count - 1) / step + 1;
+}
+
 class NumbersSource : public ISource
 {
 public:
-    NumbersSource(UInt64 block_size_, UInt64 offset_, std::optional<UInt64> limit_, UInt64 chunk_step_, const std::string & column_name, UInt64 step_)
+    NumbersSource(
+        UInt64 block_size_,
+        UInt64 offset_,
+        std::optional<UInt64> end_,
+        const std::string & column_name,
+        UInt64 step_in_chunk_,
+        UInt64 step_between_chunks_)
         : ISource(createHeader(column_name))
         , block_size(block_size_)
         , next(offset_)
-        , chunk_step(chunk_step_)
+        , end(end_)
-        , step(step_)
+        , step_in_chunk(step_in_chunk_)
+        , step_between_chunks(step_between_chunks_)
     {
-        if (limit_.has_value())
-            end = limit_.value() + offset_;
     }
     String getName() const override { return "Numbers"; }
 
@@ -64,7 +79,10 @@ protected:
         {
             if (end.value() <= next)
                 return {};
-            real_block_size = std::min(block_size, end.value() - next);
+
+            auto max_items_to_generate = itemCountInRange(next, *end, step_in_chunk);
+
+            real_block_size = std::min(block_size, max_items_to_generate);
         }
         auto column = ColumnUInt64::create(real_block_size);
         ColumnUInt64::Container & vec = column->getData();
@@ -74,21 +92,20 @@ protected:
 
         UInt64 * current_end = &vec[real_block_size];
 
-        iotaWithStepOptimized(pos, static_cast<size_t>(current_end - pos), curr, step);
+        iotaWithStepOptimized(pos, static_cast<size_t>(current_end - pos), curr, step_in_chunk);
 
-        next += chunk_step;
+        next += step_between_chunks;
 
         progress(column->size(), column->byteSize());
-
         return {Columns{std::move(column)}, real_block_size};
     }
 
 private:
     UInt64 block_size;
     UInt64 next;
-    UInt64 chunk_step;
     std::optional<UInt64> end; /// not included
-    UInt64 step;
+    UInt64 step_in_chunk;
+    UInt64 step_between_chunks;
 };
 
 struct RangeWithStep
@@ -549,20 +566,39 @@ Pipe ReadFromSystemNumbersStep::makePipe()
         return pipe;
     }
 
+    const auto end = std::invoke(
+        [&]() -> std::optional<UInt64>
+        {
+            if (numbers_storage.limit.has_value())
+                return *(numbers_storage.limit) + numbers_storage.offset;
+            return {};
+        });
+
     /// Fall back to NumbersSource
+    /// Range in a single block
+    const auto block_range = max_block_size * numbers_storage.step;
+    /// Step between chunks in a single source.
+    /// It is bigger than block_range in case of multiple threads, because we have to account for other sources as well.
+    const auto step_between_chunks = num_streams * block_range;
     for (size_t i = 0; i < num_streams; ++i)
     {
+        const auto source_offset = i * block_range;
+        if (numbers_storage.limit.has_value() && *numbers_storage.limit < source_offset)
+            break;
+
+        const auto source_start = numbers_storage.offset + source_offset;
+
         auto source = std::make_shared<NumbersSource>(
             max_block_size,
-            numbers_storage.offset + i * max_block_size * numbers_storage.step,
+            source_start,
-            numbers_storage.limit,
+            end,
-            num_streams * max_block_size * numbers_storage.step,
             numbers_storage.column_name,
-            numbers_storage.step);
+            numbers_storage.step,
+            step_between_chunks);
 
         if (numbers_storage.limit && i == 0)
         {
-            auto rows_appr = (*numbers_storage.limit - 1) / numbers_storage.step + 1;
+            auto rows_appr = itemCountInRange(numbers_storage.offset, *numbers_storage.limit, numbers_storage.step);
             if (limit > 0 && limit < rows_appr)
                 rows_appr = query_info_limit;
             source->addTotalRowsApprox(rows_appr);
@@ -571,19 +607,6 @@ Pipe ReadFromSystemNumbersStep::makePipe()
         pipe.addSource(std::move(source));
     }
 
-    if (numbers_storage.limit)
-    {
-        size_t i = 0;
-        auto storage_limit = (*numbers_storage.limit - 1) / numbers_storage.step + 1;
-        /// This formula is how to split 'limit' elements to 'num_streams' chunks almost uniformly.
-        pipe.addSimpleTransform(
-            [&](const Block & header)
-            {
-                ++i;
-                return std::make_shared<LimitTransform>(header, storage_limit * i / num_streams - storage_limit * (i - 1) / num_streams, 0);
-            });
-    }
-
     return pipe;
 }
 

src/Server/KeeperTCPHandler.cpp

@@ -622,8 +622,9 @@ void KeeperTCPHandler::updateStats(Coordination::ZooKeeperResponsePtr & response
         {
             LOG_INFO(
                 log,
-                "Total time to process a request took too long ({}ms).\nRequest info: {}",
+                "Total time to process a request in session {} took too long ({}ms).\nRequest info: {}",
-                elapsed,
+                session_id,
+                elapsed_ms,
                 request->toString(/*short_format=*/true));
         }
 

src/Storages/PostgreSQL/StorageMaterializedPostgreSQL.cpp

@@ -21,7 +21,6 @@
 
 #include <Processors/QueryPlan/QueryPlan.h>
 #include <Processors/QueryPlan/ReadFromPreparedSource.h>
-#include <Processors/Transforms/FilterTransform.h>
 #include <Parsers/ASTFunction.h>
 #include <Parsers/ASTIdentifier.h>
 #include <Parsers/ASTTablesInSelectQuery.h>

tests/ci/autoscale_runners_lambda/app.py

@@ -99,7 +99,8 @@ def set_capacity(
             continue
         raise ValueError("Queue status is not in ['in_progress', 'queued']")
 
-    scale_down, scale_up = get_scales(runner_type)
+    # scale_down, scale_up = get_scales(runner_type)
+    _, scale_up = get_scales(runner_type)
     # With lyfecycle hooks some instances are actually free because some of
     # them are in 'Terminating:Wait' state
     effective_capacity = max(
@@ -110,7 +111,7 @@ def set_capacity(
     # How much nodes are free (positive) or need to be added (negative)
     capacity_reserve = effective_capacity - running - queued
     stop = False
-    if capacity_reserve < 0:
+    if capacity_reserve <= 0:
         # This part is about scaling up
         capacity_deficit = -capacity_reserve
         # It looks that we are still OK, since no queued jobs exist
@@ -158,41 +159,43 @@ def set_capacity(
             )
         return
 
-    # Now we will calculate if we need to scale down
+    # FIXME: try decreasing capacity from runners that finished their jobs and have no job assigned
-    stop = stop or asg["DesiredCapacity"] == asg["MinSize"]
+    #   IMPORTANT: Runner init script must be of version that supports ASG decrease
-    new_capacity = asg["DesiredCapacity"] - (capacity_reserve // scale_down)
+    # # Now we will calculate if we need to scale down
-    new_capacity = max(new_capacity, asg["MinSize"])
+    # stop = stop or asg["DesiredCapacity"] == asg["MinSize"]
-    new_capacity = min(new_capacity, asg["MaxSize"])
+    # new_capacity = asg["DesiredCapacity"] - (capacity_reserve // scale_down)
-    stop = stop or asg["DesiredCapacity"] == new_capacity
+    # new_capacity = max(new_capacity, asg["MinSize"])
-    if stop:
+    # new_capacity = min(new_capacity, asg["MaxSize"])
-        logging.info(
+    # stop = stop or asg["DesiredCapacity"] == new_capacity
-            "Do not decrease ASG %s capacity, current capacity=%s, effective "
+    # if stop:
-            "capacity=%s, minimum capacity=%s, running jobs=%s, queue size=%s",
+    #     logging.info(
-            asg["AutoScalingGroupName"],
+    #         "Do not decrease ASG %s capacity, current capacity=%s, effective "
-            asg["DesiredCapacity"],
+    #         "capacity=%s, minimum capacity=%s, running jobs=%s, queue size=%s",
-            effective_capacity,
+    #         asg["AutoScalingGroupName"],
-            asg["MinSize"],
+    #         asg["DesiredCapacity"],
-            running,
+    #         effective_capacity,
-            queued,
+    #         asg["MinSize"],
-        )
+    #         running,
-        return
+    #         queued,
-
+    #     )
-    logging.info(
+    #     return
-        "The ASG %s capacity will be decreased to %s, current capacity=%s, effective "
+    #
-        "capacity=%s, minimum capacity=%s, running jobs=%s, queue size=%s",
+    # logging.info(
-        asg["AutoScalingGroupName"],
+    #     "The ASG %s capacity will be decreased to %s, current capacity=%s, effective "
-        new_capacity,
+    #     "capacity=%s, minimum capacity=%s, running jobs=%s, queue size=%s",
-        asg["DesiredCapacity"],
+    #     asg["AutoScalingGroupName"],
-        effective_capacity,
+    #     new_capacity,
-        asg["MinSize"],
+    #     asg["DesiredCapacity"],
-        running,
+    #     effective_capacity,
-        queued,
+    #     asg["MinSize"],
-    )
+    #     running,
-    if not dry_run:
+    #     queued,
-        client.set_desired_capacity(
+    # )
-            AutoScalingGroupName=asg["AutoScalingGroupName"],
+    # if not dry_run:
-            DesiredCapacity=new_capacity,
+    #     client.set_desired_capacity(
-        )
+    #         AutoScalingGroupName=asg["AutoScalingGroupName"],
+    #         DesiredCapacity=new_capacity,
+    #     )
 
 
 def main(dry_run: bool = True) -> None:

tests/ci/autoscale_runners_lambda/test_autoscale.py

@@ -97,19 +97,34 @@ class TestSetCapacity(unittest.TestCase):
             ),
             TestCase("lower-min", 10, 5, 20, [Queue("queued", 5, "lower-min")], 10),
             # Decrease capacity
-            TestCase("w/reserve", 1, 13, 20, [Queue("queued", 5, "w/reserve")], 5),
+            # FIXME: Tests changed for lambda that can only scale up
+            # TestCase("w/reserve", 1, 13, 20, [Queue("queued", 5, "w/reserve")], 5),
+            TestCase("w/reserve", 1, 13, 20, [Queue("queued", 5, "w/reserve")], -1),
+            # TestCase(
+            #     "style-checker", 1, 13, 20, [Queue("queued", 5, "style-checker")], 5
+            # ),
             TestCase(
-                "style-checker", 1, 13, 20, [Queue("queued", 5, "style-checker")], 5
+                "style-checker", 1, 13, 20, [Queue("queued", 5, "style-checker")], -1
             ),
-            TestCase("w/reserve", 1, 23, 20, [Queue("queued", 17, "w/reserve")], 17),
+            # TestCase("w/reserve", 1, 23, 20, [Queue("queued", 17, "w/reserve")], 17),
-            TestCase("decrease", 1, 13, 20, [Queue("in_progress", 3, "decrease")], 3),
+            TestCase("w/reserve", 1, 23, 20, [Queue("queued", 17, "w/reserve")], -1),
+            # TestCase("decrease", 1, 13, 20, [Queue("in_progress", 3, "decrease")], 3),
+            TestCase("decrease", 1, 13, 20, [Queue("in_progress", 3, "decrease")], -1),
+            # TestCase(
+            #     "style-checker",
+            #     1,
+            #     13,
+            #     20,
+            #     [Queue("in_progress", 5, "style-checker")],
+            #     5,
+            # ),
             TestCase(
                 "style-checker",
                 1,
                 13,
                 20,
                 [Queue("in_progress", 5, "style-checker")],
-                5,
+                -1,
             ),
         )
         for t in test_cases:

tests/ci/ci_config.py

@@ -415,7 +415,8 @@ class CI:
         JobNames.INTEGRATION_TEST_FLAKY: CommonJobConfigs.INTEGRATION_TEST.with_properties(
             required_builds=[BuildNames.PACKAGE_ASAN],
             pr_only=True,
-            reference_job_name=JobNames.INTEGRATION_TEST_TSAN,
+            # TODO: approach with reference job names does not work because digest may not be calculated if job skipped in wf
+            # reference_job_name=JobNames.INTEGRATION_TEST_TSAN,
         ),
         JobNames.COMPATIBILITY_TEST: CommonJobConfigs.COMPATIBILITY_TEST.with_properties(
             required_builds=[BuildNames.PACKAGE_RELEASE],
@@ -460,7 +461,8 @@ class CI:
             required_builds=[BuildNames.PACKAGE_ASAN],
             pr_only=True,
             timeout=3600,
-            reference_job_name=JobNames.STATELESS_TEST_RELEASE,
+            # TODO: approach with reference job names does not work because digest may not be calculated if job skipped in wf
+            # reference_job_name=JobNames.STATELESS_TEST_RELEASE,
         ),
         JobNames.JEPSEN_KEEPER: JobConfig(
             required_builds=[BuildNames.BINARY_RELEASE],

tests/ci/functional_test_check.py

@@ -17,9 +17,19 @@ from download_release_packages import download_last_release
 from env_helper import REPO_COPY, REPORT_PATH, TEMP_PATH
 from get_robot_token import get_parameter_from_ssm
 from pr_info import PRInfo
-from report import ERROR, SUCCESS, JobReport, StatusType, TestResults, read_test_results
+from report import (
+    ERROR,
+    SUCCESS,
+    JobReport,
+    StatusType,
+    TestResults,
+    read_test_results,
+    FAILURE,
+)
 from stopwatch import Stopwatch
 from tee_popen import TeePopen
+from ci_config import CI
+from ci_utils import Utils
 
 NO_CHANGES_MSG = "Nothing to run"
 
@@ -351,7 +361,23 @@ def main():
         additional_files=additional_logs,
     ).dump(to_file=args.report_to_file if args.report_to_file else None)
 
+    should_block_ci = False
     if state != SUCCESS:
+        should_block_ci = True
+
+    if state == FAILURE and CI.is_required(check_name):
+        failed_cnt = Utils.get_failed_tests_number(description)
+        print(
+            f"Job status is [{state}] with [{failed_cnt}] failed test cases. status description [{description}]"
+        )
+        if (
+            failed_cnt
+            and failed_cnt <= CI.MAX_TOTAL_FAILURES_PER_JOB_BEFORE_BLOCKING_CI
+        ):
+            print(f"Won't block the CI workflow")
+            should_block_ci = False
+
+    if should_block_ci:
         sys.exit(1)
 
 

tests/ci/integration_test_check.py

@@ -23,10 +23,13 @@ from report import (
     TestResult,
     TestResults,
     read_test_results,
+    FAILURE,
 )
 from stopwatch import Stopwatch
 
 import integration_tests_runner as runner
+from ci_config import CI
+from ci_utils import Utils
 
 
 def get_json_params_dict(
@@ -233,7 +236,23 @@ def main():
         additional_files=additional_logs,
     ).dump(to_file=args.report_to_file if args.report_to_file else None)
 
+    should_block_ci = False
     if state != SUCCESS:
+        should_block_ci = True
+
+    if state == FAILURE and CI.is_required(check_name):
+        failed_cnt = Utils.get_failed_tests_number(description)
+        print(
+            f"Job status is [{state}] with [{failed_cnt}] failed test cases. status description [{description}]"
+        )
+        if (
+            failed_cnt
+            and failed_cnt <= CI.MAX_TOTAL_FAILURES_PER_JOB_BEFORE_BLOCKING_CI
+        ):
+            print(f"Won't block the CI workflow")
+            should_block_ci = False
+
+    if should_block_ci:
         sys.exit(1)
 
 

tests/ci/lambda_shared_package/lambda_shared/pr.py

@@ -3,48 +3,27 @@
 import re
 from typing import Tuple
 
-# Individual trusted contirbutors who are not in any trusted organization.
+# Individual trusted contributors who are not in any trusted organization.
 # Can be changed in runtime: we will append users that we learned to be in
 # a trusted org, to save GitHub API calls.
 TRUSTED_CONTRIBUTORS = {
     e.lower()
     for e in [
-        "achimbab",  # Kakao corp
-        "Algunenano",  # Raúl Marín, ClickHouse, Inc
         "amosbird",
         "azat",  # SEMRush
         "bharatnc",  # Many contributions.
-        "bobrik",  # Seasoned contributor, CloudFlare
         "cwurm",  # ClickHouse, Inc
         "den-crane",  # Documentation contributor
-        "hagen1778",  # Roman Khavronenko, seasoned contributor
-        "hczhcz",
-        "hexiaoting",  # Seasoned contributor
         "ildus",  # adjust, ex-pgpro
-        "javisantana",  # a Spanish ClickHouse enthusiast, ex-Carto
-        "kreuzerkrieg",
-        "nikvas0",
         "nvartolomei",  # Seasoned contributor, CloudFlare
-        "spongedu",  # Seasoned contributor
         "taiyang-li",
         "ucasFL",  # Amos Bird's friend
-        "vdimir",  # ClickHouse, Inc
-        "YiuRULE",
-        "zlobober",  # Developer of YT
-        "ilejn",  # Arenadata, responsible for Kerberized Kafka
         "thomoco",  # ClickHouse, Inc
-        "BoloniniD",  # Seasoned contributor, HSE
         "tonickkozlov",  # Cloudflare
         "tylerhannan",  # ClickHouse, Inc
-        "myrrc",  # Mike Kot, DoubleCloud
-        "thevar1able",  # ClickHouse, Inc
-        "aalexfvk",
-        "MikhailBurdukov",
         "tsolodov",  # ClickHouse, Inc
-        "kitaisreal",
-        "k-morozov",  # Konstantin Morozov, Yandex Cloud
         "justindeguzman",  # ClickHouse, Inc
-        "jrdi",  # ClickHouse contributor, TinyBird
+        "XuJia0210",  # ClickHouse, Inc
     ]
 }
 

tests/ci/merge_pr.py

@@ -27,7 +27,6 @@ from report import SUCCESS, FAILURE
 from env_helper import GITHUB_UPSTREAM_REPOSITORY, GITHUB_REPOSITORY
 from synchronizer_utils import SYNC_BRANCH_PREFIX
 from ci_config import CI
-from ci_utils import Utils
 
 # The team name for accepted approvals
 TEAM_NAME = getenv("GITHUB_TEAM_NAME", "core")
@@ -249,74 +248,22 @@ def main():
     repo = gh.get_repo(args.repo)
 
     if args.set_ci_status:
+        # set Mergeable check status and exit
         assert args.wf_status in (FAILURE, SUCCESS)
-        # set mergeable check status and exit
         commit = get_commit(gh, args.pr_info.sha)
         statuses = get_commit_filtered_statuses(commit)
 
-        max_failed_tests_per_job = 0
-        job_name_with_max_failures = None
-        total_failed_tests = 0
-        failed_to_get_info = False
         has_failed_statuses = False
         for status in statuses:
-            if not CI.is_required(status.context) or status.context in (
-                CI.StatusNames.SYNC,
-                CI.StatusNames.PR_CHECK,
-            ):
-                # CI.StatusNames.SYNC or CI.StatusNames.PR_CHECK should not be checked
-                continue
             print(f"Check status [{status.context}], [{status.state}]")
-            if status.state == FAILURE:
+            if CI.is_required(status.context) and status.state != SUCCESS:
+                print(f"WARNING: Failed status [{status.context}], [{status.state}]")
                 has_failed_statuses = True
-                failed_cnt = Utils.get_failed_tests_number(status.description)
-                if failed_cnt is None:
-                    failed_to_get_info = True
-                    print(
-                        f"WARNING: failed to get number of failed tests from [{status.description}]"
-                    )
-                else:
-                    if failed_cnt > max_failed_tests_per_job:
-                        job_name_with_max_failures = status.context
-                        max_failed_tests_per_job = failed_cnt
-                    total_failed_tests += failed_cnt
-                    print(
-                        f"Failed test cases in [{status.context}] is [{failed_cnt}], total failures [{total_failed_tests}]"
-                    )
-            elif status.state != SUCCESS and status.context not in (
-                CI.StatusNames.SYNC,
-                CI.StatusNames.PR_CHECK,
-            ):
-                # do not block CI on failures in (CI.StatusNames.SYNC, CI.StatusNames.PR_CHECK)
-                has_failed_statuses = True
-                print(
-                    f"Unexpected status for [{status.context}]: [{status.state}] - block further testing"
-                )
-                failed_to_get_info = True
-
-        can_continue = True
-        if total_failed_tests > CI.MAX_TOTAL_FAILURES_BEFORE_BLOCKING_CI:
-            print(
-                f"Required check has [{total_failed_tests}] failed - block further testing"
-            )
-            can_continue = False
-        if max_failed_tests_per_job > CI.MAX_TOTAL_FAILURES_PER_JOB_BEFORE_BLOCKING_CI:
-            print(
-                f"Job [{job_name_with_max_failures}] has [{max_failed_tests_per_job}] failures - block further testing"
-            )
-            can_continue = False
-        if failed_to_get_info:
-            print("Unexpected commit status state - block further testing")
-            can_continue = False
-        if args.wf_status != SUCCESS and not has_failed_statuses:
-            # workflow failed but reason is unknown as no failed statuses present
-            can_continue = False
-            print(
-                "WARNING: Either the runner is faulty or the operating status is unknown. The first is self-healing, the second requires investigation."
-            )
 
         if args.wf_status == SUCCESS or has_failed_statuses:
-            # do not set mergeable check status if args.wf_status == failure, apparently it has died runners and is to be restarted
+            # set Mergeable check if workflow is successful (green)
+            # or if we have GH statuses with failures (red)
+            #    to avoid false-green on a died runner
             state = trigger_mergeable_check(
                 commit,
                 statuses,
@@ -333,10 +280,10 @@ def main():
             print(
                 "Workflow failed but no failed statuses found (died runner?) - cannot set Mergeable Check status"
             )
-
+        if args.wf_status == SUCCESS and not has_failed_statuses:
-        if not can_continue:
+            sys.exit(0)
+        else:
             sys.exit(1)
-        sys.exit(0)
 
     # An ugly and not nice fix to patch the wrong organization URL,
     # see https://github.com/PyGithub/PyGithub/issues/2395#issuecomment-1378629710

tests/ci/worker/deploy-runner-init.sh

@@ -1,5 +1,7 @@
 #!/usr/bin/env bash
 
+set -e
+
 usage() {
   echo "Usage: $0 ENVIRONMENT" >&2
   echo "Valid values for ENVIRONMENT: staging, production" >&2
@@ -55,7 +57,7 @@ EOF
 
 body() {
   local first_line
-  first_line=$(sed -n '/^# THE SCRIPT START$/{=;q}' "$SOURCE_SCRIPT")
+  first_line=$(sed -n '/^# THE SCRIPT START$/{=;q;}' "$SOURCE_SCRIPT")
   if [ -z "$first_line" ]; then
     echo "The pattern '# THE SCRIPT START' is not found in $SOURCE_SCRIPT" >&2
     exit 1

tests/ci/worker/init_runner.sh

@@ -50,7 +50,7 @@ set -uo pipefail
 # set accordingly to a runner role #
 ####################################
 
-echo "Running init script"
+echo "Running init v1"
 export DEBIAN_FRONTEND=noninteractive
 export RUNNER_HOME=/home/ubuntu/actions-runner
 
@@ -90,7 +90,6 @@ terminate_delayed() {
     # IF `sleep` IS CHANGED, CHANGE ANOTHER VALUE IN `pgrep`
     sleep=13.14159265358979323846
     echo "Going to terminate the runner's instance in $sleep seconds"
-    INSTANCE_ID=$(ec2metadata --instance-id)
     # We execute it with `at` to not have it as an orphan process, but launched independently
     # GH Runners kill all remain processes
     echo "sleep '$sleep'; aws ec2 terminate-instances --instance-ids $INSTANCE_ID" | at now || \
@@ -111,11 +110,17 @@ declare -f terminate_delayed >> /tmp/actions-hooks/common.sh
 terminate_and_exit() {
     # Terminate instance and exit from the script instantly
     echo "Going to terminate the runner's instance"
-    INSTANCE_ID=$(ec2metadata --instance-id)
     aws ec2 terminate-instances --instance-ids "$INSTANCE_ID"
     exit 0
 }
 
+terminate_decrease_and_exit() {
+    # Terminate instance and exit from the script instantly
+    echo "Going to terminate the runner's instance and decrease asg capacity"
+    aws autoscaling terminate-instance-in-auto-scaling-group --instance-id "$INSTANCE_ID" --should-decrement-desired-capacity
+    exit 0
+}
+
 declare -f terminate_and_exit >> /tmp/actions-hooks/common.sh
 
 check_spot_instance_is_old() {
@@ -324,7 +329,7 @@ while true; do
                 sudo -u ubuntu ./config.sh remove --token "$(get_runner_token)" \
                     || continue
                 echo "Runner didn't launch or have assigned jobs after ${RUNNER_AGE} seconds, shutting down"
-                terminate_and_exit
+                terminate_decrease_and_exit
             fi
         fi
     else

tests/integration/test_user_valid_until/test.py

@@ -5,7 +5,7 @@ from time import sleep
 from helpers.cluster import ClickHouseCluster
 
 cluster = ClickHouseCluster(__file__)
-node = cluster.add_instance("node")
+node = cluster.add_instance("node", stay_alive=True)
 
 
 @pytest.fixture(scope="module")
@@ -84,3 +84,22 @@ def test_details(started_cluster):
         node.query("SHOW CREATE USER user_details_time_only")
         == f"CREATE USER user_details_time_only VALID UNTIL \\'{until_year}-01-01 22:03:40\\'\n"
     )
+
+
+def test_restart(started_cluster):
+    node.query("CREATE USER user_restart VALID UNTIL '06/11/2010 08:03:20 Z+3'")
+
+    assert (
+        node.query("SHOW CREATE USER user_restart")
+        == "CREATE USER user_restart VALID UNTIL \\'2010-11-06 05:03:20\\'\n"
+    )
+
+    node.restart_clickhouse()
+
+    assert (
+        node.query("SHOW CREATE USER user_restart")
+        == "CREATE USER user_restart VALID UNTIL \\'2010-11-06 05:03:20\\'\n"
+    )
+
+    error = "Authentication failed"
+    assert error in node.query_and_get_error("SELECT 1", user="user_restart")

tests/queries/0_stateless/02903_parameterized_view_explain_ast.reference

@@ -1,4 +1,4 @@
-CreateQuery  numbers_pv (children 2)
+CreateQuery numbers_pv (children 2)
  Identifier numbers_pv
  SelectWithUnionQuery (children 1)
   ExpressionList (children 1)

tests/queries/0_stateless/02970_generate_series.reference

@@ -5,6 +5,7 @@
 501
 50
 17928
+17928
 0
 10
 13

tests/queries/0_stateless/02970_generate_series.sql

@@ -5,6 +5,7 @@ SELECT count() FROM generate_series(7, 77, 10);
 SELECT count() FROM generate_series(0, 1000, 2);
 SELECT count() FROM generate_series(0, 999, 20);
 SELECT sum(generate_series) FROM generate_series(4, 1008, 4) WHERE generate_series % 7 = 1;
+SELECT sum(generate_series) FROM generate_series(4, 1008, 4) WHERE generate_series % 7 = 1 SETTINGS max_block_size = 71;
 
 SELECT * FROM generate_series(5, 4);
 SELECT * FROM generate_series(0, 0);

tests/queries/0_stateless/03198_dynamic_read_subcolumns.sql

@@ -1,4 +1,5 @@
--- Tags: no-random-settings, no-s3-storage
+-- Tags: no-random-settings, no-object-storage
+-- Tag no-object-storage: this test relies on the number of opened files in MergeTree that can differ in object storages
 
 SET allow_experimental_dynamic_type = 1;
 DROP TABLE IF EXISTS test_dynamic;

tests/queries/0_stateless/03203_client_benchmark_options.reference

@@ -0,0 +1,6 @@
+Ok
+Ok
+Ok
+Ok
+Ok
+Ok

tests/queries/0_stateless/03203_client_benchmark_options.sh

@@ -0,0 +1,13 @@
+#!/usr/bin/env bash
+
+CUR_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
+# shellcheck source=../shell_config.sh
+. "$CUR_DIR"/../shell_config.sh
+
+${CLICKHOUSE_CLIENT} -t -q "SELECT sleepEachRow(2) FORMAT Null" 2>&1 | grep -q "^2\." && echo "Ok" || echo "Fail"
+${CLICKHOUSE_CLIENT} --time -q "SELECT sleepEachRow(2) FORMAT Null" 2>&1 | grep -q "^2\." && echo "Ok" || echo "Fail"
+${CLICKHOUSE_CLIENT} --memory-usage -q "SELECT sum(number) FROM numbers(10_000) FORMAT Null" 2>&1 | grep -q "^[0-9]\+$" && echo "Ok" || echo "Fail"
+${CLICKHOUSE_CLIENT} --memory-usage=none -q "SELECT sum(number) FROM numbers(10_000) FORMAT Null" # expected no output
+${CLICKHOUSE_CLIENT} --memory-usage=default -q "SELECT sum(number) FROM numbers(10_000) FORMAT Null" 2>&1 | grep -q "^[0-9]\+$" && echo "Ok" || echo "Fail"
+${CLICKHOUSE_CLIENT} --memory-usage=readable -q "SELECT sum(number) FROM numbers(10_000) FORMAT Null" 2>&1 | grep -q "^[0-9].*B$" && echo "Ok" || echo "Fail"
+${CLICKHOUSE_CLIENT} --memory-usage=unknown -q "SELECT sum(number) FROM numbers(10_000) FORMAT Null" 2>&1 | grep -q "BAD_ARGUMENTS" && echo "Ok" || echo "Fail"

tests/queries/0_stateless/03203_system_numbers_limit_and_offset_complex.reference

@@ -0,0 +1,14 @@
+18679	31
+0
+10
+20
+30
+40
+50
+60
+70
+80
+90
+100
+110
+4250

tests/queries/0_stateless/03203_system_numbers_limit_and_offset_complex.sql

@@ -0,0 +1,38 @@
+--- The following query was buggy before, so let's use it as a test case
+WITH
+    (num > 1) AND (arraySum(arrayMap(y -> ((y > 1) AND (y < num) AND ((num % y) = 0)), range(toUInt64(sqrt(num)) + 1))) = 0) AS is_prime_slow
+SELECT
+    num,
+    ds,
+FROM
+(
+    WITH
+        arraySum(arrayMap(y -> toUInt8(y), splitByString('', toString(num)))) AS digits_sum
+    SELECT
+        1 + (number * 2) AS num,
+        digits_sum AS ds
+    FROM numbers_mt(10000)
+    WHERE ds IN (
+        WITH
+            (number > 1) AND (arraySum(arrayMap(y -> ((y > 1) AND (y < number) AND ((number % y) = 0)), range(toUInt64(sqrt(number)) + 1))) = 0) AS is_prime_slow
+        SELECT number
+        FROM numbers(180 + 1)
+        WHERE is_prime_slow
+    )
+)
+WHERE is_prime_slow
+ORDER BY num ASC
+LIMIT 998, 1
+SETTINGS max_block_size = 64, max_threads=16;
+
+SELECT number
+FROM numbers_mt(120)
+WHERE (number % 10) = 0
+ORDER BY number ASC
+SETTINGS max_block_size = 31, max_threads = 11;
+
+SELECT number
+FROM numbers_mt(4242, 9)
+WHERE (number % 10) = 0
+ORDER BY number ASC
+SETTINGS max_block_size = 31, max_threads = 11;

tests/queries/0_stateless/03203_system_numbers_limit_and_offset_simple.reference

@@ -0,0 +1,6 @@
+case 1
+9900
+9910
+9920
+case 2
+9990

tests/queries/0_stateless/03203_system_numbers_limit_and_offset_simple.sql

@@ -0,0 +1,11 @@
+SELECT 'case 1';
+SELECT number FROM numbers_mt(10000)
+WHERE (number % 10) = 0
+ORDER BY number ASC
+LIMIT 990, 3;
+
+SELECT 'case 2';
+SELECT number FROM numbers_mt(10000)
+WHERE (number % 10) = 0
+ORDER BY number ASC
+LIMIT 999, 20 SETTINGS max_block_size = 31;

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

@@ -1653,6 +1653,7 @@ formated
 formatschema
 formatter
 formatters
+fqdn
 frac
 freezed
 fromDaysSinceYearZero