Merge branch 'master' into chesema-merge-wb

README.md

@@ -42,7 +42,6 @@ Keep an eye out for upcoming meetups and events around the world. Somewhere else
 
 Upcoming meetups
 
-* [Barcelona Meetup](https://www.meetup.com/clickhouse-spain-user-group/events/303096876/) - November 12
 * [Ghent Meetup](https://www.meetup.com/clickhouse-belgium-user-group/events/303049405/) - November 19
 * [Dubai Meetup](https://www.meetup.com/clickhouse-dubai-meetup-group/events/303096989/) - November 21
 * [Paris Meetup](https://www.meetup.com/clickhouse-france-user-group/events/303096434) - November 26
@@ -53,6 +52,7 @@ Upcoming meetups
 
 Recently completed meetups
 
+* [Barcelona Meetup](https://www.meetup.com/clickhouse-spain-user-group/events/303096876/) - November 12
 * [Madrid Meetup](https://www.meetup.com/clickhouse-spain-user-group/events/303096564/) - October 22
 * [Singapore Meetup](https://www.meetup.com/clickhouse-singapore-meetup-group/events/303212064/) - October 3
 * [Jakarta Meetup](https://www.meetup.com/clickhouse-indonesia-user-group/events/303191359/) - October 1

base/base/defines.h

@@ -145,6 +145,7 @@
 #define TSA_TRY_ACQUIRE_SHARED(...) __attribute__((try_acquire_shared_capability(__VA_ARGS__)))  /// function tries to acquire a shared capability and returns a boolean value indicating success or failure
 #define TSA_RELEASE_SHARED(...) __attribute__((release_shared_capability(__VA_ARGS__)))          /// function releases the given shared capability
 #define TSA_SCOPED_LOCKABLE __attribute__((scoped_lockable)) /// object of a class has scoped lockable capability
+#define TSA_RETURN_CAPABILITY(...) __attribute__((lock_returned(__VA_ARGS__)))             /// to return capabilities in functions
 
 /// Macros for suppressing TSA warnings for specific reads/writes (instead of suppressing it for the whole function)
 /// They use a lambda function to apply function attribute to a single statement. This enable us to suppress warnings locally instead of

docker/server/Dockerfile.ubuntu

@@ -113,7 +113,9 @@ RUN clickhouse local -q 'SELECT 1' >/dev/null 2>&1 && exit 0 || : \
         /var/lib/apt/lists/* \
         /var/cache/debconf \
         /tmp/* \
-    && apt-get autoremove --purge -yq dirmngr gnupg2
+    && apt-get autoremove --purge -yq dirmngr gnupg2 \
+    && chmod ugo+Xrw -R /etc/clickhouse-server /etc/clickhouse-client
+# The last chmod is here to make the next one is No-op in docker official library Dockerfile
 
 # post install
 # we need to allow "others" access to clickhouse folder, because docker container

docker/server/entrypoint.sh

@@ -162,7 +162,7 @@ if [ -n "${RUN_INITDB_SCRIPTS}" ]; then
         tries=${CLICKHOUSE_INIT_TIMEOUT:-1000}
         while ! wget --spider --no-check-certificate -T 1 -q "$URL" 2>/dev/null; do
             if [ "$tries" -le "0" ]; then
-                echo >&2 'ClickHouse init process failed.'
+                echo >&2 'ClickHouse init process timeout.'
                 exit 1
             fi
             tries=$(( tries-1 ))

docs/en/getting-started/example-datasets/tpch.md

@@ -46,7 +46,7 @@ Detailed table sizes with scale factor 100:
 | orders   | 150.000.000    | 6.15 GB                         |
 | lineitem | 600.00.00      | 26.69 GB                        |
 
-(The table sizes in ClickHouse are taken from `system.tables.total_bytes` and based on below table definitions.
+(Compressed sizes in ClickHouse are taken from `system.tables.total_bytes` and based on below table definitions.)
 
 Now create tables in ClickHouse.
 

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

@@ -597,6 +597,30 @@ If number of tables is greater than this value, server will throw an exception.
 <max_table_num_to_throw>400</max_table_num_to_throw>
 ```
 
+## max\_replicated\_table\_num\_to\_throw {#max-replicated-table-num-to-throw}
+If number of replicated tables is greater than this value, server will throw an exception. 0 means no limitation. Only count table in Atomic/Ordinary/Replicated/Lazy database engine.
+
+**Example**
+```xml
+<max_replicated_table_num_to_throw>400</max_replicated_table_num_to_throw>
+```
+
+## max\_dictionary\_num\_to\_throw {#max-dictionary-num-to-throw}
+If number of dictionaries is greater than this value, server will throw an exception. 0 means no limitation. Only count table in Atomic/Ordinary/Replicated/Lazy database engine.
+
+**Example**
+```xml
+<max_dictionary_num_to_throw>400</max_dictionary_num_to_throw>
+```
+
+## max\_view\_num\_to\_throw {#max-view-num-to-throw}
+If number of views is greater than this value, server will throw an exception. 0 means no limitation. Only count table in Atomic/Ordinary/Replicated/Lazy database engine.
+
+**Example**
+```xml
+<max_view_num_to_throw>400</max_view_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

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

@@ -7,115 +7,4 @@ toc_hidden: true
 
 # List of Aggregate Functions
 
-ClickHouse supports all standard SQL functions (sum, avg, min, max, count) and a wide range of aggregate functions for various applications:
-
-- [aggThrow](../reference/aggthrow.md)
-- [analysisOfVariance](../reference/analysis_of_variance.md)
-- [anyHeavy](../reference/anyheavy.md)
-- [anyLast](../reference/anylast.md)
-- [any](../reference/any.md)
-- [argMax](../reference/argmax.md)
-- [argMin](../reference/argmin.md)
-- [avgWeighted](../reference/avgweighted.md)
-- [avg](../reference/avg.md)
-- [boundingRatio](../reference/boundrat.md)
-- [categoricalInformationValue](../reference/categoricalinformationvalue.md)
-- [contingency](../reference/contingency.md)
-- [corrMatrix](../reference/corrmatrix.md)
-- [corr](../reference/corr.md)
-- [corr](../reference/corrstable.md)
-- [count](../reference/count.md)
-- [covarPopMatrix](../reference/covarpopmatrix.md)
-- [covarPop](../reference/covarpop.md)
-- [covarSampMatrix](../reference/covarsampmatrix.md)
-- [covarSampStable](../reference/covarsampstable.md)
-- [covarSamp](../reference/covarsamp.md)
-- [covarStable](../reference/covarpopstable.md)
-- [cramersVBiasCorrected](../reference/cramersvbiascorrected.md)
-- [cramersV](../reference/cramersv.md)
-- [deltaSumTimestamp](../reference/deltasumtimestamp.md)
-- [deltaSum](../reference/deltasum.md)
-- [entropy](../reference/entropy.md)
-- [exponentialMovingAverage](../reference/exponentialmovingaverage.md)
-- [first_value](../reference/first_value.md)
-- [flameGraph](../reference/flame_graph.md)
-- [groupArrayInsertAt](../reference/grouparrayinsertat.md)
-- [groupArrayIntersect](../reference/grouparrayintersect.md)
-- [groupArrayLast](../reference/grouparraylast.md)
-- [groupArrayMovingAvg](../reference/grouparraymovingavg.md)
-- [groupArrayMovingSum](../reference/grouparraymovingsum.md)
-- [groupArraySample](../reference/grouparraysample.md)
-- [groupArraySorted](../reference/grouparraysorted.md)
-- [groupArray](../reference/grouparray.md)
-- [groupBitAnd](../reference/groupbitand.md)
-- [groupBitOr](../reference/groupbitor.md)
-- [groupBitXor](../reference/groupbitxor.md)
-- [groupBitmapAnd](../reference/groupbitmapand.md)
-- [groupBitmapOr](../reference/groupbitmapor.md)
-- [groupBitmapXor](../reference/groupbitmapxor.md)
-- [groupBitmap](../reference/groupbitmap.md)
-- [groupUniqArray](../reference/groupuniqarray.md)
-- [intervalLengthSum](../reference/intervalLengthSum.md)
-- [kolmogorovSmirnovTest](../reference/kolmogorovsmirnovtest.md)
-- [kurtPop](../reference/kurtpop.md)
-- [kurtSamp](../reference/kurtsamp.md)
-- [largestTriangleThreeBuckets](../reference/largestTriangleThreeBuckets.md)
-- [last_value](../reference/last_value.md)
-- [mannwhitneyutest](../reference/mannwhitneyutest.md)
-- [maxIntersectionsPosition](../reference/maxintersectionsposition.md)
-- [maxIntersections](../reference/maxintersections.md)
-- [maxMap](../reference/maxmap.md)
-- [max](../reference/max.md)
-- [meanZTest](../reference/meanztest.md)
-- [median](../reference/median.md)
-- [minMap](../reference/minmap.md)
-- [min](../reference/min.md)
-- [quantileBFloat16Weighted](../reference/quantilebfloat16.md#quantilebfloat16weighted)
-- [quantileBFloat16](../reference/quantilebfloat16.md#quantilebfloat16)
-- [quantileDD](../reference/quantileddsketch.md#quantileddsketch)
-- [quantileDeterministic](../reference/quantiledeterministic.md)
-- [quantileExactHigh](../reference/quantileexact.md#quantileexacthigh)
-- [quantileExactLow](../reference/quantileexact.md#quantileexactlow)
-- [quantileExactWeighted](../reference/quantileexactweighted.md)
-- [quantileExact](../reference/quantileexact.md)
-- [quantileGK](../reference/quantileGK.md)
-- [quantileInterpolatedWeighted](../reference/quantileinterpolatedweighted.md)
-- [quantileTDigestWeighted](../reference/quantiletdigestweighted.md)
-- [quantileTDigest](../reference/quantiletdigest.md)
-- [quantileTimingWeighted](../reference/quantiletimingweighted.md)
-- [quantileTiming](../reference/quantiletiming.md)
-- [quantile](../reference/quantile.md)
-- [quantiles](../reference/quantiles.md)
-- [rankCorr](../reference/rankCorr.md)
-- [simpleLinearRegression](../reference/simplelinearregression.md)
-- [singleValueOrNull](../reference/singlevalueornull.md)
-- [skewPop](../reference/skewpop.md)
-- [skewSamp](../reference/skewsamp.md)
-- [sparkBar](../reference/sparkbar.md)
-- [stddevPopStable](../reference/stddevpopstable.md)
-- [stddevPop](../reference/stddevpop.md)
-- [stddevSampStable](../reference/stddevsampstable.md)
-- [stddevSamp](../reference/stddevsamp.md)
-- [stochasticLinearRegression](../reference/stochasticlinearregression.md)
-- [stochasticLogisticRegression](../reference/stochasticlogisticregression.md)
-- [studentTTest](../reference/studentttest.md)
-- [sumCount](../reference/sumcount.md)
-- [sumKahan](../reference/sumkahan.md)
-- [sumMapFilteredWithOverflow](../parametric-functions.md/#summapfilteredwithoverflow)
-- [sumMapFiltered](../parametric-functions.md/#summapfiltered)
-- [sumMapWithOverflow](../reference/summapwithoverflow.md)
-- [sumMap](../reference/summap.md)
-- [sumWithOverflow](../reference/sumwithoverflow.md)
-- [sum](../reference/sum.md)
-- [theilsU](../reference/theilsu.md)
-- [topKWeighted](../reference/topkweighted.md)
-- [topK](../reference/topk.md)
-- [uniqCombined64](../reference/uniqcombined64.md)
-- [uniqCombined](../reference/uniqcombined.md)
-- [uniqExact](../reference/uniqexact.md)
-- [uniqHLL12](../reference/uniqhll12.md)
-- [uniqTheta](../reference/uniqthetasketch.md)
-- [uniq](../reference/uniq.md)
-- [varPop](../reference/varpop.md)
-- [varSamp](../reference/varsamp.md)
-- [welchTTest](../reference/welchttest.md)
+ClickHouse supports all standard SQL aggregate functions ([sum](../reference/sum.md), [avg](../reference/avg.md), [min](../reference/min.md), [max](../reference/max.md), [count](../reference/count.md)), as well as a wide range of other aggregate functions.

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

@@ -6,7 +6,9 @@ sidebar_label: AggregateFunction
 
 # AggregateFunction
 
-Aggregate functions can have an implementation-defined intermediate state that can be serialized to an `AggregateFunction(...)` data type and stored in a table, usually, by means of [a materialized view](../../sql-reference/statements/create/view.md). The common way to produce an aggregate function state is by calling the aggregate function with the `-State` suffix. To get the final result of aggregation in the future, you must use the same aggregate function with the `-Merge`suffix.
+Aggregate functions have an implementation-defined intermediate state that can be serialized to an `AggregateFunction(...)` data type and stored in a table, usually, by means of [a materialized view](../../sql-reference/statements/create/view.md).
+The common way to produce an aggregate function state is by calling the aggregate function with the `-State` suffix.
+To get the final result of aggregation in the future, you must use the same aggregate function with the `-Merge`suffix.
 
 `AggregateFunction(name, types_of_arguments...)` — parametric data type.
 

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

@@ -6,29 +6,8 @@ sidebar_position: 1
 
 # Data Types in ClickHouse
 
-ClickHouse can store various kinds of data in table cells. This section describes the supported data types and special considerations for using and/or implementing them if any.
+This section describes the data types supported by ClickHouse, for example [integers](int-uint.md), [floats](float.md) and [strings](string.md).
 
-:::note
-You can check whether a data type name is case-sensitive in the [system.data_type_families](../../operations/system-tables/data_type_families.md#system_tables-data_type_families) table.
-:::
-
-ClickHouse data types include:
-
-- **Integer types**: [signed and unsigned integers](./int-uint.md) (`UInt8`, `UInt16`, `UInt32`, `UInt64`, `UInt128`, `UInt256`, `Int8`, `Int16`, `Int32`, `Int64`, `Int128`, `Int256`)
-- **Floating-point numbers**: [floats](./float.md)(`Float32` and `Float64`) and [`Decimal` values](./decimal.md)
-- **Boolean**: ClickHouse has a [`Boolean` type](./boolean.md)
-- **Strings**: [`String`](./string.md) and [`FixedString`](./fixedstring.md)
-- **Dates**: use [`Date`](./date.md) and [`Date32`](./date32.md) for days, and [`DateTime`](./datetime.md) and [`DateTime64`](./datetime64.md) for instances in time
-- **Object**: the [`Object`](./json.md) stores a JSON document in a single column (deprecated)
-- **JSON**: the [`JSON` object](./newjson.md) stores a JSON document in a single column
-- **UUID**: a performant option for storing [`UUID` values](./uuid.md)
-- **Low cardinality types**: use an [`Enum`](./enum.md) when you have a handful of unique values, or use [`LowCardinality`](./lowcardinality.md) when you have up to 10,000 unique values of a column
-- **Arrays**: any column can be defined as an [`Array` of values](./array.md)
-- **Maps**: use [`Map`](./map.md) for storing key/value pairs
-- **Aggregation function types**: use [`SimpleAggregateFunction`](./simpleaggregatefunction.md) and [`AggregateFunction`](./aggregatefunction.md) for storing the intermediate status of aggregate function results
-- **Nested data structures**: A [`Nested` data structure](./nested-data-structures/index.md) is like a table inside a cell
-- **Tuples**: A [`Tuple` of elements](./tuple.md), each having an individual type.
-- **Nullable**: [`Nullable`](./nullable.md) allows you to store a value as `NULL` when a value is "missing" (instead of the column settings its default value for the data type)
-- **IP addresses**: use [`IPv4`](./ipv4.md) and [`IPv6`](./ipv6.md) to efficiently store IP addresses
-- **Geo types**: for [geographical data](./geo.md), including `Point`, `Ring`, `Polygon` and `MultiPolygon`
-- **Special data types**: including [`Expression`](./special-data-types/expression.md), [`Set`](./special-data-types/set.md), [`Nothing`](./special-data-types/nothing.md) and [`Interval`](./special-data-types/interval.md)
+System table [system.data_type_families](../../operations/system-tables/data_type_families.md#system_tables-data_type_families) provides an
+overview of all available data types.
+It also shows whether a data type is an alias to another data type and its name is case-sensitive (e.g. `bool` vs. `BOOL`).

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

@@ -7,7 +7,7 @@ keywords: [object, data type]
 
 # Object Data Type (deprecated)
 
-**This feature is not production-ready and is now deprecated.** If you need to work with JSON documents, consider using [this guide](/docs/en/integrations/data-formats/json/overview) instead. A new implementation to support JSON object is in progress and can be tracked [here](https://github.com/ClickHouse/ClickHouse/issues/54864).
+**This feature is not production-ready and deprecated.** If you need to work with JSON documents, consider using [this guide](/docs/en/integrations/data-formats/json/overview) instead. A new implementation to support JSON object is in progress and can be tracked [here](https://github.com/ClickHouse/ClickHouse/issues/54864).
 
 <hr />
 

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

@@ -5,7 +5,9 @@ sidebar_label: SimpleAggregateFunction
 ---
 # SimpleAggregateFunction
 
-`SimpleAggregateFunction(name, types_of_arguments...)` data type stores current value of the aggregate function, and does not store its full state as [`AggregateFunction`](../../sql-reference/data-types/aggregatefunction.md) does. This optimization can be applied to functions for which the following property holds: the result of applying a function `f` to a row set `S1 UNION ALL S2` can be obtained by applying `f` to parts of the row set separately, and then again applying `f` to the results: `f(S1 UNION ALL S2) = f(f(S1) UNION ALL f(S2))`. This property guarantees that partial aggregation results are enough to compute the combined one, so we do not have to store and process any extra data.
+`SimpleAggregateFunction(name, types_of_arguments...)` data type stores current value (intermediate state) of the aggregate function, but not its full state as [`AggregateFunction`](../../sql-reference/data-types/aggregatefunction.md) does.
+This optimization can be applied to functions for which the following property holds: the result of applying a function `f` to a row set `S1 UNION ALL S2` can be obtained by applying `f` to parts of the row set separately, and then again applying `f` to the results: `f(S1 UNION ALL S2) = f(f(S1) UNION ALL f(S2))`.
+This property guarantees that partial aggregation results are enough to compute the combined one, so we do not have to store and process any extra data.
 
 The common way to produce an aggregate function value is by calling the aggregate function with the [-SimpleState](../../sql-reference/aggregate-functions/combinators.md#agg-functions-combinator-simplestate) suffix.
 

docs/en/sql-reference/functions/date-time-functions.md

@@ -4773,7 +4773,7 @@ Result:
 
 ## toUTCTimestamp
 
-Convert DateTime/DateTime64 type value from other time zone to UTC timezone timestamp
+Convert DateTime/DateTime64 type value from other time zone to UTC timezone timestamp. This function is mainly included for compatibility with Apache Spark and similar frameworks.
 
 **Syntax**
 
@@ -4806,7 +4806,7 @@ Result:
 
 ## fromUTCTimestamp
 
-Convert DateTime/DateTime64 type value from UTC timezone to other time zone timestamp
+Convert DateTime/DateTime64 type value from UTC timezone to other time zone timestamp.  This function is mainly included for compatibility with Apache Spark and similar frameworks.
 
 **Syntax**
 

docs/en/sql-reference/functions/geo/index.md

@@ -5,70 +5,4 @@ sidebar_position: 62
 title: "Geo Functions"
 ---
 
-
-## Geographical Coordinates Functions
-
-- [greatCircleDistance](./coordinates.md#greatcircledistance)
-- [geoDistance](./coordinates.md#geodistance)
-- [greatCircleAngle](./coordinates.md#greatcircleangle)
-- [pointInEllipses](./coordinates.md#pointinellipses)
-- [pointInPolygon](./coordinates.md#pointinpolygon)
-
-## Geohash Functions
-- [geohashEncode](./geohash.md#geohashencode)
-- [geohashDecode](./geohash.md#geohashdecode)
-- [geohashesInBox](./geohash.md#geohashesinbox)
-
-## H3 Indexes Functions
-
-- [h3IsValid](./h3.md#h3isvalid)
-- [h3GetResolution](./h3.md#h3getresolution)
-- [h3EdgeAngle](./h3.md#h3edgeangle)
-- [h3EdgeLengthM](./h3.md#h3edgelengthm)
-- [h3EdgeLengthKm](./h3.md#h3edgelengthkm)
-- [geoToH3](./h3.md#geotoh3)
-- [h3ToGeo](./h3.md#h3togeo)
-- [h3ToGeoBoundary](./h3.md#h3togeoboundary)
-- [h3kRing](./h3.md#h3kring)
-- [h3GetBaseCell](./h3.md#h3getbasecell)
-- [h3HexAreaM2](./h3.md#h3hexaream2)
-- [h3HexAreaKm2](./h3.md#h3hexareakm2)
-- [h3IndexesAreNeighbors](./h3.md#h3indexesareneighbors)
-- [h3ToChildren](./h3.md#h3tochildren)
-- [h3ToParent](./h3.md#h3toparent)
-- [h3ToString](./h3.md#h3tostring)
-- [stringToH3](./h3.md#stringtoh3)
-- [h3GetResolution](./h3.md#h3getresolution)
-- [h3IsResClassIII](./h3.md#h3isresclassiii)
-- [h3IsPentagon](./h3.md#h3ispentagon)
-- [h3GetFaces](./h3.md#h3getfaces)
-- [h3CellAreaM2](./h3.md#h3cellaream2)
-- [h3CellAreaRads2](./h3.md#h3cellarearads2)
-- [h3ToCenterChild](./h3.md#h3tocenterchild)
-- [h3ExactEdgeLengthM](./h3.md#h3exactedgelengthm)
-- [h3ExactEdgeLengthKm](./h3.md#h3exactedgelengthkm)
-- [h3ExactEdgeLengthRads](./h3.md#h3exactedgelengthrads)
-- [h3NumHexagons](./h3.md#h3numhexagons)
-- [h3Line](./h3.md#h3line)
-- [h3Distance](./h3.md#h3distance)
-- [h3HexRing](./h3.md#h3hexring)
-- [h3GetUnidirectionalEdge](./h3.md#h3getunidirectionaledge)
-- [h3UnidirectionalEdgeIsValid](./h3.md#h3unidirectionaledgeisvalid)
-- [h3GetOriginIndexFromUnidirectionalEdge](./h3.md#h3getoriginindexfromunidirectionaledge)
-- [h3GetDestinationIndexFromUnidirectionalEdge](./h3.md#h3getdestinationindexfromunidirectionaledge)
-- [h3GetIndexesFromUnidirectionalEdge](./h3.md#h3getindexesfromunidirectionaledge)
-- [h3GetUnidirectionalEdgesFromHexagon](./h3.md#h3getunidirectionaledgesfromhexagon)
-- [h3GetUnidirectionalEdgeBoundary](./h3.md#h3getunidirectionaledgeboundary)
-
-## S2 Index Functions
-
-- [geoToS2](./s2.md#geotos2)
-- [s2ToGeo](./s2.md#s2togeo)
-- [s2GetNeighbors](./s2.md#s2getneighbors)
-- [s2CellsIntersect](./s2.md#s2cellsintersect)
-- [s2CapContains](./s2.md#s2capcontains)
-- [s2CapUnion](./s2.md#s2capunion)
-- [s2RectAdd](./s2.md#s2rectadd)
-- [s2RectContains](./s2.md#s2rectcontains)
-- [s2RectUnion](./s2.md#s2rectunion)
-- [s2RectIntersection](./s2.md#s2rectintersection)
+Functions for working with geometric objects, for example [to calculate distances between points on a sphere](./coordinates.md), [compute geohashes](./geohash.md), and work with [h3 indexes](./h3.md).

docs/en/sql-reference/statements/alter/column.md

@@ -279,7 +279,7 @@ For columns with a new or updated `MATERIALIZED` value expression, all existing
 
 For columns with a new or updated `DEFAULT` value expression, the behavior depends on the ClickHouse version:
 - In ClickHouse < v24.2, all existing rows are rewritten.
-- ClickHouse >= v24.2 distinguishes if a row value in a column with `DEFAULT` value expression was explicitly specified when it was inserted, or not, i.e. calculated from the `DEFAULT` value expression. If the value was explicitly specified, ClickHouse keeps it as is. If the value was was calculated, ClickHouse changes it to the new or updated `MATERIALIZED` value expression.
+- ClickHouse >= v24.2 distinguishes if a row value in a column with `DEFAULT` value expression was explicitly specified when it was inserted, or not, i.e. calculated from the `DEFAULT` value expression. If the value was explicitly specified, ClickHouse keeps it as is. If the value was calculated, ClickHouse changes it to the new or updated `MATERIALIZED` value expression.
 
 Syntax:
 

docs/en/sql-reference/statements/create/index.md

@@ -6,16 +6,4 @@ sidebar_label: CREATE
 
 # CREATE Queries
 
-Create queries make a new entity of one of the following kinds:
-
-- [DATABASE](/docs/en/sql-reference/statements/create/database.md)
-- [TABLE](/docs/en/sql-reference/statements/create/table.md)
-- [VIEW](/docs/en/sql-reference/statements/create/view.md)
-- [DICTIONARY](/docs/en/sql-reference/statements/create/dictionary.md)
-- [FUNCTION](/docs/en/sql-reference/statements/create/function.md)
-- [USER](/docs/en/sql-reference/statements/create/user.md)
-- [ROLE](/docs/en/sql-reference/statements/create/role.md)
-- [ROW POLICY](/docs/en/sql-reference/statements/create/row-policy.md)
-- [QUOTA](/docs/en/sql-reference/statements/create/quota.md)
-- [SETTINGS PROFILE](/docs/en/sql-reference/statements/create/settings-profile.md)
-- [NAMED COLLECTION](/docs/en/sql-reference/statements/create/named-collection.md)
+CREATE queries create (for example) new [databases](/docs/en/sql-reference/statements/create/database.md), [tables](/docs/en/sql-reference/statements/create/table.md) and [views](/docs/en/sql-reference/statements/create/view.md).

docs/en/sql-reference/statements/index.md

@@ -6,27 +6,4 @@ sidebar_label: List of statements
 
 # ClickHouse SQL Statements
 
-Statements represent various kinds of action you can perform using SQL queries. Each kind of statement has it’s own syntax and usage details that are described separately:
-
-- [SELECT](/docs/en/sql-reference/statements/select/index.md)
-- [INSERT INTO](/docs/en/sql-reference/statements/insert-into.md)
-- [CREATE](/docs/en/sql-reference/statements/create/index.md)
-- [ALTER](/docs/en/sql-reference/statements/alter/index.md)
-- [SYSTEM](/docs/en/sql-reference/statements/system.md)
-- [SHOW](/docs/en/sql-reference/statements/show.md)
-- [GRANT](/docs/en/sql-reference/statements/grant.md)
-- [REVOKE](/docs/en/sql-reference/statements/revoke.md)
-- [ATTACH](/docs/en/sql-reference/statements/attach.md)
-- [CHECK TABLE](/docs/en/sql-reference/statements/check-table.md)
-- [DESCRIBE TABLE](/docs/en/sql-reference/statements/describe-table.md)
-- [DETACH](/docs/en/sql-reference/statements/detach.md)
-- [DROP](/docs/en/sql-reference/statements/drop.md)
-- [EXISTS](/docs/en/sql-reference/statements/exists.md)
-- [KILL](/docs/en/sql-reference/statements/kill.md)
-- [OPTIMIZE](/docs/en/sql-reference/statements/optimize.md)
-- [RENAME](/docs/en/sql-reference/statements/rename.md)
-- [SET](/docs/en/sql-reference/statements/set.md)
-- [SET ROLE](/docs/en/sql-reference/statements/set-role.md)
-- [TRUNCATE](/docs/en/sql-reference/statements/truncate.md)
-- [USE](/docs/en/sql-reference/statements/use.md)
-- [EXPLAIN](/docs/en/sql-reference/statements/explain.md)
+Users interact with ClickHouse using SQL statements. ClickHouse supports common SQL statements like [SELECT](select/index.md) and [CREATE](create/index.md), but it also provides specialized statements like [KILL](kill.md) and [OPTIMIZE](optimize.md).

programs/compressor/Compressor.cpp

@@ -12,9 +12,12 @@
 #include <Compression/ParallelCompressedWriteBuffer.h>
 #include <Compression/CompressedReadBuffer.h>
 #include <Compression/CompressedReadBufferFromFile.h>
+#include <Compression/getCompressionCodecForFile.h>
+#include <IO/ReadHelpers.h>
 #include <IO/WriteHelpers.h>
 #include <IO/copyData.h>
 #include <Parsers/parseQuery.h>
+#include <Parsers/queryToString.h>
 #include <Parsers/ExpressionElementParsers.h>
 #include <Compression/CompressionFactory.h>
 #include <Common/TerminalSize.h>
@@ -43,29 +46,24 @@ namespace CurrentMetrics
 namespace
 {
 
-/// Outputs sizes of uncompressed and compressed blocks for compressed file.
+/// Outputs method, sizes of uncompressed and compressed blocks for compressed file.
 void checkAndWriteHeader(DB::ReadBuffer & in, DB::WriteBuffer & out)
 {
     while (!in.eof())
     {
-        in.ignore(16);    /// checksum
-
-        char header[COMPRESSED_BLOCK_HEADER_SIZE];
-        in.readStrict(header, COMPRESSED_BLOCK_HEADER_SIZE);
-
-        UInt32 size_compressed = unalignedLoad<UInt32>(&header[1]);
+        UInt32 size_compressed;
+        UInt32 size_decompressed;
+        auto codec = DB::getCompressionCodecForFile(in, size_compressed, size_decompressed, true /* skip_to_next_block */);
 
         if (size_compressed > DBMS_MAX_COMPRESSED_SIZE)
             throw DB::Exception(DB::ErrorCodes::TOO_LARGE_SIZE_COMPRESSED, "Too large size_compressed. Most likely corrupted data.");
 
-        UInt32 size_decompressed = unalignedLoad<UInt32>(&header[5]);
-
+        DB::writeText(queryToString(codec->getFullCodecDesc()), out);
+        DB::writeChar('\t', out);
         DB::writeText(size_decompressed, out);
         DB::writeChar('\t', out);
         DB::writeText(size_compressed, out);
         DB::writeChar('\n', out);
-
-        in.ignore(size_compressed - COMPRESSED_BLOCK_HEADER_SIZE);
     }
 }
 

programs/local/LocalServer.cpp

@@ -14,6 +14,7 @@
 #include <Databases/registerDatabases.h>
 #include <Databases/DatabaseFilesystem.h>
 #include <Databases/DatabaseMemory.h>
+#include <Databases/DatabaseAtomic.h>
 #include <Databases/DatabasesOverlay.h>
 #include <Storages/System/attachSystemTables.h>
 #include <Storages/System/attachInformationSchemaTables.h>
@@ -257,12 +258,12 @@ static DatabasePtr createMemoryDatabaseIfNotExists(ContextPtr context, const Str
     return system_database;
 }
 
-static DatabasePtr createClickHouseLocalDatabaseOverlay(const String & name_, ContextPtr context_)
+static DatabasePtr createClickHouseLocalDatabaseOverlay(const String & name_, ContextPtr context)
 {
-    auto databaseCombiner = std::make_shared<DatabasesOverlay>(name_, context_);
-    databaseCombiner->registerNextDatabase(std::make_shared<DatabaseFilesystem>(name_, "", context_));
-    databaseCombiner->registerNextDatabase(std::make_shared<DatabaseMemory>(name_, context_));
-    return databaseCombiner;
+    auto overlay = std::make_shared<DatabasesOverlay>(name_, context);
+    overlay->registerNextDatabase(std::make_shared<DatabaseAtomic>(name_, fs::weakly_canonical(context->getPath()), UUIDHelpers::generateV4(), context));
+    overlay->registerNextDatabase(std::make_shared<DatabaseFilesystem>(name_, "", context));
+    return overlay;
 }
 
 /// If path is specified and not empty, will try to setup server environment and load existing metadata
@@ -811,7 +812,12 @@ void LocalServer::processConfig()
     DatabaseCatalog::instance().initializeAndLoadTemporaryDatabase();
 
     std::string default_database = server_settings[ServerSetting::default_database];
-    DatabaseCatalog::instance().attachDatabase(default_database, createClickHouseLocalDatabaseOverlay(default_database, global_context));
+    {
+        DatabasePtr database = createClickHouseLocalDatabaseOverlay(default_database, global_context);
+        if (UUID uuid = database->getUUID(); uuid != UUIDHelpers::Nil)
+            DatabaseCatalog::instance().addUUIDMapping(uuid);
+        DatabaseCatalog::instance().attachDatabase(default_database, database);
+    }
     global_context->setCurrentDatabase(default_database);
 
     if (getClientConfiguration().has("path"))

programs/main.cpp

@@ -1,27 +1,22 @@
-#include <unistd.h>
-#include <fcntl.h>
+#include <base/phdr_cache.h>
+#include <Common/EnvironmentChecks.h>
+#include <Common/StringUtils.h>
+#include <Common/getHashOfLoadedBinary.h>
 
-#include <new>
-#include <iostream>
-#include <vector>
-#include <string>
-#include <string_view>
-#include <utility> /// pair
-
-#include <fmt/format.h>
+#if defined(SANITIZE_COVERAGE)
+#    include <Common/Coverage.h>
+#endif
 
 #include "config.h"
 #include "config_tools.h"
 
-#include <Common/EnvironmentChecks.h>
-#include <Common/Coverage.h>
-#include <Common/StringUtils.h>
-#include <Common/getHashOfLoadedBinary.h>
-#include <Common/IO.h>
-
-#include <base/phdr_cache.h>
-#include <base/coverage.h>
-
+#include <filesystem>
+#include <iostream>
+#include <new>
+#include <string>
+#include <string_view>
+#include <utility> /// pair
+#include <vector>
 
 /// Universal executable for various clickhouse applications
 int mainEntryClickHouseServer(int argc, char ** argv);
@@ -238,9 +233,12 @@ int main(int argc_, char ** argv_)
     ///     clickhouse # spawn local
     ///     clickhouse local # spawn local
     ///     clickhouse "select ..." # spawn local
+    ///     clickhouse /tmp/repro --enable-analyzer
     ///
-    if (main_func == printHelp && !argv.empty() && (argv.size() == 1 || argv[1][0] == '-'
-        || std::string_view(argv[1]).contains(' ')))
+    std::error_code ec;
+    if (main_func == printHelp && !argv.empty()
+        && (argv.size() == 1 || argv[1][0] == '-' || std::string_view(argv[1]).contains(' ')
+            || std::filesystem::is_regular_file(std::filesystem::path{argv[1]}, ec)))
     {
         main_func = mainEntryClickHouseLocal;
     }

src/AggregateFunctions/AggregateFunctionDeltaSumTimestamp.cpp

@@ -22,6 +22,13 @@ namespace ErrorCodes
 namespace
 {
 
+/** Due to a lack of proper code review, this code was contributed with a multiplication of template instantiations
+  * over all pairs of data types, and we deeply regret that.
+  *
+  * We cannot remove all combinations, because the binary representation of serialized data has to remain the same,
+  * but we can partially heal the wound by treating unsigned and signed data types in the same way.
+  */
+
 template <typename ValueType, typename TimestampType>
 struct AggregationFunctionDeltaSumTimestampData
 {
@@ -37,23 +44,22 @@ template <typename ValueType, typename TimestampType>
 class AggregationFunctionDeltaSumTimestamp final
     : public IAggregateFunctionDataHelper<
         AggregationFunctionDeltaSumTimestampData<ValueType, TimestampType>,
-        AggregationFunctionDeltaSumTimestamp<ValueType, TimestampType>
-      >
+        AggregationFunctionDeltaSumTimestamp<ValueType, TimestampType>>
 {
 public:
     AggregationFunctionDeltaSumTimestamp(const DataTypes & arguments, const Array & params)
         : IAggregateFunctionDataHelper<
             AggregationFunctionDeltaSumTimestampData<ValueType, TimestampType>,
-            AggregationFunctionDeltaSumTimestamp<ValueType, TimestampType>
-        >{arguments, params, createResultType()}
-    {}
+            AggregationFunctionDeltaSumTimestamp<ValueType, TimestampType>>{arguments, params, createResultType()}
+    {
+    }
 
     AggregationFunctionDeltaSumTimestamp()
         : IAggregateFunctionDataHelper<
             AggregationFunctionDeltaSumTimestampData<ValueType, TimestampType>,
-            AggregationFunctionDeltaSumTimestamp<ValueType, TimestampType>
-        >{}
-    {}
+            AggregationFunctionDeltaSumTimestamp<ValueType, TimestampType>>{}
+    {
+    }
 
     bool allocatesMemoryInArena() const override { return false; }
 
@@ -63,8 +69,8 @@ public:
 
     void NO_SANITIZE_UNDEFINED ALWAYS_INLINE add(AggregateDataPtr __restrict place, const IColumn ** columns, size_t row_num, Arena *) const override
     {
-        auto value = assert_cast<const ColumnVector<ValueType> &>(*columns[0]).getData()[row_num];
-        auto ts = assert_cast<const ColumnVector<TimestampType> &>(*columns[1]).getData()[row_num];
+        auto value = unalignedLoad<ValueType>(columns[0]->getRawData().data() + row_num * sizeof(ValueType));
+        auto ts = unalignedLoad<TimestampType>(columns[1]->getRawData().data() + row_num * sizeof(TimestampType));
 
         auto & data = this->data(place);
 
@@ -172,10 +178,48 @@ public:
 
     void insertResultInto(AggregateDataPtr __restrict place, IColumn & to, Arena *) const override
     {
-        assert_cast<ColumnVector<ValueType> &>(to).getData().push_back(this->data(place).sum);
+        static_cast<ColumnFixedSizeHelper &>(to).template insertRawData<sizeof(ValueType)>(
+            reinterpret_cast<const char *>(&this->data(place).sum));
     }
 };
 
+
+template <typename FirstType, template <typename, typename> class AggregateFunctionTemplate, typename... TArgs>
+IAggregateFunction * createWithTwoTypesSecond(const IDataType & second_type, TArgs && ... args)
+{
+    WhichDataType which(second_type);
+
+    if (which.idx == TypeIndex::UInt32) return new AggregateFunctionTemplate<FirstType, UInt32>(args...);
+    if (which.idx == TypeIndex::UInt64) return new AggregateFunctionTemplate<FirstType, UInt64>(args...);
+    if (which.idx == TypeIndex::Int32) return new AggregateFunctionTemplate<FirstType, UInt32>(args...);
+    if (which.idx == TypeIndex::Int64) return new AggregateFunctionTemplate<FirstType, UInt64>(args...);
+    if (which.idx == TypeIndex::Float32) return new AggregateFunctionTemplate<FirstType, Float32>(args...);
+    if (which.idx == TypeIndex::Float64) return new AggregateFunctionTemplate<FirstType, Float64>(args...);
+    if (which.idx == TypeIndex::Date) return new AggregateFunctionTemplate<FirstType, UInt16>(args...);
+    if (which.idx == TypeIndex::DateTime) return new AggregateFunctionTemplate<FirstType, UInt32>(args...);
+
+    return nullptr;
+}
+
+template <template <typename, typename> class AggregateFunctionTemplate, typename... TArgs>
+IAggregateFunction * createWithTwoTypes(const IDataType & first_type, const IDataType & second_type, TArgs && ... args)
+{
+    WhichDataType which(first_type);
+
+    if (which.idx == TypeIndex::UInt8) return createWithTwoTypesSecond<UInt8, AggregateFunctionTemplate>(second_type, args...);
+    if (which.idx == TypeIndex::UInt16) return createWithTwoTypesSecond<UInt16, AggregateFunctionTemplate>(second_type, args...);
+    if (which.idx == TypeIndex::UInt32) return createWithTwoTypesSecond<UInt32, AggregateFunctionTemplate>(second_type, args...);
+    if (which.idx == TypeIndex::UInt64) return createWithTwoTypesSecond<UInt64, AggregateFunctionTemplate>(second_type, args...);
+    if (which.idx == TypeIndex::Int8) return createWithTwoTypesSecond<UInt8, AggregateFunctionTemplate>(second_type, args...);
+    if (which.idx == TypeIndex::Int16) return createWithTwoTypesSecond<UInt16, AggregateFunctionTemplate>(second_type, args...);
+    if (which.idx == TypeIndex::Int32) return createWithTwoTypesSecond<UInt32, AggregateFunctionTemplate>(second_type, args...);
+    if (which.idx == TypeIndex::Int64) return createWithTwoTypesSecond<UInt64, AggregateFunctionTemplate>(second_type, args...);
+    if (which.idx == TypeIndex::Float32) return createWithTwoTypesSecond<Float32, AggregateFunctionTemplate>(second_type, args...);
+    if (which.idx == TypeIndex::Float64) return createWithTwoTypesSecond<Float64, AggregateFunctionTemplate>(second_type, args...);
+
+    return nullptr;
+}
+
 AggregateFunctionPtr createAggregateFunctionDeltaSumTimestamp(
     const String & name,
     const DataTypes & arguments,
@@ -193,7 +237,7 @@ AggregateFunctionPtr createAggregateFunctionDeltaSumTimestamp(
         throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Illegal type {} of argument for aggregate function {}, "
                         "must be Int, Float, Date, DateTime", arguments[1]->getName(), name);
 
-    return AggregateFunctionPtr(createWithTwoNumericOrDateTypes<AggregationFunctionDeltaSumTimestamp>(
+    return AggregateFunctionPtr(createWithTwoTypes<AggregationFunctionDeltaSumTimestamp>(
         *arguments[0], *arguments[1], arguments, params));
 }
 }

src/AggregateFunctions/Helpers.h

@@ -184,36 +184,8 @@ static IAggregateFunction * createWithDecimalType(const IDataType & argument_typ
 }
 
 /** For template with two arguments.
+  * This is an extremely dangerous for code bloat - do not use.
   */
-template <typename FirstType, template <typename, typename> class AggregateFunctionTemplate, typename... TArgs>
-static IAggregateFunction * createWithTwoNumericTypesSecond(const IDataType & second_type, TArgs && ... args)
-{
-    WhichDataType which(second_type);
-#define DISPATCH(TYPE) \
-    if (which.idx == TypeIndex::TYPE) return new AggregateFunctionTemplate<FirstType, TYPE>(args...);
-    FOR_NUMERIC_TYPES(DISPATCH)
-#undef DISPATCH
-    if (which.idx == TypeIndex::Enum8) return new AggregateFunctionTemplate<FirstType, Int8>(args...);
-    if (which.idx == TypeIndex::Enum16) return new AggregateFunctionTemplate<FirstType, Int16>(args...);
-    return nullptr;
-}
-
-template <template <typename, typename> class AggregateFunctionTemplate, typename... TArgs>
-static IAggregateFunction * createWithTwoNumericTypes(const IDataType & first_type, const IDataType & second_type, TArgs && ... args)
-{
-    WhichDataType which(first_type);
-#define DISPATCH(TYPE) \
-    if (which.idx == TypeIndex::TYPE) \
-        return createWithTwoNumericTypesSecond<TYPE, AggregateFunctionTemplate>(second_type, args...);
-    FOR_NUMERIC_TYPES(DISPATCH)
-#undef DISPATCH
-    if (which.idx == TypeIndex::Enum8)
-        return createWithTwoNumericTypesSecond<Int8, AggregateFunctionTemplate>(second_type, args...);
-    if (which.idx == TypeIndex::Enum16)
-        return createWithTwoNumericTypesSecond<Int16, AggregateFunctionTemplate>(second_type, args...);
-    return nullptr;
-}
-
 template <typename FirstType, template <typename, typename> class AggregateFunctionTemplate, typename... TArgs>
 static IAggregateFunction * createWithTwoBasicNumericTypesSecond(const IDataType & second_type, TArgs && ... args)
 {
@@ -237,46 +209,6 @@ static IAggregateFunction * createWithTwoBasicNumericTypes(const IDataType & fir
     return nullptr;
 }
 
-template <typename FirstType, template <typename, typename> class AggregateFunctionTemplate, typename... TArgs>
-static IAggregateFunction * createWithTwoNumericOrDateTypesSecond(const IDataType & second_type, TArgs && ... args)
-{
-    WhichDataType which(second_type);
-#define DISPATCH(TYPE) \
-    if (which.idx == TypeIndex::TYPE) return new AggregateFunctionTemplate<FirstType, TYPE>(args...);
-    FOR_NUMERIC_TYPES(DISPATCH)
-#undef DISPATCH
-    if (which.idx == TypeIndex::Enum8) return new AggregateFunctionTemplate<FirstType, Int8>(args...);
-    if (which.idx == TypeIndex::Enum16) return new AggregateFunctionTemplate<FirstType, Int16>(args...);
-
-    /// expects that DataTypeDate based on UInt16, DataTypeDateTime based on UInt32
-    if (which.idx == TypeIndex::Date) return new AggregateFunctionTemplate<FirstType, UInt16>(args...);
-    if (which.idx == TypeIndex::DateTime) return new AggregateFunctionTemplate<FirstType, UInt32>(args...);
-
-    return nullptr;
-}
-
-template <template <typename, typename> class AggregateFunctionTemplate, typename... TArgs>
-static IAggregateFunction * createWithTwoNumericOrDateTypes(const IDataType & first_type, const IDataType & second_type, TArgs && ... args)
-{
-    WhichDataType which(first_type);
-#define DISPATCH(TYPE) \
-    if (which.idx == TypeIndex::TYPE) \
-        return createWithTwoNumericOrDateTypesSecond<TYPE, AggregateFunctionTemplate>(second_type, args...);
-    FOR_NUMERIC_TYPES(DISPATCH)
-#undef DISPATCH
-    if (which.idx == TypeIndex::Enum8)
-        return createWithTwoNumericOrDateTypesSecond<Int8, AggregateFunctionTemplate>(second_type, args...);
-    if (which.idx == TypeIndex::Enum16)
-        return createWithTwoNumericOrDateTypesSecond<Int16, AggregateFunctionTemplate>(second_type, args...);
-
-    /// expects that DataTypeDate based on UInt16, DataTypeDateTime based on UInt32
-    if (which.idx == TypeIndex::Date)
-        return createWithTwoNumericOrDateTypesSecond<UInt16, AggregateFunctionTemplate>(second_type, args...);
-    if (which.idx == TypeIndex::DateTime)
-        return createWithTwoNumericOrDateTypesSecond<UInt32, AggregateFunctionTemplate>(second_type, args...);
-    return nullptr;
-}
-
 template <template <typename> class AggregateFunctionTemplate, typename... TArgs>
 static IAggregateFunction * createWithStringType(const IDataType & argument_type, TArgs && ... args)
 {

src/Analyzer/FunctionNode.cpp

@@ -88,6 +88,7 @@ void FunctionNode::resolveAsFunction(FunctionBasePtr function_value)
     function_name = function_value->getName();
     function = std::move(function_value);
     kind = FunctionKind::ORDINARY;
+    nulls_action = NullsAction::EMPTY;
 }
 
 void FunctionNode::resolveAsAggregateFunction(AggregateFunctionPtr aggregate_function_value)
@@ -95,6 +96,12 @@ void FunctionNode::resolveAsAggregateFunction(AggregateFunctionPtr aggregate_fun
     function_name = aggregate_function_value->getName();
     function = std::move(aggregate_function_value);
     kind = FunctionKind::AGGREGATE;
+    /**  When the function is resolved, we do not need the nulls action anymore.
+      * The only thing that the nulls action does is map from one function to another.
+      * Thus, the nulls action is encoded in the function name and does not make sense anymore.
+      * Keeping the nulls action may lead to incorrect comparison of functions, e.g., count() and count() IGNORE NULLS are the same function.
+      */
+    nulls_action = NullsAction::EMPTY;
 }
 
 void FunctionNode::resolveAsWindowFunction(AggregateFunctionPtr window_function_value)

src/Analyzer/JoinNode.cpp

@@ -48,9 +48,15 @@ ASTPtr JoinNode::toASTTableJoin() const
         auto join_expression_ast = children[join_expression_child_index]->toAST();
 
         if (is_using_join_expression)
-            join_ast->using_expression_list = std::move(join_expression_ast);
+        {
+            join_ast->using_expression_list = join_expression_ast;
+            join_ast->children.push_back(join_ast->using_expression_list);
+        }
         else
-            join_ast->on_expression = std::move(join_expression_ast);
+        {
+            join_ast->on_expression = join_expression_ast;
+            join_ast->children.push_back(join_ast->on_expression);
+        }
     }
 
     return join_ast;

src/Analyzer/Passes/FuseFunctionsPass.cpp

@@ -85,10 +85,9 @@ QueryTreeNodePtr createResolvedFunction(const ContextPtr & context, const String
 }
 
 FunctionNodePtr createResolvedAggregateFunction(
-    const String & name, const QueryTreeNodePtr & argument, const Array & parameters = {}, NullsAction action = NullsAction::EMPTY)
+    const String & name, const QueryTreeNodePtr & argument, const Array & parameters = {})
 {
     auto function_node = std::make_shared<FunctionNode>(name);
-    function_node->setNullsAction(action);
 
     if (!parameters.empty())
     {
@@ -100,7 +99,7 @@ FunctionNodePtr createResolvedAggregateFunction(
     function_node->getArguments().getNodes() = { argument };
 
     AggregateFunctionProperties properties;
-    auto aggregate_function = AggregateFunctionFactory::instance().get(name, action, {argument->getResultType()}, parameters, properties);
+    auto aggregate_function = AggregateFunctionFactory::instance().get(name, NullsAction::EMPTY, {argument->getResultType()}, parameters, properties);
     function_node->resolveAsAggregateFunction(std::move(aggregate_function));
 
     return function_node;

src/Analyzer/QueryTreePassManager.cpp

@@ -3,7 +3,6 @@
 #include <memory>
 
 #include <Common/Exception.h>
-#include "Analyzer/Passes/OptimizeGroupByInjectiveFunctionsPass.h"
 
 #include <IO/WriteHelpers.h>
 #include <IO/Operators.h>
@@ -16,39 +15,39 @@
 #include <Analyzer/ColumnNode.h>
 #include <Analyzer/FunctionNode.h>
 #include <Analyzer/InDepthQueryTreeVisitor.h>
-#include <Analyzer/Utils.h>
+#include <Analyzer/Passes/AggregateFunctionOfGroupByKeysPass.h>
+#include <Analyzer/Passes/AggregateFunctionsArithmericOperationsPass.h>
+#include <Analyzer/Passes/ArrayExistsToHasPass.h>
+#include <Analyzer/Passes/AutoFinalOnQueryPass.h>
+#include <Analyzer/Passes/ComparisonTupleEliminationPass.h>
+#include <Analyzer/Passes/ConvertOrLikeChainPass.h>
+#include <Analyzer/Passes/ConvertQueryToCNFPass.h>
+#include <Analyzer/Passes/CountDistinctPass.h>
+#include <Analyzer/Passes/CrossToInnerJoinPass.h>
+#include <Analyzer/Passes/FunctionToSubcolumnsPass.h>
+#include <Analyzer/Passes/FuseFunctionsPass.h>
+#include <Analyzer/Passes/GroupingFunctionsResolvePass.h>
+#include <Analyzer/Passes/IfChainToMultiIfPass.h>
+#include <Analyzer/Passes/IfConstantConditionPass.h>
+#include <Analyzer/Passes/IfTransformStringsToEnumPass.h>
+#include <Analyzer/Passes/LogicalExpressionOptimizerPass.h>
+#include <Analyzer/Passes/MultiIfToIfPass.h>
+#include <Analyzer/Passes/NormalizeCountVariantsPass.h>
+#include <Analyzer/Passes/OptimizeDateOrDateTimeConverterWithPreimagePass.h>
+#include <Analyzer/Passes/OptimizeGroupByFunctionKeysPass.h>
+#include <Analyzer/Passes/OptimizeGroupByInjectiveFunctionsPass.h>
+#include <Analyzer/Passes/OptimizeRedundantFunctionsInOrderByPass.h>
+#include <Analyzer/Passes/OrderByLimitByDuplicateEliminationPass.h>
+#include <Analyzer/Passes/OrderByTupleEliminationPass.h>
 #include <Analyzer/Passes/QueryAnalysisPass.h>
 #include <Analyzer/Passes/RemoveUnusedProjectionColumnsPass.h>
-#include <Analyzer/Passes/RewriteSumFunctionWithSumAndCountPass.h>
-#include <Analyzer/Passes/CountDistinctPass.h>
-#include <Analyzer/Passes/UniqToCountPass.h>
-#include <Analyzer/Passes/FunctionToSubcolumnsPass.h>
 #include <Analyzer/Passes/RewriteAggregateFunctionWithIfPass.h>
-#include <Analyzer/Passes/SumIfToCountIfPass.h>
-#include <Analyzer/Passes/MultiIfToIfPass.h>
-#include <Analyzer/Passes/IfConstantConditionPass.h>
-#include <Analyzer/Passes/IfChainToMultiIfPass.h>
-#include <Analyzer/Passes/OrderByTupleEliminationPass.h>
-#include <Analyzer/Passes/NormalizeCountVariantsPass.h>
-#include <Analyzer/Passes/AggregateFunctionsArithmericOperationsPass.h>
-#include <Analyzer/Passes/UniqInjectiveFunctionsEliminationPass.h>
-#include <Analyzer/Passes/OrderByLimitByDuplicateEliminationPass.h>
-#include <Analyzer/Passes/FuseFunctionsPass.h>
-#include <Analyzer/Passes/OptimizeGroupByFunctionKeysPass.h>
-#include <Analyzer/Passes/IfTransformStringsToEnumPass.h>
-#include <Analyzer/Passes/ConvertOrLikeChainPass.h>
-#include <Analyzer/Passes/OptimizeRedundantFunctionsInOrderByPass.h>
-#include <Analyzer/Passes/GroupingFunctionsResolvePass.h>
-#include <Analyzer/Passes/AutoFinalOnQueryPass.h>
-#include <Analyzer/Passes/ArrayExistsToHasPass.h>
-#include <Analyzer/Passes/ComparisonTupleEliminationPass.h>
-#include <Analyzer/Passes/LogicalExpressionOptimizerPass.h>
-#include <Analyzer/Passes/CrossToInnerJoinPass.h>
+#include <Analyzer/Passes/RewriteSumFunctionWithSumAndCountPass.h>
 #include <Analyzer/Passes/ShardNumColumnToFunctionPass.h>
-#include <Analyzer/Passes/ConvertQueryToCNFPass.h>
-#include <Analyzer/Passes/AggregateFunctionOfGroupByKeysPass.h>
-#include <Analyzer/Passes/OptimizeDateOrDateTimeConverterWithPreimagePass.h>
-
+#include <Analyzer/Passes/SumIfToCountIfPass.h>
+#include <Analyzer/Passes/UniqInjectiveFunctionsEliminationPass.h>
+#include <Analyzer/Passes/UniqToCountPass.h>
+#include <Analyzer/Utils.h>
 
 namespace DB
 {

src/Analyzer/Resolve/QueryAnalyzer.cpp

@@ -1,3 +1,4 @@
+#include <Interpreters/ProcessorsProfileLog.h>
 #include <Common/FieldVisitorToString.h>
 
 #include <DataTypes/DataTypesNumber.h>
@@ -676,6 +677,8 @@ void QueryAnalyzer::evaluateScalarSubqueryIfNeeded(QueryTreeNodePtr & node, Iden
                         "tuple"});
                 }
             }
+
+            logProcessorProfile(context, io.pipeline.getProcessors());
         }
 
         scalars_cache.emplace(node_with_hash, scalar_block);

src/Backups/BackupConcurrencyCheck.cpp

@@ -14,12 +14,12 @@ namespace ErrorCodes
 
 
 BackupConcurrencyCheck::BackupConcurrencyCheck(
-    const UUID & backup_or_restore_uuid_,
     bool is_restore_,
     bool on_cluster_,
+    const String & zookeeper_path_,
     bool allow_concurrency_,
     BackupConcurrencyCounters & counters_)
-    : is_restore(is_restore_), backup_or_restore_uuid(backup_or_restore_uuid_), on_cluster(on_cluster_), counters(counters_)
+    : is_restore(is_restore_), on_cluster(on_cluster_), zookeeper_path(zookeeper_path_), counters(counters_)
 {
     std::lock_guard lock{counters.mutex};
 
@@ -32,7 +32,7 @@ BackupConcurrencyCheck::BackupConcurrencyCheck(
             size_t num_on_cluster_restores = counters.on_cluster_restores.size();
             if (on_cluster)
             {
-                if (!counters.on_cluster_restores.contains(backup_or_restore_uuid))
+                if (!counters.on_cluster_restores.contains(zookeeper_path))
                     ++num_on_cluster_restores;
             }
             else
@@ -47,7 +47,7 @@ BackupConcurrencyCheck::BackupConcurrencyCheck(
             size_t num_on_cluster_backups = counters.on_cluster_backups.size();
             if (on_cluster)
             {
-                if (!counters.on_cluster_backups.contains(backup_or_restore_uuid))
+                if (!counters.on_cluster_backups.contains(zookeeper_path))
                     ++num_on_cluster_backups;
             }
             else
@@ -64,9 +64,9 @@ BackupConcurrencyCheck::BackupConcurrencyCheck(
     if (on_cluster)
     {
         if (is_restore)
-            ++counters.on_cluster_restores[backup_or_restore_uuid];
+            ++counters.on_cluster_restores[zookeeper_path];
         else
-            ++counters.on_cluster_backups[backup_or_restore_uuid];
+            ++counters.on_cluster_backups[zookeeper_path];
     }
     else
     {
@@ -86,7 +86,7 @@ BackupConcurrencyCheck::~BackupConcurrencyCheck()
     {
         if (is_restore)
         {
-            auto it = counters.on_cluster_restores.find(backup_or_restore_uuid);
+            auto it = counters.on_cluster_restores.find(zookeeper_path);
             if (it != counters.on_cluster_restores.end())
             {
                 if (!--it->second)
@@ -95,7 +95,7 @@ BackupConcurrencyCheck::~BackupConcurrencyCheck()
         }
         else
         {
-            auto it = counters.on_cluster_backups.find(backup_or_restore_uuid);
+            auto it = counters.on_cluster_backups.find(zookeeper_path);
             if (it != counters.on_cluster_backups.end())
             {
                 if (!--it->second)

src/Backups/BackupConcurrencyCheck.h

@@ -1,7 +1,8 @@
 #pragma once
 
-#include <Core/UUID.h>
+#include <base/defines.h>
 #include <base/scope_guard.h>
+#include <base/types.h>
 #include <mutex>
 #include <unordered_map>
 
@@ -19,9 +20,9 @@ public:
     /// Checks concurrency of a BACKUP operation or a RESTORE operation.
     /// Keep a constructed instance of BackupConcurrencyCheck until the operation is done.
     BackupConcurrencyCheck(
-        const UUID & backup_or_restore_uuid_,
         bool is_restore_,
         bool on_cluster_,
+        const String & zookeeper_path_,
         bool allow_concurrency_,
         BackupConcurrencyCounters & counters_);
 
@@ -31,8 +32,8 @@ public:
 
 private:
     const bool is_restore;
-    const UUID backup_or_restore_uuid;
     const bool on_cluster;
+    const String zookeeper_path;
     BackupConcurrencyCounters & counters;
 };
 
@@ -47,8 +48,8 @@ private:
     friend class BackupConcurrencyCheck;
     size_t local_backups TSA_GUARDED_BY(mutex) = 0;
     size_t local_restores TSA_GUARDED_BY(mutex) = 0;
-    std::unordered_map<UUID /* backup_uuid */, size_t /* num_refs */> on_cluster_backups TSA_GUARDED_BY(mutex);
-    std::unordered_map<UUID /* restore_uuid */, size_t /* num_refs */> on_cluster_restores TSA_GUARDED_BY(mutex);
+    std::unordered_map<String /* zookeeper_path */, size_t /* num_refs */> on_cluster_backups TSA_GUARDED_BY(mutex);
+    std::unordered_map<String /* zookeeper_path */, size_t /* num_refs */> on_cluster_restores TSA_GUARDED_BY(mutex);
     std::mutex mutex;
 };
 

src/Backups/BackupCoordinationCleaner.cpp

@@ -4,31 +4,29 @@
 namespace DB
 {
 
-BackupCoordinationCleaner::BackupCoordinationCleaner(const String & zookeeper_path_, const WithRetries & with_retries_, LoggerPtr log_)
-    : zookeeper_path(zookeeper_path_), with_retries(with_retries_), log(log_)
+BackupCoordinationCleaner::BackupCoordinationCleaner(bool is_restore_, const String & zookeeper_path_, const WithRetries & with_retries_, LoggerPtr log_)
+    : is_restore(is_restore_), zookeeper_path(zookeeper_path_), with_retries(with_retries_), log(log_)
 {
 }
 
-void BackupCoordinationCleaner::cleanup()
+bool BackupCoordinationCleaner::cleanup(bool throw_if_error)
 {
-    tryRemoveAllNodes(/* throw_if_error = */ true, /* retries_kind = */ WithRetries::kNormal);
+    WithRetries::Kind retries_kind = throw_if_error ? WithRetries::kNormal : WithRetries::kErrorHandling;
+    return cleanupImpl(throw_if_error, retries_kind);
 }
 
-bool BackupCoordinationCleaner::tryCleanupAfterError() noexcept
-{
-    return tryRemoveAllNodes(/* throw_if_error = */ false, /* retries_kind = */ WithRetries::kNormal);
-}
-
-bool BackupCoordinationCleaner::tryRemoveAllNodes(bool throw_if_error, WithRetries::Kind retries_kind)
+bool BackupCoordinationCleaner::cleanupImpl(bool throw_if_error, WithRetries::Kind retries_kind)
 {
     {
         std::lock_guard lock{mutex};
-        if (cleanup_result.succeeded)
-            return true;
-        if (cleanup_result.exception)
+        if (succeeded)
         {
-            if (throw_if_error)
-                std::rethrow_exception(cleanup_result.exception);
+            LOG_TRACE(log, "Nodes from ZooKeeper are already removed");
+            return true;
+        }
+        if (tried)
+        {
+            LOG_INFO(log, "Skipped removing nodes from ZooKeeper because because earlier we failed to do that");
             return false;
         }
     }
@@ -44,16 +42,18 @@ bool BackupCoordinationCleaner::tryRemoveAllNodes(bool throw_if_error, WithRetri
         });
 
         std::lock_guard lock{mutex};
-        cleanup_result.succeeded = true;
+        tried = true;
+        succeeded = true;
         return true;
     }
     catch (...)
     {
-        LOG_TRACE(log, "Caught exception while removing nodes from ZooKeeper for this restore: {}",
+        LOG_TRACE(log, "Caught exception while removing nodes from ZooKeeper for this {}: {}",
+                  is_restore ? "restore" : "backup",
                   getCurrentExceptionMessage(/* with_stacktrace= */ false, /* check_embedded_stacktrace= */ true));
 
         std::lock_guard lock{mutex};
-        cleanup_result.exception = std::current_exception();
+        tried = true;
 
         if (throw_if_error)
             throw;

src/Backups/BackupCoordinationCleaner.h

@@ -12,14 +12,14 @@ namespace DB
 class BackupCoordinationCleaner
 {
 public:
-    BackupCoordinationCleaner(const String & zookeeper_path_, const WithRetries & with_retries_, LoggerPtr log_);
+    BackupCoordinationCleaner(bool is_restore_, const String & zookeeper_path_, const WithRetries & with_retries_, LoggerPtr log_);
 
-    void cleanup();
-    bool tryCleanupAfterError() noexcept;
+    bool cleanup(bool throw_if_error);
 
 private:
-    bool tryRemoveAllNodes(bool throw_if_error, WithRetries::Kind retries_kind);
+    bool cleanupImpl(bool throw_if_error, WithRetries::Kind retries_kind);
 
+    const bool is_restore;
     const String zookeeper_path;
 
     /// A reference to a field of the parent object which is either BackupCoordinationOnCluster or RestoreCoordinationOnCluster.
@@ -27,13 +27,8 @@ private:
 
     const LoggerPtr log;
 
-    struct CleanupResult
-    {
-        bool succeeded = false;
-        std::exception_ptr exception;
-    };
-    CleanupResult cleanup_result TSA_GUARDED_BY(mutex);
-
+    bool tried TSA_GUARDED_BY(mutex) = false;
+    bool succeeded TSA_GUARDED_BY(mutex) = false;
     std::mutex mutex;
 };
 

src/Backups/BackupCoordinationLocal.cpp

@@ -11,12 +11,11 @@ namespace DB
 {
 
 BackupCoordinationLocal::BackupCoordinationLocal(
-    const UUID & backup_uuid_,
     bool is_plain_backup_,
     bool allow_concurrent_backup_,
     BackupConcurrencyCounters & concurrency_counters_)
     : log(getLogger("BackupCoordinationLocal"))
-    , concurrency_check(backup_uuid_, /* is_restore = */ false, /* on_cluster = */ false, allow_concurrent_backup_, concurrency_counters_)
+    , concurrency_check(/* is_restore = */ false, /* on_cluster = */ false, /* zookeeper_path = */ "", allow_concurrent_backup_, concurrency_counters_)
     , file_infos(is_plain_backup_)
 {
 }

src/Backups/BackupCoordinationLocal.h

@@ -23,20 +23,19 @@ class BackupCoordinationLocal : public IBackupCoordination
 {
 public:
     explicit BackupCoordinationLocal(
-        const UUID & backup_uuid_,
         bool is_plain_backup_,
         bool allow_concurrent_backup_,
         BackupConcurrencyCounters & concurrency_counters_);
 
     ~BackupCoordinationLocal() override;
 
+    void setBackupQueryIsSentToOtherHosts() override {}
+    bool isBackupQuerySentToOtherHosts() const override { return false; }
     Strings setStage(const String &, const String &, bool) override { return {}; }
-    void setBackupQueryWasSentToOtherHosts() override {}
-    bool trySetError(std::exception_ptr) override { return true; }
-    void finish() override {}
-    bool tryFinishAfterError() noexcept override { return true; }
-    void waitForOtherHostsToFinish() override {}
-    bool tryWaitForOtherHostsToFinishAfterError() noexcept override { return true; }
+    bool setError(std::exception_ptr, bool) override { return true; }
+    bool waitOtherHostsFinish(bool) const override { return true; }
+    bool finish(bool) override { return true; }
+    bool cleanup(bool) override { return true; }
 
     void addReplicatedPartNames(const String & table_zk_path, const String & table_name_for_logs, const String & replica_name,
                                 const std::vector<PartNameAndChecksum> & part_names_and_checksums) override;

src/Backups/BackupCoordinationOnCluster.cpp

@@ -184,17 +184,21 @@ BackupCoordinationOnCluster::BackupCoordinationOnCluster(
     , plain_backup(is_plain_backup_)
     , log(getLogger("BackupCoordinationOnCluster"))
     , with_retries(log, get_zookeeper_, keeper_settings, process_list_element_, [root_zookeeper_path_](Coordination::ZooKeeperWithFaultInjection::Ptr zk) { zk->sync(root_zookeeper_path_); })
-    , concurrency_check(backup_uuid_, /* is_restore = */ false, /* on_cluster = */ true, allow_concurrent_backup_, concurrency_counters_)
-    , stage_sync(/* is_restore = */ false, fs::path{zookeeper_path} / "stage", current_host, all_hosts, allow_concurrent_backup_, with_retries, schedule_, process_list_element_, log)
-    , cleaner(zookeeper_path, with_retries, log)
+    , cleaner(/* is_restore = */ false, zookeeper_path, with_retries, log)
+    , stage_sync(/* is_restore = */ false, fs::path{zookeeper_path} / "stage", current_host, all_hosts, allow_concurrent_backup_, concurrency_counters_, with_retries, schedule_, process_list_element_, log)
+{
+    try
     {
         createRootNodes();
     }
-
-BackupCoordinationOnCluster::~BackupCoordinationOnCluster()
+    catch (...)
     {
-    tryFinishImpl();
+        stage_sync.setError(std::current_exception(), /* throw_if_error = */ false);
+        throw;
     }
+}
+
+BackupCoordinationOnCluster::~BackupCoordinationOnCluster() = default;
 
 void BackupCoordinationOnCluster::createRootNodes()
 {
@@ -217,69 +221,52 @@ void BackupCoordinationOnCluster::createRootNodes()
     });
 }
 
+void BackupCoordinationOnCluster::setBackupQueryIsSentToOtherHosts()
+{
+    stage_sync.setQueryIsSentToOtherHosts();
+}
+
+bool BackupCoordinationOnCluster::isBackupQuerySentToOtherHosts() const
+{
+    return stage_sync.isQuerySentToOtherHosts();
+}
+
 Strings BackupCoordinationOnCluster::setStage(const String & new_stage, const String & message, bool sync)
 {
     stage_sync.setStage(new_stage, message);
-
-    if (!sync)
+    if (sync)
+        return stage_sync.waitHostsReachStage(all_hosts_without_initiator, new_stage);
     return {};
-
-    return stage_sync.waitForHostsToReachStage(new_stage, all_hosts_without_initiator);
 }
 
-void BackupCoordinationOnCluster::setBackupQueryWasSentToOtherHosts()
+bool BackupCoordinationOnCluster::setError(std::exception_ptr exception, bool throw_if_error)
 {
-    backup_query_was_sent_to_other_hosts = true;
+    return stage_sync.setError(exception, throw_if_error);
 }
 
-bool BackupCoordinationOnCluster::trySetError(std::exception_ptr exception)
+bool BackupCoordinationOnCluster::waitOtherHostsFinish(bool throw_if_error) const
 {
-    return stage_sync.trySetError(exception);
+    return stage_sync.waitOtherHostsFinish(throw_if_error);
 }
 
-void BackupCoordinationOnCluster::finish()
+bool BackupCoordinationOnCluster::finish(bool throw_if_error)
 {
-    bool other_hosts_also_finished = false;
-    stage_sync.finish(other_hosts_also_finished);
-
-    if ((current_host == kInitiator) && (other_hosts_also_finished || !backup_query_was_sent_to_other_hosts))
-        cleaner.cleanup();
+    return stage_sync.finish(throw_if_error);
 }
 
-bool BackupCoordinationOnCluster::tryFinishAfterError() noexcept
+bool BackupCoordinationOnCluster::cleanup(bool throw_if_error)
 {
-    return tryFinishImpl();
-}
-
-bool BackupCoordinationOnCluster::tryFinishImpl() noexcept
+    /// All the hosts must finish before we remove the coordination nodes.
+    bool expect_other_hosts_finished = stage_sync.isQuerySentToOtherHosts() || !stage_sync.isErrorSet();
+    bool all_hosts_finished = stage_sync.finished() && (stage_sync.otherHostsFinished() || !expect_other_hosts_finished);
+    if (!all_hosts_finished)
     {
-    bool other_hosts_also_finished = false;
-    if (!stage_sync.tryFinishAfterError(other_hosts_also_finished))
-        return false;
-
-    if ((current_host == kInitiator) && (other_hosts_also_finished || !backup_query_was_sent_to_other_hosts))
-    {
-        if (!cleaner.tryCleanupAfterError())
+        auto unfinished_hosts = expect_other_hosts_finished ? stage_sync.getUnfinishedHosts() : Strings{current_host};
+        LOG_INFO(log, "Skipping removing nodes from ZooKeeper because hosts {} didn't finish",
+                 BackupCoordinationStageSync::getHostsDesc(unfinished_hosts));
         return false;
     }
-
-    return true;
-}
-
-void BackupCoordinationOnCluster::waitForOtherHostsToFinish()
-{
-    if ((current_host != kInitiator) || !backup_query_was_sent_to_other_hosts)
-        return;
-    stage_sync.waitForOtherHostsToFinish();
-}
-
-bool BackupCoordinationOnCluster::tryWaitForOtherHostsToFinishAfterError() noexcept
-{
-    if (current_host != kInitiator)
-        return false;
-    if (!backup_query_was_sent_to_other_hosts)
-        return true;
-    return stage_sync.tryWaitForOtherHostsToFinishAfterError();
+    return cleaner.cleanup(throw_if_error);
 }
 
 ZooKeeperRetriesInfo BackupCoordinationOnCluster::getOnClusterInitializationKeeperRetriesInfo() const

src/Backups/BackupCoordinationOnCluster.h

@@ -1,7 +1,6 @@
 #pragma once
 
 #include <Backups/IBackupCoordination.h>
-#include <Backups/BackupConcurrencyCheck.h>
 #include <Backups/BackupCoordinationCleaner.h>
 #include <Backups/BackupCoordinationFileInfos.h>
 #include <Backups/BackupCoordinationReplicatedAccess.h>
@@ -20,7 +19,7 @@ class BackupCoordinationOnCluster : public IBackupCoordination
 {
 public:
     /// Empty string as the current host is used to mark the initiator of a BACKUP ON CLUSTER query.
-    static const constexpr std::string_view kInitiator;
+    static const constexpr std::string_view kInitiator = BackupCoordinationStageSync::kInitiator;
 
     BackupCoordinationOnCluster(
         const UUID & backup_uuid_,
@@ -37,13 +36,13 @@ public:
 
     ~BackupCoordinationOnCluster() override;
 
+    void setBackupQueryIsSentToOtherHosts() override;
+    bool isBackupQuerySentToOtherHosts() const override;
     Strings setStage(const String & new_stage, const String & message, bool sync) override;
-    void setBackupQueryWasSentToOtherHosts() override;
-    bool trySetError(std::exception_ptr exception) override;
-    void finish() override;
-    bool tryFinishAfterError() noexcept override;
-    void waitForOtherHostsToFinish() override;
-    bool tryWaitForOtherHostsToFinishAfterError() noexcept override;
+    bool setError(std::exception_ptr exception, bool throw_if_error) override;
+    bool waitOtherHostsFinish(bool throw_if_error) const override;
+    bool finish(bool throw_if_error) override;
+    bool cleanup(bool throw_if_error) override;
 
     void addReplicatedPartNames(
         const String & table_zk_path,
@@ -110,11 +109,10 @@ private:
     const bool plain_backup;
     LoggerPtr const log;
 
+    /// The order is important: `stage_sync` must be initialized after `with_retries` and `cleaner`.
     const WithRetries with_retries;
-    BackupConcurrencyCheck concurrency_check;
-    BackupCoordinationStageSync stage_sync;
     BackupCoordinationCleaner cleaner;
-    std::atomic<bool> backup_query_was_sent_to_other_hosts = false;
+    BackupCoordinationStageSync stage_sync;
 
     mutable std::optional<BackupCoordinationReplicatedTables> replicated_tables TSA_GUARDED_BY(replicated_tables_mutex);
     mutable std::optional<BackupCoordinationReplicatedAccess> replicated_access TSA_GUARDED_BY(replicated_access_mutex);

src/Backups/BackupCoordinationStageSync.h

@@ -1,7 +1,9 @@
 #pragma once
 
+#include <Backups/BackupConcurrencyCheck.h>
 #include <Backups/WithRetries.h>
 
+
 namespace DB
 {
 
@@ -9,12 +11,16 @@ namespace DB
 class BackupCoordinationStageSync
 {
 public:
+    /// Empty string as the current host is used to mark the initiator of a BACKUP ON CLUSTER or RESTORE ON CLUSTER query.
+    static const constexpr std::string_view kInitiator;
+
     BackupCoordinationStageSync(
         bool is_restore_,                    /// true if this is a RESTORE ON CLUSTER command, false if this is a BACKUP ON CLUSTER command
         const String & zookeeper_path_,      /// path to the "stage" folder in ZooKeeper
         const String & current_host_,        /// the current host, or an empty string if it's the initiator of the BACKUP/RESTORE ON CLUSTER command
         const Strings & all_hosts_,          /// all the hosts (including the initiator and the current host) performing the BACKUP/RESTORE ON CLUSTER command
         bool allow_concurrency_,             /// whether it's allowed to have concurrent backups or restores.
+        BackupConcurrencyCounters & concurrency_counters_,
         const WithRetries & with_retries_,
         ThreadPoolCallbackRunnerUnsafe<void> schedule_,
         QueryStatusPtr process_list_element_,
@@ -22,30 +28,37 @@ public:
 
     ~BackupCoordinationStageSync();
 
+    /// Sets that the BACKUP or RESTORE query was sent to other hosts.
+    void setQueryIsSentToOtherHosts();
+    bool isQuerySentToOtherHosts() const;
+
     /// Sets the stage of the current host and signal other hosts if there were other hosts waiting for that.
     void setStage(const String & stage, const String & stage_result = {});
 
-    /// Waits until all the specified hosts come to the specified stage.
-    /// The function returns the results which specified hosts set when they came to the required stage.
-    /// If it doesn't happen before the timeout then the function will stop waiting and throw an exception.
-    Strings waitForHostsToReachStage(const String & stage_to_wait, const Strings & hosts, std::optional<std::chrono::milliseconds> timeout = {}) const;
-
-    /// Waits until all the other hosts finish their work.
-    /// Stops waiting and throws an exception if another host encounters an error or if some host gets cancelled.
-    void waitForOtherHostsToFinish() const;
-
-    /// Lets other host know that the current host has finished its work.
-    void finish(bool & other_hosts_also_finished);
+    /// Waits until specified hosts come to the specified stage.
+    /// The function returns the results which the specified hosts set when they came to the required stage.
+    Strings waitHostsReachStage(const Strings & hosts, const String & stage_to_wait) const;
 
     /// Lets other hosts know that the current host has encountered an error.
-    bool trySetError(std::exception_ptr exception) noexcept;
+    /// The function returns true if it successfully created the error node or if the error node was found already exist.
+    bool setError(std::exception_ptr exception, bool throw_if_error);
+    bool isErrorSet() const;
 
-    /// Waits until all the other hosts finish their work (as a part of error-handling process).
-    /// Doesn't stops waiting if some host encounters an error or gets cancelled.
-    bool tryWaitForOtherHostsToFinishAfterError() const noexcept;
+    /// Waits until the hosts other than the current host finish their work. Must be called before finish().
+    /// Stops waiting and throws an exception if another host encounters an error or if some host gets cancelled.
+    bool waitOtherHostsFinish(bool throw_if_error) const;
+    bool otherHostsFinished() const;
 
-    /// Lets other host know that the current host has finished its work (as a part of error-handling process).
-    bool tryFinishAfterError(bool & other_hosts_also_finished) noexcept;
+    /// Lets other hosts know that the current host has finished its work.
+    bool finish(bool throw_if_error);
+    bool finished() const;
+
+    /// Returns true if all the hosts have finished.
+    bool allHostsFinished() const { return finished() && otherHostsFinished(); }
+
+    /// Returns a list of the hosts which haven't finished yet.
+    Strings getUnfinishedHosts() const;
+    Strings getUnfinishedOtherHosts() const;
 
     /// Returns a printable name of a specific host. For empty host the function returns "initiator".
     static String getHostDesc(const String & host);
@@ -78,14 +91,17 @@ private:
 
     /// Reads the current state from ZooKeeper without throwing exceptions.
     void readCurrentState(Coordination::ZooKeeperWithFaultInjection::Ptr zookeeper);
+
+    /// Creates a stage node to let other hosts know we've reached the specified stage.
+    void createStageNode(const String & stage, const String & stage_result, Coordination::ZooKeeperWithFaultInjection::Ptr zookeeper);
     String getStageNodePath(const String & stage) const;
 
     /// Lets other hosts know that the current host has encountered an error.
-    bool trySetError(const Exception & exception);
-    void setError(const Exception & exception);
+    bool setError(const Exception & exception, bool throw_if_error);
+    void createErrorNode(const Exception & exception, Coordination::ZooKeeperWithFaultInjection::Ptr zookeeper);
 
     /// Deserializes an error stored in the error node.
-    static std::pair<std::exception_ptr, String> parseErrorNode(const String & error_node_contents);
+    std::pair<std::exception_ptr, String> parseErrorNode(const String & error_node_contents) const;
 
     /// Reset the `connected` flag for each host.
     void resetConnectedFlag();
@@ -102,19 +118,27 @@ private:
     void cancelQueryIfDisconnectedTooLong();
 
     /// Used by waitForHostsToReachStage() to check if everything is ready to return.
-    bool checkIfHostsReachStage(const Strings & hosts, const String & stage_to_wait, bool time_is_out, std::optional<std::chrono::milliseconds> timeout, Strings & results) const TSA_REQUIRES(mutex);
+    bool checkIfHostsReachStage(const Strings & hosts, const String & stage_to_wait, Strings & results) const TSA_REQUIRES(mutex);
 
     /// Creates the 'finish' node.
-    bool tryFinishImpl();
-    bool tryFinishImpl(bool & other_hosts_also_finished, bool throw_if_error, WithRetries::Kind retries_kind);
-    void createFinishNodeAndRemoveAliveNode(Coordination::ZooKeeperWithFaultInjection::Ptr zookeeper);
+    bool finishImpl(bool throw_if_error, WithRetries::Kind retries_kind);
+    void createFinishNodeAndRemoveAliveNode(Coordination::ZooKeeperWithFaultInjection::Ptr zookeeper, bool throw_if_error);
 
     /// Returns the version used by the initiator.
     int getInitiatorVersion() const;
 
     /// Waits until all the other hosts finish their work.
-    bool tryWaitForOtherHostsToFinishImpl(const String & reason, bool throw_if_error, std::optional<std::chrono::seconds> timeout) const;
-    bool checkIfOtherHostsFinish(const String & reason, bool throw_if_error, bool time_is_out, std::optional<std::chrono::milliseconds> timeout) const TSA_REQUIRES(mutex);
+    bool waitOtherHostsFinishImpl(const String & reason, std::optional<std::chrono::seconds> timeout, bool throw_if_error) const;
+    bool checkIfOtherHostsFinish(const String & reason, std::optional<std::chrono::milliseconds> timeout, bool time_is_out, bool & result, bool throw_if_error) const TSA_REQUIRES(mutex);
+
+    /// Returns true if all the hosts have finished.
+    bool allHostsFinishedNoLock() const TSA_REQUIRES(mutex);
+    bool finishedNoLock() const TSA_REQUIRES(mutex);
+    bool otherHostsFinishedNoLock() const TSA_REQUIRES(mutex);
+
+    /// Returns a list of the hosts which haven't finished yet.
+    Strings getUnfinishedHostsNoLock() const TSA_REQUIRES(mutex);
+    Strings getUnfinishedOtherHostsNoLock() const TSA_REQUIRES(mutex);
 
     const bool is_restore;
     const String operation_name;
@@ -138,15 +162,16 @@ private:
     /// Paths in ZooKeeper.
     const std::filesystem::path zookeeper_path;
     const String root_zookeeper_path;
-    const String operation_node_path;
+    const String operation_zookeeper_path;
     const String operation_node_name;
-    const String stage_node_path;
     const String start_node_path;
     const String finish_node_path;
     const String num_hosts_node_path;
+    const String error_node_path;
     const String alive_node_path;
     const String alive_tracker_node_path;
-    const String error_node_path;
+
+    std::optional<BackupConcurrencyCheck> concurrency_check;
 
     std::shared_ptr<Poco::Event> zk_nodes_changed;
 
@@ -176,25 +201,21 @@ private:
     {
         std::map<String /* host */, HostInfo> hosts; /// std::map because we need to compare states
         std::optional<String> host_with_error;
-        bool cancelled = false;
 
         bool operator ==(const State & other) const;
         bool operator !=(const State & other) const;
+        void merge(const State & other);
     };
 
     State state TSA_GUARDED_BY(mutex);
     mutable std::condition_variable state_changed;
 
     std::future<void> watching_thread_future;
-    std::atomic<bool> should_stop_watching_thread = false;
+    bool should_stop_watching_thread TSA_GUARDED_BY(mutex) = false;
 
-    struct FinishResult
-    {
-        bool succeeded = false;
-        std::exception_ptr exception;
-        bool other_hosts_also_finished = false;
-    };
-    FinishResult finish_result TSA_GUARDED_BY(mutex);
+    bool query_is_sent_to_other_hosts TSA_GUARDED_BY(mutex) = false;
+    bool tried_to_finish TSA_GUARDED_BY(mutex) = false;
+    bool tried_to_set_error TSA_GUARDED_BY(mutex) = false;
 
     mutable std::mutex mutex;
 };

src/Backups/BackupsWorker.cpp

@@ -330,6 +330,7 @@ std::pair<OperationID, BackupStatus> BackupsWorker::start(const ASTPtr & backup_
 struct BackupsWorker::BackupStarter
 {
     BackupsWorker & backups_worker;
+    LoggerPtr log;
     std::shared_ptr<ASTBackupQuery> backup_query;
     ContextPtr query_context; /// We have to keep `query_context` until the end of the operation because a pointer to it is stored inside the ThreadGroup we're using.
     ContextMutablePtr backup_context;
@@ -346,6 +347,7 @@ struct BackupsWorker::BackupStarter
 
     BackupStarter(BackupsWorker & backups_worker_, const ASTPtr & query_, const ContextPtr & context_)
         : backups_worker(backups_worker_)
+        , log(backups_worker.log)
         , backup_query(std::static_pointer_cast<ASTBackupQuery>(query_->clone()))
         , query_context(context_)
         , backup_context(Context::createCopy(query_context))
@@ -400,9 +402,20 @@ struct BackupsWorker::BackupStarter
         chassert(!backup);
         backup = backups_worker.openBackupForWriting(backup_info, backup_settings, backup_coordination, backup_context);
 
-        backups_worker.doBackup(
-            backup, backup_query, backup_id, backup_name_for_logging, backup_settings, backup_coordination, backup_context,
+        backups_worker.doBackup(backup, backup_query, backup_id, backup_settings, backup_coordination, backup_context,
                                 on_cluster, cluster);
+
+        backup_coordination->finish(/* throw_if_error = */ true);
+        backup.reset();
+
+        /// The backup coordination is not needed anymore.
+        if (!is_internal_backup)
+            backup_coordination->cleanup(/* throw_if_error = */ true);
+        backup_coordination.reset();
+
+        /// NOTE: setStatus is called after setNumFilesAndSize in order to have actual information in a backup log record
+        LOG_INFO(log, "{} {} was created successfully", (is_internal_backup ? "Internal backup" : "Backup"), backup_name_for_logging);
+        backups_worker.setStatus(backup_id, BackupStatus::BACKUP_CREATED);
     }
 
     void onException()
@@ -417,15 +430,28 @@ struct BackupsWorker::BackupStarter
         if (backup && !backup->setIsCorrupted())
             should_remove_files_in_backup = false;
 
-        if (backup_coordination && backup_coordination->trySetError(std::current_exception()))
-        {
-            bool other_hosts_finished = backup_coordination->tryWaitForOtherHostsToFinishAfterError();
+        bool all_hosts_finished = false;
 
-            if (should_remove_files_in_backup && other_hosts_finished)
+        if (backup_coordination && backup_coordination->setError(std::current_exception(), /* throw_if_error = */ false))
+        {
+            bool other_hosts_finished = !is_internal_backup
+                && (!backup_coordination->isBackupQuerySentToOtherHosts() || backup_coordination->waitOtherHostsFinish(/* throw_if_error = */ false));
+
+            all_hosts_finished = backup_coordination->finish(/* throw_if_error = */ false) && other_hosts_finished;
+        }
+
+        if (!all_hosts_finished)
+            should_remove_files_in_backup = false;
+
+        if (backup && should_remove_files_in_backup)
             backup->tryRemoveAllFiles();
 
-            backup_coordination->tryFinishAfterError();
-        }
+        backup.reset();
+
+        if (backup_coordination && all_hosts_finished)
+            backup_coordination->cleanup(/* throw_if_error = */ false);
+
+        backup_coordination.reset();
 
         backups_worker.setStatusSafe(backup_id, getBackupStatusFromCurrentException());
     }
@@ -498,7 +524,6 @@ void BackupsWorker::doBackup(
     BackupMutablePtr backup,
     const std::shared_ptr<ASTBackupQuery> & backup_query,
     const OperationID & backup_id,
-    const String & backup_name_for_logging,
     const BackupSettings & backup_settings,
     std::shared_ptr<IBackupCoordination> backup_coordination,
     ContextMutablePtr context,
@@ -522,10 +547,10 @@ void BackupsWorker::doBackup(
         backup_settings.copySettingsToQuery(*backup_query);
         sendQueryToOtherHosts(*backup_query, cluster, backup_settings.shard_num, backup_settings.replica_num,
                               context, required_access, backup_coordination->getOnClusterInitializationKeeperRetriesInfo());
-        backup_coordination->setBackupQueryWasSentToOtherHosts();
+        backup_coordination->setBackupQueryIsSentToOtherHosts();
 
         /// Wait until all the hosts have written their backup entries.
-        backup_coordination->waitForOtherHostsToFinish();
+        backup_coordination->waitOtherHostsFinish(/* throw_if_error = */ true);
     }
     else
     {
@@ -570,18 +595,8 @@ void BackupsWorker::doBackup(
         compressed_size = backup->getCompressedSize();
     }
 
-    /// Close the backup.
-    backup.reset();
-
-    /// The backup coordination is not needed anymore.
-    backup_coordination->finish();
-
     /// NOTE: we need to update metadata again after backup->finalizeWriting(), because backup metadata is written there.
     setNumFilesAndSize(backup_id, num_files, total_size, num_entries, uncompressed_size, compressed_size, 0, 0);
-
-    /// NOTE: setStatus is called after setNumFilesAndSize in order to have actual information in a backup log record
-    LOG_INFO(log, "{} {} was created successfully", (is_internal_backup ? "Internal backup" : "Backup"), backup_name_for_logging);
-    setStatus(backup_id, BackupStatus::BACKUP_CREATED);
 }
 
 
@@ -688,6 +703,7 @@ void BackupsWorker::writeBackupEntries(
 struct BackupsWorker::RestoreStarter
 {
     BackupsWorker & backups_worker;
+    LoggerPtr log;
     std::shared_ptr<ASTBackupQuery> restore_query;
     ContextPtr query_context; /// We have to keep `query_context` until the end of the operation because a pointer to it is stored inside the ThreadGroup we're using.
     ContextMutablePtr restore_context;
@@ -703,6 +719,7 @@ struct BackupsWorker::RestoreStarter
 
     RestoreStarter(BackupsWorker & backups_worker_, const ASTPtr & query_, const ContextPtr & context_)
         : backups_worker(backups_worker_)
+        , log(backups_worker.log)
         , restore_query(std::static_pointer_cast<ASTBackupQuery>(query_->clone()))
         , query_context(context_)
         , restore_context(Context::createCopy(query_context))
@@ -754,16 +771,17 @@ struct BackupsWorker::RestoreStarter
         }
         restore_coordination = backups_worker.makeRestoreCoordination(on_cluster, restore_settings, restore_context);
 
-        backups_worker.doRestore(
-            restore_query,
-            restore_id,
-            backup_name_for_logging,
-            backup_info,
-            restore_settings,
-            restore_coordination,
-            restore_context,
-            on_cluster,
-            cluster);
+        backups_worker.doRestore(restore_query, restore_id, backup_info, restore_settings, restore_coordination, restore_context,
+                                 on_cluster, cluster);
+
+        /// The restore coordination is not needed anymore.
+        restore_coordination->finish(/* throw_if_error = */ true);
+        if (!is_internal_restore)
+            restore_coordination->cleanup(/* throw_if_error = */ true);
+        restore_coordination.reset();
+
+        LOG_INFO(log, "Restored from {} {} successfully", (is_internal_restore ? "internal backup" : "backup"), backup_name_for_logging);
+        backups_worker.setStatus(restore_id, BackupStatus::RESTORED);
     }
 
     void onException()
@@ -771,12 +789,16 @@ struct BackupsWorker::RestoreStarter
         /// Something bad happened, some data were not restored.
         tryLogCurrentException(backups_worker.log, fmt::format("Failed to restore from {} {}", (is_internal_restore ? "internal backup" : "backup"), backup_name_for_logging));
 
-        if (restore_coordination && restore_coordination->trySetError(std::current_exception()))
+        if (restore_coordination && restore_coordination->setError(std::current_exception(), /* throw_if_error = */ false))
         {
-            restore_coordination->tryWaitForOtherHostsToFinishAfterError();
-            restore_coordination->tryFinishAfterError();
+            bool other_hosts_finished = !is_internal_restore
+                && (!restore_coordination->isRestoreQuerySentToOtherHosts() || restore_coordination->waitOtherHostsFinish(/* throw_if_error = */ false));
+            if (restore_coordination->finish(/* throw_if_error = */ false) && other_hosts_finished)
+                restore_coordination->cleanup(/* throw_if_error = */ false);
         }
 
+        restore_coordination.reset();
+
         backups_worker.setStatusSafe(restore_id, getRestoreStatusFromCurrentException());
     }
 };
@@ -839,7 +861,6 @@ BackupPtr BackupsWorker::openBackupForReading(const BackupInfo & backup_info, co
 void BackupsWorker::doRestore(
     const std::shared_ptr<ASTBackupQuery> & restore_query,
     const OperationID & restore_id,
-    const String & backup_name_for_logging,
     const BackupInfo & backup_info,
     RestoreSettings restore_settings,
     std::shared_ptr<IRestoreCoordination> restore_coordination,
@@ -883,10 +904,10 @@ void BackupsWorker::doRestore(
         restore_settings.copySettingsToQuery(*restore_query);
         sendQueryToOtherHosts(*restore_query, cluster, restore_settings.shard_num, restore_settings.replica_num,
                               context, {}, restore_coordination->getOnClusterInitializationKeeperRetriesInfo());
-        restore_coordination->setRestoreQueryWasSentToOtherHosts();
+        restore_coordination->setRestoreQueryIsSentToOtherHosts();
 
         /// Wait until all the hosts have done with their restoring work.
-        restore_coordination->waitForOtherHostsToFinish();
+        restore_coordination->waitOtherHostsFinish(/* throw_if_error = */ true);
     }
     else
     {
@@ -906,12 +927,6 @@ void BackupsWorker::doRestore(
                                     backup, context, getThreadPool(ThreadPoolId::RESTORE), after_task_callback};
         restorer.run(RestorerFromBackup::RESTORE);
     }
-
-    /// The restore coordination is not needed anymore.
-    restore_coordination->finish();
-
-    LOG_INFO(log, "Restored from {} {} successfully", (is_internal_restore ? "internal backup" : "backup"), backup_name_for_logging);
-    setStatus(restore_id, BackupStatus::RESTORED);
 }
 
 
@@ -944,7 +959,7 @@ BackupsWorker::makeBackupCoordination(bool on_cluster, const BackupSettings & ba
     if (!on_cluster)
     {
         return std::make_shared<BackupCoordinationLocal>(
-            *backup_settings.backup_uuid, !backup_settings.deduplicate_files, allow_concurrent_backups, *concurrency_counters);
+            !backup_settings.deduplicate_files, allow_concurrent_backups, *concurrency_counters);
     }
 
     bool is_internal_backup = backup_settings.internal;
@@ -982,8 +997,7 @@ BackupsWorker::makeRestoreCoordination(bool on_cluster, const RestoreSettings &
 {
     if (!on_cluster)
     {
-        return std::make_shared<RestoreCoordinationLocal>(
-            *restore_settings.restore_uuid, allow_concurrent_restores, *concurrency_counters);
+        return std::make_shared<RestoreCoordinationLocal>(allow_concurrent_restores, *concurrency_counters);
     }
 
     bool is_internal_restore = restore_settings.internal;

src/Backups/BackupsWorker.h

@@ -81,7 +81,6 @@ private:
         BackupMutablePtr backup,
         const std::shared_ptr<ASTBackupQuery> & backup_query,
         const BackupOperationID & backup_id,
-        const String & backup_name_for_logging,
         const BackupSettings & backup_settings,
         std::shared_ptr<IBackupCoordination> backup_coordination,
         ContextMutablePtr context,
@@ -102,7 +101,6 @@ private:
     void doRestore(
         const std::shared_ptr<ASTBackupQuery> & restore_query,
         const BackupOperationID & restore_id,
-        const String & backup_name_for_logging,
         const BackupInfo & backup_info,
         RestoreSettings restore_settings,
         std::shared_ptr<IRestoreCoordination> restore_coordination,

src/Backups/IBackupCoordination.h

@@ -20,29 +20,27 @@ class IBackupCoordination
 public:
     virtual ~IBackupCoordination() = default;
 
+    /// Sets that the backup query was sent to other hosts.
+    /// Function waitOtherHostsFinish() will check that to find out if it should really wait or not.
+    virtual void setBackupQueryIsSentToOtherHosts() = 0;
+    virtual bool isBackupQuerySentToOtherHosts() const = 0;
+
     /// Sets the current stage and waits for other hosts to come to this stage too.
     virtual Strings setStage(const String & new_stage, const String & message, bool sync) = 0;
 
-    /// Sets that the backup query was sent to other hosts.
-    /// Function waitForOtherHostsToFinish() will check that to find out if it should really wait or not.
-    virtual void setBackupQueryWasSentToOtherHosts() = 0;
-
     /// Lets other hosts know that the current host has encountered an error.
-    virtual bool trySetError(std::exception_ptr exception) = 0;
-
-    /// Lets other hosts know that the current host has finished its work.
-    virtual void finish() = 0;
-
-    /// Lets other hosts know that the current host has finished its work (as a part of error-handling process).
-    virtual bool tryFinishAfterError() noexcept = 0;
+    /// Returns true if the information is successfully passed so other hosts can read it.
+    virtual bool setError(std::exception_ptr exception, bool throw_if_error) = 0;
 
     /// Waits until all the other hosts finish their work.
     /// Stops waiting and throws an exception if another host encounters an error or if some host gets cancelled.
-    virtual void waitForOtherHostsToFinish() = 0;
+    virtual bool waitOtherHostsFinish(bool throw_if_error) const = 0;
 
-    /// Waits until all the other hosts finish their work (as a part of error-handling process).
-    /// Doesn't stops waiting if some host encounters an error or gets cancelled.
-    virtual bool tryWaitForOtherHostsToFinishAfterError() noexcept = 0;
+    /// Lets other hosts know that the current host has finished its work.
+    virtual bool finish(bool throw_if_error) = 0;
+
+    /// Removes temporary nodes in ZooKeeper.
+    virtual bool cleanup(bool throw_if_error) = 0;
 
     struct PartNameAndChecksum
     {

src/Backups/IRestoreCoordination.h

@@ -18,29 +18,27 @@ class IRestoreCoordination
 public:
     virtual ~IRestoreCoordination() = default;
 
+    /// Sets that the restore query was sent to other hosts.
+    /// Function waitOtherHostsFinish() will check that to find out if it should really wait or not.
+    virtual void setRestoreQueryIsSentToOtherHosts() = 0;
+    virtual bool isRestoreQuerySentToOtherHosts() const = 0;
+
     /// Sets the current stage and waits for other hosts to come to this stage too.
     virtual Strings setStage(const String & new_stage, const String & message, bool sync) = 0;
 
-    /// Sets that the restore query was sent to other hosts.
-    /// Function waitForOtherHostsToFinish() will check that to find out if it should really wait or not.
-    virtual void setRestoreQueryWasSentToOtherHosts() = 0;
-
     /// Lets other hosts know that the current host has encountered an error.
-    virtual bool trySetError(std::exception_ptr exception) = 0;
-
-    /// Lets other hosts know that the current host has finished its work.
-    virtual void finish() = 0;
-
-    /// Lets other hosts know that the current host has finished its work (as a part of error-handling process).
-    virtual bool tryFinishAfterError() noexcept = 0;
+    /// Returns true if the information is successfully passed so other hosts can read it.
+    virtual bool setError(std::exception_ptr exception, bool throw_if_error) = 0;
 
     /// Waits until all the other hosts finish their work.
     /// Stops waiting and throws an exception if another host encounters an error or if some host gets cancelled.
-    virtual void waitForOtherHostsToFinish() = 0;
+    virtual bool waitOtherHostsFinish(bool throw_if_error) const = 0;
 
-    /// Waits until all the other hosts finish their work (as a part of error-handling process).
-    /// Doesn't stops waiting if some host encounters an error or gets cancelled.
-    virtual bool tryWaitForOtherHostsToFinishAfterError() noexcept = 0;
+    /// Lets other hosts know that the current host has finished its work.
+    virtual bool finish(bool throw_if_error) = 0;
+
+    /// Removes temporary nodes in ZooKeeper.
+    virtual bool cleanup(bool throw_if_error) = 0;
 
     /// Starts creating a table in a replicated database. Returns false if there is another host which is already creating this table.
     virtual bool acquireCreatingTableInReplicatedDatabase(const String & database_zk_path, const String & table_name) = 0;

src/Backups/RestoreCoordinationLocal.cpp

@@ -10,9 +10,9 @@ namespace DB
 {
 
 RestoreCoordinationLocal::RestoreCoordinationLocal(
-    const UUID & restore_uuid, bool allow_concurrent_restore_, BackupConcurrencyCounters & concurrency_counters_)
+    bool allow_concurrent_restore_, BackupConcurrencyCounters & concurrency_counters_)
     : log(getLogger("RestoreCoordinationLocal"))
-    , concurrency_check(restore_uuid, /* is_restore = */ true, /* on_cluster = */ false, allow_concurrent_restore_, concurrency_counters_)
+    , concurrency_check(/* is_restore = */ true, /* on_cluster = */ false, /* zookeeper_path = */ "", allow_concurrent_restore_, concurrency_counters_)
 {
 }
 

src/Backups/RestoreCoordinationLocal.h

@@ -17,16 +17,16 @@ class ASTCreateQuery;
 class RestoreCoordinationLocal : public IRestoreCoordination
 {
 public:
-    RestoreCoordinationLocal(const UUID & restore_uuid_, bool allow_concurrent_restore_, BackupConcurrencyCounters & concurrency_counters_);
+    RestoreCoordinationLocal(bool allow_concurrent_restore_, BackupConcurrencyCounters & concurrency_counters_);
     ~RestoreCoordinationLocal() override;
 
+    void setRestoreQueryIsSentToOtherHosts() override {}
+    bool isRestoreQuerySentToOtherHosts() const override { return false; }
     Strings setStage(const String &, const String &, bool) override { return {}; }
-    void setRestoreQueryWasSentToOtherHosts() override {}
-    bool trySetError(std::exception_ptr) override { return true; }
-    void finish() override {}
-    bool tryFinishAfterError() noexcept override { return true; }
-    void waitForOtherHostsToFinish() override {}
-    bool tryWaitForOtherHostsToFinishAfterError() noexcept override { return true; }
+    bool setError(std::exception_ptr, bool) override { return true; }
+    bool waitOtherHostsFinish(bool) const override { return true; }
+    bool finish(bool) override { return true; }
+    bool cleanup(bool) override { return true; }
 
     /// Starts creating a table in a replicated database. Returns false if there is another host which is already creating this table.
     bool acquireCreatingTableInReplicatedDatabase(const String & database_zk_path, const String & table_name) override;

src/Backups/RestoreCoordinationOnCluster.cpp

@@ -35,17 +35,21 @@ RestoreCoordinationOnCluster::RestoreCoordinationOnCluster(
     , current_host_index(BackupCoordinationOnCluster::findCurrentHostIndex(current_host, all_hosts))
     , log(getLogger("RestoreCoordinationOnCluster"))
     , with_retries(log, get_zookeeper_, keeper_settings, process_list_element_, [root_zookeeper_path_](Coordination::ZooKeeperWithFaultInjection::Ptr zk) { zk->sync(root_zookeeper_path_); })
-    , concurrency_check(restore_uuid_, /* is_restore = */ true, /* on_cluster = */ true, allow_concurrent_restore_, concurrency_counters_)
-    , stage_sync(/* is_restore = */ true, fs::path{zookeeper_path} / "stage", current_host, all_hosts, allow_concurrent_restore_, with_retries, schedule_, process_list_element_, log)
-    , cleaner(zookeeper_path, with_retries, log)
+    , cleaner(/* is_restore = */ true, zookeeper_path, with_retries, log)
+    , stage_sync(/* is_restore = */ true, fs::path{zookeeper_path} / "stage", current_host, all_hosts, allow_concurrent_restore_, concurrency_counters_, with_retries, schedule_, process_list_element_, log)
+{
+    try
     {
         createRootNodes();
     }
-
-RestoreCoordinationOnCluster::~RestoreCoordinationOnCluster()
+    catch (...)
     {
-    tryFinishImpl();
+        stage_sync.setError(std::current_exception(), /* throw_if_error = */ false);
+        throw;
     }
+}
+
+RestoreCoordinationOnCluster::~RestoreCoordinationOnCluster() = default;
 
 void RestoreCoordinationOnCluster::createRootNodes()
 {
@@ -66,69 +70,52 @@ void RestoreCoordinationOnCluster::createRootNodes()
         });
 }
 
+void RestoreCoordinationOnCluster::setRestoreQueryIsSentToOtherHosts()
+{
+    stage_sync.setQueryIsSentToOtherHosts();
+}
+
+bool RestoreCoordinationOnCluster::isRestoreQuerySentToOtherHosts() const
+{
+    return stage_sync.isQuerySentToOtherHosts();
+}
+
 Strings RestoreCoordinationOnCluster::setStage(const String & new_stage, const String & message, bool sync)
 {
     stage_sync.setStage(new_stage, message);
-
-    if (!sync)
+    if (sync)
+        return stage_sync.waitHostsReachStage(all_hosts_without_initiator, new_stage);
     return {};
-
-    return stage_sync.waitForHostsToReachStage(new_stage, all_hosts_without_initiator);
 }
 
-void RestoreCoordinationOnCluster::setRestoreQueryWasSentToOtherHosts()
+bool RestoreCoordinationOnCluster::setError(std::exception_ptr exception, bool throw_if_error)
 {
-    restore_query_was_sent_to_other_hosts = true;
+    return stage_sync.setError(exception, throw_if_error);
 }
 
-bool RestoreCoordinationOnCluster::trySetError(std::exception_ptr exception)
+bool RestoreCoordinationOnCluster::waitOtherHostsFinish(bool throw_if_error) const
 {
-    return stage_sync.trySetError(exception);
+    return stage_sync.waitOtherHostsFinish(throw_if_error);
 }
 
-void RestoreCoordinationOnCluster::finish()
+bool RestoreCoordinationOnCluster::finish(bool throw_if_error)
 {
-    bool other_hosts_also_finished = false;
-    stage_sync.finish(other_hosts_also_finished);
-
-    if ((current_host == kInitiator) && (other_hosts_also_finished || !restore_query_was_sent_to_other_hosts))
-        cleaner.cleanup();
+    return stage_sync.finish(throw_if_error);
 }
 
-bool RestoreCoordinationOnCluster::tryFinishAfterError() noexcept
+bool RestoreCoordinationOnCluster::cleanup(bool throw_if_error)
 {
-    return tryFinishImpl();
-}
-
-bool RestoreCoordinationOnCluster::tryFinishImpl() noexcept
+    /// All the hosts must finish before we remove the coordination nodes.
+    bool expect_other_hosts_finished = stage_sync.isQuerySentToOtherHosts() || !stage_sync.isErrorSet();
+    bool all_hosts_finished = stage_sync.finished() && (stage_sync.otherHostsFinished() || !expect_other_hosts_finished);
+    if (!all_hosts_finished)
     {
-    bool other_hosts_also_finished = false;
-    if (!stage_sync.tryFinishAfterError(other_hosts_also_finished))
-        return false;
-
-    if ((current_host == kInitiator) && (other_hosts_also_finished || !restore_query_was_sent_to_other_hosts))
-    {
-        if (!cleaner.tryCleanupAfterError())
+        auto unfinished_hosts = expect_other_hosts_finished ? stage_sync.getUnfinishedHosts() : Strings{current_host};
+        LOG_INFO(log, "Skipping removing nodes from ZooKeeper because hosts {} didn't finish",
+                 BackupCoordinationStageSync::getHostsDesc(unfinished_hosts));
         return false;
     }
-
-    return true;
-}
-
-void RestoreCoordinationOnCluster::waitForOtherHostsToFinish()
-{
-    if ((current_host != kInitiator) || !restore_query_was_sent_to_other_hosts)
-        return;
-    stage_sync.waitForOtherHostsToFinish();
-}
-
-bool RestoreCoordinationOnCluster::tryWaitForOtherHostsToFinishAfterError() noexcept
-{
-    if (current_host != kInitiator)
-        return false;
-    if (!restore_query_was_sent_to_other_hosts)
-        return true;
-    return stage_sync.tryWaitForOtherHostsToFinishAfterError();
+    return cleaner.cleanup(throw_if_error);
 }
 
 ZooKeeperRetriesInfo RestoreCoordinationOnCluster::getOnClusterInitializationKeeperRetriesInfo() const

src/Backups/RestoreCoordinationOnCluster.h

@@ -1,7 +1,6 @@
 #pragma once
 
 #include <Backups/IRestoreCoordination.h>
-#include <Backups/BackupConcurrencyCheck.h>
 #include <Backups/BackupCoordinationCleaner.h>
 #include <Backups/BackupCoordinationStageSync.h>
 #include <Backups/WithRetries.h>
@@ -15,7 +14,7 @@ class RestoreCoordinationOnCluster : public IRestoreCoordination
 {
 public:
     /// Empty string as the current host is used to mark the initiator of a RESTORE ON CLUSTER query.
-    static const constexpr std::string_view kInitiator;
+    static const constexpr std::string_view kInitiator = BackupCoordinationStageSync::kInitiator;
 
     RestoreCoordinationOnCluster(
         const UUID & restore_uuid_,
@@ -31,13 +30,13 @@ public:
 
     ~RestoreCoordinationOnCluster() override;
 
+    void setRestoreQueryIsSentToOtherHosts() override;
+    bool isRestoreQuerySentToOtherHosts() const override;
     Strings setStage(const String & new_stage, const String & message, bool sync) override;
-    void setRestoreQueryWasSentToOtherHosts() override;
-    bool trySetError(std::exception_ptr exception) override;
-    void finish() override;
-    bool tryFinishAfterError() noexcept override;
-    void waitForOtherHostsToFinish() override;
-    bool tryWaitForOtherHostsToFinishAfterError() noexcept override;
+    bool setError(std::exception_ptr exception, bool throw_if_error) override;
+    bool waitOtherHostsFinish(bool throw_if_error) const override;
+    bool finish(bool throw_if_error) override;
+    bool cleanup(bool throw_if_error) override;
 
     /// Starts creating a table in a replicated database. Returns false if there is another host which is already creating this table.
     bool acquireCreatingTableInReplicatedDatabase(const String & database_zk_path, const String & table_name) override;
@@ -78,11 +77,10 @@ private:
     const size_t current_host_index;
     LoggerPtr const log;
 
+    /// The order is important: `stage_sync` must be initialized after `with_retries` and `cleaner`.
     const WithRetries with_retries;
-    BackupConcurrencyCheck concurrency_check;
-    BackupCoordinationStageSync stage_sync;
     BackupCoordinationCleaner cleaner;
-    std::atomic<bool> restore_query_was_sent_to_other_hosts = false;
+    BackupCoordinationStageSync stage_sync;
 };
 
 }

src/Client/ClientBaseOptimizedParts.cpp

@@ -1,5 +1,7 @@
 #include <Client/ClientApplicationBase.h>
 
+#include <filesystem>
+
 namespace DB
 {
 
@@ -108,6 +110,7 @@ void ClientApplicationBase::parseAndCheckOptions(OptionsDescription & options_de
         {
             /// Two special cases for better usability:
             /// - if the option contains a whitespace, it might be a query: clickhouse "SELECT 1"
+            /// - if the option is a filesystem file, then it's likely a queries file (clickhouse repro.sql)
             /// These are relevant for interactive usage - user-friendly, but questionable in general.
             /// In case of ambiguity or for scripts, prefer using proper options.
 
@@ -115,8 +118,11 @@ void ClientApplicationBase::parseAndCheckOptions(OptionsDescription & options_de
             po::variable_value value(boost::any(op.value), false);
 
             const char * option;
+            std::error_code ec;
             if (token.contains(' '))
                 option = "query";
+            else if (std::filesystem::is_regular_file(std::filesystem::path{token}, ec))
+                option = "queries-file";
             else
                 throw Exception(ErrorCodes::BAD_ARGUMENTS, "Positional option `{}` is not supported.", token);
 

src/Common/AsynchronousMetrics.cpp

@@ -331,7 +331,7 @@ AsynchronousMetrics::~AsynchronousMetrics()
 
 AsynchronousMetricValues AsynchronousMetrics::getValues() const
 {
-    std::lock_guard lock(data_mutex);
+    SharedLockGuard lock(values_mutex);
     return values;
 }
 
@@ -1807,7 +1807,10 @@ void AsynchronousMetrics::update(TimePoint update_time, bool force_update)
     first_run = false;
 
     // Finally, update the current metrics.
-    values = new_values;
+    {
+        std::lock_guard values_lock(values_mutex);
+        values.swap(new_values);
+    }
 }
 
 }

src/Common/AsynchronousMetrics.h

@@ -4,6 +4,7 @@
 #include <Common/MemoryStatisticsOS.h>
 #include <Common/ThreadPool.h>
 #include <Common/Stopwatch.h>
+#include <Common/SharedMutex.h>
 #include <IO/ReadBufferFromFile.h>
 
 #include <condition_variable>
@@ -100,6 +101,7 @@ private:
     std::condition_variable wait_cond;
     bool quit TSA_GUARDED_BY(thread_mutex) = false;
 
+    /// Protects all raw data and serializes multiple updates.
     mutable std::mutex data_mutex;
 
     /// Some values are incremental and we have to calculate the difference.
@@ -107,7 +109,15 @@ private:
     bool first_run TSA_GUARDED_BY(data_mutex) = true;
     TimePoint previous_update_time TSA_GUARDED_BY(data_mutex);
 
-    AsynchronousMetricValues values TSA_GUARDED_BY(data_mutex);
+    /// Protects saved values.
+    mutable SharedMutex values_mutex;
+    /// Values store the result of the last update prepared for reading.
+#ifdef OS_LINUX
+    AsynchronousMetricValues values TSA_GUARDED_BY(values_mutex);
+#else
+    /// When SharedMutex == std::shared_mutex it may not be annotated with the 'capability'.
+    AsynchronousMetricValues values;
+#endif
 
 #if defined(OS_LINUX) || defined(OS_FREEBSD)
     MemoryStatisticsOS memory_stat TSA_GUARDED_BY(data_mutex);

src/Common/CurrentMetrics.cpp

@@ -255,6 +255,7 @@
     M(PartsActive, "Active data part, used by current and upcoming SELECTs.") \
     M(AttachedDatabase, "Active databases.") \
     M(AttachedTable, "Active tables.") \
+    M(AttachedReplicatedTable, "Active replicated tables.") \
     M(AttachedView, "Active views.") \
     M(AttachedDictionary, "Active dictionaries.") \
     M(PartsOutdated, "Not active data part, but could be used by only current SELECTs, could be deleted after SELECTs finishes.") \

src/Common/LockGuard.h

@@ -1,23 +1,47 @@
 #pragma once
 
-#include <Common/OvercommitTracker.h>
 #include <base/defines.h>
+#include <Common/Exception.h>
+#include <Common/OvercommitTracker.h>
 
 namespace DB
 {
 
+namespace ErrorCodes
+{
+    extern const int LOGICAL_ERROR;
+};
+
 /** LockGuard provides RAII-style locking mechanism for a mutex.
- ** It's intended to be used like std::unique_ptr but with TSA annotations
+ ** It's intended to be used like std::unique_lock but with TSA annotations
   */
 template <typename Mutex>
 class TSA_SCOPED_LOCKABLE LockGuard
 {
 public:
-    explicit LockGuard(Mutex & mutex_) TSA_ACQUIRE(mutex_) : mutex(mutex_) { mutex.lock(); }
-    ~LockGuard() TSA_RELEASE() { mutex.unlock(); }
+    explicit LockGuard(Mutex & mutex_) TSA_ACQUIRE(mutex_) : mutex(mutex_) { lock(); }
+    ~LockGuard() TSA_RELEASE() { if (locked) unlock(); }
+
+    void lock() TSA_ACQUIRE()
+    {
+        /// Don't allow recursive_mutex for now.
+        if (locked)
+            throw Exception(ErrorCodes::LOGICAL_ERROR, "Can't lock twice the same mutex");
+        mutex.lock();
+        locked = true;
+    }
+
+    void unlock() TSA_RELEASE()
+    {
+        if (!locked)
+            throw Exception(ErrorCodes::LOGICAL_ERROR, "Can't unlock the mutex without locking it first");
+        mutex.unlock();
+        locked = false;
+    }
 
 private:
     Mutex & mutex;
+    bool locked = false;
 };
 
 template <template<typename> typename TLockGuard, typename Mutex>

src/Compression/getCompressionCodecForFile.cpp

@@ -10,33 +10,50 @@
 namespace DB
 {
 
-
 using Checksum = CityHash_v1_0_2::uint128;
 
-CompressionCodecPtr getCompressionCodecForFile(const IDataPartStorage & data_part_storage, const String & relative_path)
+CompressionCodecPtr
+getCompressionCodecForFile(ReadBuffer & read_buffer, UInt32 & size_compressed, UInt32 & size_decompressed, bool skip_to_next_block)
 {
-    auto read_buffer = data_part_storage.readFile(relative_path, {}, std::nullopt, std::nullopt);
-    read_buffer->ignore(sizeof(Checksum));
+    read_buffer.ignore(sizeof(Checksum));
 
     UInt8 header_size = ICompressionCodec::getHeaderSize();
+    size_t starting_bytes = read_buffer.count();
     PODArray<char> compressed_buffer;
     compressed_buffer.resize(header_size);
-    read_buffer->readStrict(compressed_buffer.data(), header_size);
+    read_buffer.readStrict(compressed_buffer.data(), header_size);
     uint8_t method = ICompressionCodec::readMethod(compressed_buffer.data());
+    size_compressed = unalignedLoad<UInt32>(&compressed_buffer[1]);
+    size_decompressed = unalignedLoad<UInt32>(&compressed_buffer[5]);
     if (method == static_cast<uint8_t>(CompressionMethodByte::Multiple))
     {
         compressed_buffer.resize(1);
-        read_buffer->readStrict(compressed_buffer.data(), 1);
+        read_buffer.readStrict(compressed_buffer.data(), 1);
         compressed_buffer.resize(1 + compressed_buffer[0]);
-        read_buffer->readStrict(compressed_buffer.data() + 1, compressed_buffer[0]);
+        read_buffer.readStrict(compressed_buffer.data() + 1, compressed_buffer[0]);
         auto codecs_bytes = CompressionCodecMultiple::getCodecsBytesFromData(compressed_buffer.data());
         Codecs codecs;
         for (auto byte : codecs_bytes)
             codecs.push_back(CompressionCodecFactory::instance().get(byte));
 
+        if (skip_to_next_block)
+            read_buffer.ignore(size_compressed - (read_buffer.count() - starting_bytes));
+
         return std::make_shared<CompressionCodecMultiple>(codecs);
     }
+
+    if (skip_to_next_block)
+        read_buffer.ignore(size_compressed - (read_buffer.count() - starting_bytes));
+
     return CompressionCodecFactory::instance().get(method);
 }
 
+CompressionCodecPtr getCompressionCodecForFile(const IDataPartStorage & data_part_storage, const String & relative_path)
+{
+    auto read_buffer = data_part_storage.readFile(relative_path, {}, std::nullopt, std::nullopt);
+    UInt32 size_compressed;
+    UInt32 size_decompressed;
+    return getCompressionCodecForFile(*read_buffer, size_compressed, size_decompressed, false);
+}
+
 }

src/Compression/getCompressionCodecForFile.h

@@ -13,4 +13,8 @@ namespace DB
 /// from metadata.
 CompressionCodecPtr getCompressionCodecForFile(const IDataPartStorage & data_part_storage, const String & relative_path);
 
+/// Same as above which is used by clickhouse-compressor to print compression statistics of each data block.
+CompressionCodecPtr
+getCompressionCodecForFile(ReadBuffer & read_buffer, UInt32 & size_compressed, UInt32 & size_decompressed, bool skip_to_next_block);
+
 }

src/Core/ServerSettings.cpp

@@ -131,6 +131,9 @@ namespace DB
     DECLARE(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) \
     DECLARE(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) \
     DECLARE(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) \
+    DECLARE(UInt64, max_replicated_table_num_to_throw, 0lu, "If number of replicated tables is greater than this value, server will throw an exception. 0 means no limitation. Only count table in Atomic/Ordinary/Replicated/Lazy database engine.", 0) \
+    DECLARE(UInt64, max_dictionary_num_to_throw, 0lu, "If number of dictionaries is greater than this value, server will throw an exception. 0 means no limitation. Only count table in Atomic/Ordinary/Replicated/Lazy database engine.", 0) \
+    DECLARE(UInt64, max_view_num_to_throw, 0lu, "If number of views is greater than this value, server will throw an exception. 0 means no limitation. Only count table in Atomic/Ordinary/Replicated/Lazy database engine.", 0) \
     DECLARE(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) \
     DECLARE(UInt64, max_authentication_methods_per_user, 100, "The maximum number of authentication methods a user can be created with or altered. Changing this setting does not affect existing users. Zero means unlimited", 0) \
     DECLARE(UInt64, concurrent_threads_soft_limit_num, 0, "Sets how many concurrent thread can be allocated before applying CPU pressure. Zero means unlimited.", 0) \

src/Core/Settings.cpp

@@ -2892,6 +2892,9 @@ Possible values:
 **See Also**
 
 - [ORDER BY Clause](../../sql-reference/statements/select/order-by.md/#optimize_read_in_order)
+)", 0) \
+    DECLARE(Bool, read_in_order_use_virtual_row, false, R"(
+Use virtual row while reading in order of primary key or its monotonic function fashion. It is useful when searching over multiple parts as only relevant ones are touched.
 )", 0) \
     DECLARE(Bool, optimize_read_in_window_order, true, R"(
 Enable ORDER BY optimization in window clause for reading data in corresponding order in MergeTree tables.
@@ -3666,6 +3669,11 @@ Given that, for example, dictionaries, can be out of sync across nodes, mutation
 
 </profiles>
 ```
+)", 0) \
+ DECLARE(Bool, validate_mutation_query, true, R"(
+Validate mutation queries before accepting them. Mutations are executed in the background, and running an invalid query will cause mutations to get stuck, requiring manual intervention.
+
+Only change this setting if you encounter a backward-incompatible bug.
 )", 0) \
     DECLARE(Seconds, lock_acquire_timeout, DBMS_DEFAULT_LOCK_ACQUIRE_TIMEOUT_SEC, R"(
 Defines how many seconds a locking request waits before failing.
@@ -4557,7 +4565,7 @@ Possible values:
 - 0 - Disable
 - 1 - Enable
 )", 0) \
-    DECLARE(Bool, query_plan_merge_filters, false, R"(
+    DECLARE(Bool, query_plan_merge_filters, true, R"(
 Allow to merge filters in the query plan
 )", 0) \
     DECLARE(Bool, query_plan_filter_push_down, true, R"(
@@ -4858,9 +4866,9 @@ Allows to record the filesystem caching log for each query
     DECLARE(Bool, read_from_filesystem_cache_if_exists_otherwise_bypass_cache, false, R"(
 Allow to use the filesystem cache in passive mode - benefit from the existing cache entries, but don't put more entries into the cache. If you set this setting for heavy ad-hoc queries and leave it disabled for short real-time queries, this will allows to avoid cache threshing by too heavy queries and to improve the overall system efficiency.
 )", 0) \
-    DECLARE(Bool, skip_download_if_exceeds_query_cache, true, R"(
+    DECLARE(Bool, filesystem_cache_skip_download_if_exceeds_per_query_cache_write_limit, true, R"(
 Skip download from remote filesystem if exceeds query cache size
-)", 0) \
+)", 0)  ALIAS(skip_download_if_exceeds_query_cache) \
     DECLARE(UInt64, filesystem_cache_max_download_size, (128UL * 1024 * 1024 * 1024), R"(
 Max remote filesystem cache size that can be downloaded by a single query
 )", 0) \
@@ -4872,6 +4880,9 @@ Limit on size of a single batch of file segments that a read buffer can request
 )", 0) \
     DECLARE(UInt64, filesystem_cache_reserve_space_wait_lock_timeout_milliseconds, 1000, R"(
 Wait time to lock cache for space reservation in filesystem cache
+)", 0) \
+    DECLARE(Bool, filesystem_cache_prefer_bigger_buffer_size, true, R"(
+Prefer bigger buffer size if filesystem cache is enabled to avoid writing small file segments which deteriorate cache performance. On the other hand, enabling this setting might increase memory usage.
 )", 0) \
     DECLARE(UInt64, temporary_data_in_cache_reserve_space_wait_lock_timeout_milliseconds, (10 * 60 * 1000), R"(
 Wait time to lock cache for space reservation for temporary data in filesystem cache

src/Core/SettingsChangesHistory.cpp

@@ -64,6 +64,7 @@ static std::initializer_list<std::pair<ClickHouseVersion, SettingsChangesHistory
     },
     {"24.11",
         {
+            {"validate_mutation_query", false, true, "New setting to validate mutation queries by default."},
             {"enable_job_stack_trace", false, true, "Enable by default collecting stack traces from job's scheduling."},
             {"allow_suspicious_types_in_group_by", true, false, "Don't allow Variant/Dynamic types in GROUP BY by default"},
             {"allow_suspicious_types_in_order_by", true, false, "Don't allow Variant/Dynamic types in ORDER BY by default"},
@@ -76,7 +77,11 @@ static std::initializer_list<std::pair<ClickHouseVersion, SettingsChangesHistory
             {"backup_restore_keeper_max_retries_while_initializing", 0, 20, "New setting."},
             {"backup_restore_keeper_max_retries_while_handling_error", 0, 20, "New setting."},
             {"backup_restore_finish_timeout_after_error_sec", 0, 180, "New setting."},
+            {"query_plan_merge_filters", false, true, "Allow to merge filters in the query plan. This is required to properly support filter-push-down with a new analyzer."},
             {"parallel_replicas_local_plan", false, true, "Use local plan for local replica in a query with parallel replicas"},
+            {"filesystem_cache_skip_download_if_exceeds_per_query_cache_write_limit", 1, 1, "Rename of setting skip_download_if_exceeds_query_cache_limit"},
+            {"filesystem_cache_prefer_bigger_buffer_size", true, true, "New setting"},
+            {"read_in_order_use_virtual_row", false, false, "Use virtual row while reading in order of primary key or its monotonic function fashion. It is useful when searching over multiple parts as only relevant ones are touched."},
         }
     },
     {"24.10",

src/Databases/DatabaseAtomic.cpp

@@ -6,7 +6,6 @@
 #include <Databases/DatabaseReplicated.h>
 #include <IO/ReadBufferFromFile.h>
 #include <IO/ReadHelpers.h>
-#include <IO/WriteHelpers.h>
 #include <Interpreters/Context.h>
 #include <Interpreters/DDLTask.h>
 #include <Interpreters/DatabaseCatalog.h>
@@ -19,6 +18,7 @@
 #include <Common/filesystemHelpers.h>
 #include <Core/Settings.h>
 
+
 namespace fs = std::filesystem;
 
 namespace DB
@@ -60,9 +60,6 @@ DatabaseAtomic::DatabaseAtomic(String name_, String metadata_path_, UUID uuid, c
     , db_uuid(uuid)
 {
     assert(db_uuid != UUIDHelpers::Nil);
-    fs::create_directories(fs::path(getContext()->getPath()) / "metadata");
-    fs::create_directories(path_to_table_symlinks);
-    tryCreateMetadataSymlink();
 }
 
 DatabaseAtomic::DatabaseAtomic(String name_, String metadata_path_, UUID uuid, ContextPtr context_)
@@ -70,6 +67,20 @@ DatabaseAtomic::DatabaseAtomic(String name_, String metadata_path_, UUID uuid, C
 {
 }
 
+void DatabaseAtomic::createDirectories()
+{
+    std::lock_guard lock(mutex);
+    createDirectoriesUnlocked();
+}
+
+void DatabaseAtomic::createDirectoriesUnlocked()
+{
+    DatabaseOnDisk::createDirectoriesUnlocked();
+    fs::create_directories(fs::path(getContext()->getPath()) / "metadata");
+    fs::create_directories(path_to_table_symlinks);
+    tryCreateMetadataSymlink();
+}
+
 String DatabaseAtomic::getTableDataPath(const String & table_name) const
 {
     std::lock_guard lock(mutex);
@@ -108,6 +119,7 @@ void DatabaseAtomic::attachTable(ContextPtr /* context_ */, const String & name,
     assert(relative_table_path != data_path && !relative_table_path.empty());
     DetachedTables not_in_use;
     std::lock_guard lock(mutex);
+    createDirectoriesUnlocked();
     not_in_use = cleanupDetachedTables();
     auto table_id = table->getStorageID();
     assertDetachedTableNotInUse(table_id.uuid);
@@ -208,11 +220,15 @@ void DatabaseAtomic::renameTable(ContextPtr local_context, const String & table_
     if (exchange && !supportsAtomicRename(&message))
         throw Exception(ErrorCodes::NOT_IMPLEMENTED, "RENAME EXCHANGE is not supported because exchanging files is not supported by the OS ({})", message);
 
+    createDirectories();
     waitDatabaseStarted();
 
     auto & other_db = dynamic_cast<DatabaseAtomic &>(to_database);
     bool inside_database = this == &other_db;
 
+    if (!inside_database)
+        other_db.createDirectories();
+
     String old_metadata_path = getObjectMetadataPath(table_name);
     String new_metadata_path = to_database.getObjectMetadataPath(to_table_name);
 
@@ -333,6 +349,7 @@ void DatabaseAtomic::commitCreateTable(const ASTCreateQuery & query, const Stora
                                        const String & table_metadata_tmp_path, const String & table_metadata_path,
                                        ContextPtr query_context)
 {
+    createDirectories();
     DetachedTables not_in_use;
     auto table_data_path = getTableDataPath(query);
     try
@@ -469,6 +486,9 @@ void DatabaseAtomic::beforeLoadingMetadata(ContextMutablePtr /*context*/, Loadin
     if (mode < LoadingStrictnessLevel::FORCE_RESTORE)
         return;
 
+    if (!fs::exists(path_to_table_symlinks))
+        return;
+
     /// Recreate symlinks to table data dirs in case of force restore, because some of them may be broken
     for (const auto & table_path : fs::directory_iterator(path_to_table_symlinks))
     {
@@ -611,6 +631,7 @@ void DatabaseAtomic::renameDatabase(ContextPtr query_context, const String & new
 {
     /// CREATE, ATTACH, DROP, DETACH and RENAME DATABASE must hold DDLGuard
 
+    createDirectories();
     waitDatabaseStarted();
 
     bool check_ref_deps = query_context->getSettingsRef()[Setting::check_referential_table_dependencies];
@@ -702,4 +723,5 @@ void registerDatabaseAtomic(DatabaseFactory & factory)
     };
     factory.registerDatabase("Atomic", create_fn);
 }
+
 }

src/Databases/DatabaseAtomic.h

@@ -76,6 +76,9 @@ protected:
     using DetachedTables = std::unordered_map<UUID, StoragePtr>;
     [[nodiscard]] DetachedTables cleanupDetachedTables() TSA_REQUIRES(mutex);
 
+    void createDirectories();
+    void createDirectoriesUnlocked() TSA_REQUIRES(mutex);
+
     void tryCreateMetadataSymlink();
 
     virtual bool allowMoveTableToOtherDatabaseEngine(IDatabase & /*to_database*/) const { return false; }

src/Databases/DatabaseLazy.cpp

@@ -47,6 +47,7 @@ DatabaseLazy::DatabaseLazy(const String & name_, const String & metadata_path_,
     : DatabaseOnDisk(name_, metadata_path_, std::filesystem::path("data") / escapeForFileName(name_) / "", "DatabaseLazy (" + name_ + ")", context_)
     , expiration_time(expiration_time_)
 {
+    createDirectories();
 }
 
 

src/Databases/DatabaseOnDisk.cpp

@@ -180,7 +180,18 @@ DatabaseOnDisk::DatabaseOnDisk(
     , metadata_path(metadata_path_)
     , data_path(data_path_)
 {
-    fs::create_directories(local_context->getPath() + data_path);
+}
+
+
+void DatabaseOnDisk::createDirectories()
+{
+    std::lock_guard lock(mutex);
+    createDirectoriesUnlocked();
+}
+
+void DatabaseOnDisk::createDirectoriesUnlocked()
+{
+    fs::create_directories(std::filesystem::path(getContext()->getPath()) / data_path);
     fs::create_directories(metadata_path);
 }
 
@@ -198,6 +209,8 @@ void DatabaseOnDisk::createTable(
     const StoragePtr & table,
     const ASTPtr & query)
 {
+    createDirectories();
+
     const auto & settings = local_context->getSettingsRef();
     const auto & create = query->as<ASTCreateQuery &>();
     assert(table_name == create.getTable());
@@ -265,7 +278,6 @@ void DatabaseOnDisk::createTable(
     }
 
     commitCreateTable(create, table, table_metadata_tmp_path, table_metadata_path, local_context);
-
     removeDetachedPermanentlyFlag(local_context, table_name, table_metadata_path, false);
 }
 
@@ -293,6 +305,8 @@ void DatabaseOnDisk::commitCreateTable(const ASTCreateQuery & query, const Stora
 {
     try
     {
+        createDirectories();
+
         /// Add a table to the map of known tables.
         attachTable(query_context, query.getTable(), table, getTableDataPath(query));
 
@@ -426,6 +440,7 @@ void DatabaseOnDisk::renameTable(
             throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Moving tables between databases of different engines is not supported");
     }
 
+    createDirectories();
     waitDatabaseStarted();
 
     auto table_data_relative_path = getTableDataPath(table_name);
@@ -621,6 +636,9 @@ time_t DatabaseOnDisk::getObjectMetadataModificationTime(const String & object_n
 
 void DatabaseOnDisk::iterateMetadataFiles(const IteratingFunction & process_metadata_file) const
 {
+    if (!fs::exists(metadata_path))
+        return;
+
     auto process_tmp_drop_metadata_file = [&](const String & file_name)
     {
         assert(getUUID() == UUIDHelpers::Nil);

src/Databases/DatabaseOnDisk.h

@@ -99,6 +99,9 @@ protected:
     virtual void removeDetachedPermanentlyFlag(ContextPtr context, const String & table_name, const String & table_metadata_path, bool attach);
     virtual void setDetachedTableNotInUseForce(const UUID & /*uuid*/) {}
 
+    void createDirectories();
+    void createDirectoriesUnlocked() TSA_REQUIRES(mutex);
+
     const String metadata_path;
     const String data_path;
 };

src/Databases/DatabasesCommon.cpp

@@ -382,7 +382,8 @@ StoragePtr DatabaseWithOwnTablesBase::detachTableUnlocked(const String & table_n
     if (!table_storage->isSystemStorage() && !DatabaseCatalog::isPredefinedDatabase(database_name))
     {
         LOG_TEST(log, "Counting detached table {} to database {}", table_name, database_name);
-        CurrentMetrics::sub(getAttachedCounterForStorage(table_storage));
+        for (auto metric : getAttachedCountersForStorage(table_storage))
+            CurrentMetrics::sub(metric);
     }
 
     auto table_id = table_storage->getStorageID();
@@ -430,7 +431,8 @@ void DatabaseWithOwnTablesBase::attachTableUnlocked(const String & table_name, c
     if (!table->isSystemStorage() && !DatabaseCatalog::isPredefinedDatabase(database_name))
     {
         LOG_TEST(log, "Counting attached table {} to database {}", table_name, database_name);
-        CurrentMetrics::add(getAttachedCounterForStorage(table));
+        for (auto metric : getAttachedCountersForStorage(table))
+            CurrentMetrics::add(metric);
     }
 }
 

src/Databases/IDatabase.h

@@ -416,6 +416,7 @@ public:
         std::lock_guard lock{mutex};
         return database_name;
     }
+
     /// Get UUID of database.
     virtual UUID getUUID() const { return UUIDHelpers::Nil; }
 

src/Databases/MySQL/DatabaseMaterializedMySQL.cpp

@@ -62,6 +62,7 @@ DatabaseMaterializedMySQL::DatabaseMaterializedMySQL(
     , settings(std::move(settings_))
     , materialize_thread(context_, database_name_, mysql_database_name_, std::move(pool_), std::move(client_), binlog_client_, settings.get())
 {
+    createDirectories();
 }
 
 DatabaseMaterializedMySQL::~DatabaseMaterializedMySQL() = default;

src/Disks/ObjectStorages/AzureBlobStorage/AzureObjectStorage.cpp

@@ -277,19 +277,6 @@ void AzureObjectStorage::removeObjectImpl(const StoredObject & object, const Sha
     }
 }
 
-/// Remove file. Throws exception if file doesn't exists or it's a directory.
-void AzureObjectStorage::removeObject(const StoredObject & object)
-{
-    removeObjectImpl(object, client.get(), false);
-}
-
-void AzureObjectStorage::removeObjects(const StoredObjects & objects)
-{
-    auto client_ptr = client.get();
-    for (const auto & object : objects)
-        removeObjectImpl(object, client_ptr, false);
-}
-
 void AzureObjectStorage::removeObjectIfExists(const StoredObject & object)
 {
     removeObjectImpl(object, client.get(), true);

src/Disks/ObjectStorages/AzureBlobStorage/AzureObjectStorage.h

@@ -59,11 +59,6 @@ public:
         size_t buf_size = DBMS_DEFAULT_BUFFER_SIZE,
         const WriteSettings & write_settings = {}) override;
 
-    /// Remove file. Throws exception if file doesn't exists or it's a directory.
-    void removeObject(const StoredObject & object) override;
-
-    void removeObjects(const StoredObjects & objects) override;
-
     void removeObjectIfExists(const StoredObject & object) override;
 
     void removeObjectsIfExist(const StoredObjects & objects) override;

src/Disks/ObjectStorages/Cached/CachedObjectStorage.cpp

@@ -148,20 +148,6 @@ void CachedObjectStorage::removeCacheIfExists(const std::string & path_key_for_c
         cache->removeKeyIfExists(getCacheKey(path_key_for_cache), FileCache::getCommonUser().user_id);
 }
 
-void CachedObjectStorage::removeObject(const StoredObject & object)
-{
-    removeCacheIfExists(object.remote_path);
-    object_storage->removeObject(object);
-}
-
-void CachedObjectStorage::removeObjects(const StoredObjects & objects)
-{
-    for (const auto & object : objects)
-        removeCacheIfExists(object.remote_path);
-
-    object_storage->removeObjects(objects);
-}
-
 void CachedObjectStorage::removeObjectIfExists(const StoredObject & object)
 {
     removeCacheIfExists(object.remote_path);

src/Disks/ObjectStorages/Cached/CachedObjectStorage.h

@@ -45,10 +45,6 @@ public:
         size_t buf_size = DBMS_DEFAULT_BUFFER_SIZE,
         const WriteSettings & write_settings = {}) override;
 
-    void removeObject(const StoredObject & object) override;
-
-    void removeObjects(const StoredObjects & objects) override;
-
     void removeObjectIfExists(const StoredObject & object) override;
 
     void removeObjectsIfExist(const StoredObjects & objects) override;

src/Disks/ObjectStorages/DiskObjectStorage.cpp

@@ -642,7 +642,10 @@ std::unique_ptr<ReadBufferFromFileBase> DiskObjectStorage::readFile(
     };
 
     /// Avoid cache fragmentation by choosing bigger buffer size.
-    bool prefer_bigger_buffer_size = object_storage->supportsCache() && read_settings.enable_filesystem_cache;
+    bool prefer_bigger_buffer_size = read_settings.filesystem_cache_prefer_bigger_buffer_size
+        && object_storage->supportsCache()
+        && read_settings.enable_filesystem_cache;
+
     size_t buffer_size = prefer_bigger_buffer_size
         ? std::max<size_t>(settings.remote_fs_buffer_size, DBMS_DEFAULT_BUFFER_SIZE)
         : settings.remote_fs_buffer_size;

src/Disks/ObjectStorages/DiskObjectStorageTransaction.cpp

@@ -480,8 +480,7 @@ struct WriteFileObjectStorageOperation final : public IDiskObjectStorageOperatio
 
     void undo() override
     {
-        if (object_storage.exists(object))
-            object_storage.removeObject(object);
+        object_storage.removeObjectIfExists(object);
     }
 
     void finalize() override
@@ -543,8 +542,7 @@ struct CopyFileObjectStorageOperation final : public IDiskObjectStorageOperation
 
     void undo() override
     {
-        for (const auto & object : created_objects)
-            destination_object_storage.removeObject(object);
+         destination_object_storage.removeObjectsIfExist(created_objects);
     }
 
     void finalize() override

src/Disks/ObjectStorages/HDFS/HDFSObjectStorage.h

@@ -77,11 +77,6 @@ public:
         size_t buf_size = DBMS_DEFAULT_BUFFER_SIZE,
         const WriteSettings & write_settings = {}) override;
 
-    /// Remove file. Throws exception if file doesn't exists or it's a directory.
-    void removeObject(const StoredObject & object) override;
-
-    void removeObjects(const StoredObjects & objects) override;
-
     void removeObjectIfExists(const StoredObject & object) override;
 
     void removeObjectsIfExist(const StoredObjects & objects) override;
@@ -117,6 +112,11 @@ private:
     void initializeHDFSFS() const;
     std::string extractObjectKeyFromURL(const StoredObject & object) const;
 
+    /// Remove file. Throws exception if file doesn't exists or it's a directory.
+    void removeObject(const StoredObject & object);
+
+    void removeObjects(const StoredObjects & objects);
+
     const Poco::Util::AbstractConfiguration & config;
 
     mutable HDFSBuilderWrapper hdfs_builder;

src/Disks/ObjectStorages/IObjectStorage.h

@@ -161,11 +161,11 @@ public:
     virtual bool isRemote() const = 0;
 
     /// Remove object. Throws exception if object doesn't exists.
-    virtual void removeObject(const StoredObject & object) = 0;
+    // virtual void removeObject(const StoredObject & object) = 0;
 
     /// Remove multiple objects. Some object storages can do batch remove in a more
     /// optimal way.
-    virtual void removeObjects(const StoredObjects & objects) = 0;
+    // virtual void removeObjects(const StoredObjects & objects) = 0;
 
     /// Remove object on path if exists
     virtual void removeObjectIfExists(const StoredObject & object) = 0;

src/Disks/ObjectStorages/Local/LocalObjectStorage.cpp

@@ -81,7 +81,7 @@ std::unique_ptr<WriteBufferFromFileBase> LocalObjectStorage::writeObject( /// NO
     return std::make_unique<WriteBufferFromFile>(object.remote_path, buf_size);
 }
 
-void LocalObjectStorage::removeObject(const StoredObject & object)
+void LocalObjectStorage::removeObject(const StoredObject & object) const
 {
     /// For local object storage files are actually removed when "metadata" is removed.
     if (!exists(object))
@@ -91,7 +91,7 @@ void LocalObjectStorage::removeObject(const StoredObject & object)
         ErrnoException::throwFromPath(ErrorCodes::CANNOT_UNLINK, object.remote_path, "Cannot unlink file {}", object.remote_path);
 }
 
-void LocalObjectStorage::removeObjects(const StoredObjects & objects)
+void LocalObjectStorage::removeObjects(const StoredObjects & objects) const
 {
     for (const auto & object : objects)
         removeObject(object);

src/Disks/ObjectStorages/Local/LocalObjectStorage.h

@@ -42,10 +42,6 @@ public:
         size_t buf_size = DBMS_DEFAULT_BUFFER_SIZE,
         const WriteSettings & write_settings = {}) override;
 
-    void removeObject(const StoredObject & object) override;
-
-    void removeObjects(const StoredObjects &  objects) override;
-
     void removeObjectIfExists(const StoredObject & object) override;
 
     void removeObjectsIfExist(const StoredObjects & objects) override;
@@ -82,6 +78,10 @@ public:
     ReadSettings patchSettings(const ReadSettings & read_settings) const override;
 
 private:
+    void removeObject(const StoredObject & object) const;
+
+    void removeObjects(const StoredObjects &  objects) const;
+
     String key_prefix;
     LoggerPtr log;
     std::string description;

src/Disks/ObjectStorages/MetadataStorageFromPlainObjectStorage.cpp

@@ -203,7 +203,7 @@ void MetadataStorageFromPlainObjectStorageTransaction::unlinkFile(const std::str
 {
     auto object_key = metadata_storage.object_storage->generateObjectKeyForPath(path, std::nullopt /* key_prefix */);
     auto object = StoredObject(object_key.serialize());
-    metadata_storage.object_storage->removeObject(object);
+    metadata_storage.object_storage->removeObjectIfExists(object);
 }
 
 void MetadataStorageFromPlainObjectStorageTransaction::removeDirectory(const std::string & path)
@@ -211,7 +211,7 @@ void MetadataStorageFromPlainObjectStorageTransaction::removeDirectory(const std
     if (metadata_storage.object_storage->isWriteOnce())
     {
         for (auto it = metadata_storage.iterateDirectory(path); it->isValid(); it->next())
-            metadata_storage.object_storage->removeObject(StoredObject(it->path()));
+            metadata_storage.object_storage->removeObjectIfExists(StoredObject(it->path()));
     }
     else
     {

src/Disks/ObjectStorages/MetadataStorageFromPlainObjectStorageOperations.cpp

@@ -107,7 +107,7 @@ void MetadataStorageFromPlainObjectStorageCreateDirectoryOperation::undo(std::un
         auto metric = object_storage->getMetadataStorageMetrics().directory_map_size;
         CurrentMetrics::sub(metric, 1);
 
-        object_storage->removeObject(StoredObject(metadata_object_key.serialize(), path / PREFIX_PATH_FILE_NAME));
+        object_storage->removeObjectIfExists(StoredObject(metadata_object_key.serialize(), path / PREFIX_PATH_FILE_NAME));
     }
     else if (write_created)
         object_storage->removeObjectIfExists(StoredObject(metadata_object_key.serialize(), path / PREFIX_PATH_FILE_NAME));
@@ -247,7 +247,7 @@ void MetadataStorageFromPlainObjectStorageRemoveDirectoryOperation::execute(std:
 
     auto metadata_object_key = createMetadataObjectKey(key_prefix, metadata_key_prefix);
     auto metadata_object = StoredObject(/*remote_path*/ metadata_object_key.serialize(), /*local_path*/ path / PREFIX_PATH_FILE_NAME);
-    object_storage->removeObject(metadata_object);
+    object_storage->removeObjectIfExists(metadata_object);
 
     {
         std::lock_guard lock(path_map.mutex);

src/Disks/ObjectStorages/S3/S3ObjectStorage.cpp

@@ -326,21 +326,11 @@ void S3ObjectStorage::removeObjectsImpl(const StoredObjects & objects, bool if_e
                       ProfileEvents::DiskS3DeleteObjects);
 }
 
-void S3ObjectStorage::removeObject(const StoredObject & object)
-{
-    removeObjectImpl(object, false);
-}
-
 void S3ObjectStorage::removeObjectIfExists(const StoredObject & object)
 {
     removeObjectImpl(object, true);
 }
 
-void S3ObjectStorage::removeObjects(const StoredObjects & objects)
-{
-    removeObjectsImpl(objects, false);
-}
-
 void S3ObjectStorage::removeObjectsIfExist(const StoredObjects & objects)
 {
     removeObjectsImpl(objects, true);

src/Disks/ObjectStorages/S3/S3ObjectStorage.h

@@ -101,13 +101,6 @@ public:
 
     ObjectStorageIteratorPtr iterate(const std::string & path_prefix, size_t max_keys) const override;
 
-    /// Uses `DeleteObjectRequest`.
-    void removeObject(const StoredObject & object) override;
-
-    /// Uses `DeleteObjectsRequest` if it is allowed by `s3_capabilities`, otherwise `DeleteObjectRequest`.
-    /// `DeleteObjectsRequest` is not supported on GCS, see https://issuetracker.google.com/issues/162653700 .
-    void removeObjects(const StoredObjects & objects) override;
-
     /// Uses `DeleteObjectRequest`.
     void removeObjectIfExists(const StoredObject & object) override;
 

src/Disks/ObjectStorages/Web/WebObjectStorage.cpp

@@ -254,16 +254,6 @@ std::unique_ptr<WriteBufferFromFileBase> WebObjectStorage::writeObject( /// NOLI
     throwNotAllowed();
 }
 
-void WebObjectStorage::removeObject(const StoredObject &)
-{
-    throwNotAllowed();
-}
-
-void WebObjectStorage::removeObjects(const StoredObjects &)
-{
-    throwNotAllowed();
-}
-
 void WebObjectStorage::removeObjectIfExists(const StoredObject &)
 {
     throwNotAllowed();

src/Disks/ObjectStorages/Web/WebObjectStorage.h

@@ -47,10 +47,6 @@ public:
         size_t buf_size = DBMS_DEFAULT_BUFFER_SIZE,
         const WriteSettings & write_settings = {}) override;
 
-    void removeObject(const StoredObject & object) override;
-
-    void removeObjects(const StoredObjects &  objects) override;
-
     void removeObjectIfExists(const StoredObject & object) override;
 
     void removeObjectsIfExist(const StoredObjects & objects) override;

src/Functions/FunctionsComparison.h

@@ -1033,6 +1033,9 @@ private:
             size_t tuple_size,
             size_t input_rows_count) const
     {
+        if (0 == tuple_size)
+            throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Comparison of zero-sized tuples is not implemented");
+
         ColumnsWithTypeAndName less_columns(tuple_size);
         ColumnsWithTypeAndName equal_columns(tuple_size - 1);
         ColumnsWithTypeAndName tmp_columns(2);

src/Functions/FunctionsConversion.cpp

@@ -4410,7 +4410,7 @@ private:
                     variant_column = IColumn::mutate(column);
                 /// Otherwise we should filter column.
                 else
-                    variant_column = column->filter(filter, variant_size_hint)->assumeMutable();
+                    variant_column = IColumn::mutate(column->filter(filter, variant_size_hint));
 
                 assert_cast<ColumnLowCardinality &>(*variant_column).nestedRemoveNullable();
                 return createVariantFromDescriptorsAndOneNonEmptyVariant(variant_types, std::move(discriminators), std::move(variant_column), variant_discr);

src/Functions/if.cpp

@@ -668,6 +668,9 @@ private:
         temporary_columns[0] = arguments[0];
 
         size_t tuple_size = type1.getElements().size();
+        if (tuple_size == 0)
+            return ColumnTuple::create(input_rows_count);
+
         Columns tuple_columns(tuple_size);
 
         for (size_t i = 0; i < tuple_size; ++i)

src/IO/ReadSettings.h

@@ -61,6 +61,7 @@ struct ReadSettings
     bool filesystem_cache_allow_background_download = true;
     bool filesystem_cache_allow_background_download_for_metadata_files_in_packed_storage = true;
     bool filesystem_cache_allow_background_download_during_fetch = true;
+    bool filesystem_cache_prefer_bigger_buffer_size = true;
 
     bool use_page_cache_for_disks_without_file_cache = false;
     bool read_from_page_cache_if_exists_otherwise_bypass_cache = false;
@@ -68,7 +69,7 @@ struct ReadSettings
     std::shared_ptr<PageCache> page_cache;
 
     size_t filesystem_cache_max_download_size = (128UL * 1024 * 1024 * 1024);
-    bool skip_download_if_exceeds_query_cache = true;
+    bool filesystem_cache_skip_download_if_exceeds_per_query_cache_write_limit = true;
 
     size_t remote_read_min_bytes_for_seek = DBMS_DEFAULT_BUFFER_SIZE;
 

src/IO/S3/URI.cpp

@@ -37,7 +37,7 @@ URI::URI(const std::string & uri_, bool allow_archive_path_syntax)
     /// Case when bucket name represented in domain name of S3 URL.
     /// E.g. (https://bucket-name.s3.region.amazonaws.com/key)
     /// https://docs.aws.amazon.com/AmazonS3/latest/dev/VirtualHosting.html#virtual-hosted-style-access
-    static const RE2 virtual_hosted_style_pattern(R"((.+)\.(s3express[\-a-z0-9]+|s3|cos|obs|oss|eos)([.\-][a-z0-9\-.:]+))");
+    static const RE2 virtual_hosted_style_pattern(R"((.+)\.(s3express[\-a-z0-9]+|s3|cos|obs|oss-data-acc|oss|eos)([.\-][a-z0-9\-.:]+))");
 
     /// Case when AWS Private Link Interface is being used
     /// E.g. (bucket.vpce-07a1cd78f1bd55c5f-j3a3vg6w.s3.us-east-1.vpce.amazonaws.com/bucket-name/key)
@@ -115,7 +115,15 @@ URI::URI(const std::string & uri_, bool allow_archive_path_syntax)
         && re2::RE2::FullMatch(uri.getAuthority(), virtual_hosted_style_pattern, &bucket, &name, &endpoint_authority_from_uri))
     {
         is_virtual_hosted_style = true;
+        if (name == "oss-data-acc")
+        {
+            bucket = bucket.substr(0, bucket.find('.'));
+            endpoint = uri.getScheme() + "://" + uri.getHost().substr(bucket.length() + 1);
+        }
+        else
+        {
             endpoint = uri.getScheme() + "://" + name + endpoint_authority_from_uri;
+        }
         validateBucket(bucket, uri);
 
         if (!uri.getPath().empty())

src/IO/tests/gtest_s3_uri.cpp

@@ -212,6 +212,22 @@ TEST(S3UriTest, validPatterns)
         ASSERT_EQ("", uri.version_id);
         ASSERT_EQ(true, uri.is_virtual_hosted_style);
     }
+    {
+        S3::URI uri("https://bucket-test1.oss-cn-beijing-internal.aliyuncs.com/ab-test");
+        ASSERT_EQ("https://oss-cn-beijing-internal.aliyuncs.com", uri.endpoint);
+        ASSERT_EQ("bucket-test1", uri.bucket);
+        ASSERT_EQ("ab-test", uri.key);
+        ASSERT_EQ("", uri.version_id);
+        ASSERT_EQ(true, uri.is_virtual_hosted_style);
+    }
+    {
+        S3::URI uri("https://bucket-test.cn-beijing-internal.oss-data-acc.aliyuncs.com/ab-test");
+        ASSERT_EQ("https://cn-beijing-internal.oss-data-acc.aliyuncs.com", uri.endpoint);
+        ASSERT_EQ("bucket-test", uri.bucket);
+        ASSERT_EQ("ab-test", uri.key);
+        ASSERT_EQ("", uri.version_id);
+        ASSERT_EQ(true, uri.is_virtual_hosted_style);
+    }
 }
 
 TEST(S3UriTest, versionIdChecks)

src/Interpreters/Cache/QueryLimit.cpp

@@ -53,7 +53,7 @@ FileCacheQueryLimit::QueryContextPtr FileCacheQueryLimit::getOrSetQueryContext(
     {
         it->second = std::make_shared<QueryContext>(
             settings.filesystem_cache_max_download_size,
-            !settings.skip_download_if_exceeds_query_cache);
+            !settings.filesystem_cache_skip_download_if_exceeds_per_query_cache_write_limit);
     }
 
     return it->second;

src/Interpreters/Context.cpp

@@ -196,6 +196,7 @@ namespace Setting
     extern const SettingsUInt64 filesystem_cache_segments_batch_size;
     extern const SettingsBool filesystem_cache_enable_background_download_for_metadata_files_in_packed_storage;
     extern const SettingsBool filesystem_cache_enable_background_download_during_fetch;
+    extern const SettingsBool filesystem_cache_prefer_bigger_buffer_size;
     extern const SettingsBool http_make_head_request;
     extern const SettingsUInt64 http_max_fields;
     extern const SettingsUInt64 http_max_field_name_size;
@@ -236,7 +237,7 @@ namespace Setting
     extern const SettingsUInt64 remote_fs_read_backoff_max_tries;
     extern const SettingsUInt64 remote_read_min_bytes_for_seek;
     extern const SettingsBool throw_on_error_from_cache_on_write_operations;
-    extern const SettingsBool skip_download_if_exceeds_query_cache;
+    extern const SettingsBool filesystem_cache_skip_download_if_exceeds_per_query_cache_write_limit;
     extern const SettingsBool s3_allow_parallel_part_upload;
     extern const SettingsBool use_page_cache_for_disks_without_file_cache;
     extern const SettingsUInt64 use_structure_from_insertion_table_in_table_functions;
@@ -5751,9 +5752,10 @@ ReadSettings Context::getReadSettings() const
     res.filesystem_cache_allow_background_download_for_metadata_files_in_packed_storage
         = settings_ref[Setting::filesystem_cache_enable_background_download_for_metadata_files_in_packed_storage];
     res.filesystem_cache_allow_background_download_during_fetch = settings_ref[Setting::filesystem_cache_enable_background_download_during_fetch];
+    res.filesystem_cache_prefer_bigger_buffer_size = settings_ref[Setting::filesystem_cache_prefer_bigger_buffer_size];
 
     res.filesystem_cache_max_download_size = settings_ref[Setting::filesystem_cache_max_download_size];
-    res.skip_download_if_exceeds_query_cache = settings_ref[Setting::skip_download_if_exceeds_query_cache];
+    res.filesystem_cache_skip_download_if_exceeds_per_query_cache_write_limit = settings_ref[Setting::filesystem_cache_skip_download_if_exceeds_per_query_cache_write_limit];
 
     res.page_cache = getPageCache();
     res.use_page_cache_for_disks_without_file_cache = settings_ref[Setting::use_page_cache_for_disks_without_file_cache];

src/Interpreters/ExecuteScalarSubqueriesVisitor.cpp

@@ -5,9 +5,11 @@
 #include <Core/Settings.h>
 #include <DataTypes/DataTypeNullable.h>
 #include <DataTypes/DataTypeTuple.h>
+#include <IO/WriteBufferFromString.h>
 #include <IO/WriteHelpers.h>
 #include <Interpreters/Context.h>
 #include <Interpreters/InterpreterSelectWithUnionQuery.h>
+#include <Interpreters/ProcessorsProfileLog.h>
 #include <Interpreters/addTypeConversionToAST.h>
 #include <Interpreters/misc.h>
 #include <Parsers/ASTExpressionList.h>
@@ -19,9 +21,8 @@
 #include <Parsers/ASTWithElement.h>
 #include <Parsers/queryToString.h>
 #include <Processors/Executors/PullingAsyncPipelineExecutor.h>
-#include <Common/ProfileEvents.h>
 #include <Common/FieldVisitorToString.h>
-#include <IO/WriteBufferFromString.h>
+#include <Common/ProfileEvents.h>
 
 namespace ProfileEvents
 {
@@ -67,6 +68,18 @@ bool ExecuteScalarSubqueriesMatcher::needChildVisit(ASTPtr & node, const ASTPtr
             return false;
     }
 
+    if (auto * tables = node->as<ASTTablesInSelectQueryElement>())
+    {
+        /// Contrary to what's said in the code block above, ARRAY JOIN needs to resolve the subquery if possible
+        /// and assign an alias for 02367_optimize_trivial_count_with_array_join to pass. Otherwise it will fail in
+        /// ArrayJoinedColumnsVisitor (`No alias for non-trivial value in ARRAY JOIN: _a`)
+        /// This looks 100% as a incomplete code working on top of a bug, but this code has already been made obsolete
+        /// by the new analyzer, so it's an inconvenience we can live with until we deprecate it.
+        if (child == tables->array_join)
+            return true;
+        return false;
+    }
+
     return true;
 }
 
@@ -246,6 +259,8 @@ void ExecuteScalarSubqueriesMatcher::visit(const ASTSubquery & subquery, ASTPtr
 
             if (tmp_block.rows() != 0)
                 throw Exception(ErrorCodes::INCORRECT_RESULT_OF_SCALAR_SUBQUERY, "Scalar subquery returned more than one row");
+
+            logProcessorProfile(data.getContext(), io.pipeline.getProcessors());
         }
 
         block = materializeBlock(block);

src/Interpreters/InterpreterCreateQuery.cpp

@@ -98,6 +98,9 @@
 namespace CurrentMetrics
 {
     extern const Metric AttachedTable;
+    extern const Metric AttachedReplicatedTable;
+    extern const Metric AttachedDictionary;
+    extern const Metric AttachedView;
 }
 
 namespace DB
@@ -145,7 +148,10 @@ namespace ServerSetting
 {
     extern const ServerSettingsBool ignore_empty_sql_security_in_create_view_query;
     extern const ServerSettingsUInt64 max_database_num_to_throw;
+    extern const ServerSettingsUInt64 max_dictionary_num_to_throw;
     extern const ServerSettingsUInt64 max_table_num_to_throw;
+    extern const ServerSettingsUInt64 max_replicated_table_num_to_throw;
+    extern const ServerSettingsUInt64 max_view_num_to_throw;
 }
 
 namespace ErrorCodes
@@ -1912,16 +1918,8 @@ bool InterpreterCreateQuery::doCreateTable(ASTCreateQuery & create,
         }
     }
 
-    UInt64 table_num_limit = getContext()->getGlobalContext()->getServerSettings()[ServerSetting::max_table_num_to_throw];
-    if (table_num_limit > 0 && !internal)
-    {
-        UInt64 table_count = CurrentMetrics::get(CurrentMetrics::AttachedTable);
-        if (table_count >= table_num_limit)
-            throw Exception(ErrorCodes::TOO_MANY_TABLES,
-                            "Too many tables. "
-                            "The limit (server configuration parameter `max_table_num_to_throw`) is set to {}, the current number of tables is {}",
-                            table_num_limit, table_count);
-    }
+    if (!internal)
+        throwIfTooManyEntities(create, res);
 
     database->createTable(getContext(), create.getTable(), res, query_ptr);
 
@@ -1948,6 +1946,30 @@ bool InterpreterCreateQuery::doCreateTable(ASTCreateQuery & create,
 }
 
 
+void InterpreterCreateQuery::throwIfTooManyEntities(ASTCreateQuery & create, StoragePtr storage) const
+{
+    auto check_and_throw = [&](auto setting, CurrentMetrics::Metric metric, String setting_name, String entity_name)
+        {
+            UInt64 num_limit = getContext()->getGlobalContext()->getServerSettings()[setting];
+            UInt64 attached_count = CurrentMetrics::get(metric);
+            if (num_limit > 0 && attached_count >= num_limit)
+                throw Exception(ErrorCodes::TOO_MANY_TABLES,
+                                "Too many {}. "
+                                "The limit (server configuration parameter `{}`) is set to {}, the current number is {}",
+                                entity_name, setting_name, num_limit, attached_count);
+        };
+
+    if (auto * replicated_storage = typeid_cast<StorageReplicatedMergeTree *>(storage.get()))
+        check_and_throw(ServerSetting::max_replicated_table_num_to_throw, CurrentMetrics::AttachedReplicatedTable, "max_replicated_table_num_to_throw", "replicated tables");
+    else if (create.is_dictionary)
+        check_and_throw(ServerSetting::max_dictionary_num_to_throw, CurrentMetrics::AttachedDictionary, "max_dictionary_num_to_throw", "dictionaries");
+    else if (create.isView())
+        check_and_throw(ServerSetting::max_view_num_to_throw, CurrentMetrics::AttachedView, "max_view_num_to_throw", "views");
+    else
+        check_and_throw(ServerSetting::max_table_num_to_throw, CurrentMetrics::AttachedTable, "max_table_num_to_throw", "tables");
+}
+
+
 BlockIO InterpreterCreateQuery::doCreateOrReplaceTable(ASTCreateQuery & create,
                                                        const InterpreterCreateQuery::TableProperties & properties, LoadingStrictnessLevel mode)
 {

src/Interpreters/InterpreterCreateQuery.h

@@ -122,6 +122,8 @@ private:
 
     BlockIO executeQueryOnCluster(ASTCreateQuery & create);
 
+    void throwIfTooManyEntities(ASTCreateQuery & create, StoragePtr storage) const;
+
     ASTPtr query_ptr;
 
     /// Skip safety threshold when loading tables.

src/Interpreters/MutationsInterpreter.cpp

@@ -53,6 +53,7 @@ namespace Setting
     extern const SettingsBool allow_nondeterministic_mutations;
     extern const SettingsUInt64 max_block_size;
     extern const SettingsBool use_concurrency_control;
+    extern const SettingsBool validate_mutation_query;
 }
 
 namespace MergeTreeSetting
@@ -1386,6 +1387,18 @@ void MutationsInterpreter::validate()
         }
     }
 
+    // Make sure the mutation query is valid
+    if (context->getSettingsRef()[Setting::validate_mutation_query])
+    {
+        if (context->getSettingsRef()[Setting::allow_experimental_analyzer])
+            prepareQueryAffectedQueryTree(commands, source.getStorage(), context);
+        else
+        {
+            ASTPtr select_query = prepareQueryAffectedAST(commands, source.getStorage(), context);
+            InterpreterSelectQuery(select_query, context, source.getStorage(), metadata_snapshot);
+        }
+    }
+
     QueryPlan plan;
 
     initQueryPlan(stages.front(), plan);

src/Interpreters/PreparedSets.cpp

@@ -1,21 +1,22 @@
 #include <chrono>
 #include <variant>
-#include <Interpreters/PreparedSets.h>
-#include <Interpreters/Set.h>
-#include <Interpreters/InterpreterSelectWithUnionQuery.h>
-#include <IO/Operators.h>
-#include <Common/logger_useful.h>
+#include <Core/Block.h>
 #include <Core/Settings.h>
-#include <Processors/QueryPlan/CreatingSetsStep.h>
+#include <IO/Operators.h>
+#include <Interpreters/InterpreterSelectWithUnionQuery.h>
+#include <Interpreters/PreparedSets.h>
+#include <Interpreters/ProcessorsProfileLog.h>
+#include <Interpreters/Set.h>
 #include <Processors/Executors/CompletedPipelineExecutor.h>
 #include <Processors/QueryPlan/BuildQueryPipelineSettings.h>
+#include <Processors/QueryPlan/CreatingSetsStep.h>
 #include <Processors/QueryPlan/Optimizations/QueryPlanOptimizationSettings.h>
+#include <Processors/QueryPlan/QueryPlan.h>
 #include <Processors/Sinks/EmptySink.h>
 #include <Processors/Sinks/NullSink.h>
-#include <Processors/QueryPlan/QueryPlan.h>
-#include <Core/Block.h>
 #include <QueryPipeline/QueryPipelineBuilder.h>
 #include <QueryPipeline/SizeLimits.h>
+#include <Common/logger_useful.h>
 
 namespace DB
 {
@@ -239,6 +240,8 @@ SetPtr FutureSetFromSubquery::buildOrderedSetInplace(const ContextPtr & context)
     if (!set_and_key->set->isCreated())
         return nullptr;
 
+    logProcessorProfile(context, pipeline.getProcessors());
+
     return set_and_key->set;
 }
 

src/Interpreters/ProcessorsProfileLog.cpp

@@ -1,5 +1,6 @@
 #include <Interpreters/ProcessorsProfileLog.h>
 
+#include <Core/Settings.h>
 #include <DataTypes/DataTypeArray.h>
 #include <DataTypes/DataTypeDate.h>
 #include <DataTypes/DataTypeDateTime.h>
@@ -8,16 +9,19 @@
 #include <DataTypes/DataTypeNullable.h>
 #include <DataTypes/DataTypeString.h>
 #include <DataTypes/DataTypesNumber.h>
+#include <Interpreters/Context.h>
 #include <base/getFQDNOrHostName.h>
 #include <Common/ClickHouseRevision.h>
 #include <Common/DateLUTImpl.h>
-#include <Common/logger_useful.h>
-
-#include <array>
 
 namespace DB
 {
 
+namespace Setting
+{
+extern const SettingsBool log_processors_profiles;
+}
+
 ColumnsDescription ProcessorProfileLogElement::getColumnsDescription()
 {
     return ColumnsDescription
@@ -81,5 +85,57 @@ void ProcessorProfileLogElement::appendToBlock(MutableColumns & columns) const
     columns[i++]->insert(output_bytes);
 }
 
+void logProcessorProfile(ContextPtr context, const Processors & processors)
+{
+    const Settings & settings = context->getSettingsRef();
+    if (settings[Setting::log_processors_profiles])
+    {
+        if (auto processors_profile_log = context->getProcessorsProfileLog())
+        {
+            ProcessorProfileLogElement processor_elem;
 
+            const auto time_now = std::chrono::system_clock::now();
+            processor_elem.event_time = timeInSeconds(time_now);
+            processor_elem.event_time_microseconds = timeInMicroseconds(time_now);
+            processor_elem.initial_query_id = context->getInitialQueryId();
+            processor_elem.query_id = context->getCurrentQueryId();
+
+            auto get_proc_id = [](const IProcessor & proc) -> UInt64 { return reinterpret_cast<std::uintptr_t>(&proc); };
+
+            for (const auto & processor : processors)
+            {
+                std::vector<UInt64> parents;
+                for (const auto & port : processor->getOutputs())
+                {
+                    if (!port.isConnected())
+                        continue;
+                    const IProcessor & next = port.getInputPort().getProcessor();
+                    parents.push_back(get_proc_id(next));
+                }
+
+                processor_elem.id = get_proc_id(*processor);
+                processor_elem.parent_ids = std::move(parents);
+
+                processor_elem.plan_step = reinterpret_cast<std::uintptr_t>(processor->getQueryPlanStep());
+                processor_elem.plan_step_name = processor->getPlanStepName();
+                processor_elem.plan_step_description = processor->getPlanStepDescription();
+                processor_elem.plan_group = processor->getQueryPlanStepGroup();
+
+                processor_elem.processor_name = processor->getName();
+
+                processor_elem.elapsed_us = static_cast<UInt64>(processor->getElapsedNs() / 1000U);
+                processor_elem.input_wait_elapsed_us = static_cast<UInt64>(processor->getInputWaitElapsedNs() / 1000U);
+                processor_elem.output_wait_elapsed_us = static_cast<UInt64>(processor->getOutputWaitElapsedNs() / 1000U);
+
+                auto stats = processor->getProcessorDataStats();
+                processor_elem.input_rows = stats.input_rows;
+                processor_elem.input_bytes = stats.input_bytes;
+                processor_elem.output_rows = stats.output_rows;
+                processor_elem.output_bytes = stats.output_bytes;
+
+                processors_profile_log->add(processor_elem);
+            }
+        }
+    }
+}
 }

src/Interpreters/ProcessorsProfileLog.h

@@ -50,4 +50,5 @@ public:
     using SystemLog<ProcessorProfileLogElement>::SystemLog;
 };
 
+void logProcessorProfile(ContextPtr context, const Processors & processors);
 }

src/Interpreters/QueryMetricLog.cpp

@@ -1,6 +1,7 @@
 #include <base/getFQDNOrHostName.h>
 #include <Common/DateLUT.h>
 #include <Common/DateLUTImpl.h>
+#include <Common/LockGuard.h>
 #include <DataTypes/DataTypeDate.h>
 #include <DataTypes/DataTypeDateTime.h>
 #include <DataTypes/DataTypeDateTime64.h>
@@ -16,7 +17,6 @@
 
 #include <chrono>
 #include <fmt/chrono.h>
-#include <mutex>
 
 
 namespace DB
@@ -24,6 +24,15 @@ namespace DB
 
 static auto logger = getLogger("QueryMetricLog");
 
+String timePointToString(QueryMetricLog::TimePoint time)
+{
+    /// fmtlib supports subsecond formatting in 10.0.0. We're in 9.1.0, so we need to add the milliseconds ourselves.
+    auto seconds = std::chrono::time_point_cast<std::chrono::seconds>(time);
+    auto microseconds = std::chrono::duration_cast<std::chrono::microseconds>(time - seconds).count();
+
+    return fmt::format("{:%Y.%m.%d %H:%M:%S}.{:06}", seconds, microseconds);
+}
+
 ColumnsDescription QueryMetricLogElement::getColumnsDescription()
 {
     ColumnsDescription result;
@@ -87,36 +96,73 @@ void QueryMetricLog::shutdown()
     Base::shutdown();
 }
 
-void QueryMetricLog::startQuery(const String & query_id, TimePoint start_time, UInt64 interval_milliseconds)
+void QueryMetricLog::collectMetric(const ProcessList & process_list, String query_id)
 {
-    QueryMetricLogStatus status;
-    status.interval_milliseconds = interval_milliseconds;
-    status.next_collect_time = start_time + std::chrono::milliseconds(interval_milliseconds);
-
-    auto context = getContext();
-    const auto & process_list = context->getProcessList();
-    status.task = context->getSchedulePool().createTask("QueryMetricLog", [this, &process_list, query_id] {
     auto current_time = std::chrono::system_clock::now();
     const auto query_info = process_list.getQueryInfo(query_id, false, true, false);
     if (!query_info)
     {
+        /// TODO: remove trace before 24.11 release after checking everything is fine on the CI
         LOG_TRACE(logger, "Query {} is not running anymore, so we couldn't get its QueryStatusInfo", query_id);
         return;
     }
 
-        auto elem = createLogMetricElement(query_id, *query_info, current_time);
+    LockGuard global_lock(queries_mutex);
+    auto it = queries.find(query_id);
+
+    /// The query might have finished while the scheduled task is running.
+    if (it == queries.end())
+    {
+        global_lock.unlock();
+        /// TODO: remove trace before 24.11 release after checking everything is fine on the CI
+        LOG_TRACE(logger, "Query {} not found in the list. Finished while this collecting task was running", query_id);
+        return;
+    }
+
+    auto & query_status = it->second;
+    if (!query_status.mutex)
+    {
+        global_lock.unlock();
+        /// TODO: remove trace before 24.11 release after checking everything is fine on the CI
+        LOG_TRACE(logger, "Query {} finished while this collecting task was running", query_id);
+        return;
+    }
+
+    LockGuard query_lock(query_status.getMutex());
+    global_lock.unlock();
+
+    auto elem = query_status.createLogMetricElement(query_id, *query_info, current_time);
     if (elem)
         add(std::move(elem.value()));
+}
+
+/// We use TSA_NO_THREAD_SAFETY_ANALYSIS to prevent TSA complaining that we're modifying the query_status fields
+/// without locking the mutex. Since we're building it from scratch, there's no harm in not holding it.
+/// If we locked it to make TSA happy, TSAN build would falsely complain about
+///     lock-order-inversion (potential deadlock)
+/// which is not a real issue since QueryMetricLogStatus's mutex cannot be locked by anything else
+/// until we add it to the queries map.
+void QueryMetricLog::startQuery(const String & query_id, TimePoint start_time, UInt64 interval_milliseconds) TSA_NO_THREAD_SAFETY_ANALYSIS
+{
+    QueryMetricLogStatus query_status;
+    QueryMetricLogStatusInfo & info = query_status.info;
+    info.interval_milliseconds = interval_milliseconds;
+    info.next_collect_time = start_time;
+
+    auto context = getContext();
+    const auto & process_list = context->getProcessList();
+    info.task = context->getSchedulePool().createTask("QueryMetricLog", [this, &process_list, query_id] {
+        collectMetric(process_list, query_id);
     });
 
-    std::lock_guard lock(queries_mutex);
-    status.task->scheduleAfter(interval_milliseconds);
-    queries.emplace(query_id, std::move(status));
+    LockGuard global_lock(queries_mutex);
+    query_status.scheduleNext(query_id);
+    queries.emplace(query_id, std::move(query_status));
 }
 
 void QueryMetricLog::finishQuery(const String & query_id, TimePoint finish_time, QueryStatusInfoPtr query_info)
 {
-    std::unique_lock lock(queries_mutex);
+    LockGuard global_lock(queries_mutex);
     auto it = queries.find(query_id);
 
     /// finishQuery may be called from logExceptionBeforeStart when the query has not even started
@@ -124,9 +170,19 @@ void QueryMetricLog::finishQuery(const String & query_id, TimePoint finish_time,
     if (it == queries.end())
         return;
 
+    auto & query_status = it->second;
+    decltype(query_status.mutex) query_mutex;
+    LockGuard query_lock(query_status.getMutex());
+
+    /// Move the query mutex here so that we hold it until the end, after removing the query from queries.
+    query_mutex = std::move(query_status.mutex);
+    query_status.mutex = {};
+
+    global_lock.unlock();
+
     if (query_info)
     {
-        auto elem = createLogMetricElement(query_id, *query_info, finish_time, false);
+        auto elem = query_status.createLogMetricElement(query_id, *query_info, finish_time, false);
         if (elem)
             add(std::move(elem.value()));
     }
@@ -139,51 +195,58 @@ void QueryMetricLog::finishQuery(const String & query_id, TimePoint finish_time,
     /// that order.
     {
         /// Take ownership of the task so that we can destroy it in this scope after unlocking `queries_mutex`.
-        auto task = std::move(it->second.task);
+        auto task = std::move(query_status.info.task);
 
         /// Build an empty task for the old task to make sure it does not lock any mutex on its destruction.
-        it->second.task = {};
+        query_status.info.task = {};
+        query_lock.unlock();
 
+        global_lock.lock();
         queries.erase(query_id);
 
         /// Ensure `queries_mutex` is unlocked before calling task's destructor at the end of this
         /// scope which will lock `exec_mutex`.
-        lock.unlock();
+        global_lock.unlock();
     }
 }
 
-std::optional<QueryMetricLogElement> QueryMetricLog::createLogMetricElement(const String & query_id, const QueryStatusInfo & query_info, TimePoint query_info_time, bool schedule_next)
+void QueryMetricLogStatus::scheduleNext(String query_id)
 {
-    /// fmtlib supports subsecond formatting in 10.0.0. We're in 9.1.0, so we need to add the milliseconds ourselves.
-    auto seconds = std::chrono::time_point_cast<std::chrono::seconds>(query_info_time);
-    auto microseconds = std::chrono::duration_cast<std::chrono::microseconds>(query_info_time - seconds).count();
-    LOG_DEBUG(logger, "Collecting query_metric_log for query {} with QueryStatusInfo from {:%Y.%m.%d %H:%M:%S}.{:06}. Schedule next: {}", query_id, seconds, microseconds, schedule_next);
-
-    std::unique_lock lock(queries_mutex);
-    auto query_status_it = queries.find(query_id);
-
-    /// The query might have finished while the scheduled task is running.
-    if (query_status_it == queries.end())
+    info.next_collect_time += std::chrono::milliseconds(info.interval_milliseconds);
+    const auto now = std::chrono::system_clock::now();
+    if (info.next_collect_time > now)
     {
-        lock.unlock();
-        LOG_TRACE(logger, "Query {} finished already while this collecting task was running", query_id);
-        return {};
+        const auto wait_time = std::chrono::duration_cast<std::chrono::milliseconds>(info.next_collect_time - now).count();
+        info.task->scheduleAfter(wait_time);
+    }
+    else
+    {
+        LOG_TRACE(logger, "The next collecting task for query {} should have already run at {}. Scheduling it right now",
+            query_id, timePointToString(info.next_collect_time));
+        info.task->schedule();
+    }
 }
 
-    auto & query_status = query_status_it->second;
-    if (query_info_time <= query_status.last_collect_time)
+std::optional<QueryMetricLogElement> QueryMetricLogStatus::createLogMetricElement(const String & query_id, const QueryStatusInfo & query_info, TimePoint query_info_time, bool schedule_next)
 {
-        lock.unlock();
+    /// TODO: remove trace before 24.11 release after checking everything is fine on the CI
+    LOG_TRACE(logger, "Collecting query_metric_log for query {} and interval {} ms with QueryStatusInfo from {}. Next collection time: {}",
+        query_id, info.interval_milliseconds, timePointToString(query_info_time),
+        schedule_next ? timePointToString(info.next_collect_time + std::chrono::milliseconds(info.interval_milliseconds)) : "finished");
+
+    if (query_info_time <= info.last_collect_time)
+    {
+        /// TODO: remove trace before 24.11 release after checking everything is fine on the CI
         LOG_TRACE(logger, "Query {} has a more recent metrics collected. Skipping this one", query_id);
         return {};
     }
 
-    query_status.last_collect_time = query_info_time;
+    info.last_collect_time = query_info_time;
 
     QueryMetricLogElement elem;
     elem.event_time = timeInSeconds(query_info_time);
     elem.event_time_microseconds = timeInMicroseconds(query_info_time);
-    elem.query_id = query_status_it->first;
+    elem.query_id = query_id;
     elem.memory_usage = query_info.memory_usage > 0 ? query_info.memory_usage : 0;
     elem.peak_memory_usage = query_info.peak_memory_usage > 0 ? query_info.peak_memory_usage : 0;
 
@@ -192,7 +255,7 @@ std::optional<QueryMetricLogElement> QueryMetricLog::createLogMetricElement(cons
         for (ProfileEvents::Event i = ProfileEvents::Event(0), end = ProfileEvents::end(); i < end; ++i)
         {
             const auto & new_value = (*(query_info.profile_counters))[i];
-            auto & old_value = query_status.last_profile_events[i];
+            auto & old_value = info.last_profile_events[i];
 
             /// Profile event counters are supposed to be monotonic. However, at least the `NetworkReceiveBytes` can be inaccurate.
             /// So, since in the future the counter should always have a bigger value than in the past, we skip this event.
@@ -208,16 +271,13 @@ std::optional<QueryMetricLogElement> QueryMetricLog::createLogMetricElement(cons
     }
     else
     {
-        LOG_TRACE(logger, "Query {} has no profile counters", query_id);
+        /// TODO: remove trace before 24.11 release after checking everything is fine on the CI
+        LOG_DEBUG(logger, "Query {} has no profile counters", query_id);
         elem.profile_events = std::vector<ProfileEvents::Count>(ProfileEvents::end());
     }
 
     if (schedule_next)
-    {
-        query_status.next_collect_time += std::chrono::milliseconds(query_status.interval_milliseconds);
-        const auto wait_time = std::chrono::duration_cast<std::chrono::milliseconds>(query_status.next_collect_time - std::chrono::system_clock::now()).count();
-        query_status.task->scheduleAfter(wait_time);
-    }
+        scheduleNext(query_id);
 
     return elem;
 }

src/Interpreters/QueryMetricLog.h

@@ -1,5 +1,6 @@
 #pragma once
 
+#include <base/defines.h>
 #include <Common/ProfileEvents.h>
 #include <Common/CurrentMetrics.h>
 #include <Core/BackgroundSchedulePool.h>
@@ -11,11 +12,17 @@
 
 #include <chrono>
 #include <ctime>
+#include <mutex>
 
 
 namespace DB
 {
 
+namespace ErrorCodes
+{
+    extern const int LOGICAL_ERROR;
+};
+
 /** QueryMetricLogElement is a log of query metric values measured at regular time interval.
   */
 
@@ -34,7 +41,7 @@ struct QueryMetricLogElement
     void appendToBlock(MutableColumns & columns) const;
 };
 
-struct QueryMetricLogStatus
+struct QueryMetricLogStatusInfo
 {
     UInt64 interval_milliseconds;
     std::chrono::system_clock::time_point last_collect_time;
@@ -43,24 +50,47 @@ struct QueryMetricLogStatus
     BackgroundSchedulePool::TaskHolder task;
 };
 
+struct QueryMetricLogStatus
+{
+    using TimePoint = std::chrono::system_clock::time_point;
+    using Mutex = std::mutex;
+
+    QueryMetricLogStatusInfo info TSA_GUARDED_BY(getMutex());
+
+    /// We need to be able to move it for the hash map, so we need to add an indirection here.
+    std::unique_ptr<Mutex> mutex = std::make_unique<Mutex>();
+
+    /// Return a reference to the mutex, used for Thread Sanitizer annotations.
+    Mutex & getMutex() const TSA_RETURN_CAPABILITY(mutex)
+    {
+        if (!mutex)
+            throw Exception(ErrorCodes::LOGICAL_ERROR, "Mutex cannot be NULL");
+        return *mutex;
+    }
+
+    void scheduleNext(String query_id) TSA_REQUIRES(getMutex());
+    std::optional<QueryMetricLogElement> createLogMetricElement(const String & query_id, const QueryStatusInfo & query_info, TimePoint query_info_time, bool schedule_next = true) TSA_REQUIRES(getMutex());
+};
+
 class QueryMetricLog : public SystemLog<QueryMetricLogElement>
 {
     using SystemLog<QueryMetricLogElement>::SystemLog;
-    using TimePoint = std::chrono::system_clock::time_point;
     using Base = SystemLog<QueryMetricLogElement>;
 
 public:
+    using TimePoint = std::chrono::system_clock::time_point;
+
     void shutdown() final;
 
-    // Both startQuery and finishQuery are called from the thread that executes the query
+    /// Both startQuery and finishQuery are called from the thread that executes the query.
     void startQuery(const String & query_id, TimePoint start_time, UInt64 interval_milliseconds);
     void finishQuery(const String & query_id, TimePoint finish_time, QueryStatusInfoPtr query_info = nullptr);
 
 private:
-    std::optional<QueryMetricLogElement> createLogMetricElement(const String & query_id, const QueryStatusInfo & query_info, TimePoint query_info_time, bool schedule_next = true);
+    void collectMetric(const ProcessList & process_list, String query_id);
 
-    std::recursive_mutex queries_mutex;
-    std::unordered_map<String, QueryMetricLogStatus> queries;
+    std::mutex queries_mutex;
+    std::unordered_map<String, QueryMetricLogStatus> queries TSA_GUARDED_BY(queries_mutex);
 };
 
 }

src/Interpreters/QueryNormalizer.cpp

@@ -161,7 +161,13 @@ void QueryNormalizer::visit(ASTTablesInSelectQueryElement & node, const ASTPtr &
     {
         auto & join = node.table_join->as<ASTTableJoin &>();
         if (join.on_expression)
+        {
+            ASTPtr original_on_expression = join.on_expression;
             visit(join.on_expression, data);
+            if (join.on_expression != original_on_expression)
+                join.children = { join.on_expression };
+        }
+
     }
 }
 

src/Interpreters/RewriteArrayExistsFunctionVisitor.cpp

@@ -6,6 +6,12 @@
 
 namespace DB
 {
+
+namespace ErrorCode
+{
+extern const int LOGICAL_ERROR;
+}
+
 void RewriteArrayExistsFunctionMatcher::visit(ASTPtr & ast, Data & data)
 {
     if (auto * func = ast->as<ASTFunction>())
@@ -20,20 +26,20 @@ void RewriteArrayExistsFunctionMatcher::visit(ASTPtr & ast, Data & data)
         if (join->using_expression_list)
         {
             auto * it = std::find(join->children.begin(), join->children.end(), join->using_expression_list);
+            if (it == join->children.end())
+                throw Exception(ErrorCodes::LOGICAL_ERROR, "Could not find join->using_expression_list in '{}'", join->formatForLogging());
 
             visit(join->using_expression_list, data);
-
-            if (it && *it != join->using_expression_list)
             *it = join->using_expression_list;
         }
 
         if (join->on_expression)
         {
             auto * it = std::find(join->children.begin(), join->children.end(), join->on_expression);
+            if (it == join->children.end())
+                throw Exception(ErrorCodes::LOGICAL_ERROR, "Could not find join->on_expression in '{}'", join->formatForLogging());
 
             visit(join->on_expression, data);
-
-            if (it && *it != join->on_expression)
             *it = join->on_expression;
         }
     }

src/Interpreters/executeQuery.cpp

@@ -120,7 +120,6 @@ namespace Setting
     extern const SettingsOverflowMode join_overflow_mode;
     extern const SettingsString log_comment;
     extern const SettingsBool log_formatted_queries;
-    extern const SettingsBool log_processors_profiles;
     extern const SettingsBool log_profile_events;
     extern const SettingsUInt64 log_queries_cut_to_length;
     extern const SettingsBool log_queries;
@@ -510,6 +509,7 @@ void logQueryFinish(
 
         auto time_now = std::chrono::system_clock::now();
         QueryStatusInfo info = process_list_elem->getInfo(true, settings[Setting::log_profile_events]);
+        logQueryMetricLogFinish(context, internal, elem.client_info.current_query_id, time_now, std::make_shared<QueryStatusInfo>(info));
         elem.type = QueryLogElementType::QUERY_FINISH;
 
         addStatusInfoToQueryLogElement(elem, info, query_ast, context);
@@ -555,53 +555,8 @@ void logQueryFinish(
             if (auto query_log = context->getQueryLog())
                 query_log->add(elem);
         }
-        if (settings[Setting::log_processors_profiles])
-        {
-            if (auto processors_profile_log = context->getProcessorsProfileLog())
-            {
-                ProcessorProfileLogElement processor_elem;
-                processor_elem.event_time = elem.event_time;
-                processor_elem.event_time_microseconds = elem.event_time_microseconds;
-                processor_elem.initial_query_id = elem.client_info.initial_query_id;
-                processor_elem.query_id = elem.client_info.current_query_id;
 
-                auto get_proc_id = [](const IProcessor & proc) -> UInt64 { return reinterpret_cast<std::uintptr_t>(&proc); };
-
-                for (const auto & processor : query_pipeline.getProcessors())
-                {
-                    std::vector<UInt64> parents;
-                    for (const auto & port : processor->getOutputs())
-                    {
-                        if (!port.isConnected())
-                            continue;
-                        const IProcessor & next = port.getInputPort().getProcessor();
-                        parents.push_back(get_proc_id(next));
-                    }
-
-                    processor_elem.id = get_proc_id(*processor);
-                    processor_elem.parent_ids = std::move(parents);
-
-                    processor_elem.plan_step = reinterpret_cast<std::uintptr_t>(processor->getQueryPlanStep());
-                    processor_elem.plan_step_name = processor->getPlanStepName();
-                    processor_elem.plan_step_description = processor->getPlanStepDescription();
-                    processor_elem.plan_group = processor->getQueryPlanStepGroup();
-
-                    processor_elem.processor_name = processor->getName();
-
-                    processor_elem.elapsed_us = static_cast<UInt64>(processor->getElapsedNs() / 1000U);
-                    processor_elem.input_wait_elapsed_us = static_cast<UInt64>(processor->getInputWaitElapsedNs() / 1000U);
-                    processor_elem.output_wait_elapsed_us = static_cast<UInt64>(processor->getOutputWaitElapsedNs() / 1000U);
-
-                    auto stats = processor->getProcessorDataStats();
-                    processor_elem.input_rows = stats.input_rows;
-                    processor_elem.input_bytes = stats.input_bytes;
-                    processor_elem.output_rows = stats.output_rows;
-                    processor_elem.output_bytes = stats.output_bytes;
-
-                    processors_profile_log->add(processor_elem);
-                }
-            }
-        }
+        logProcessorProfile(context, query_pipeline.getProcessors());
 
         logQueryMetricLogFinish(context, internal, elem.client_info.current_query_id, time_now, std::make_shared<QueryStatusInfo>(info));
     }
@@ -673,6 +628,7 @@ void logQueryException(
     {
         elem.query_duration_ms = start_watch.elapsedMilliseconds();
     }
+    logQueryMetricLogFinish(context, internal, elem.client_info.current_query_id, time_now, info);
 
     elem.query_cache_usage = QueryCache::Usage::None;
 
@@ -702,8 +658,6 @@ void logQueryException(
         query_span->addAttribute("clickhouse.exception_code", elem.exception_code);
         query_span->finish();
     }
-
-    logQueryMetricLogFinish(context, internal, elem.client_info.current_query_id, time_now, info);
 }
 
 void logExceptionBeforeStart(
@@ -757,6 +711,8 @@ void logExceptionBeforeStart(
 
     elem.client_info = context->getClientInfo();
 
+    logQueryMetricLogFinish(context, false, elem.client_info.current_query_id, std::chrono::system_clock::now(), nullptr);
+
     elem.log_comment = settings[Setting::log_comment];
     if (elem.log_comment.size() > settings[Setting::max_query_size])
         elem.log_comment.resize(settings[Setting::max_query_size]);
@@ -803,8 +759,6 @@ void logExceptionBeforeStart(
             ProfileEvents::increment(ProfileEvents::FailedInsertQuery);
         }
     }
-
-    logQueryMetricLogFinish(context, false, elem.client_info.current_query_id, std::chrono::system_clock::now(), nullptr);
 }
 
 void validateAnalyzerSettings(ASTPtr ast, bool context_value)

src/Processors/Formats/Impl/NativeORCBlockInputFormat.cpp

@@ -1534,6 +1534,8 @@ static ColumnWithTypeAndName readColumnWithDateData(
     column_data.reserve(orc_int_column->numElements);
 
     for (size_t i = 0; i < orc_int_column->numElements; ++i)
+    {
+        if (!orc_int_column->hasNulls || orc_int_column->notNull[i])
         {
             Int32 days_num = static_cast<Int32>(orc_int_column->data[i]);
             if (check_date_range && (days_num > DATE_LUT_MAX_EXTEND_DAY_NUM || days_num < -DAYNUM_OFFSET_EPOCH))
@@ -1545,6 +1547,12 @@ static ColumnWithTypeAndName readColumnWithDateData(
 
             column_data.push_back(days_num);
         }
+        else
+        {
+            /// ORC library doesn't guarantee that orc_int_column->data[i] is initialized to zero when orc_int_column->notNull[i] is false since https://github.com/ClickHouse/ClickHouse/pull/69473
+            column_data.push_back(0);
+        }
+    }
 
     return {std::move(internal_column), internal_type, column_name};
 }

src/Processors/Merges/Algorithms/IMergingAlgorithmWithDelayedChunk.cpp

@@ -1,5 +1,5 @@
 #include <Processors/Merges/Algorithms/IMergingAlgorithmWithDelayedChunk.h>
-#include <Processors/Merges/Algorithms/MergeTreePartLevelInfo.h>
+#include <Processors/Merges/Algorithms/MergeTreeReadInfo.h>
 
 
 namespace DB