Merge remote-tracking branch 'origin/master' into pr-local-plan

.editorconfig

@@ -19,3 +19,7 @@ charset = utf-8
 indent_style = space
 indent_size = 4
 trim_trailing_whitespace = true
+
+# Some SQL results have trailing whitespace which is removed by IDEs
+[tests/queries/**.reference]
+trim_trailing_whitespace = false

.github/PULL_REQUEST_TEMPLATE.md

@@ -48,19 +48,17 @@ At a minimum, the following information should be added (but add more as needed)
 - [ ] <!---ci_include_stateful--> Allow: Stateful tests
 - [ ] <!---ci_include_integration--> Allow: Integration Tests
 - [ ] <!---ci_include_performance--> Allow: Performance tests
+- [ ] <!---ci_set_normal_builds--> Allow: Normal Builds
+- [ ] <!---ci_set_special_builds--> Allow: Special Builds
 - [ ] <!---ci_set_non_required--> Allow: All NOT Required Checks
 - [ ] <!---batch_0_1--> Allow: batch 1, 2 for multi-batch jobs
 - [ ] <!---batch_2_3--> Allow: batch 3, 4, 5, 6 for multi-batch jobs
 ---
 - [ ] <!---ci_exclude_style--> Exclude: Style check
 - [ ] <!---ci_exclude_fast--> Exclude: Fast test
-- [ ] <!---ci_exclude_integration--> Exclude: Integration Tests
-- [ ] <!---ci_exclude_stateless--> Exclude: Stateless tests
-- [ ] <!---ci_exclude_stateful--> Exclude: Stateful tests
-- [ ] <!---ci_exclude_performance--> Exclude: Performance tests
 - [ ] <!---ci_exclude_asan--> Exclude: All with ASAN
-- [ ] <!---ci_exclude_aarch64--> Exclude: All with Aarch64
 - [ ] <!---ci_exclude_tsan|msan|ubsan|coverage--> Exclude: All with TSAN, MSAN, UBSAN, Coverage
+- [ ] <!---ci_exclude_aarch64|release|debug--> Exclude: All with aarch64, release, debug
 ---
 - [ ] <!---do_not_test--> Do not test
 - [ ] <!---upload_all--> Upload binaries for special builds

contrib/cld2

@@ -1 +1 @@
-Subproject commit bc6d493a2f64ed1fc1c4c4b4294a542a04e04217
+Subproject commit 217ba8b8805b41557faadaa47bb6e99f2242eea3

contrib/orc

@@ -1 +1 @@
-Subproject commit e24f2c2a3ca0769c96704ab20ad6f512a83ea2ad
+Subproject commit 947cebaf9432d708253ac08dc3012daa6b4ede6f

docker/test/style/Dockerfile

@@ -30,6 +30,7 @@ RUN pip3 install \
         mypy==1.8.0 \
         pylint==3.1.0 \
         python-magic==0.4.24 \
+        flake8==4.0.1 \
         requests \
         thefuzz \
         types-requests \

docker/test/style/run.sh

@@ -9,6 +9,8 @@ echo "Check style" | ts
 ./check-style -n              |& tee /test_output/style_output.txt
 echo "Check python formatting with black" | ts
 ./check-black -n              |& tee /test_output/black_output.txt
+echo "Check python with flake8" | ts
+./check-flake8                |& tee /test_output/flake8_output.txt
 echo "Check python type hinting with mypy" | ts
 ./check-mypy -n               |& tee /test_output/mypy_output.txt
 echo "Check typos" | ts

docs/changelogs/v24.4.2.141-stable.md

@@ -0,0 +1,101 @@
+---
+sidebar_position: 1
+sidebar_label: 2024
+---
+
+# 2024 Changelog
+
+### ClickHouse release v24.4.2.141-stable (9e23d27bd11) FIXME as compared to v24.4.1.2088-stable (6d4b31322d1)
+
+#### Improvement
+* Backported in [#63467](https://github.com/ClickHouse/ClickHouse/issues/63467): Make rabbitmq nack broken messages. Closes [#45350](https://github.com/ClickHouse/ClickHouse/issues/45350). [#60312](https://github.com/ClickHouse/ClickHouse/pull/60312) ([Kseniia Sumarokova](https://github.com/kssenii)).
+
+#### Build/Testing/Packaging Improvement
+* Backported in [#63612](https://github.com/ClickHouse/ClickHouse/issues/63612): The Dockerfile is reviewed by the docker official library in https://github.com/docker-library/official-images/pull/15846. [#63400](https://github.com/ClickHouse/ClickHouse/pull/63400) ([Mikhail f. Shiryaev](https://github.com/Felixoid)).
+
+#### Bug Fix (user-visible misbehavior in an official stable release)
+
+* Backported in [#64279](https://github.com/ClickHouse/ClickHouse/issues/64279): Fix queries with FINAL give wrong result when table does not use adaptive granularity. [#62432](https://github.com/ClickHouse/ClickHouse/pull/62432) ([Duc Canh Le](https://github.com/canhld94)).
+* Backported in [#63295](https://github.com/ClickHouse/ClickHouse/issues/63295): Fix crash with untuple and unresolved lambda. [#63131](https://github.com/ClickHouse/ClickHouse/pull/63131) ([Raúl Marín](https://github.com/Algunenano)).
+* Backported in [#63978](https://github.com/ClickHouse/ClickHouse/issues/63978): Fix intersect parts when restart after drop range. [#63202](https://github.com/ClickHouse/ClickHouse/pull/63202) ([Han Fei](https://github.com/hanfei1991)).
+* Backported in [#63413](https://github.com/ClickHouse/ClickHouse/issues/63413): Fix a misbehavior when SQL security defaults don't load for old tables during server startup. [#63209](https://github.com/ClickHouse/ClickHouse/pull/63209) ([pufit](https://github.com/pufit)).
+* Backported in [#63388](https://github.com/ClickHouse/ClickHouse/issues/63388): JOIN filter push down filled join fix. Closes [#63228](https://github.com/ClickHouse/ClickHouse/issues/63228). [#63234](https://github.com/ClickHouse/ClickHouse/pull/63234) ([Maksim Kita](https://github.com/kitaisreal)).
+* Backported in [#63618](https://github.com/ClickHouse/ClickHouse/issues/63618): Fix bug which could potentially lead to rare LOGICAL_ERROR during SELECT query with message: `Unexpected return type from materialize. Expected type_XXX. Got type_YYY.` Introduced in [#59379](https://github.com/ClickHouse/ClickHouse/issues/59379). [#63353](https://github.com/ClickHouse/ClickHouse/pull/63353) ([alesapin](https://github.com/alesapin)).
+* Backported in [#63451](https://github.com/ClickHouse/ClickHouse/issues/63451): Fix `X-ClickHouse-Timezone` header returning wrong timezone when using `session_timezone` as query level setting. [#63377](https://github.com/ClickHouse/ClickHouse/pull/63377) ([Andrey Zvonov](https://github.com/zvonand)).
+* Backported in [#63605](https://github.com/ClickHouse/ClickHouse/issues/63605): Fix backup of projection part in case projection was removed from table metadata, but part still has projection. [#63426](https://github.com/ClickHouse/ClickHouse/pull/63426) ([Kseniia Sumarokova](https://github.com/kssenii)).
+* Backported in [#63510](https://github.com/ClickHouse/ClickHouse/issues/63510): Fix 'Every derived table must have its own alias' error for MYSQL dictionary source, close [#63341](https://github.com/ClickHouse/ClickHouse/issues/63341). [#63481](https://github.com/ClickHouse/ClickHouse/pull/63481) ([vdimir](https://github.com/vdimir)).
+* Backported in [#63592](https://github.com/ClickHouse/ClickHouse/issues/63592): Avoid segafult in `MergeTreePrefetchedReadPool` while fetching projection parts. [#63513](https://github.com/ClickHouse/ClickHouse/pull/63513) ([Antonio Andelic](https://github.com/antonio2368)).
+* Backported in [#63750](https://github.com/ClickHouse/ClickHouse/issues/63750): Read only the necessary columns from VIEW (new analyzer). Closes [#62594](https://github.com/ClickHouse/ClickHouse/issues/62594). [#63688](https://github.com/ClickHouse/ClickHouse/pull/63688) ([Maksim Kita](https://github.com/kitaisreal)).
+* Backported in [#63772](https://github.com/ClickHouse/ClickHouse/issues/63772): Fix [#63539](https://github.com/ClickHouse/ClickHouse/issues/63539). Forbid WINDOW redefinition in new analyzer. [#63694](https://github.com/ClickHouse/ClickHouse/pull/63694) ([Dmitry Novik](https://github.com/novikd)).
+* Backported in [#63872](https://github.com/ClickHouse/ClickHouse/issues/63872): Flatten_nested is broken with replicated database. [#63695](https://github.com/ClickHouse/ClickHouse/pull/63695) ([Nikolai Kochetov](https://github.com/KochetovNicolai)).
+* Backported in [#63854](https://github.com/ClickHouse/ClickHouse/issues/63854): Fix `Not found column` and `CAST AS Map from array requires nested tuple of 2 elements` exceptions for distributed queries which use `Map(Nothing, Nothing)` type. Fixes [#63637](https://github.com/ClickHouse/ClickHouse/issues/63637). [#63753](https://github.com/ClickHouse/ClickHouse/pull/63753) ([Nikolai Kochetov](https://github.com/KochetovNicolai)).
+* Backported in [#63847](https://github.com/ClickHouse/ClickHouse/issues/63847): Fix possible `ILLEGAL_COLUMN` error in `partial_merge` join, close [#37928](https://github.com/ClickHouse/ClickHouse/issues/37928). [#63755](https://github.com/ClickHouse/ClickHouse/pull/63755) ([vdimir](https://github.com/vdimir)).
+* Backported in [#63908](https://github.com/ClickHouse/ClickHouse/issues/63908): `query_plan_remove_redundant_distinct` can break queries with WINDOW FUNCTIONS (with `allow_experimental_analyzer` is on). Fixes [#62820](https://github.com/ClickHouse/ClickHouse/issues/62820). [#63776](https://github.com/ClickHouse/ClickHouse/pull/63776) ([Igor Nikonov](https://github.com/devcrafter)).
+* Backported in [#63955](https://github.com/ClickHouse/ClickHouse/issues/63955): Fix possible crash with SYSTEM UNLOAD PRIMARY KEY. [#63778](https://github.com/ClickHouse/ClickHouse/pull/63778) ([Raúl Marín](https://github.com/Algunenano)).
+* Backported in [#63938](https://github.com/ClickHouse/ClickHouse/issues/63938): Allow JOIN filter push down to both streams if only single equivalent column is used in query. Closes [#63799](https://github.com/ClickHouse/ClickHouse/issues/63799). [#63819](https://github.com/ClickHouse/ClickHouse/pull/63819) ([Maksim Kita](https://github.com/kitaisreal)).
+* Backported in [#63991](https://github.com/ClickHouse/ClickHouse/issues/63991): Fix incorrect select query result when parallel replicas were used to read from a Materialized View. [#63861](https://github.com/ClickHouse/ClickHouse/pull/63861) ([Nikita Taranov](https://github.com/nickitat)).
+* Backported in [#64033](https://github.com/ClickHouse/ClickHouse/issues/64033): Fix a error `Database name is empty` for remote queries with lambdas over the cluster with modified default database. Fixes [#63471](https://github.com/ClickHouse/ClickHouse/issues/63471). [#63864](https://github.com/ClickHouse/ClickHouse/pull/63864) ([Nikolai Kochetov](https://github.com/KochetovNicolai)).
+* Backported in [#64561](https://github.com/ClickHouse/ClickHouse/issues/64561): Fix SIGSEGV due to CPU/Real (`query_profiler_real_time_period_ns`/`query_profiler_cpu_time_period_ns`) profiler (has been an issue since 2022, that leads to periodic server crashes, especially if you were using distributed engine). [#63865](https://github.com/ClickHouse/ClickHouse/pull/63865) ([Azat Khuzhin](https://github.com/azat)).
+* Backported in [#64011](https://github.com/ClickHouse/ClickHouse/issues/64011): Fix analyzer - IN function with arbitrary deep sub-selects in materialized view to use insertion block. [#63930](https://github.com/ClickHouse/ClickHouse/pull/63930) ([Yakov Olkhovskiy](https://github.com/yakov-olkhovskiy)).
+* Backported in [#64238](https://github.com/ClickHouse/ClickHouse/issues/64238): Fix resolve of unqualified COLUMNS matcher. Preserve the input columns order and forbid usage of unknown identifiers. [#63962](https://github.com/ClickHouse/ClickHouse/pull/63962) ([Dmitry Novik](https://github.com/novikd)).
+* Backported in [#64103](https://github.com/ClickHouse/ClickHouse/issues/64103): Deserialize untrusted binary inputs in a safer way. [#64024](https://github.com/ClickHouse/ClickHouse/pull/64024) ([Robert Schulze](https://github.com/rschu1ze)).
+* Backported in [#64170](https://github.com/ClickHouse/ClickHouse/issues/64170): Add missing settings to recoverLostReplica. [#64040](https://github.com/ClickHouse/ClickHouse/pull/64040) ([Raúl Marín](https://github.com/Algunenano)).
+* Backported in [#64322](https://github.com/ClickHouse/ClickHouse/issues/64322): This fix will use a proper redefined context with the correct definer for each individual view in the query pipeline Closes [#63777](https://github.com/ClickHouse/ClickHouse/issues/63777). [#64079](https://github.com/ClickHouse/ClickHouse/pull/64079) ([pufit](https://github.com/pufit)).
+* Backported in [#64382](https://github.com/ClickHouse/ClickHouse/issues/64382): Fix analyzer: "Not found column" error is fixed when using INTERPOLATE. [#64096](https://github.com/ClickHouse/ClickHouse/pull/64096) ([Yakov Olkhovskiy](https://github.com/yakov-olkhovskiy)).
+* Backported in [#64568](https://github.com/ClickHouse/ClickHouse/issues/64568): Fix creating backups to S3 buckets with different credentials from the disk containing the file. [#64153](https://github.com/ClickHouse/ClickHouse/pull/64153) ([Antonio Andelic](https://github.com/antonio2368)).
+* Backported in [#64272](https://github.com/ClickHouse/ClickHouse/issues/64272): Prevent LOGICAL_ERROR on CREATE TABLE as MaterializedView. [#64174](https://github.com/ClickHouse/ClickHouse/pull/64174) ([Raúl Marín](https://github.com/Algunenano)).
+* Backported in [#64330](https://github.com/ClickHouse/ClickHouse/issues/64330): The query cache now considers two identical queries against different databases as different. The previous behavior could be used to bypass missing privileges to read from a table. [#64199](https://github.com/ClickHouse/ClickHouse/pull/64199) ([Robert Schulze](https://github.com/rschu1ze)).
+* Backported in [#64254](https://github.com/ClickHouse/ClickHouse/issues/64254): Ignore `text_log` config when using Keeper. [#64218](https://github.com/ClickHouse/ClickHouse/pull/64218) ([Antonio Andelic](https://github.com/antonio2368)).
+* Backported in [#64690](https://github.com/ClickHouse/ClickHouse/issues/64690): Fix Query Tree size validation. Closes [#63701](https://github.com/ClickHouse/ClickHouse/issues/63701). [#64377](https://github.com/ClickHouse/ClickHouse/pull/64377) ([Dmitry Novik](https://github.com/novikd)).
+* Backported in [#64409](https://github.com/ClickHouse/ClickHouse/issues/64409): Fix `Logical error: Bad cast` for `Buffer` table with `PREWHERE`. Fixes [#64172](https://github.com/ClickHouse/ClickHouse/issues/64172). [#64388](https://github.com/ClickHouse/ClickHouse/pull/64388) ([Nikolai Kochetov](https://github.com/KochetovNicolai)).
+* Backported in [#64727](https://github.com/ClickHouse/ClickHouse/issues/64727): Fixed `CREATE TABLE AS` queries for tables with default expressions. [#64455](https://github.com/ClickHouse/ClickHouse/pull/64455) ([Anton Popov](https://github.com/CurtizJ)).
+* Backported in [#64623](https://github.com/ClickHouse/ClickHouse/issues/64623): Fix an error `Cannot find column` in distributed queries with constant CTE in the `GROUP BY` key. [#64519](https://github.com/ClickHouse/ClickHouse/pull/64519) ([Nikolai Kochetov](https://github.com/KochetovNicolai)).
+* Backported in [#64680](https://github.com/ClickHouse/ClickHouse/issues/64680): Fix [#64612](https://github.com/ClickHouse/ClickHouse/issues/64612). Do not rewrite aggregation if `-If` combinator is already used. [#64638](https://github.com/ClickHouse/ClickHouse/pull/64638) ([Dmitry Novik](https://github.com/novikd)).
+* Backported in [#64942](https://github.com/ClickHouse/ClickHouse/issues/64942): Fix OrderByLimitByDuplicateEliminationVisitor across subqueries. [#64766](https://github.com/ClickHouse/ClickHouse/pull/64766) ([Raúl Marín](https://github.com/Algunenano)).
+* Backported in [#64871](https://github.com/ClickHouse/ClickHouse/issues/64871): Fixed memory possible incorrect memory tracking in several kinds of queries: queries that read any data from S3, queries via http protocol, asynchronous inserts. [#64844](https://github.com/ClickHouse/ClickHouse/pull/64844) ([Anton Popov](https://github.com/CurtizJ)).
+
+#### CI Fix or Improvement (changelog entry is not required)
+
+* Backported in [#63364](https://github.com/ClickHouse/ClickHouse/issues/63364): Implement cumulative A Sync status. [#61464](https://github.com/ClickHouse/ClickHouse/pull/61464) ([Mikhail f. Shiryaev](https://github.com/Felixoid)).
+* Backported in [#63338](https://github.com/ClickHouse/ClickHouse/issues/63338): Use `/commit/` to have the URLs in [reports](https://play.clickhouse.com/play?user=play#c2VsZWN0IGRpc3RpbmN0IGNvbW1pdF91cmwgZnJvbSBjaGVja3Mgd2hlcmUgY2hlY2tfc3RhcnRfdGltZSA+PSBub3coKSAtIGludGVydmFsIDEgbW9udGggYW5kIHB1bGxfcmVxdWVzdF9udW1iZXI9NjA1MzI=) like https://github.com/ClickHouse/ClickHouse/commit/44f8bc5308b53797bec8cccc3bd29fab8a00235d and not like https://github.com/ClickHouse/ClickHouse/commits/44f8bc5308b53797bec8cccc3bd29fab8a00235d. [#63331](https://github.com/ClickHouse/ClickHouse/pull/63331) ([Mikhail f. Shiryaev](https://github.com/Felixoid)).
+* Backported in [#63376](https://github.com/ClickHouse/ClickHouse/issues/63376):. [#63366](https://github.com/ClickHouse/ClickHouse/pull/63366) ([Aleksei Filatov](https://github.com/aalexfvk)).
+* Backported in [#63571](https://github.com/ClickHouse/ClickHouse/issues/63571):. [#63551](https://github.com/ClickHouse/ClickHouse/pull/63551) ([Konstantin Bogdanov](https://github.com/thevar1able)).
+* Backported in [#63651](https://github.com/ClickHouse/ClickHouse/issues/63651): Fix 02362_part_log_merge_algorithm flaky test. [#63635](https://github.com/ClickHouse/ClickHouse/pull/63635) ([Miсhael Stetsyuk](https://github.com/mstetsyuk)).
+* Backported in [#63828](https://github.com/ClickHouse/ClickHouse/issues/63828): Fix test_odbc_interaction from aarch64 [#61457](https://github.com/ClickHouse/ClickHouse/issues/61457). [#63787](https://github.com/ClickHouse/ClickHouse/pull/63787) ([alesapin](https://github.com/alesapin)).
+* Backported in [#63897](https://github.com/ClickHouse/ClickHouse/issues/63897): Fix test `test_catboost_evaluate` for aarch64. [#61457](https://github.com/ClickHouse/ClickHouse/issues/61457). [#63789](https://github.com/ClickHouse/ClickHouse/pull/63789) ([alesapin](https://github.com/alesapin)).
+* Backported in [#63889](https://github.com/ClickHouse/ClickHouse/issues/63889): Remove HDFS from disks config for one integration test for arm. [#61457](https://github.com/ClickHouse/ClickHouse/issues/61457). [#63832](https://github.com/ClickHouse/ClickHouse/pull/63832) ([alesapin](https://github.com/alesapin)).
+* Backported in [#63881](https://github.com/ClickHouse/ClickHouse/issues/63881): Bump version for old image in test_short_strings_aggregation to make it work on arm. [#61457](https://github.com/ClickHouse/ClickHouse/issues/61457). [#63836](https://github.com/ClickHouse/ClickHouse/pull/63836) ([alesapin](https://github.com/alesapin)).
+* Backported in [#63919](https://github.com/ClickHouse/ClickHouse/issues/63919): Disable test `test_non_default_compression/test.py::test_preconfigured_deflateqpl_codec` on arm. [#61457](https://github.com/ClickHouse/ClickHouse/issues/61457). [#63839](https://github.com/ClickHouse/ClickHouse/pull/63839) ([alesapin](https://github.com/alesapin)).
+* Backported in [#63971](https://github.com/ClickHouse/ClickHouse/issues/63971): Fix 02124_insert_deduplication_token_multiple_blocks. [#63950](https://github.com/ClickHouse/ClickHouse/pull/63950) ([Han Fei](https://github.com/hanfei1991)).
+* Backported in [#64049](https://github.com/ClickHouse/ClickHouse/issues/64049): Add `ClickHouseVersion.copy` method. Create a branch release in advance without spinning out the release to increase the stability. [#64039](https://github.com/ClickHouse/ClickHouse/pull/64039) ([Mikhail f. Shiryaev](https://github.com/Felixoid)).
+* Backported in [#64078](https://github.com/ClickHouse/ClickHouse/issues/64078): The mime type is not 100% reliable for Python and shell scripts without shebangs; add a check for file extension. [#64062](https://github.com/ClickHouse/ClickHouse/pull/64062) ([Mikhail f. Shiryaev](https://github.com/Felixoid)).
+* Backported in [#64161](https://github.com/ClickHouse/ClickHouse/issues/64161): Add retries in git submodule update. [#64125](https://github.com/ClickHouse/ClickHouse/pull/64125) ([Alexey Milovidov](https://github.com/alexey-milovidov)).
+
+#### Critical Bug Fix (crash, LOGICAL_ERROR, data loss, RBAC)
+
+* Backported in [#64589](https://github.com/ClickHouse/ClickHouse/issues/64589): Disabled `enable_vertical_final` setting by default. This feature should not be used because it has a bug: [#64543](https://github.com/ClickHouse/ClickHouse/issues/64543). [#64544](https://github.com/ClickHouse/ClickHouse/pull/64544) ([Alexander Tokmakov](https://github.com/tavplubix)).
+* Backported in [#64880](https://github.com/ClickHouse/ClickHouse/issues/64880): This PR fixes an error when a user in a specific situation can escalate their privileges on the default database without necessary grants. [#64769](https://github.com/ClickHouse/ClickHouse/pull/64769) ([pufit](https://github.com/pufit)).
+
+#### NO CL CATEGORY
+
+* Backported in [#63306](https://github.com/ClickHouse/ClickHouse/issues/63306):. [#63297](https://github.com/ClickHouse/ClickHouse/pull/63297) ([Alexey Milovidov](https://github.com/alexey-milovidov)).
+* Backported in [#63710](https://github.com/ClickHouse/ClickHouse/issues/63710):. [#63415](https://github.com/ClickHouse/ClickHouse/pull/63415) ([Nikolai Kochetov](https://github.com/KochetovNicolai)).
+
+#### NO CL ENTRY
+
+* NO CL ENTRY:  'Revert "Backport [#64363](https://github.com/ClickHouse/ClickHouse/issues/64363) to 24.4: Split tests 03039_dynamic_all_merge_algorithms to avoid timeouts"'. [#64905](https://github.com/ClickHouse/ClickHouse/pull/64905) ([Raúl Marín](https://github.com/Algunenano)).
+
+#### NOT FOR CHANGELOG / INSIGNIFICANT
+
+* group_by_use_nulls strikes back [#62922](https://github.com/ClickHouse/ClickHouse/pull/62922) ([Nikolai Kochetov](https://github.com/KochetovNicolai)).
+* Add `FROM` keyword to `TRUNCATE ALL TABLES` [#63241](https://github.com/ClickHouse/ClickHouse/pull/63241) ([Yarik Briukhovetskyi](https://github.com/yariks5s)).
+* More checks for concurrently deleted files and dirs in system.remote_data_paths [#63274](https://github.com/ClickHouse/ClickHouse/pull/63274) ([Alexander Gololobov](https://github.com/davenger)).
+* Try fix segfault in `MergeTreeReadPoolBase::createTask` [#63323](https://github.com/ClickHouse/ClickHouse/pull/63323) ([Antonio Andelic](https://github.com/antonio2368)).
+* Skip unaccessible table dirs in system.remote_data_paths [#63330](https://github.com/ClickHouse/ClickHouse/pull/63330) ([Alexander Gololobov](https://github.com/davenger)).
+* Workaround for `oklch()` inside canvas bug for firefox [#63404](https://github.com/ClickHouse/ClickHouse/pull/63404) ([Sergei Trifonov](https://github.com/serxa)).
+* Cancel S3 reads properly when parallel reads are used [#63687](https://github.com/ClickHouse/ClickHouse/pull/63687) ([Antonio Andelic](https://github.com/antonio2368)).
+* Userspace page cache: don't collect stats if cache is unused [#63730](https://github.com/ClickHouse/ClickHouse/pull/63730) ([Michael Kolupaev](https://github.com/al13n321)).
+* Fix sanitizers [#64090](https://github.com/ClickHouse/ClickHouse/pull/64090) ([Azat Khuzhin](https://github.com/azat)).
+* Split tests 03039_dynamic_all_merge_algorithms to avoid timeouts [#64363](https://github.com/ClickHouse/ClickHouse/pull/64363) ([Kruglov Pavel](https://github.com/Avogar)).
+* CI: Critical bugfix category in PR template [#64480](https://github.com/ClickHouse/ClickHouse/pull/64480) ([Max K.](https://github.com/maxknv)).
+

docs/en/development/continuous-integration.md

@@ -91,6 +91,9 @@ cd ./utils/check-style
 # Check python type hinting with mypy
 ./check-mypy
 
+# Check python with flake8
+./check-flake8
+
 # Check code with codespell
 ./check-typos
 

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

@@ -54,6 +54,7 @@ SELECT * FROM test_table;
 - `_path` — Path to the file. Type: `LowCardinalty(String)`.
 - `_file` — Name of the file. Type: `LowCardinalty(String)`.
 - `_size` — Size of the file in bytes. Type: `Nullable(UInt64)`. If the size is unknown, the value is `NULL`.
+- `_time` — Last modified time of the file. Type: `Nullable(DateTime)`. If the time is unknown, the value is `NULL`.
 
 ## See also
 

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

@@ -235,6 +235,7 @@ libhdfs3 support HDFS namenode HA.
 - `_path` — Path to the file. Type: `LowCardinalty(String)`.
 - `_file` — Name of the file. Type: `LowCardinalty(String)`.
 - `_size` — Size of the file in bytes. Type: `Nullable(UInt64)`. If the size is unknown, the value is `NULL`.
+- `_time` — Last modified time of the file. Type: `Nullable(DateTime)`. If the time is unknown, the value is `NULL`.
 
 ## Storage Settings {#storage-settings}
 

docs/en/engines/table-engines/integrations/s3.md

@@ -53,14 +53,14 @@ For partitioning by month, use the `toYYYYMM(date_column)` expression, where `da
 
 This example uses the [docker compose recipe](https://github.com/ClickHouse/examples/tree/5fdc6ff72f4e5137e23ea075c88d3f44b0202490/docker-compose-recipes/recipes/ch-and-minio-S3), which integrates ClickHouse and MinIO.  You should be able to reproduce the same queries using S3 by replacing the endpoint and authentication values.
 
-Notice that the S3 endpoint in the `ENGINE` configuration uses the parameter token `{_partition_id}` as part of the S3 object (filename), and that the SELECT queries select against those resulting object names (e.g., `test_3.csv`). 
+Notice that the S3 endpoint in the `ENGINE` configuration uses the parameter token `{_partition_id}` as part of the S3 object (filename), and that the SELECT queries select against those resulting object names (e.g., `test_3.csv`).
 
 :::note
 As shown in the example, querying from S3 tables that are partitioned is
 not directly supported at this time, but can be accomplished by querying the individual partitions
 using the S3 table function.
 
-The primary use-case for writing 
+The primary use-case for writing
 partitioned data in S3 is to enable transferring that data into another
 ClickHouse system (for example, moving from on-prem systems to ClickHouse
 Cloud).  Because ClickHouse datasets are often very large, and network
@@ -78,9 +78,9 @@ CREATE TABLE p
 )
 ENGINE = S3(
 # highlight-next-line
-           'http://minio:10000/clickhouse//test_{_partition_id}.csv', 
+           'http://minio:10000/clickhouse//test_{_partition_id}.csv',
-           'minioadmin', 
+           'minioadmin',
-           'minioadminpassword', 
+           'minioadminpassword',
            'CSV')
 PARTITION BY column3
 ```
@@ -145,6 +145,7 @@ Code: 48. DB::Exception: Received from localhost:9000. DB::Exception: Reading fr
 - `_path` — Path to the file. Type: `LowCardinalty(String)`.
 - `_file` — Name of the file. Type: `LowCardinalty(String)`.
 - `_size` — Size of the file in bytes. Type: `Nullable(UInt64)`. If the size is unknown, the value is `NULL`.
+- `_time` — Last modified time of the file. Type: `Nullable(DateTime)`. If the time is unknown, the value is `NULL`.
 
 For more information about virtual columns see [here](../../../engines/table-engines/index.md#table_engines-virtual_columns).
 

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

@@ -267,7 +267,7 @@ For introspection use `system.s3queue` stateless table and `system.s3queue_log`
     `exception` String
 )
 ENGINE = SystemS3Queue
-COMMENT 'SYSTEM TABLE is built on the fly.' │
+COMMENT 'Contains in-memory state of S3Queue metadata and currently processed rows per file.' │
 └────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┘
 ```
 

docs/en/engines/table-engines/mergetree-family/mergetree.md

@@ -6,35 +6,26 @@ sidebar_label:  MergeTree
 
 # MergeTree
 
-The `MergeTree` engine and other engines of this family (`*MergeTree`) are the most commonly used and most robust ClickHouse table engines.
+The `MergeTree` engine and other engines of the `MergeTree` family (e.g. `ReplacingMergeTree`, `AggregatingMergeTree` ) are the most commonly used and most robust table engines in ClickHouse.
 
-Engines in the `MergeTree` family are designed for inserting a very large amount of data into a table. The data is quickly written to the table part by part, then rules are applied for merging the parts in the background. This method is much more efficient than continually rewriting the data in storage during insert.
+`MergeTree`-family table engines are designed for high data ingest rates and huge data volumes.
+Insert operations create table parts which are merged by a background process with other table parts.
 
-Main features:
+Main features of `MergeTree`-family table engines.
 
-- Stores data sorted by primary key.
+- The table's primary key determines the sort order within each table part (clustered index). The primary key also does not reference individual rows but blocks of 8192 rows called granules. This makes primary keys of huge data sets small enough to remain loaded in main memory, while still providing fast access to on-disk data.
 
-    This allows you to create a small sparse index that helps find data faster.
+- Tables can be partitioned using an arbitrary partition expression. Partition pruning ensures partitions are omitted from reading when the query allows it.
 
-- Partitions can be used if the [partitioning key](/docs/en/engines/table-engines/mergetree-family/custom-partitioning-key.md) is specified.
+- Data can be replicated across multiple cluster nodes for high availability, failover, and zero downtime upgrades. See [Data replication](/docs/en/engines/table-engines/mergetree-family/replication.md).
 
-    ClickHouse supports certain operations with partitions that are more efficient than general operations on the same data with the same result. ClickHouse also automatically cuts off the partition data where the partitioning key is specified in the query.
+- `MergeTree` table engines support various statistics kinds and sampling methods to help query optimization.
 
-- Data replication support.
+:::note
-
+Despite a similar name, the [Merge](/docs/en/engines/table-engines/special/merge.md/#merge) engine is different from `*MergeTree` engines.
-    The family of `ReplicatedMergeTree` tables provides data replication. For more information, see [Data replication](/docs/en/engines/table-engines/mergetree-family/replication.md).
-
-- Data sampling support.
-
-    If necessary, you can set the data sampling method in the table.
-
-:::info
-The [Merge](/docs/en/engines/table-engines/special/merge.md/#merge) engine does not belong to the `*MergeTree` family.
 :::
 
-If you need to update rows frequently, we recommend using the [`ReplacingMergeTree`](/docs/en/engines/table-engines/mergetree-family/replacingmergetree.md) table engine. Using `ALTER TABLE my_table UPDATE` to update rows triggers a mutation, which causes parts to be re-written and uses IO/resources. With `ReplacingMergeTree`, you can simply insert the updated rows and the old rows will be replaced according to the table sorting key.
+## Creating Tables {#table_engine-mergetree-creating-a-table}
-
-## Creating a Table {#table_engine-mergetree-creating-a-table}
 
 ``` sql
 CREATE TABLE [IF NOT EXISTS] [db.]table_name [ON CLUSTER cluster]
@@ -59,23 +50,24 @@ ORDER BY expr
 [SETTINGS name = value, ...]
 ```
 
-For a description of parameters, see the [CREATE query description](/docs/en/sql-reference/statements/create/table.md).
+For a detailed description of the parameters, see the [CREATE TABLE](/docs/en/sql-reference/statements/create/table.md) statement
 
 ### Query Clauses {#mergetree-query-clauses}
 
 #### ENGINE
 
-`ENGINE` — Name and parameters of the engine. `ENGINE = MergeTree()`. The `MergeTree` engine does not have parameters.
+`ENGINE` — Name and parameters of the engine. `ENGINE = MergeTree()`. The `MergeTree` engine has no parameters.
 
 #### ORDER_BY
 
 `ORDER BY` — The sorting key.
 
-A tuple of column names or arbitrary expressions. Example: `ORDER BY (CounterID, EventDate)`.
+A tuple of column names or arbitrary expressions. Example: `ORDER BY (CounterID + 1, EventDate)`.
 
-ClickHouse uses the sorting key as a primary key if the primary key is not defined explicitly by the `PRIMARY KEY` clause.
+If no primary key is defined (i.e. `PRIMARY KEY` was not specified), ClickHouse uses the the sorting key as primary key.
 
-Use the `ORDER BY tuple()` syntax, if you do not need sorting, or set `create_table_empty_primary_key_by_default` to `true` to use the `ORDER BY tuple()` syntax by default. See [Selecting the Primary Key](#selecting-the-primary-key).
+If no sorting is required, you can use syntax `ORDER BY tuple()`.
+Alternatively, if setting `create_table_empty_primary_key_by_default` is enabled, `ORDER BY tuple()` is implicitly added to `CREATE TABLE` statements. See [Selecting a Primary Key](#selecting-a-primary-key).
 
 #### PARTITION BY
 
@@ -87,100 +79,32 @@ For partitioning by month, use the `toYYYYMM(date_column)` expression, where `da
 
 `PRIMARY KEY` — The primary key if it [differs from the sorting key](#choosing-a-primary-key-that-differs-from-the-sorting-key). Optional.
 
-By default the primary key is the same as the sorting key (which is specified by the `ORDER BY` clause). Thus in most cases it is unnecessary to specify a separate `PRIMARY KEY` clause.
+Specifying a sorting key (using `ORDER BY` clause) implicitly specifies a primary key.
+It is usually not necessary to specify the primary key in addition to the primary key.
 
 #### SAMPLE BY
 
-`SAMPLE BY` — An expression for sampling. Optional.
+`SAMPLE BY` — A sampling expression. Optional.
 
-If a sampling expression is used, the primary key must contain it. The result of a sampling expression must be an unsigned integer. Example: `SAMPLE BY intHash32(UserID) ORDER BY (CounterID, EventDate, intHash32(UserID))`.
+If specified, it must be contained in the primary key.
+The sampling expression must result in an unsigned integer.
+
+Example: `SAMPLE BY intHash32(UserID) ORDER BY (CounterID, EventDate, intHash32(UserID))`.
 
 ####  TTL
 
-`TTL` — A list of rules specifying storage duration of rows and defining logic of automatic parts movement [between disks and volumes](#table_engine-mergetree-multiple-volumes). Optional.
+`TTL` — A list of rules that specify the storage duration of rows and the logic of automatic parts movement [between disks and volumes](#table_engine-mergetree-multiple-volumes). Optional.
 
-Expression must have one `Date` or `DateTime` column as a result. Example:
+Expression must result in a `Date` or `DateTime`, e.g. `TTL date + INTERVAL 1 DAY`.
-```
-TTL date + INTERVAL 1 DAY
-```
 
 Type of the rule `DELETE|TO DISK 'xxx'|TO VOLUME 'xxx'|GROUP BY` specifies an action to be done with the part if the expression is satisfied (reaches current time): removal of expired rows, moving a part (if expression is satisfied for all rows in a part) to specified disk (`TO DISK 'xxx'`) or to volume (`TO VOLUME 'xxx'`), or aggregating values in expired rows. Default type of the rule is removal (`DELETE`). List of multiple rules can be specified, but there should be no more than one `DELETE` rule.
 
+
 For more details, see [TTL for columns and tables](#table_engine-mergetree-ttl)
 
-### SETTINGS
+#### SETTINGS
-Additional parameters that control the behavior of the `MergeTree` (optional):
 
-#### index_granularity
+See [MergeTree Settings](../../../operations/settings/merge-tree-settings.md).
-
-`index_granularity` — Maximum number of data rows between the marks of an index. Default value: 8192. See [Data Storage](#mergetree-data-storage).
-
-#### index_granularity_bytes
-
-`index_granularity_bytes` — Maximum size of data granules in bytes. Default value: 10Mb. To restrict the granule size only by number of rows, set to 0 (not recommended). See [Data Storage](#mergetree-data-storage).
-
-#### min_index_granularity_bytes
-
-`min_index_granularity_bytes` — Min allowed size of data granules in bytes. Default value: 1024b. To provide a safeguard against accidentally creating tables with very low index_granularity_bytes. See [Data Storage](#mergetree-data-storage).
-
-#### enable_mixed_granularity_parts
-
-`enable_mixed_granularity_parts` — Enables or disables transitioning to control the granule size with the `index_granularity_bytes` setting. Before version 19.11, there was only the `index_granularity` setting for restricting granule size. The `index_granularity_bytes` setting improves ClickHouse performance when selecting data from tables with big rows (tens and hundreds of megabytes). If you have tables with big rows, you can enable this setting for the tables to improve the efficiency of `SELECT` queries.
-
-#### use_minimalistic_part_header_in_zookeeper
-
-`use_minimalistic_part_header_in_zookeeper` — Storage method of the data parts headers in ZooKeeper. If `use_minimalistic_part_header_in_zookeeper=1`, then ZooKeeper stores less data. For more information, see the [setting description](/docs/en/operations/server-configuration-parameters/settings.md/#server-settings-use_minimalistic_part_header_in_zookeeper) in “Server configuration parameters”.
-
-#### min_merge_bytes_to_use_direct_io
-
-`min_merge_bytes_to_use_direct_io` — The minimum data volume for merge operation that is required for using direct I/O access to the storage disk. When merging data parts, ClickHouse calculates the total storage volume of all the data to be merged. If the volume exceeds `min_merge_bytes_to_use_direct_io` bytes, ClickHouse reads and writes the data to the storage disk using the direct I/O interface (`O_DIRECT` option). If `min_merge_bytes_to_use_direct_io = 0`, then direct I/O is disabled. Default value: `10 * 1024 * 1024 * 1024` bytes.
-
-#### merge_with_ttl_timeout
-
-`merge_with_ttl_timeout` — Minimum delay in seconds before repeating a merge with delete TTL. Default value: `14400` seconds (4 hours).
-#### merge_with_recompression_ttl_timeout
-
-`merge_with_recompression_ttl_timeout` — Minimum delay in seconds before repeating a merge with recompression TTL. Default value: `14400` seconds (4 hours).
-
-#### try_fetch_recompressed_part_timeout
-
-`try_fetch_recompressed_part_timeout` — Timeout (in seconds) before starting merge with recompression. During this time ClickHouse tries to fetch recompressed part from replica which assigned this merge with recompression. Default value: `7200` seconds (2 hours).
-
-#### write_final_mark
-
-`write_final_mark` — Enables or disables writing the final index mark at the end of data part (after the last byte). Default value: 1. Don’t turn it off.
-
-#### merge_max_block_size
-
-`merge_max_block_size` — Maximum number of rows in block for merge operations. Default value: 8192.
-
-#### storage_policy
-
-`storage_policy` — Storage policy. See [Using Multiple Block Devices for Data Storage](#table_engine-mergetree-multiple-volumes).
-
-#### min_bytes_for_wide_part
-
-`min_bytes_for_wide_part`, `min_rows_for_wide_part` — Minimum number of bytes/rows in a data part that can be stored in `Wide` format. You can set one, both or none of these settings. See [Data Storage](#mergetree-data-storage).
-
-#### max_parts_in_total
-
-`max_parts_in_total` — Maximum number of parts in all partitions.
-
-#### max_compress_block_size
-
-`max_compress_block_size` — Maximum size of blocks of uncompressed data before compressing for writing to a table. You can also specify this setting in the global settings (see [max_compress_block_size](/docs/en/operations/settings/settings.md/#max-compress-block-size) setting). The value specified when table is created overrides the global value for this setting.
-
-#### min_compress_block_size
-
-`min_compress_block_size` — Minimum size of blocks of uncompressed data required for compression when writing the next mark. You can also specify this setting in the global settings (see [min_compress_block_size](/docs/en/operations/settings/settings.md/#min-compress-block-size) setting). The value specified when table is created overrides the global value for this setting.
-
-#### max_partitions_to_read
-
-`max_partitions_to_read` — Limits the maximum number of partitions that can be accessed in one query. You can also specify setting [max_partitions_to_read](/docs/en/operations/settings/merge-tree-settings.md/#max-partitions-to-read) in the global setting.
-
-#### allow_experimental_optimized_row_order
-
-`allow_experimental_optimized_row_order` - Experimental. Enables the optimization of the row order during inserts to improve the compressability of the data for compression codecs (e.g. LZ4). Analyzes and reorders the data, and thus increases the CPU overhead of inserts.
 
 **Example of Sections Setting**
 
@@ -270,7 +194,7 @@ ClickHouse does not require a unique primary key. You can insert multiple rows w
 
 You can use `Nullable`-typed expressions in the `PRIMARY KEY` and `ORDER BY` clauses but it is strongly discouraged. To allow this feature, turn on the [allow_nullable_key](/docs/en/operations/settings/settings.md/#allow-nullable-key) setting. The [NULLS_LAST](/docs/en/sql-reference/statements/select/order-by.md/#sorting-of-special-values) principle applies for `NULL` values in the `ORDER BY` clause.
 
-### Selecting the Primary Key {#selecting-the-primary-key}
+### Selecting a Primary Key {#selecting-a-primary-key}
 
 The number of columns in the primary key is not explicitly limited. Depending on the data structure, you can include more or fewer columns in the primary key. This may:
 

docs/en/engines/table-engines/special/file.md

@@ -102,6 +102,7 @@ For partitioning by month, use the `toYYYYMM(date_column)` expression, where `da
 - `_path` — Path to the file. Type: `LowCardinalty(String)`.
 - `_file` — Name of the file. Type: `LowCardinalty(String)`.
 - `_size` — Size of the file in bytes. Type: `Nullable(UInt64)`. If the size is unknown, the value is `NULL`.
+- `_time` — Last modified time of the file. Type: `Nullable(DateTime)`. If the time is unknown, the value is `NULL`.
 
 ## Settings {#settings}
 

docs/en/engines/table-engines/special/url.md

@@ -108,6 +108,7 @@ For partitioning by month, use the `toYYYYMM(date_column)` expression, where `da
 - `_path` — Path to the `URL`. Type: `LowCardinalty(String)`.
 - `_file` — Resource name of the `URL`. Type: `LowCardinalty(String)`.
 - `_size` — Size of the resource in bytes. Type: `Nullable(UInt64)`. If the size is unknown, the value is `NULL`.
+- `_time` — Last modified time of the file. Type: `Nullable(DateTime)`. If the time is unknown, the value is `NULL`.
 
 ## Storage Settings {#storage-settings}
 

docs/en/operations/analyzer.md

@@ -0,0 +1,194 @@
+---
+slug: /en/operations/analyzer
+sidebar_label: Analyzer
+title: Analyzer
+description: Details about ClickHouse's query analyzer
+keywords: [analyzer]
+---
+
+# Analyzer
+
+<BetaBadge />
+
+## Known incompatibilities
+
+In ClickHouse version `24.3`, the new query analyzer was enabled by default.
+Despite fixing a large number of bugs and introducing new optimizations, it also introduces some breaking changes in ClickHouse behaviour. Please read the following changes to determine how to rewrite your queries for the new analyzer.
+
+### Invalid queries are no longer optimized
+
+The previous query planning infrastructure applied AST-level optimizations before the query validation step.
+Optimizations could rewrite the initial query so it becomes valid and can be executed.
+
+In the new analyzer, query validation takes place before the optimization step.
+This means that invalid queries that were possible to execute before are now unsupported.
+In such cases, the query must be fixed manually.
+
+**Example 1:**
+
+```sql
+SELECT number
+FROM numbers(1)
+GROUP BY toString(number)
+```
+
+The following query uses column `number` in the projection list when only `toString(number)` is available after the aggregation.
+In the old analyzer, `GROUP BY toString(number)` was optimized into `GROUP BY number,` making the query valid.
+
+**Example 2:**
+
+```sql
+SELECT
+    number % 2 AS n,
+    sum(number)
+FROM numbers(10)
+GROUP BY n
+HAVING number > 5
+```
+
+The same problem occurs in this query: column `number` is used after aggregation with another key.
+The previous query analyzer fixed this query by moving the `number > 5` filter from the `HAVING` clause to the `WHERE` clause.
+
+To fix the query, you should move all conditions that apply to non-aggregated columns to the `WHERE` section to conform to standard SQL syntax:
+```sql
+SELECT
+    number % 2 AS n,
+    sum(number)
+FROM numbers(10)
+WHERE number > 5
+GROUP BY n
+```
+
+### CREATE VIEW with invalid query
+
+The new analyzer always performs type-checking.
+Previously, it was possible to create a `VIEW` with an invalid `SELECT` query. It would then fail during the first `SELECT` or `INSERT` (in the case of `MATERIALIZED VIEW`).
+
+Now, it's not possible to create such `VIEW`s anymore.
+
+**Example:**
+
+```sql
+CREATE TABLE source (data String) ENGINE=MergeTree ORDER BY tuple();
+
+CREATE VIEW some_view
+AS SELECT JSONExtract(data, 'test', 'DateTime64(3)')
+FROM source;
+```
+
+### Known incompatibilities of the `JOIN` clause
+
+#### Join using column from projection
+
+Alias from the `SELECT` list can not be used as a `JOIN USING` key by default.
+
+A new setting, `analyzer_compatibility_join_using_top_level_identifier`, when enabled, alters the behavior of `JOIN USING` to prefer to resolve identifiers based on expressions from the projection list of the `SELECT` query, rather than using the columns from left table directly.
+
+**Example:**
+
+```sql
+SELECT a + 1 AS b, t2.s
+FROM Values('a UInt64, b UInt64', (1, 1)) AS t1
+JOIN Values('b UInt64, s String', (1, 'one'), (2, 'two')) t2
+USING (b);
+```
+
+With `analyzer_compatibility_join_using_top_level_identifier` set to `true`, the join condition is interpreted as `t1.a + 1 = t2.b`, matching the behavior of earlier versions. So, the result will be `2, 'two'`.
+When the setting is `false`, the join condition defaults to `t1.b = t2.b`, and the query will return `2, 'one'`.
+If `b` is not present in `t1`, the query will fail with an error.
+
+#### Changes in behavior with `JOIN USING` and `ALIAS`/`MATERIALIZED` columns
+
+In the new analyzer, using `*` in a `JOIN USING` query that involves `ALIAS` or `MATERIALIZED` columns will include those columns in the result set by default.
+
+**Example:**
+
+```sql
+CREATE TABLE t1 (id UInt64, payload ALIAS sipHash64(id)) ENGINE = MergeTree ORDER BY id;
+INSERT INTO t1 VALUES (1), (2);
+
+CREATE TABLE t2 (id UInt64, payload ALIAS sipHash64(id)) ENGINE = MergeTree ORDER BY id;
+INSERT INTO t2 VALUES (2), (3);
+
+SELECT * FROM t1
+FULL JOIN t2 USING (payload);
+```
+
+In the new analyzer, the result of this query will include the `payload` column along with `id` from both tables. In contrast, the previous analyzer would only include these `ALIAS` columns if specific settings (`asterisk_include_alias_columns` or `asterisk_include_materialized_columns`) were enabled, and the columns might appear in a different order.
+
+To ensure consistent and expected results, especially when migrating old queries to the new analyzer, it is advisable to specify columns explicitly in the `SELECT` clause rather than using `*`.
+
+#### Handling of Type Modifiers for columns in `USING` Clause
+
+In the new version of the analyzer, the rules for determining the common supertype for columns specified in the `USING` clause have been standardized to produce more predictable outcomes, especially when dealing with type modifiers like `LowCardinality` and `Nullable`.
+
+- `LowCardinality(T)` and `T`: When a column of type `LowCardinality(T)` is joined with a column of type `T`, the resulting common supertype will be `T`, effectively discarding the `LowCardinality` modifier.
+  
+- `Nullable(T)` and `T`: When a column of type `Nullable(T)` is joined with a column of type `T`, the resulting common supertype will be `Nullable(T)`, ensuring that the nullable property is preserved.
+
+**Example:**
+
+```sql
+SELECT id, toTypeName(id) FROM Values('id LowCardinality(String)', ('a')) AS t1
+FULL OUTER JOIN Values('id String', ('b')) AS t2
+USING (id);
+```
+
+In this query, the common supertype for `id` is determined as `String`, discarding the `LowCardinality` modifier from `t1`.
+
+### Projection column names changes
+
+During projection names computation, aliases are not substituted.
+
+```sql
+SELECT
+    1 + 1 AS x,
+    x + 1
+SETTINGS allow_experimental_analyzer = 0
+FORMAT PrettyCompact
+
+   ┌─x─┬─plus(plus(1, 1), 1)─┐
+1. │ 2 │                   3 │
+   └───┴─────────────────────┘
+
+SELECT
+    1 + 1 AS x,
+    x + 1
+SETTINGS allow_experimental_analyzer = 1
+FORMAT PrettyCompact
+
+   ┌─x─┬─plus(x, 1)─┐
+1. │ 2 │          3 │
+   └───┴────────────┘
+```
+
+### Incompatible function arguments types
+
+In the new analyzer, type inference happens during initial query analysis.
+This change means that type checks are done before short-circuit evaluation; thus, `if` function arguments must always have a common supertype.
+
+**Example:**
+
+The following query fails with `There is no supertype for types Array(UInt8), String because some of them are Array and some of them are not`:
+
+```sql
+SELECT toTypeName(if(0, [2, 3, 4], 'String'))
+```
+
+### Heterogeneous clusters
+
+The new analyzer significantly changed the communication protocol between servers in the cluster. Thus, it's impossible to run distributed queries on servers with different `allow_experimental_analyzer` setting values.
+
+### Mutations are interpreted by previous analyzer
+
+Mutations are still using the old analyzer.
+This means some new ClickHouse SQL features can't be used in mutations. For example, the `QUALIFY` clause.
+Status can be checked [here](https://github.com/ClickHouse/ClickHouse/issues/61563).
+
+### Unsupported features
+
+The list of features new analyzer currently doesn't support:
+
+- Annoy index.
+- Hypothesis index. Work in progress [here](https://github.com/ClickHouse/ClickHouse/pull/48381).
+- Window view is not supported. There are no plans to support it in the future.

docs/en/operations/named-collections.md

@@ -443,3 +443,59 @@ SELECT dictGet('dict', 'b', 1);
 │ a                       │
 └─────────────────────────┘
 ```
+
+## Named collections for accessing Kafka
+
+The description of parameters see [Kafka](../engines/table-engines/integrations/kafka.md).
+
+### DDL example
+
+```sql
+CREATE NAMED COLLECTION my_kafka_cluster AS
+kafka_broker_list = 'localhost:9092',
+kafka_topic_list = 'kafka_topic',
+kafka_group_name = 'consumer_group',
+kafka_format = 'JSONEachRow',
+kafka_max_block_size = '1048576';
+       
+```
+### XML example
+
+```xml
+<clickhouse>
+    <named_collections>
+        <my_kafka_cluster>
+            <kafka_broker_list>localhost:9092</kafka_broker_list>
+            <kafka_topic_list>kafka_topic</kafka_topic_list>
+            <kafka_group_name>consumer_group</kafka_group_name>
+            <kafka_format>JSONEachRow</kafka_format>
+            <kafka_max_block_size>1048576</kafka_max_block_size>
+        </my_kafka_cluster>
+    </named_collections>
+</clickhouse>
+```
+
+### Example of using named collections with a Kafka table
+
+Both of the following examples use the same named collection `my_kafka_cluster`:
+
+
+```sql
+CREATE TABLE queue
+(
+    timestamp UInt64,
+    level String,
+    message String
+)
+ENGINE = Kafka(my_kafka_cluster)
+
+CREATE TABLE queue
+(
+    timestamp UInt64,
+    level String,
+    message String
+)
+ENGINE = Kafka(my_kafka_cluster)
+SETTINGS kafka_num_consumers = 4,
+         kafka_thread_per_consumer = 1;
+```

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

@@ -3,9 +3,126 @@ slug: /en/operations/settings/merge-tree-settings
 title: "MergeTree tables settings"
 ---
 
-The values of `merge_tree` settings (for all MergeTree tables) can be viewed in the table `system.merge_tree_settings`, they can be overridden in `config.xml` in the `merge_tree` section, or set in the `SETTINGS` section of each table.
+System table `system.merge_tree_settings` shows the globally set MergeTree settings.
 
-These are example overrides for `max_suspicious_broken_parts`:
+MergeTree settings can be set in the `merge_tree` section of the server config file, or specified for each `MergeTree` table individually in
+the `SETTINGS` clause of the `CREATE TABLE` statement.
+
+Example for customizing setting `max_suspicious_broken_parts`:
+
+Configure the default for all `MergeTree` tables in the server configuration file:
+
+``` text
+<merge_tree>
+    <max_suspicious_broken_parts>5</max_suspicious_broken_parts>
+</merge_tree>
+```
+
+Set for a particular table:
+
+``` sql
+CREATE TABLE tab
+(
+    `A` Int64
+)
+ENGINE = MergeTree
+ORDER BY tuple()
+SETTINGS max_suspicious_broken_parts = 500;
+```
+
+Change the settings for a particular table using `ALTER TABLE ... MODIFY SETTING`:
+
+```sql
+ALTER TABLE tab MODIFY SETTING max_suspicious_broken_parts = 100;
+
+-- reset to global default (value from system.merge_tree_settings)
+ALTER TABLE tab RESET SETTING max_suspicious_broken_parts;
+```
+
+## index_granularity
+
+Maximum number of data rows between the marks of an index.
+
+Default value: 8192.
+
+## index_granularity_bytes
+
+Maximum size of data granules in bytes.
+
+Default value: 10Mb.
+
+To restrict the granule size only by number of rows, set to 0 (not recommended).
+
+## min_index_granularity_bytes
+
+Min allowed size of data granules in bytes.
+
+Default value: 1024b.
+
+To provide a safeguard against accidentally creating tables with very low index_granularity_bytes.
+
+## enable_mixed_granularity_parts
+
+Enables or disables transitioning to control the granule size with the `index_granularity_bytes` setting. Before version 19.11, there was only the `index_granularity` setting for restricting granule size. The `index_granularity_bytes` setting improves ClickHouse performance when selecting data from tables with big rows (tens and hundreds of megabytes). If you have tables with big rows, you can enable this setting for the tables to improve the efficiency of `SELECT` queries.
+
+## use_minimalistic_part_header_in_zookeeper
+
+Storage method of the data parts headers in ZooKeeper. If enabled, ZooKeeper stores less data. For details, see [here](../server-configuration-parameters/settings.md/#server-settings-use_minimalistic_part_header_in_zookeeper).
+
+## min_merge_bytes_to_use_direct_io
+
+The minimum data volume for merge operation that is required for using direct I/O access to the storage disk.
+When merging data parts, ClickHouse calculates the total storage volume of all the data to be merged.
+If the volume exceeds `min_merge_bytes_to_use_direct_io` bytes, ClickHouse reads and writes the data to the storage disk using the direct I/O interface (`O_DIRECT` option).
+If `min_merge_bytes_to_use_direct_io = 0`, then direct I/O is disabled.
+
+Default value: `10 * 1024 * 1024 * 1024` bytes.
+
+## merge_with_ttl_timeout
+
+Minimum delay in seconds before repeating a merge with delete TTL.
+
+Default value: `14400` seconds (4 hours).
+
+## merge_with_recompression_ttl_timeout
+
+Minimum delay in seconds before repeating a merge with recompression TTL.
+
+Default value: `14400` seconds (4 hours).
+
+## write_final_mark
+
+Enables or disables writing the final index mark at the end of data part (after the last byte).
+
+Default value: 1.
+
+Don’t change or bad things will happen.
+
+## storage_policy
+
+Storage policy.
+
+## min_bytes_for_wide_part
+
+Minimum number of bytes/rows in a data part that can be stored in `Wide` format.
+You can set one, both or none of these settings.
+
+## max_compress_block_size
+
+Maximum size of blocks of uncompressed data before compressing for writing to a table.
+You can also specify this setting in the global settings (see [max_compress_block_size](/docs/en/operations/settings/settings.md/#max-compress-block-size) setting).
+The value specified when table is created overrides the global value for this setting.
+
+## min_compress_block_size
+
+Minimum size of blocks of uncompressed data required for compression when writing the next mark.
+You can also specify this setting in the global settings (see [min_compress_block_size](/docs/en/operations/settings/settings.md/#min-compress-block-size) setting).
+The value specified when table is created overrides the global value for this setting.
+
+## max_partitions_to_read
+
+Limits the maximum number of partitions that can be accessed in one query.
+You can also specify setting [max_partitions_to_read](/docs/en/operations/settings/merge-tree-settings.md/#max-partitions-to-read) in the global setting.
 
 ## max_suspicious_broken_parts
 
@@ -17,37 +134,6 @@ Possible values:
 
 Default value: 100.
 
-Override example in `config.xml`:
-
-``` text
-<merge_tree>
-    <max_suspicious_broken_parts>5</max_suspicious_broken_parts>
-</merge_tree>
-```
-
-An example to set in `SETTINGS` for a particular table:
-
-``` sql
-CREATE TABLE foo
-(
-    `A` Int64
-)
-ENGINE = MergeTree
-ORDER BY tuple()
-SETTINGS max_suspicious_broken_parts = 500;
-```
-
-An example of changing the settings for a specific table with the `ALTER TABLE ... MODIFY SETTING` command:
-
-``` sql
-ALTER TABLE foo
-    MODIFY SETTING max_suspicious_broken_parts = 100;
-
--- reset to default (use value from system.merge_tree_settings)
-ALTER TABLE foo
-    RESET SETTING max_suspicious_broken_parts;
-```
-
 ## parts_to_throw_insert {#parts-to-throw-insert}
 
 If the number of active parts in a single partition exceeds the `parts_to_throw_insert` value, `INSERT` is interrupted with the `Too many parts (N). Merges are processing significantly slower than inserts` exception.
@@ -301,6 +387,8 @@ Default value: 10800
 
 ## try_fetch_recompressed_part_timeout
 
+Timeout (in seconds) before starting merge with recompression. During this time ClickHouse tries to fetch recompressed part from replica which assigned this merge with recompression.
+
 Recompression works slow in most cases, so we don't start merge with recompression until this timeout and trying to fetch recompressed part from replica which assigned this merge with recompression.
 
 Possible values:

docs/en/operations/settings/settings.md

@@ -1590,6 +1590,22 @@ Possible values:
 
 Default value: `default`.
 
+## parallel_replicas_custom_key_range_lower {#parallel_replicas_custom_key_range_lower}
+
+Allows the filter type `range` to split the work evenly between replicas based on the custom range `[parallel_replicas_custom_key_range_lower, INT_MAX]`. 
+
+When used in conjuction with [parallel_replicas_custom_key_range_upper](#parallel_replicas_custom_key_range_upper), it lets the filter evenly split the work over replicas for the range `[parallel_replicas_custom_key_range_lower, parallel_replicas_custom_key_range_upper]`. 
+
+Note: This setting will not cause any additional data to be filtered during query processing, rather it changes the points at which the range filter breaks up the range `[0, INT_MAX]` for parallel processing. 
+
+## parallel_replicas_custom_key_range_upper {#parallel_replicas_custom_key_range_upper}
+
+Allows the filter type `range` to split the work evenly between replicas based on the custom range `[0, parallel_replicas_custom_key_range_upper]`. A value of 0 disables the upper bound, setting it the max value of the custom key expression.
+
+When used in conjuction with [parallel_replicas_custom_key_range_lower](#parallel_replicas_custom_key_range_lower), it lets the filter evenly split the work over replicas for the range `[parallel_replicas_custom_key_range_lower, parallel_replicas_custom_key_range_upper]`. 
+
+Note: This setting will not cause any additional data to be filtered during query processing, rather it changes the points at which the range filter breaks up the range `[0, INT_MAX]` for parallel processing. 
+
 ## allow_experimental_parallel_reading_from_replicas
 
 Enables or disables sending SELECT queries to all replicas of a table (up to `max_parallel_replicas`). Reading is parallelized and coordinated dynamically. It will work for any kind of MergeTree table.
@@ -3862,6 +3878,10 @@ Possible values:
 
 Default value: 30.
 
+:::note
+It's applicable only to the default profile. A server reboot is required for the changes to take effect.
+:::
+
 ## http_receive_timeout {#http_receive_timeout}
 
 HTTP receive timeout (in seconds).

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

@@ -142,6 +142,34 @@ SELECT readWKTPoint('POINT (1.2 3.4)');
 (1.2,3.4)
 ```
 
+## readWKTLineString
+
+Parses a Well-Known Text (WKT) representation of a LineString geometry and returns it in the internal ClickHouse format.
+
+### Syntax
+
+```sql
+readWKTLineString(wkt_string)
+```
+
+### Arguments
+
+- `wkt_string`: The input WKT string representing a LineString geometry.
+
+### Returned value
+
+The function returns a ClickHouse internal representation of the linestring geometry.
+
+### Example
+
+```sql
+SELECT readWKTLineString('LINESTRING (1 1, 2 2, 3 3, 1 1)');
+```
+
+```response
+[(1,1),(2,2),(3,3),(1,1)]
+```
+
 ## readWKTRing
 
 Parses a Well-Known Text (WKT) representation of a Polygon geometry and returns a ring (closed linestring) in the internal ClickHouse format.
@@ -163,7 +191,7 @@ The function returns a ClickHouse internal representation of the ring (closed li
 ### Example
 
 ```sql
-SELECT readWKTRing('LINESTRING (1 1, 2 2, 3 3, 1 1)');
+SELECT readWKTRing('POLYGON ((1 1, 2 2, 3 3, 1 1))');
 ```
 
 ```response

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

@@ -212,7 +212,7 @@ toTypeName(x)
 
 ## blockSize {#blockSize}
 
-In ClickHouse, queries are processed in blocks (chunks).
+In ClickHouse, queries are processed in [blocks](../../development/architecture.md/#block-block) (chunks).
 This function returns the size (row count) of the block the function is called on.
 
 **Syntax**
@@ -221,6 +221,33 @@ This function returns the size (row count) of the block the function is called o
 blockSize()
 ```
 
+**Example**
+
+Query:
+
+```sql
+DROP TABLE IF EXISTS test;
+CREATE TABLE test (n UInt8) ENGINE = Memory;
+
+INSERT INTO test
+SELECT * FROM system.numbers LIMIT 5;
+
+SELECT blockSize()
+FROM test;
+```
+
+Result:
+
+```response
+   ┌─blockSize()─┐
+1. │           5 │
+2. │           5 │
+3. │           5 │
+4. │           5 │
+5. │           5 │
+   └─────────────┘ 
+```
+
 ## byteSize
 
 Returns an estimation of uncompressed byte size of its arguments in memory.
@@ -3688,3 +3715,108 @@ Result:
 ```response
 {'version':'1','serial_number':'2D9071D64530052D48308473922C7ADAFA85D6C5','signature_algo':'sha256WithRSAEncryption','issuer':'/CN=marsnet.local CA','not_before':'May  7 17:01:21 2024 GMT','not_after':'May  7 17:01:21 2025 GMT','subject':'/CN=chnode1','pkey_algo':'rsaEncryption'}
 ```
+
+## lowCardinalityIndices
+
+Returns the position of a value in the dictionary of a [LowCardinality](../data-types/lowcardinality.md) column. Positions start at 1. Since LowCardinality have per-part dictionaries, this function may return different positions for the same value in different parts.
+
+**Syntax**
+
+```sql
+lowCardinalityIndices(col)
+```
+
+**Arguments**
+
+- `col` — a low cardinality column. [LowCardinality](../data-types/lowcardinality.md).
+
+**Returned value**
+
+- The position of the value in the dictionary of the current part. [UInt64](../data-types/int-uint.md).
+
+**Example**
+
+Query:
+
+```sql
+DROP TABLE IF EXISTS test;
+CREATE TABLE test (s LowCardinality(String)) ENGINE = Memory;
+
+-- create two parts:
+
+INSERT INTO test VALUES ('ab'), ('cd'), ('ab'), ('ab'), ('df');
+INSERT INTO test VALUES ('ef'), ('cd'), ('ab'), ('cd'), ('ef');
+
+SELECT s, lowCardinalityIndices(s) FROM test;
+```
+
+Result:
+
+```response
+   ┌─s──┬─lowCardinalityIndices(s)─┐
+1. │ ab │                        1 │
+2. │ cd │                        2 │
+3. │ ab │                        1 │
+4. │ ab │                        1 │
+5. │ df │                        3 │
+   └────┴──────────────────────────┘
+    ┌─s──┬─lowCardinalityIndices(s)─┐
+ 6. │ ef │                        1 │
+ 7. │ cd │                        2 │
+ 8. │ ab │                        3 │
+ 9. │ cd │                        2 │
+10. │ ef │                        1 │
+    └────┴──────────────────────────┘
+```
+## lowCardinalityKeys
+
+Returns the dictionary values of a [LowCardinality](../data-types/lowcardinality.md) column. If the block is smaller or larger than the dictionary size, the result will be truncated or extended with default values. Since LowCardinality have per-part dictionaries, this function may return different dictionary values in different parts.
+
+**Syntax**
+
+```sql
+lowCardinalityIndices(col)
+```
+
+**Arguments**
+
+- `col` — a low cardinality column. [LowCardinality](../data-types/lowcardinality.md).
+
+**Returned value**
+
+- The dictionary keys. [UInt64](../data-types/int-uint.md).
+
+**Example**
+
+Query:
+
+```sql
+DROP TABLE IF EXISTS test;
+CREATE TABLE test (s LowCardinality(String)) ENGINE = Memory;
+
+-- create two parts:
+
+INSERT INTO test VALUES ('ab'), ('cd'), ('ab'), ('ab'), ('df');
+INSERT INTO test VALUES ('ef'), ('cd'), ('ab'), ('cd'), ('ef');
+
+SELECT s, lowCardinalityKeys(s) FROM test;
+```
+
+Result:
+
+```response
+   ┌─s──┬─lowCardinalityKeys(s)─┐
+1. │ ef │                       │
+2. │ cd │ ef                    │
+3. │ ab │ cd                    │
+4. │ cd │ ab                    │
+5. │ ef │                       │
+   └────┴───────────────────────┘
+    ┌─s──┬─lowCardinalityKeys(s)─┐
+ 6. │ ab │                       │
+ 7. │ cd │ ab                    │
+ 8. │ ab │ cd                    │
+ 9. │ ab │ df                    │
+10. │ df │                       │
+    └────┴───────────────────────┘
+```

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

@@ -6,49 +6,90 @@ sidebar_label: Rounding
 
 # Rounding Functions
 
-## floor(x\[, N\])
+## floor
 
-Returns the largest round number that is less than or equal to `x`. A round number is a multiple of 1/10N, or the nearest number of the appropriate data type if 1 / 10N isn’t exact.
+Returns the largest rounded number less than or equal `x`.
-‘N’ is an integer constant, optional parameter. By default it is zero, which means to round to an integer.
+A rounded number is a multiple of 1 / 10 * N, or the nearest number of the appropriate data type if 1 / 10 * N isn’t exact.
-‘N’ may be negative.
 
-Examples: `floor(123.45, 1) = 123.4, floor(123.45, -1) = 120.`
+Integer arguments may be rounded with negative `N` argument, with non-negative `N` the function returns `x`, i.e. does nothing.
 
-`x` is any numeric type. The result is a number of the same type.
+If rounding causes an overflow (for example, `floor(-128, -1)`), the result is undefined.
-For integer arguments, it makes sense to round with a negative `N` value (for non-negative `N`, the function does not do anything).
-If rounding causes overflow (for example, floor(-128, -1)), an implementation-specific result is returned.
 
-## ceil(x\[, N\]), ceiling(x\[, N\])
+**Syntax**
 
-Returns the smallest round number that is greater than or equal to `x`. In every other way, it is the same as the `floor` function (see above).
+``` sql
+floor(x[, N])
+```
 
-## trunc(x\[, N\]), truncate(x\[, N\])
+**Parameters**
 
-Returns the round number with largest absolute value that has an absolute value less than or equal to `x`‘s. In every other way, it is the same as the ’floor’ function (see above).
+- `x` - The value to round. [Float*](../data-types/float.md), [Decimal*](../data-types/decimal.md), or [(U)Int*](../data-types/int-uint.md).
+- `N` . [(U)Int*](../data-types/int-uint.md). The default is zero, which means rounding to an integer. Can be negative.
+
+**Returned value**
+
+A rounded number of the same type as `x`.
+
+**Examples**
+
+Query:
+
+```sql
+SELECT floor(123.45, 1) AS rounded
+```
+
+Result:
+
+```
+┌─rounded─┐
+│   123.4 │
+└─────────┘
+```
+
+Query:
+
+```sql
+SELECT floor(123.45, -1)
+```
+
+Result:
+
+```
+┌─rounded─┐
+│     120 │
+└─────────┘
+```
+
+## ceiling
+
+Like `floor` but returns the smallest rounded number greater than or equal `x`.
+
+**Syntax**
+
+``` sql
+ceiling(x[, N])
+```
+
+Alias: `ceil`
+
+## truncate
+
+Like `floor` but returns the rounded number with largest absolute value that has an absolute value less than or equal to `x`‘s.
 
 **Syntax**
 
 ```sql
-trunc(input, precision)
+truncate(x[, N])
 ```
 
-Alias: `truncate`.
+Alias: `trunc`.
-
-**Parameters**
-
-- `input`: A numeric type ([Float](../data-types/float.md), [Decimal](../data-types/decimal.md) or [Integer](../data-types/int-uint.md)).
-- `precision`: An [Integer](../data-types/int-uint.md) type.
-
-**Returned value**
-
-- A data type of `input`.
 
 **Example**
 
 Query:
 
 ```sql
-SELECT trunc(123.499, 1) as res;
+SELECT truncate(123.499, 1) as res;
 ```
 
 ```response
@@ -57,37 +98,40 @@ SELECT trunc(123.499, 1) as res;
 └───────┘
 ```
 
-## round(x\[, N\])
+## round
 
 Rounds a value to a specified number of decimal places.
 
-The function returns the nearest number of the specified order. In case when given number has equal distance to surrounding numbers, the function uses banker’s rounding for float number types and rounds away from zero for the other number types (Decimal).
+The function returns the nearest number of the specified order.
+If the input value has equal distance to two neighboring numbers, the function uses banker’s rounding for [Float*](../data-types/float.md) inputs and rounds away from zero for the other number types ([Decimal*](../data-types/decimal.md).
+
+**Syntax**
 
 ``` sql
-round(expression [, decimal_places])
+round(x[, N])
 ```
 
 **Arguments**
 
-- `expression` — A number to be rounded. Can be any [expression](../../sql-reference/syntax.md#syntax-expressions) returning the numeric [data type](../data-types/index.md#data_types).
+- `x` — A number to round. [Float*](../data-types/float.md), [Decimal*](../data-types/decimal.md), or [(U)Int*](../data-types/int-uint.md).
-- `decimal-places` — An integer value.
+- `N` — The number of decimal places to round to. Integer. Defaults to `0`.
-    - If `decimal-places > 0` then the function rounds the value to the right of the decimal point.
+    - If `N > 0`, the function rounds to the right of the decimal point.
-    - If `decimal-places < 0` then the function rounds the value to the left of the decimal point.
+    - If `N < 0`, the function rounds to the left of the decimal point.
-    - If `decimal-places = 0` then the function rounds the value to integer. In this case the argument can be omitted.
+    - If `N = 0`, the function rounds to the next integer.
 
 **Returned value:**
 
-The rounded number of the same type as the input number.
+A rounded number of the same type as `x`.
 
 **Examples**
 
-Example of usage with Float:
+Example with `Float` inputs:
 
-``` sql
+```sql
 SELECT number / 2 AS x, round(x) FROM system.numbers LIMIT 3;
 ```
 
-``` text
+```
 ┌───x─┬─round(divide(number, 2))─┐
 │   0 │                        0 │
 │ 0.5 │                        0 │
@@ -95,13 +139,13 @@ SELECT number / 2 AS x, round(x) FROM system.numbers LIMIT 3;
 └─────┴──────────────────────────┘
 ```
 
-Example of usage with Decimal:
+Example with `Decimal` inputs:
 
-``` sql
+```sql
 SELECT cast(number / 2 AS  Decimal(10,4)) AS x, round(x) FROM system.numbers LIMIT 3;
 ```
 
-``` text
+```
 ┌───x─┬─round(CAST(divide(number, 2), 'Decimal(10, 4)'))─┐
 │   0 │                                                0 │
 │ 0.5 │                                                1 │
@@ -109,14 +153,14 @@ SELECT cast(number / 2 AS  Decimal(10,4)) AS x, round(x) FROM system.numbers LIM
 └─────┴──────────────────────────────────────────────────┘
 ```
 
-If you want to keep the trailing zeros, you need to enable `output_format_decimal_trailing_zeros`
+To retain trailing zeros, enable setting `output_format_decimal_trailing_zeros`:
 
-``` sql
+```sql
 SELECT cast(number / 2 AS  Decimal(10,4)) AS x, round(x) FROM system.numbers LIMIT 3 settings output_format_decimal_trailing_zeros=1;
 
 ```
 
-``` text
+```
 ┌──────x─┬─round(CAST(divide(number, 2), 'Decimal(10, 4)'))─┐
 │ 0.0000 │                                           0.0000 │
 │ 0.5000 │                                           1.0000 │
@@ -151,9 +195,15 @@ round(3.65, 1) = 3.6
 
 Rounds a number to a specified decimal position.
 
-- If the rounding number is halfway between two numbers, the function uses banker’s rounding. Banker's rounding is a method of rounding fractional numbers. When the rounding number is halfway between two numbers, it's rounded to the nearest even digit at the specified decimal position. For example: 3.5 rounds up to 4, 2.5 rounds down to 2. It's the default rounding method for floating point numbers defined in [IEEE 754](https://en.wikipedia.org/wiki/IEEE_754#Roundings_to_nearest). The [round](#rounding_functions-round) function performs the same rounding for floating point numbers. The `roundBankers` function also rounds integers the same way, for example, `roundBankers(45, -1) = 40`.
+If the rounding number is halfway between two numbers, the function uses banker’s rounding.
+Banker's rounding is a method of rounding fractional numbers
+When the rounding number is halfway between two numbers, it's rounded to the nearest even digit at the specified decimal position.
+For example: 3.5 rounds up to 4, 2.5 rounds down to 2.
+It's the default rounding method for floating point numbers defined in [IEEE 754](https://en.wikipedia.org/wiki/IEEE_754#Roundings_to_nearest).
+The [round](#rounding_functions-round) function performs the same rounding for floating point numbers.
+The `roundBankers` function also rounds integers the same way, for example, `roundBankers(45, -1) = 40`.
 
-- In other cases, the function rounds numbers to the nearest integer.
+In other cases, the function rounds numbers to the nearest integer.
 
 Using banker’s rounding, you can reduce the effect that rounding numbers has on the results of summing or subtracting these numbers.
 
@@ -166,16 +216,20 @@ For example, sum numbers 1.5, 2.5, 3.5, 4.5 with different rounding:
 **Syntax**
 
 ``` sql
-roundBankers(expression [, decimal_places])
+roundBankers(x [, N])
 ```
 
 **Arguments**
 
-- `expression` — A number to be rounded. Can be any [expression](../../sql-reference/syntax.md#syntax-expressions) returning the numeric [data type](../data-types/index.md#data_types).
+    - `N > 0` — The function rounds the number to the given position right of the decimal point. Example: `roundBankers(3.55, 1) = 3.6`.
-- `decimal-places` — Decimal places. An integer number.
+    - `N < 0` — The function rounds the number to the given position left of the decimal point. Example: `roundBankers(24.55, -1) = 20`.
-    - `decimal-places > 0` — The function rounds the number to the given position right of the decimal point. Example: `roundBankers(3.55, 1) = 3.6`.
+    - `N = 0` — The function rounds the number to an integer. In this case the argument can be omitted. Example: `roundBankers(2.5) = 2`.
-    - `decimal-places < 0` — The function rounds the number to the given position left of the decimal point. Example: `roundBankers(24.55, -1) = 20`.
+
-    - `decimal-places = 0` — The function rounds the number to an integer. In this case the argument can be omitted. Example: `roundBankers(2.5) = 2`.
+- `x` — A number to round. [Float*](../data-types/float.md), [Decimal*](../data-types/decimal.md), or [(U)Int*](../data-types/int-uint.md).
+- `N` — The number of decimal places to round to. Integer. Defaults to `0`.
+    - If `N > 0`, the function rounds to the right of the decimal point.
+    - If `N < 0`, the function rounds to the left of the decimal point.
+    - If `N = 0`, the function rounds to the next integer.
 
 **Returned value**
 
@@ -185,13 +239,13 @@ A value rounded by the banker’s rounding method.
 
 Query:
 
-``` sql
+```sql
  SELECT number / 2 AS x, roundBankers(x, 0) AS b fROM system.numbers limit 10
 ```
 
 Result:
 
-``` text
+```
 ┌───x─┬─b─┐
 │   0 │ 0 │
 │ 0.5 │ 0 │
@@ -208,7 +262,7 @@ Result:
 
 Examples of Banker’s rounding:
 
-``` text
+```
 roundBankers(0.4) = 0
 roundBankers(-3.5) = -4
 roundBankers(4.5) = 4
@@ -264,7 +318,7 @@ Result:
 
 ## roundDuration
 
-Accepts a number. If the number is less than one, it returns `0`. Otherwise, it rounds the number down to numbers from the set of commonly used durations: `1, 10, 30, 60, 120, 180, 240, 300, 600, 1200, 1800, 3600, 7200, 18000, 36000`. 
+Accepts a number. If the number is less than one, it returns `0`. Otherwise, it rounds the number down to numbers from the set of commonly used durations: `1, 10, 30, 60, 120, 180, 240, 300, 600, 1200, 1800, 3600, 7200, 18000, 36000`.
 
 **Syntax**
 

docs/en/sql-reference/table-functions/azureBlobStorage.md

@@ -72,6 +72,7 @@ SELECT count(*) FROM azureBlobStorage('DefaultEndpointsProtocol=https;AccountNam
 - `_path` — Path to the file. Type: `LowCardinalty(String)`.
 - `_file` — Name of the file. Type: `LowCardinalty(String)`.
 - `_size` — Size of the file in bytes. Type: `Nullable(UInt64)`. If the file size is unknown, the value is `NULL`.
+- `_time` — Last modified time of the file. Type: `Nullable(DateTime)`. If the time is unknown, the value is `NULL`.
 
 **See Also**
 

docs/en/sql-reference/table-functions/file.md

@@ -196,6 +196,7 @@ SELECT count(*) FROM file('big_dir/**/file002', 'CSV', 'name String, value UInt3
 - `_path` — Path to the file. Type: `LowCardinalty(String)`.
 - `_file` — Name of the file. Type: `LowCardinalty(String)`.
 - `_size` — Size of the file in bytes. Type: `Nullable(UInt64)`. If the file size is unknown, the value is `NULL`.
+- `_time` — Last modified time of the file. Type: `Nullable(DateTime)`. If the time is unknown, the value is `NULL`.
 
 ## Settings {#settings}
 

docs/en/sql-reference/table-functions/hdfs.md

@@ -97,6 +97,7 @@ FROM hdfs('hdfs://hdfs1:9000/big_dir/file{0..9}{0..9}{0..9}', 'CSV', 'name Strin
 - `_path` — Path to the file. Type: `LowCardinalty(String)`.
 - `_file` — Name of the file. Type: `LowCardinalty(String)`.
 - `_size` — Size of the file in bytes. Type: `Nullable(UInt64)`. If the size is unknown, the value is `NULL`.
+- `_time` — Last modified time of the file. Type: `Nullable(DateTime)`. If the time is unknown, the value is `NULL`.
 
 ## Storage Settings {#storage-settings}
 

docs/en/sql-reference/table-functions/s3.md

@@ -272,6 +272,7 @@ FROM s3(
 - `_path` — Path to the file. Type: `LowCardinalty(String)`.
 - `_file` — Name of the file. Type: `LowCardinalty(String)`.
 - `_size` — Size of the file in bytes. Type: `Nullable(UInt64)`. If the file size is unknown, the value is `NULL`.
+- `_time` — Last modified time of the file. Type: `Nullable(DateTime)`. If the time is unknown, the value is `NULL`.
 
 ## Storage Settings {#storage-settings}
 

docs/en/sql-reference/table-functions/url.md

@@ -53,6 +53,7 @@ Character `|` inside patterns is used to specify failover addresses. They are it
 - `_path` — Path to the `URL`. Type: `LowCardinalty(String)`.
 - `_file` — Resource name of the `URL`. Type: `LowCardinalty(String)`.
 - `_size` — Size of the resource in bytes. Type: `Nullable(UInt64)`. If the size is unknown, the value is `NULL`.
+- `_time` — Last modified time of the file. Type: `Nullable(DateTime)`. If the time is unknown, the value is `NULL`.
 
 ## Storage Settings {#storage-settings}
 

docs/ru/sql-reference/functions/rounding-functions.md

@@ -10,7 +10,7 @@ sidebar_label: "Функции округления"
 
 Возвращает наибольшее круглое число, которое меньше или равно, чем x.
 Круглым называется число, кратное 1 / 10N или ближайшее к нему число соответствующего типа данных, если 1 / 10N не представимо точно.
-N - целочисленная константа, не обязательный параметр. По умолчанию - ноль, что означает - округлять до целого числа.
+N - целочисленный аргумент, не обязательный параметр. По умолчанию - ноль, что означает - округлять до целого числа.
 N может быть отрицательным.
 
 Примеры: `floor(123.45, 1) = 123.4, floor(123.45, -1) = 120.`

programs/server/Server.cpp

@@ -773,7 +773,27 @@ try
     LOG_INFO(log, "Available CPU instruction sets: {}", cpu_info);
 #endif
 
-    bool will_have_trace_collector = hasPHDRCache() && config().has("trace_log");
+    bool has_trace_collector = false;
+    /// Disable it if we collect test coverage information, because it will work extremely slow.
+#if !WITH_COVERAGE
+    /// Profilers cannot work reliably with any other libunwind or without PHDR cache.
+    has_trace_collector = hasPHDRCache() && config().has("trace_log");
+#endif
+
+    /// Describe multiple reasons when query profiler cannot work.
+
+#if WITH_COVERAGE
+    LOG_INFO(log, "Query Profiler and TraceCollector are disabled because they work extremely slow with test coverage.");
+#endif
+
+#if defined(SANITIZER)
+    LOG_INFO(log, "Query Profiler disabled because they cannot work under sanitizers"
+        " when two different stack unwinding methods will interfere with each other.");
+#endif
+
+    if (!hasPHDRCache())
+        LOG_INFO(log, "Query Profiler and TraceCollector are disabled because they require PHDR cache to be created"
+            " (otherwise the function 'dl_iterate_phdr' is not lock free and not async-signal safe).");
 
     // Initialize global thread pool. Do it before we fetch configs from zookeeper
     // nodes (`from_zk`), because ZooKeeper interface uses the pool. We will
@@ -782,8 +802,27 @@ try
         server_settings.max_thread_pool_size,
         server_settings.max_thread_pool_free_size,
         server_settings.thread_pool_queue_size,
-        will_have_trace_collector ? server_settings.global_profiler_real_time_period_ns : 0,
+        has_trace_collector ? server_settings.global_profiler_real_time_period_ns : 0,
-        will_have_trace_collector ? server_settings.global_profiler_cpu_time_period_ns : 0);
+        has_trace_collector ? server_settings.global_profiler_cpu_time_period_ns : 0);
+
+    if (has_trace_collector)
+    {
+        global_context->createTraceCollector();
+
+        /// Set up server-wide memory profiler (for total memory tracker).
+        if (server_settings.total_memory_profiler_step)
+            total_memory_tracker.setProfilerStep(server_settings.total_memory_profiler_step);
+
+        if (server_settings.total_memory_tracker_sample_probability > 0.0)
+            total_memory_tracker.setSampleProbability(server_settings.total_memory_tracker_sample_probability);
+
+        if (server_settings.total_memory_profiler_sample_min_allocation_size)
+            total_memory_tracker.setSampleMinAllocationSize(server_settings.total_memory_profiler_sample_min_allocation_size);
+
+        if (server_settings.total_memory_profiler_sample_max_allocation_size)
+            total_memory_tracker.setSampleMaxAllocationSize(server_settings.total_memory_profiler_sample_max_allocation_size);
+    }
+
     /// Wait for all threads to avoid possible use-after-free (for example logging objects can be already destroyed).
     SCOPE_EXIT({
         Stopwatch watch;
@@ -1950,52 +1989,9 @@ try
 
     LOG_DEBUG(log, "Loaded metadata.");
 
-    /// Init trace collector only after trace_log system table was created
+    if (has_trace_collector)
-    /// Disable it if we collect test coverage information, because it will work extremely slow.
-#if !WITH_COVERAGE
-    /// Profilers cannot work reliably with any other libunwind or without PHDR cache.
-    if (hasPHDRCache())
-    {
         global_context->initializeTraceCollector();
 
-        /// Set up server-wide memory profiler (for total memory tracker).
-        if (server_settings.total_memory_profiler_step)
-        {
-            total_memory_tracker.setProfilerStep(server_settings.total_memory_profiler_step);
-        }
-
-        if (server_settings.total_memory_tracker_sample_probability > 0.0)
-        {
-            total_memory_tracker.setSampleProbability(server_settings.total_memory_tracker_sample_probability);
-        }
-
-        if (server_settings.total_memory_profiler_sample_min_allocation_size)
-        {
-            total_memory_tracker.setSampleMinAllocationSize(server_settings.total_memory_profiler_sample_min_allocation_size);
-        }
-
-        if (server_settings.total_memory_profiler_sample_max_allocation_size)
-        {
-            total_memory_tracker.setSampleMaxAllocationSize(server_settings.total_memory_profiler_sample_max_allocation_size);
-        }
-    }
-#endif
-
-    /// Describe multiple reasons when query profiler cannot work.
-
-#if WITH_COVERAGE
-    LOG_INFO(log, "Query Profiler and TraceCollector are disabled because they work extremely slow with test coverage.");
-#endif
-
-#if defined(SANITIZER)
-    LOG_INFO(log, "Query Profiler disabled because they cannot work under sanitizers"
-        " when two different stack unwinding methods will interfere with each other.");
-#endif
-
-    if (!hasPHDRCache())
-        LOG_INFO(log, "Query Profiler and TraceCollector are disabled because they require PHDR cache to be created"
-            " (otherwise the function 'dl_iterate_phdr' is not lock free and not async-signal safe).");
-
 #if defined(OS_LINUX)
     auto tasks_stats_provider = TasksStatsCounters::findBestAvailableProvider();
     if (tasks_stats_provider == TasksStatsCounters::MetricsProvider::None)

programs/server/config.xml

@@ -1561,6 +1561,7 @@
     <rocksdb>
         <options>
             <max_background_jobs>8</max_background_jobs>
+            <info_log_level>DEBUG_LEVEL</info_log_level>
         </options>
         <column_family_options>
             <num_levels>2</num_levels>

src/Analyzer/ArrayJoinNode.cpp

@@ -1,5 +1,6 @@
 #include <Analyzer/ArrayJoinNode.h>
 #include <Analyzer/ColumnNode.h>
+#include <Analyzer/FunctionNode.h>
 #include <Analyzer/Utils.h>
 #include <IO/Operators.h>
 #include <IO/WriteBuffer.h>
@@ -64,7 +65,12 @@ ASTPtr ArrayJoinNode::toASTImpl(const ConvertToASTOptions & options) const
 
         auto * column_node = array_join_expression->as<ColumnNode>();
         if (column_node && column_node->getExpression())
-            array_join_expression_ast = column_node->getExpression()->toAST(options);
+        {
+            if (const auto * function_node = column_node->getExpression()->as<FunctionNode>(); function_node && function_node->getFunctionName() == "nested")
+                array_join_expression_ast = array_join_expression->toAST(options);
+            else
+                array_join_expression_ast = column_node->getExpression()->toAST(options);
+        }
         else
             array_join_expression_ast = array_join_expression->toAST(options);
 

src/Analyzer/Passes/OrderByLimitByDuplicateEliminationPass.cpp

@@ -22,6 +22,7 @@ public:
 
         if (query_node->hasOrderBy())
         {
+            QueryTreeNodeConstRawPtrWithHashSet unique_expressions_nodes_set;
             QueryTreeNodes result_nodes;
 
             auto & query_order_by_nodes = query_node->getOrderBy().getNodes();
@@ -45,10 +46,9 @@ public:
             query_order_by_nodes = std::move(result_nodes);
         }
 
-        unique_expressions_nodes_set.clear();
-
         if (query_node->hasLimitBy())
         {
+            QueryTreeNodeConstRawPtrWithHashSet unique_expressions_nodes_set;
             QueryTreeNodes result_nodes;
 
             auto & query_limit_by_nodes = query_node->getLimitBy().getNodes();
@@ -63,9 +63,6 @@ public:
             query_limit_by_nodes = std::move(result_nodes);
         }
     }
-
-private:
-    QueryTreeNodeConstRawPtrWithHashSet unique_expressions_nodes_set;
 };
 
 }

src/Columns/MaskOperations.cpp

@@ -289,10 +289,14 @@ void executeColumnIfNeeded(ColumnWithTypeAndName & column, bool empty)
     if (!column_function)
         return;
 
+    size_t original_size = column.column->size();
+
     if (!empty)
         column = column_function->reduce();
     else
-        column.column = column_function->getResultType()->createColumn();
+        column.column = column_function->getResultType()->createColumnConstWithDefaultValue(original_size)->convertToFullColumnIfConst();
+
+    chassert(column.column->size() == original_size);
 }
 
 int checkShortCircuitArguments(const ColumnsWithTypeAndName & arguments)

src/Common/QueryProfiler.cpp

@@ -228,9 +228,9 @@ void Timer::cleanup()
 #endif
 
 template <typename ProfilerImpl>
-QueryProfilerBase<ProfilerImpl>::QueryProfilerBase([[maybe_unused]] UInt64 thread_id, [[maybe_unused]] int clock_type, [[maybe_unused]] UInt32 period, [[maybe_unused]] int pause_signal_)
+QueryProfilerBase<ProfilerImpl>::QueryProfilerBase(
-    : log(getLogger("QueryProfiler"))
+    [[maybe_unused]] UInt64 thread_id, [[maybe_unused]] int clock_type, [[maybe_unused]] UInt32 period, [[maybe_unused]] int pause_signal_)
-    , pause_signal(pause_signal_)
+    : log(getLogger("QueryProfiler")), pause_signal(pause_signal_)
 {
 #if defined(SANITIZER)
     throw Exception(ErrorCodes::NOT_IMPLEMENTED, "QueryProfiler disabled because they cannot work under sanitizers");

src/Common/StringUtils.h

@@ -140,6 +140,18 @@ inline bool isPrintableASCII(char c)
     return uc >= 32 && uc <= 126;   /// 127 is ASCII DEL.
 }
 
+inline bool isCSIParameterByte(char c)
+{
+    uint8_t uc = c;
+    return uc >= 0x30 && uc <= 0x3F; /// ASCII 0–9:;<=>?
+}
+
+inline bool isCSIIntermediateByte(char c)
+{
+    uint8_t uc = c;
+    return uc >= 0x20 && uc <= 0x2F; /// ASCII !"#$%&'()*+,-./
+}
+
 inline bool isCSIFinalByte(char c)
 {
     uint8_t uc = c;

src/Common/UTF8Helpers.cpp

@@ -103,7 +103,7 @@ template <ComputeWidthMode mode>
 size_t computeWidthImpl(const UInt8 * data, size_t size, size_t prefix, size_t limit) noexcept
 {
     UTF8Decoder decoder;
-    int isEscapeSequence = false;
+    bool is_escape_sequence = false;
     size_t width = 0;
     size_t rollback = 0;
     for (size_t i = 0; i < size; ++i)
@@ -116,6 +116,9 @@ size_t computeWidthImpl(const UInt8 * data, size_t size, size_t prefix, size_t l
 
         while (i + 15 < size)
         {
+            if (is_escape_sequence)
+                break;
+
             __m128i bytes = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&data[i]));
 
             const uint16_t non_regular_width_mask = _mm_movemask_epi8(
@@ -132,25 +135,28 @@ size_t computeWidthImpl(const UInt8 * data, size_t size, size_t prefix, size_t l
             }
             else
             {
-                if (isEscapeSequence)
+                i += 16;
-                {
+                width += 16;
-                    break;
-                }
-                else
-                {
-                    i += 16;
-                    width += 16;
-                }
             }
         }
 #endif
 
         while (i < size && isPrintableASCII(data[i]))
         {
-            if (!isEscapeSequence)
+            bool ignore_width = is_escape_sequence && (isCSIParameterByte(data[i]) || isCSIIntermediateByte(data[i]));
+
+            if (ignore_width || (data[i] == '[' && is_escape_sequence))
+            {
+                /// don't count the width
+            }
+            else if (is_escape_sequence && isCSIFinalByte(data[i]))
+            {
+                is_escape_sequence = false;
+            }
+            else
+            {
                 ++width;
-            else if (isCSIFinalByte(data[i]) && data[i - 1] != '\x1b')
+            }
-                isEscapeSequence = false; /// end of CSI escape sequence reached
             ++i;
         }
 
@@ -178,7 +184,7 @@ size_t computeWidthImpl(const UInt8 * data, size_t size, size_t prefix, size_t l
                 // special treatment for '\t' and for ESC
                 size_t next_width = width;
                 if (decoder.codepoint == '\x1b')
-                    isEscapeSequence = true;
+                    is_escape_sequence = true;
                 else if (decoder.codepoint == '\t')
                     next_width += 8 - (prefix + width) % 8;
                 else

src/Core/Settings.h

@@ -202,6 +202,8 @@ class IColumn;
     M(UInt64, parallel_replica_offset, 0, "This is internal setting that should not be used directly and represents an implementation detail of the 'parallel replicas' mode. This setting will be automatically set up by the initiator server for distributed queries to the index of the replica participating in query processing among parallel replicas.", 0) \
     M(String, parallel_replicas_custom_key, "", "Custom key assigning work to replicas when parallel replicas are used.", 0) \
     M(ParallelReplicasCustomKeyFilterType, parallel_replicas_custom_key_filter_type, ParallelReplicasCustomKeyFilterType::DEFAULT, "Type of filter to use with custom key for parallel replicas. default - use modulo operation on the custom key, range - use range filter on custom key using all possible values for the value type of custom key.", 0) \
+    M(UInt64, parallel_replicas_custom_key_range_lower, 0, "Lower bound for the universe that the parallel replicas custom range filter is calculated over", 0) \
+    M(UInt64, parallel_replicas_custom_key_range_upper, 0, "Upper bound for the universe that the parallel replicas custom range filter is calculated over. A value of 0 disables the upper bound, setting it to the max value of the custom key expression", 0) \
     \
     M(String, cluster_for_parallel_replicas, "", "Cluster for a shard in which current server is located", 0) \
     M(UInt64, allow_experimental_parallel_reading_from_replicas, 0, "Use all the replicas from a shard for SELECT query execution. Reading is parallelized and coordinated dynamically. 0 - disabled, 1 - enabled, silently disable them in case of failure, 2 - enabled, throw an exception in case of failure", 0) \

src/Core/SettingsChangesHistory.h

@@ -100,7 +100,9 @@ static const std::map<ClickHouseVersion, SettingsChangesHistory::SettingsChanges
               {"allow_statistic_optimize", false, false, "Old setting which popped up here being renamed."},
               {"allow_experimental_statistic", false, false, "Old setting which popped up here being renamed."},
               {"allow_statistics_optimize", false, false, "The setting was renamed. The previous name is `allow_statistic_optimize`."},
-              {"allow_experimental_statistics", false, false, "The setting was renamed. The previous name is `allow_experimental_statistic`."}
+              {"allow_experimental_statistics", false, false, "The setting was renamed. The previous name is `allow_experimental_statistic`."},
+              {"parallel_replicas_custom_key_range_lower", 0, 0, "Add settings to control the range filter when using parallel replicas with dynamic shards"},
+              {"parallel_replicas_custom_key_range_upper", 0, 0, "Add settings to control the range filter when using parallel replicas with dynamic shards. A value of 0 disables the upper limit"},
               }},
     {"24.5", {{"allow_deprecated_error_prone_window_functions", true, false, "Allow usage of deprecated error prone window functions (neighbor, runningAccumulate, runningDifferenceStartingWithFirstValue, runningDifference)"},
               {"allow_experimental_join_condition", false, false, "Support join with inequal conditions which involve columns from both left and right table. e.g. t1.y < t2.y."},

src/DataTypes/DataTypeCustomGeo.cpp

@@ -17,6 +17,13 @@ void registerDataTypeDomainGeo(DataTypeFactory & factory)
             std::make_unique<DataTypeCustomDesc>(std::make_unique<DataTypePointName>()));
     });
 
+    // Custom type for simple line which consists from several segments.
+    factory.registerSimpleDataTypeCustom("LineString", []
+    {
+        return std::make_pair(DataTypeFactory::instance().get("Array(Point)"),
+            std::make_unique<DataTypeCustomDesc>(std::make_unique<DataTypeLineStringName>()));
+    });
+
     // Custom type for simple polygon without holes stored as Array(Point)
     factory.registerSimpleDataTypeCustom("Ring", []
     {

src/DataTypes/DataTypeCustomGeo.h

@@ -11,6 +11,12 @@ public:
     DataTypePointName() : DataTypeCustomFixedName("Point") {}
 };
 
+class DataTypeLineStringName : public DataTypeCustomFixedName
+{
+public:
+    DataTypeLineStringName() : DataTypeCustomFixedName("LineString") {}
+};
+
 class DataTypeRingName : public DataTypeCustomFixedName
 {
 public:

src/Databases/DatabaseAtomic.cpp

@@ -1,20 +1,21 @@
+#include <filesystem>
 #include <Databases/DatabaseAtomic.h>
+#include <Databases/DatabaseFactory.h>
 #include <Databases/DatabaseOnDisk.h>
 #include <Databases/DatabaseReplicated.h>
-#include <Databases/DatabaseFactory.h>
+#include <IO/ReadBufferFromFile.h>
 #include <IO/ReadHelpers.h>
 #include <IO/WriteHelpers.h>
-#include <IO/ReadBufferFromFile.h>
+#include <Interpreters/Context.h>
+#include <Interpreters/DDLTask.h>
+#include <Interpreters/DatabaseCatalog.h>
+#include <Interpreters/ExternalDictionariesLoader.h>
 #include <Parsers/formatAST.h>
+#include <Storages/StorageMaterializedView.h>
+#include "Common/logger_useful.h"
 #include <Common/PoolId.h>
 #include <Common/atomicRename.h>
 #include <Common/filesystemHelpers.h>
-#include <Storages/StorageMaterializedView.h>
-#include <Interpreters/Context.h>
-#include <Interpreters/DatabaseCatalog.h>
-#include <Interpreters/ExternalDictionariesLoader.h>
-#include <filesystem>
-#include <Interpreters/DDLTask.h>
 
 namespace fs = std::filesystem;
 
@@ -393,6 +394,7 @@ DatabaseAtomic::DetachedTables DatabaseAtomic::cleanupDetachedTables()
 {
     DetachedTables not_in_use;
     auto it = detached_tables.begin();
+    LOG_DEBUG(log, "There are {} detached tables. Start searching non used tables.", detached_tables.size());
     while (it != detached_tables.end())
     {
         if (it->second.unique())
@@ -403,6 +405,7 @@ DatabaseAtomic::DetachedTables DatabaseAtomic::cleanupDetachedTables()
         else
             ++it;
     }
+    LOG_DEBUG(log, "Found {} non used tables in detached tables.", not_in_use.size());
     /// It should be destroyed in caller with released database mutex
     return not_in_use;
 }

src/Databases/DatabaseOnDisk.cpp

@@ -794,7 +794,7 @@ ASTPtr DatabaseOnDisk::getCreateQueryFromStorage(const String & table_name, cons
         throw_on_error);
 
     create_table_query->set(create_table_query->as<ASTCreateQuery>()->comment,
-                            std::make_shared<ASTLiteral>("SYSTEM TABLE is built on the fly."));
+                            std::make_shared<ASTLiteral>(storage->getInMemoryMetadata().comment));
 
     return create_table_query;
 }

src/Databases/DatabaseReplicated.cpp

@@ -122,6 +122,13 @@ DatabaseReplicated::DatabaseReplicated(
         fillClusterAuthInfo(db_settings.collection_name.value, context_->getConfigRef());
 
     replica_group_name = context_->getConfigRef().getString("replica_group_name", "");
+
+    if (!replica_group_name.empty() && database_name.starts_with(DatabaseReplicated::ALL_GROUPS_CLUSTER_PREFIX))
+    {
+        context_->addWarningMessage(fmt::format("There's a Replicated database with a name starting from '{}', "
+                                                "and replica_group_name is configured. It may cause collisions in cluster names.",
+                                                ALL_GROUPS_CLUSTER_PREFIX));
+    }
 }
 
 String DatabaseReplicated::getFullReplicaName(const String & shard, const String & replica)
@@ -173,13 +180,40 @@ ClusterPtr DatabaseReplicated::tryGetCluster() const
     return cluster;
 }
 
-void DatabaseReplicated::setCluster(ClusterPtr && new_cluster)
+ClusterPtr DatabaseReplicated::tryGetAllGroupsCluster() const
 {
     std::lock_guard lock{mutex};
-    cluster = std::move(new_cluster);
+    if (replica_group_name.empty())
+        return nullptr;
+
+    if (cluster_all_groups)
+        return cluster_all_groups;
+
+    /// Database is probably not created or not initialized yet, it's ok to return nullptr
+    if (is_readonly)
+        return cluster_all_groups;
+
+    try
+    {
+        cluster_all_groups = getClusterImpl(/*all_groups*/ true);
+    }
+    catch (...)
+    {
+        tryLogCurrentException(log);
+    }
+    return cluster_all_groups;
 }
 
-ClusterPtr DatabaseReplicated::getClusterImpl() const
+void DatabaseReplicated::setCluster(ClusterPtr && new_cluster, bool all_groups)
+{
+    std::lock_guard lock{mutex};
+    if (all_groups)
+        cluster_all_groups = std::move(new_cluster);
+    else
+        cluster = std::move(new_cluster);
+}
+
+ClusterPtr DatabaseReplicated::getClusterImpl(bool all_groups) const
 {
     Strings unfiltered_hosts;
     Strings hosts;
@@ -199,17 +233,24 @@ ClusterPtr DatabaseReplicated::getClusterImpl() const
                             "It's possible if the first replica is not fully created yet "
                             "or if the last replica was just dropped or due to logical error", zookeeper_path);
 
-        hosts.clear();
+        if (all_groups)
-        std::vector<String> paths;
-        for (const auto & host : unfiltered_hosts)
-            paths.push_back(zookeeper_path + "/replicas/" + host + "/replica_group");
-
-        auto replica_groups = zookeeper->tryGet(paths);
-
-        for (size_t i = 0; i < paths.size(); ++i)
         {
-            if (replica_groups[i].data == replica_group_name)
+            hosts = unfiltered_hosts;
-                hosts.push_back(unfiltered_hosts[i]);
+        }
+        else
+        {
+            hosts.clear();
+            std::vector<String> paths;
+            for (const auto & host : unfiltered_hosts)
+                paths.push_back(zookeeper_path + "/replicas/" + host + "/replica_group");
+
+            auto replica_groups = zookeeper->tryGet(paths);
+
+            for (size_t i = 0; i < paths.size(); ++i)
+            {
+                if (replica_groups[i].data == replica_group_name)
+                    hosts.push_back(unfiltered_hosts[i]);
+            }
         }
 
         Int32 cversion = stat.cversion;
@@ -274,6 +315,11 @@ ClusterPtr DatabaseReplicated::getClusterImpl() const
 
     bool treat_local_as_remote = false;
     bool treat_local_port_as_remote = getContext()->getApplicationType() == Context::ApplicationType::LOCAL;
+
+    String cluster_name = TSA_SUPPRESS_WARNING_FOR_READ(database_name);     /// FIXME
+    if (all_groups)
+        cluster_name = ALL_GROUPS_CLUSTER_PREFIX + cluster_name;
+
     ClusterConnectionParameters params{
         cluster_auth_info.cluster_username,
         cluster_auth_info.cluster_password,
@@ -282,7 +328,7 @@ ClusterPtr DatabaseReplicated::getClusterImpl() const
         treat_local_port_as_remote,
         cluster_auth_info.cluster_secure_connection,
         Priority{1},
-        TSA_SUPPRESS_WARNING_FOR_READ(database_name),     /// FIXME
+        cluster_name,
         cluster_auth_info.cluster_secret};
 
     return std::make_shared<Cluster>(getContext()->getSettingsRef(), shards, params);

src/Databases/DatabaseReplicated.h

@@ -20,6 +20,8 @@ using ClusterPtr = std::shared_ptr<Cluster>;
 class DatabaseReplicated : public DatabaseAtomic
 {
 public:
+    static constexpr auto ALL_GROUPS_CLUSTER_PREFIX = "all_groups.";
+
     DatabaseReplicated(const String & name_, const String & metadata_path_, UUID uuid,
                        const String & zookeeper_path_, const String & shard_name_, const String & replica_name_,
                        DatabaseReplicatedSettings db_settings_,
@@ -65,6 +67,7 @@ public:
 
     /// Returns cluster consisting of database replicas
     ClusterPtr tryGetCluster() const;
+    ClusterPtr tryGetAllGroupsCluster() const;
 
     void drop(ContextPtr /*context*/) override;
 
@@ -113,8 +116,8 @@ private:
     ASTPtr parseQueryFromMetadataInZooKeeper(const String & node_name, const String & query);
     String readMetadataFile(const String & table_name) const;
 
-    ClusterPtr getClusterImpl() const;
+    ClusterPtr getClusterImpl(bool all_groups = false) const;
-    void setCluster(ClusterPtr && new_cluster);
+    void setCluster(ClusterPtr && new_cluster, bool all_groups = false);
 
     void createEmptyLogEntry(const ZooKeeperPtr & current_zookeeper);
 
@@ -155,6 +158,7 @@ private:
     UInt64 tables_metadata_digest TSA_GUARDED_BY(metadata_mutex);
 
     mutable ClusterPtr cluster;
+    mutable ClusterPtr cluster_all_groups;
 
     LoadTaskPtr startup_replicated_database_task TSA_GUARDED_BY(mutex);
 };

src/Databases/DatabaseReplicatedWorker.cpp

@@ -421,6 +421,8 @@ DDLTaskPtr DatabaseReplicatedDDLWorker::initAndCheckTask(const String & entry_na
     {
         /// Some replica is added or removed, let's update cached cluster
         database->setCluster(database->getClusterImpl());
+        if (!database->replica_group_name.empty())
+            database->setCluster(database->getClusterImpl(/*all_groups*/ true), /*all_groups*/ true);
         out_reason = fmt::format("Entry {} is a dummy task", entry_name);
         return {};
     }

src/Disks/IO/WriteBufferFromAzureBlobStorage.cpp

@@ -19,11 +19,15 @@ namespace ProfileEvents
 namespace DB
 {
 
+namespace ErrorCodes
+{
+    extern const int LOGICAL_ERROR;
+}
+
 struct WriteBufferFromAzureBlobStorage::PartData
 {
     Memory<> memory;
     size_t data_size = 0;
-    std::string block_id;
 };
 
 BufferAllocationPolicyPtr createBufferAllocationPolicy(const AzureObjectStorageSettings & settings)
@@ -119,22 +123,30 @@ void WriteBufferFromAzureBlobStorage::preFinalize()
     // This function should not be run again
     is_prefinalized = true;
 
+    hidePartialData();
+
+    if (hidden_size > 0)
+        detachBuffer();
+
+    setFakeBufferWhenPreFinalized();
+
     /// If there is only one block and size is less than or equal to max_single_part_upload_size
     /// then we use single part upload instead of multi part upload
-    if (buffer_allocation_policy->getBufferNumber() == 1)
+    if (block_ids.empty() && detached_part_data.size() == 1 && detached_part_data.front().data_size <= max_single_part_upload_size)
     {
-        size_t data_size = size_t(position() - memory.data());
+        auto part_data = std::move(detached_part_data.front());
-        if (data_size <= max_single_part_upload_size)
+        auto block_blob_client = blob_container_client->GetBlockBlobClient(blob_path);
-        {
+        Azure::Core::IO::MemoryBodyStream memory_stream(reinterpret_cast<const uint8_t *>(part_data.memory.data()), part_data.data_size);
-            auto block_blob_client = blob_container_client->GetBlockBlobClient(blob_path);
+        execWithRetry([&](){ block_blob_client.Upload(memory_stream); }, max_unexpected_write_error_retries, part_data.data_size);
-            Azure::Core::IO::MemoryBodyStream memory_stream(reinterpret_cast<const uint8_t *>(memory.data()), data_size);
+        LOG_TRACE(log, "Committed single block for blob `{}`", blob_path);
-            execWithRetry([&](){ block_blob_client.Upload(memory_stream); }, max_unexpected_write_error_retries, data_size);
-            LOG_TRACE(log, "Committed single block for blob `{}`", blob_path);
-            return;
-        }
-    }
 
-    writePart();
+        detached_part_data.pop_front();
+        return;
+    }
+    else
+    {
+        writeMultipartUpload();
+    }
 }
 
 void WriteBufferFromAzureBlobStorage::finalizeImpl()
@@ -144,9 +156,13 @@ void WriteBufferFromAzureBlobStorage::finalizeImpl()
     if (!is_prefinalized)
         preFinalize();
 
+    chassert(offset() == 0);
+    chassert(hidden_size == 0);
+
+    task_tracker->waitAll();
+
     if (!block_ids.empty())
     {
-        task_tracker->waitAll();
         auto block_blob_client = blob_container_client->GetBlockBlobClient(blob_path);
         execWithRetry([&](){ block_blob_client.CommitBlockList(block_ids); }, max_unexpected_write_error_retries);
         LOG_TRACE(log, "Committed {} blocks for blob `{}`", block_ids.size(), blob_path);
@@ -155,14 +171,66 @@ void WriteBufferFromAzureBlobStorage::finalizeImpl()
 
 void WriteBufferFromAzureBlobStorage::nextImpl()
 {
+    if (is_prefinalized)
+        throw Exception(
+            ErrorCodes::LOGICAL_ERROR,
+            "Cannot write to prefinalized buffer for Azure Blob Storage, the file could have been created");
+
     task_tracker->waitIfAny();
-    writePart();
+
+    hidePartialData();
+
+    reallocateFirstBuffer();
+
+    if (available() > 0)
+        return;
+
+    detachBuffer();
+
+    if (detached_part_data.size() > 1)
+        writeMultipartUpload();
+
     allocateBuffer();
 }
 
+void WriteBufferFromAzureBlobStorage::hidePartialData()
+{
+    if (write_settings.remote_throttler)
+        write_settings.remote_throttler->add(offset(), ProfileEvents::RemoteWriteThrottlerBytes, ProfileEvents::RemoteWriteThrottlerSleepMicroseconds);
+
+    chassert(memory.size() >= hidden_size + offset());
+
+    hidden_size += offset();
+    chassert(memory.data() + hidden_size == working_buffer.begin() + offset());
+    chassert(memory.data() + hidden_size == position());
+
+    WriteBuffer::set(memory.data() + hidden_size, memory.size() - hidden_size);
+    chassert(offset() == 0);
+}
+
+void WriteBufferFromAzureBlobStorage::reallocateFirstBuffer()
+{
+    chassert(offset() == 0);
+
+    if (buffer_allocation_policy->getBufferNumber() > 1 || available() > 0)
+        return;
+
+    const size_t max_first_buffer = buffer_allocation_policy->getBufferSize();
+    if (memory.size() == max_first_buffer)
+        return;
+
+    size_t size = std::min(memory.size() * 2, max_first_buffer);
+    memory.resize(size);
+
+    WriteBuffer::set(memory.data() + hidden_size, memory.size() - hidden_size);
+    chassert(offset() == 0);
+}
+
 void WriteBufferFromAzureBlobStorage::allocateBuffer()
 {
     buffer_allocation_policy->nextBuffer();
+    chassert(0 == hidden_size);
+
     auto size = buffer_allocation_policy->getBufferSize();
 
     if (buffer_allocation_policy->getBufferNumber() == 1)
@@ -172,30 +240,56 @@ void WriteBufferFromAzureBlobStorage::allocateBuffer()
     WriteBuffer::set(memory.data(), memory.size());
 }
 
-void WriteBufferFromAzureBlobStorage::writePart()
+void WriteBufferFromAzureBlobStorage::detachBuffer()
 {
-    auto data_size = size_t(position() - memory.data());
+    size_t data_size = size_t(position() - memory.data());
     if (data_size == 0)
         return;
 
-    const std::string & block_id = block_ids.emplace_back(getRandomASCIIString(64));
+    chassert(data_size == hidden_size);
-    std::shared_ptr<PartData> part_data = std::make_shared<PartData>(std::move(memory), data_size, block_id);
-    WriteBuffer::set(nullptr, 0);
 
-    auto upload_worker = [this, part_data] ()
+    auto buf = std::move(memory);
+
+    WriteBuffer::set(nullptr, 0);
+    total_size += hidden_size;
+    hidden_size = 0;
+
+    detached_part_data.push_back({std::move(buf), data_size});
+    WriteBuffer::set(nullptr, 0);
+}
+
+void WriteBufferFromAzureBlobStorage::writePart(WriteBufferFromAzureBlobStorage::PartData && part_data)
+{
+    const std::string & block_id = block_ids.emplace_back(getRandomASCIIString(64));
+    auto worker_data = std::make_shared<std::tuple<std::string, WriteBufferFromAzureBlobStorage::PartData>>(block_id, std::move(part_data));
+
+    auto upload_worker = [this, worker_data] ()
     {
+        auto & data_size = std::get<1>(*worker_data).data_size;
+        auto & data_block_id = std::get<0>(*worker_data);
         auto block_blob_client = blob_container_client->GetBlockBlobClient(blob_path);
 
-        Azure::Core::IO::MemoryBodyStream memory_stream(reinterpret_cast<const uint8_t *>(part_data->memory.data()), part_data->data_size);
+        Azure::Core::IO::MemoryBodyStream memory_stream(reinterpret_cast<const uint8_t *>(std::get<1>(*worker_data).memory.data()), data_size);
-        execWithRetry([&](){ block_blob_client.StageBlock(part_data->block_id, memory_stream); }, max_unexpected_write_error_retries, part_data->data_size);
+        execWithRetry([&](){ block_blob_client.StageBlock(data_block_id, memory_stream); }, max_unexpected_write_error_retries, data_size);
-
-        if (write_settings.remote_throttler)
-            write_settings.remote_throttler->add(part_data->data_size, ProfileEvents::RemoteWriteThrottlerBytes, ProfileEvents::RemoteWriteThrottlerSleepMicroseconds);
     };
 
     task_tracker->add(std::move(upload_worker));
 }
 
+void WriteBufferFromAzureBlobStorage::setFakeBufferWhenPreFinalized()
+{
+    WriteBuffer::set(fake_buffer_when_prefinalized, sizeof(fake_buffer_when_prefinalized));
+}
+
+void WriteBufferFromAzureBlobStorage::writeMultipartUpload()
+{
+    while (!detached_part_data.empty())
+    {
+        writePart(std::move(detached_part_data.front()));
+        detached_part_data.pop_front();
+    }
+}
+
 }
 
 #endif

src/Disks/IO/WriteBufferFromAzureBlobStorage.h

@@ -48,8 +48,13 @@ public:
 private:
     struct PartData;
 
-    void writePart();
+    void writeMultipartUpload();
+    void writePart(PartData && part_data);
+    void detachBuffer();
+    void reallocateFirstBuffer();
     void allocateBuffer();
+    void hidePartialData();
+    void setFakeBufferWhenPreFinalized();
 
     void finalizeImpl() override;
     void execWithRetry(std::function<void()> func, size_t num_tries, size_t cost = 0);
@@ -77,9 +82,16 @@ private:
 
     MemoryBufferPtr allocateBuffer() const;
 
+    char fake_buffer_when_prefinalized[1] = {};
+
     bool first_buffer=true;
 
+    size_t total_size = 0;
+    size_t hidden_size = 0;
+
     std::unique_ptr<TaskTracker> task_tracker;
+
+    std::deque<PartData> detached_part_data;
 };
 
 }

src/Disks/ObjectStorages/AzureBlobStorage/AzureObjectStorage.h

@@ -166,6 +166,8 @@ public:
         return client.get();
     }
 
+    bool supportParallelWrite() const override { return true; }
+
 private:
     using SharedAzureClientPtr = std::shared_ptr<const Azure::Storage::Blobs::BlobContainerClient>;
     void removeObjectImpl(const StoredObject & object, const SharedAzureClientPtr & client_ptr, bool if_exists);

src/Functions/FunctionHelpers.cpp

@@ -314,7 +314,7 @@ void checkFunctionArgumentSizes(const ColumnsWithTypeAndName & arguments, size_t
         if (current_size != input_rows_count)
             throw Exception(
                 ErrorCodes::LOGICAL_ERROR,
-                "Expected the argument nº#{} ('{}' of type {}) to have {} rows, but it has {}",
+                "Expected the argument №{} ('{}' of type {}) to have {} rows, but it has {}",
                 i + 1,
                 arguments[i].name,
                 arguments[i].type->getName(),

src/Functions/FunctionsRound.h

@@ -31,7 +31,6 @@ namespace DB
 
 namespace ErrorCodes
 {
-    extern const int NUMBER_OF_ARGUMENTS_DOESNT_MATCH;
     extern const int ILLEGAL_TYPE_OF_ARGUMENT;
     extern const int ARGUMENT_OUT_OF_BOUND;
     extern const int ILLEGAL_COLUMN;
@@ -40,26 +39,22 @@ namespace ErrorCodes
 }
 
 
-/** Rounding Functions:
+/// Rounding Functions:
-    * round(x, N) - rounding to nearest (N = 0 by default). Use banker's rounding for floating point numbers.
+/// - round(x, N) - rounding to nearest (N = 0 by default). Use banker's rounding for floating point numbers.
-    * roundBankers(x, N) - rounding to nearest (N = 0 by default). Use banker's rounding for all numbers.
+/// - roundBankers(x, N) - rounding to nearest (N = 0 by default). Use banker's rounding for all numbers.
-    * floor(x, N) is the largest number <= x (N = 0 by default).
+/// - floor(x, N) is the largest number <= x (N = 0 by default).
-    * ceil(x, N) is the smallest number >= x (N = 0 by default).
+/// - ceil(x, N) is the smallest number >= x (N = 0 by default).
-    * trunc(x, N) - is the largest by absolute value number that is not greater than x by absolute value (N = 0 by default).
+/// - trunc(x, N) - is the largest by absolute value number that is not greater than x by absolute value (N = 0 by default).
-    *
-    * The value of the parameter N (scale):
-    * - N > 0: round to the number with N decimal places after the decimal point
-    * - N < 0: round to an integer with N zero characters
-    * - N = 0: round to an integer
-    *
-    * Type of the result is the type of argument.
-    * For integer arguments, when passing negative scale, overflow can occur.
-    * In that case, the behavior is implementation specific.
-    */
 
+/// The value of the parameter N (scale):
+/// - N > 0: round to the number with N decimal places after the decimal point
+/// - N < 0: round to an integer with N zero characters
+/// - N = 0: round to an integer
 
-/** This parameter controls the behavior of the rounding functions.
+/// Type of the result is the type of argument.
-  */
+/// For integer arguments, when passing negative scale, overflow can occur. In that case, the behavior is undefined.
+
+/// Controls the behavior of the rounding functions.
 enum class ScaleMode : uint8_t
 {
     Positive,   // round to a number with N decimal places after the decimal point
@@ -75,7 +70,7 @@ enum class RoundingMode : uint8_t
     Ceil    = _MM_FROUND_TO_POS_INF | _MM_FROUND_NO_EXC,
     Trunc   = _MM_FROUND_TO_ZERO | _MM_FROUND_NO_EXC,
 #else
-    Round   = 8,    /// Values are correspond to above just in case.
+    Round   = 8,    /// Values correspond to above values, just in case.
     Floor   = 9,
     Ceil    = 10,
     Trunc   = 11,
@@ -84,16 +79,21 @@ enum class RoundingMode : uint8_t
 
 enum class TieBreakingMode : uint8_t
 {
-    Auto, // use banker's rounding for floating point numbers, round up otherwise
+    Auto,    /// banker's rounding for floating point numbers, round up otherwise
-    Bankers, // use banker's rounding
+    Bankers, /// banker's rounding
+};
+
+enum class Vectorize : uint8_t
+{
+    No,
+    Yes
 };
 
 /// For N, no more than the number of digits in the largest type.
 using Scale = Int16;
 
 
-/** Rounding functions for integer values.
+/// Rounding functions for integer values.
-  */
 template <typename T, RoundingMode rounding_mode, ScaleMode scale_mode, TieBreakingMode tie_breaking_mode>
 struct IntegerRoundingComputation
 {
@@ -149,6 +149,8 @@ struct IntegerRoundingComputation
                 return x;
             }
         }
+
+        std::unreachable();
     }
 
     static ALWAYS_INLINE T compute(T x, T scale)
@@ -161,9 +163,12 @@ struct IntegerRoundingComputation
             case ScaleMode::Negative:
                 return computeImpl(x, scale);
         }
+
+        std::unreachable();
     }
 
-    static ALWAYS_INLINE void compute(const T * __restrict in, size_t scale, T * __restrict out) requires std::integral<T>
+    static ALWAYS_INLINE void compute(const T * __restrict in, size_t scale, T * __restrict out)
+    requires std::integral<T>
     {
         if constexpr (sizeof(T) <= sizeof(scale) && scale_mode == ScaleMode::Negative)
         {
@@ -176,20 +181,23 @@ struct IntegerRoundingComputation
         *out = compute(*in, static_cast<T>(scale));
     }
 
-    static ALWAYS_INLINE void compute(const T * __restrict in, T scale, T * __restrict out) requires(!std::integral<T>)
+    static ALWAYS_INLINE void compute(const T * __restrict in, T scale, T * __restrict out)
+    requires(!std::integral<T>)
     {
         *out = compute(*in, scale);
     }
 };
 
 
+template <typename T, Vectorize vectorize>
+class FloatRoundingComputationBase;
+
 #ifdef __SSE4_1__
 
-template <typename T>
+/// Vectorized implementation for x86.
-class BaseFloatRoundingComputation;
 
 template <>
-class BaseFloatRoundingComputation<Float32>
+class FloatRoundingComputationBase<Float32, Vectorize::Yes>
 {
 public:
     using ScalarType = Float32;
@@ -210,7 +218,7 @@ public:
 };
 
 template <>
-class BaseFloatRoundingComputation<Float64>
+class FloatRoundingComputationBase<Float64, Vectorize::Yes>
 {
 public:
     using ScalarType = Float64;
@@ -230,9 +238,9 @@ public:
     }
 };
 
-#else
+#endif
 
-/// Implementation for ARM. Not vectorized.
+/// Sequential implementation for ARM. Also used for scalar arguments.
 
 inline float roundWithMode(float x, RoundingMode mode)
 {
@@ -243,6 +251,8 @@ inline float roundWithMode(float x, RoundingMode mode)
         case RoundingMode::Ceil: return ceilf(x);
         case RoundingMode::Trunc: return truncf(x);
     }
+
+    std::unreachable();
 }
 
 inline double roundWithMode(double x, RoundingMode mode)
@@ -254,10 +264,12 @@ inline double roundWithMode(double x, RoundingMode mode)
         case RoundingMode::Ceil: return ceil(x);
         case RoundingMode::Trunc: return trunc(x);
     }
+
+    std::unreachable();
 }
 
 template <typename T>
-class BaseFloatRoundingComputation
+class FloatRoundingComputationBase<T, Vectorize::No>
 {
 public:
     using ScalarType = T;
@@ -277,15 +289,13 @@ public:
     }
 };
 
-#endif
-
 
 /** Implementation of low-level round-off functions for floating-point values.
   */
-template <typename T, RoundingMode rounding_mode, ScaleMode scale_mode>
+template <typename T, RoundingMode rounding_mode, ScaleMode scale_mode, Vectorize vectorize>
-class FloatRoundingComputation : public BaseFloatRoundingComputation<T>
+class FloatRoundingComputation : public FloatRoundingComputationBase<T, vectorize>
 {
-    using Base = BaseFloatRoundingComputation<T>;
+    using Base = FloatRoundingComputationBase<T, vectorize>;
 
 public:
     static void compute(const T * __restrict in, const typename Base::VectorType & scale, T * __restrict out)
@@ -317,15 +327,22 @@ struct FloatRoundingImpl
 private:
     static_assert(!is_decimal<T>);
 
-    using Op = FloatRoundingComputation<T, rounding_mode, scale_mode>;
+    template <Vectorize vectorize =
-    using Data = std::array<T, Op::data_count>;
+#ifdef __SSE4_1__
+        Vectorize::Yes
+#else
+        Vectorize::No
+#endif
+    >
+    using Op = FloatRoundingComputation<T, rounding_mode, scale_mode, vectorize>;
+    using Data = std::array<T, Op<>::data_count>;
     using ColumnType = ColumnVector<T>;
     using Container = typename ColumnType::Container;
 
 public:
     static NO_INLINE void apply(const Container & in, size_t scale, Container & out)
     {
-        auto mm_scale = Op::prepare(scale);
+        auto mm_scale = Op<>::prepare(scale);
 
         const size_t data_count = std::tuple_size<Data>();
 
@@ -337,7 +354,7 @@ public:
 
         while (p_in < limit)
         {
-            Op::compute(p_in, mm_scale, p_out);
+            Op<>::compute(p_in, mm_scale, p_out);
             p_in += data_count;
             p_out += data_count;
         }
@@ -350,10 +367,17 @@ public:
             size_t tail_size_bytes = (end_in - p_in) * sizeof(*p_in);
 
             memcpy(&tmp_src, p_in, tail_size_bytes);
-            Op::compute(reinterpret_cast<T *>(&tmp_src), mm_scale, reinterpret_cast<T *>(&tmp_dst));
+            Op<>::compute(reinterpret_cast<T *>(&tmp_src), mm_scale, reinterpret_cast<T *>(&tmp_dst));
             memcpy(p_out, &tmp_dst, tail_size_bytes);
         }
     }
+
+    static void applyOne(T in, size_t scale, T& out)
+    {
+        using ScalarOp = Op<Vectorize::No>;
+        auto s = ScalarOp::prepare(scale);
+        ScalarOp::compute(&in, s, &out);
+    }
 };
 
 template <typename T, RoundingMode rounding_mode, ScaleMode scale_mode, TieBreakingMode tie_breaking_mode>
@@ -409,6 +433,11 @@ public:
                 throw Exception(ErrorCodes::BAD_ARGUMENTS, "Unexpected 'scale' parameter passed to function");
         }
     }
+
+    static void applyOne(T in, size_t scale, T& out)
+    {
+        Op::compute(&in, scale, &out);
+    }
 };
 
 
@@ -444,11 +473,40 @@ public:
             memcpy(out.data(), in.data(), in.size() * sizeof(T));
         }
     }
+
+    static void applyOne(NativeType in, UInt32 in_scale, NativeType& out, Scale scale_arg)
+    {
+        scale_arg = in_scale - scale_arg;
+        if (scale_arg > 0)
+        {
+            auto scale = intExp10OfSize<NativeType>(scale_arg);
+            Op::compute(&in, scale, &out);
+        }
+        else
+        {
+            memcpy(&out, &in, sizeof(T));
+        }
+    }
 };
 
+/// Select the appropriate processing algorithm depending on the scale.
+inline void validateScale(Int64 scale64)
+{
+    if (scale64 > std::numeric_limits<Scale>::max() || scale64 < std::numeric_limits<Scale>::min())
+        throw Exception(ErrorCodes::ARGUMENT_OUT_OF_BOUND, "Scale argument for rounding function is too large");
+}
 
-/** Select the appropriate processing algorithm depending on the scale.
+inline Scale getScaleArg(const ColumnConst* scale_col)
-  */
+{
+    const auto & scale_field = scale_col->getField();
+
+    Int64 scale64 = scale_field.get<Int64>();
+    validateScale(scale64);
+
+    return scale64;
+}
+
+/// Generic dispatcher
 template <typename T, RoundingMode rounding_mode, TieBreakingMode tie_breaking_mode>
 struct Dispatcher
 {
@@ -457,30 +515,65 @@ struct Dispatcher
         FloatRoundingImpl<T, rounding_mode, scale_mode>,
         IntegerRoundingImpl<T, rounding_mode, scale_mode, tie_breaking_mode>>;
 
-    static ColumnPtr apply(const IColumn * col_general, Scale scale_arg)
+    template <typename ScaleType>
+    static ColumnPtr apply(const IColumn * value_col, const IColumn * scale_col = nullptr)
     {
-        const auto & col = checkAndGetColumn<ColumnVector<T>>(*col_general);
+        const auto & value_col_typed = checkAndGetColumn<ColumnVector<T>>(*value_col);
         auto col_res = ColumnVector<T>::create();
 
         typename ColumnVector<T>::Container & vec_res = col_res->getData();
-        vec_res.resize(col.getData().size());
+        vec_res.resize(value_col_typed.getData().size());
 
         if (!vec_res.empty())
         {
-            if (scale_arg == 0)
+            if (scale_col == nullptr || isColumnConst(*scale_col))
             {
-                size_t scale = 1;
+                auto scale_arg = (scale_col == nullptr) ? 0 : getScaleArg(checkAndGetColumnConst<ColumnVector<ScaleType>>(scale_col));
-                FunctionRoundingImpl<ScaleMode::Zero>::apply(col.getData(), scale, vec_res);
+                if (scale_arg == 0)
+                {
+                    size_t scale = 1;
+                    FunctionRoundingImpl<ScaleMode::Zero>::apply(value_col_typed.getData(), scale, vec_res);
+                }
+                else if (scale_arg > 0)
+                {
+                    size_t scale = intExp10(scale_arg);
+                    FunctionRoundingImpl<ScaleMode::Positive>::apply(value_col_typed.getData(), scale, vec_res);
+                }
+                else
+                {
+                    size_t scale = intExp10(-scale_arg);
+                    FunctionRoundingImpl<ScaleMode::Negative>::apply(value_col_typed.getData(), scale, vec_res);
+                }
             }
-            else if (scale_arg > 0)
+            /// Non-const scale argument:
+            else if (const auto * scale_col_typed = checkAndGetColumn<ColumnVector<ScaleType>>(scale_col))
             {
-                size_t scale = intExp10(scale_arg);
+                const auto & value_data = value_col_typed.getData();
-                FunctionRoundingImpl<ScaleMode::Positive>::apply(col.getData(), scale, vec_res);
+                const auto & scale_data = scale_col_typed->getData();
-            }
+                const size_t rows = value_data.size();
-            else
+
-            {
+                for (size_t i = 0; i < rows; ++i)
-                size_t scale = intExp10(-scale_arg);
+                {
-                FunctionRoundingImpl<ScaleMode::Negative>::apply(col.getData(), scale, vec_res);
+                    Int64 scale64 = scale_data[i];
+                    validateScale(scale64);
+                    Scale raw_scale = scale64;
+
+                    if (raw_scale == 0)
+                    {
+                        size_t scale = 1;
+                        FunctionRoundingImpl<ScaleMode::Zero>::applyOne(value_data[i], scale, vec_res[i]);
+                    }
+                    else if (raw_scale > 0)
+                    {
+                        size_t scale = intExp10(raw_scale);
+                        FunctionRoundingImpl<ScaleMode::Positive>::applyOne(value_data[i], scale, vec_res[i]);
+                    }
+                    else
+                    {
+                        size_t scale = intExp10(-raw_scale);
+                        FunctionRoundingImpl<ScaleMode::Negative>::applyOne(value_data[i], scale, vec_res[i]);
+                    }
+                }
             }
         }
 
@@ -488,28 +581,51 @@ struct Dispatcher
     }
 };
 
+/// Dispatcher for Decimal inputs
 template <is_decimal T, RoundingMode rounding_mode, TieBreakingMode tie_breaking_mode>
 struct Dispatcher<T, rounding_mode, tie_breaking_mode>
 {
 public:
-    static ColumnPtr apply(const IColumn * col_general, Scale scale_arg)
+    template <typename ScaleType>
+    static ColumnPtr apply(const IColumn * value_col, const IColumn * scale_col = nullptr)
     {
-        const auto & col = checkAndGetColumn<ColumnDecimal<T>>(*col_general);
+        const auto & value_col_typed = checkAndGetColumn<ColumnDecimal<T>>(*value_col);
-        const typename ColumnDecimal<T>::Container & vec_src = col.getData();
+        const typename ColumnDecimal<T>::Container & vec_src = value_col_typed.getData();
 
-        auto col_res = ColumnDecimal<T>::create(vec_src.size(), col.getScale());
+        auto col_res = ColumnDecimal<T>::create(vec_src.size(), value_col_typed.getScale());
         auto & vec_res = col_res->getData();
 
         if (!vec_res.empty())
-            DecimalRoundingImpl<T, rounding_mode, tie_breaking_mode>::apply(col.getData(), col.getScale(), vec_res, scale_arg);
+        {
+            if (scale_col == nullptr || isColumnConst(*scale_col))
+            {
+                auto scale_arg = scale_col == nullptr ? 0 : getScaleArg(checkAndGetColumnConst<ColumnVector<ScaleType>>(scale_col));
+                DecimalRoundingImpl<T, rounding_mode, tie_breaking_mode>::apply(value_col_typed.getData(), value_col_typed.getScale(), vec_res, scale_arg);
+            }
+            /// Non-const scale argument
+            else if (const auto * scale_col_typed = checkAndGetColumn<ColumnVector<ScaleType>>(scale_col))
+            {
+                const auto & scale = scale_col_typed->getData();
+                const size_t rows = vec_src.size();
+
+                for (size_t i = 0; i < rows; ++i)
+                {
+                    Int64 scale64 = scale[i];
+                    validateScale(scale64);
+                    Scale raw_scale = scale64;
+
+                    DecimalRoundingImpl<T, rounding_mode, tie_breaking_mode>::applyOne(value_col_typed.getElement(i), value_col_typed.getScale(),
+                        reinterpret_cast<ColumnDecimal<T>::NativeT&>(col_res->getElement(i)), raw_scale);
+                }
+            }
+        }
 
         return col_res;
     }
 };
 
-/** A template for functions that round the value of an input parameter of type
+/// Functions that round the value of an input parameter of type (U)Int8/16/32/64, Float32/64 or Decimal32/64/128.
-  * (U)Int8/16/32/64, Float32/64 or Decimal32/64/128, and accept an additional optional parameter (default is 0).
+/// Accept an additional optional parameter of type (U)Int8/16/32/64 (0 by default).
-  */
 template <typename Name, RoundingMode rounding_mode, TieBreakingMode tie_breaking_mode>
 class FunctionRounding : public IFunction
 {
@@ -517,75 +633,58 @@ public:
     static constexpr auto name = Name::name;
     static FunctionPtr create(ContextPtr) { return std::make_shared<FunctionRounding>(); }
 
-    String getName() const override
+    String getName() const override { return name; }
-    {
-        return name;
-    }
-
     bool isVariadic() const override { return true; }
     size_t getNumberOfArguments() const override { return 0; }
     bool isSuitableForShortCircuitArgumentsExecution(const DataTypesWithConstInfo & /*arguments*/) const override { return false; }
-
-    /// Get result types by argument types. If the function does not apply to these arguments, throw an exception.
-    DataTypePtr getReturnTypeImpl(const DataTypes & arguments) const override
-    {
-        if ((arguments.empty()) || (arguments.size() > 2))
-            throw Exception(ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH,
-                "Number of arguments for function {} doesn't match: passed {}, should be 1 or 2.",
-                getName(), arguments.size());
-
-        for (const auto & type : arguments)
-            if (!isNumber(type))
-                throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Illegal type {} of argument of function {}",
-                    arguments[0]->getName(), getName());
-
-        return arguments[0];
-    }
-
-    static Scale getScaleArg(const ColumnsWithTypeAndName & arguments)
-    {
-        if (arguments.size() == 2)
-        {
-            const IColumn & scale_column = *arguments[1].column;
-            if (!isColumnConst(scale_column))
-                throw Exception(ErrorCodes::ILLEGAL_COLUMN, "Scale argument for rounding functions must be constant");
-
-            Field scale_field = assert_cast<const ColumnConst &>(scale_column).getField();
-            if (scale_field.getType() != Field::Types::UInt64
-                && scale_field.getType() != Field::Types::Int64)
-                throw Exception(ErrorCodes::ILLEGAL_COLUMN, "Scale argument for rounding functions must have integer type");
-
-            Int64 scale64 = scale_field.get<Int64>();
-            if (scale64 > std::numeric_limits<Scale>::max()
-                || scale64 < std::numeric_limits<Scale>::min())
-                throw Exception(ErrorCodes::ARGUMENT_OUT_OF_BOUND, "Scale argument for rounding function is too large");
-
-            return scale64;
-        }
-        return 0;
-    }
-
     bool useDefaultImplementationForConstants() const override { return true; }
-    ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {1}; }
+
+    DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override
+    {
+        FunctionArgumentDescriptors mandatory_args{
+            {"x", static_cast<FunctionArgumentDescriptor::TypeValidator>(&isNumber), nullptr, "A number to round"},
+        };
+        FunctionArgumentDescriptors optional_args{
+            {"N", static_cast<FunctionArgumentDescriptor::TypeValidator>(&isNativeInteger), nullptr, "The number of decimal places to round to"},
+        };
+        validateFunctionArgumentTypes(*this, arguments, mandatory_args, optional_args);
+
+        return arguments[0].type;
+    }
 
     ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override
     {
-        const ColumnWithTypeAndName & column = arguments[0];
+        const ColumnWithTypeAndName & value_arg = arguments[0];
-        Scale scale_arg = getScaleArg(arguments);
 
         ColumnPtr res;
-        auto call = [&](const auto & types) -> bool
+        auto call_data = [&](const auto & types) -> bool
         {
             using Types = std::decay_t<decltype(types)>;
-            using DataType = typename Types::LeftType;
+            using DataType = typename Types::RightType;
 
-            if constexpr (IsDataTypeNumber<DataType> || IsDataTypeDecimal<DataType>)
+            if (arguments.size() > 1)
             {
-                using FieldType = typename DataType::FieldType;
+                const ColumnWithTypeAndName & scale_column = arguments[1];
-                res = Dispatcher<FieldType, rounding_mode, tie_breaking_mode>::apply(column.column.get(), scale_arg);
+
+                auto call_scale = [&](const auto & scaleTypes) -> bool
+                {
+                    using ScaleTypes = std::decay_t<decltype(scaleTypes)>;
+                    using ScaleType = typename ScaleTypes::RightType;
+
+                    if (isColumnConst(*value_arg.column) && !isColumnConst(*scale_column.column))
+                        throw Exception(ErrorCodes::ILLEGAL_COLUMN, "Scale column must be const for const data column");
+
+                    res = Dispatcher<DataType, rounding_mode, tie_breaking_mode>::template apply<ScaleType>(value_arg.column.get(), scale_column.column.get());
+                    return true;
+                };
+
+                TypeIndex right_index = scale_column.type->getTypeId();
+                if (!callOnBasicType<void, true, false, false, false>(right_index, call_scale))
+                    throw Exception(ErrorCodes::ILLEGAL_COLUMN, "Scale argument for rounding functions must have integer type");
                 return true;
             }
-            return false;
+            res = Dispatcher<DataType, rounding_mode, tie_breaking_mode>::template apply<int>(value_arg.column.get());
+            return true;
         };
 
 #if !defined(__SSE4_1__)
@@ -597,10 +696,9 @@ public:
                 throw Exception(ErrorCodes::CANNOT_SET_ROUNDING_MODE, "Cannot set floating point rounding mode");
 #endif
 
-        if (!callOnIndexAndDataType<void>(column.type->getTypeId(), call))
+        TypeIndex left_index = value_arg.type->getTypeId();
-        {
+        if (!callOnBasicType<void, true, true, true, false>(left_index, call_data))
-            throw Exception(ErrorCodes::ILLEGAL_COLUMN, "Illegal column {} of argument of function {}", column.name, getName());
+            throw Exception(ErrorCodes::ILLEGAL_COLUMN, "Illegal column {} of argument of function {}", value_arg.name, getName());
-        }
 
         return res;
     }
@@ -617,9 +715,8 @@ public:
 };
 
 
-/** Rounds down to a number within explicitly specified array.
+/// Rounds down to a number within explicitly specified array.
-  * If the value is less than the minimal bound - returns the minimal bound.
+/// If the value is less than the minimal bound - returns the minimal bound.
-  */
 class FunctionRoundDown : public IFunction
 {
 public:
@@ -627,7 +724,6 @@ public:
     static FunctionPtr create(ContextPtr) { return std::make_shared<FunctionRoundDown>(); }
 
     String getName() const override { return name; }
-
     bool isVariadic() const override { return false; }
     size_t getNumberOfArguments() const override { return 2; }
     bool useDefaultImplementationForConstants() const override { return true; }

src/Functions/geometryConverters.h

@@ -28,6 +28,9 @@ namespace ErrorCodes
     extern const int ILLEGAL_TYPE_OF_ARGUMENT;
 }
 
+template <typename Point>
+using LineString = boost::geometry::model::linestring<Point>;
+
 template <typename Point>
 using Ring = boost::geometry::model::ring<Point>;
 
@@ -38,11 +41,13 @@ template <typename Point>
 using MultiPolygon = boost::geometry::model::multi_polygon<Polygon<Point>>;
 
 using CartesianPoint = boost::geometry::model::d2::point_xy<Float64>;
+using CartesianLineString = LineString<CartesianPoint>;
 using CartesianRing = Ring<CartesianPoint>;
 using CartesianPolygon = Polygon<CartesianPoint>;
 using CartesianMultiPolygon = MultiPolygon<CartesianPoint>;
 
 using SphericalPoint = boost::geometry::model::point<Float64, 2, boost::geometry::cs::spherical_equatorial<boost::geometry::degree>>;
+using SphericalLineString = LineString<SphericalPoint>;
 using SphericalRing = Ring<SphericalPoint>;
 using SphericalPolygon = Polygon<SphericalPoint>;
 using SphericalMultiPolygon = MultiPolygon<SphericalPoint>;
@@ -85,6 +90,29 @@ struct ColumnToPointsConverter
     }
 };
 
+
+/**
+ * Class which converts Column with type Array(Tuple(Float64, Float64)) to a vector of boost linestring type.
+*/
+template <typename Point>
+struct ColumnToLineStringsConverter
+{
+    static std::vector<LineString<Point>> convert(ColumnPtr col)
+    {
+        const IColumn::Offsets & offsets = typeid_cast<const ColumnArray &>(*col).getOffsets();
+        size_t prev_offset = 0;
+        std::vector<LineString<Point>> answer;
+        answer.reserve(offsets.size());
+        auto tmp = ColumnToPointsConverter<Point>::convert(typeid_cast<const ColumnArray &>(*col).getDataPtr());
+        for (size_t offset : offsets)
+        {
+            answer.emplace_back(tmp.begin() + prev_offset, tmp.begin() + offset);
+            prev_offset = offset;
+        }
+        return answer;
+    }
+};
+
 /**
  * Class which converts Column with type Array(Tuple(Float64, Float64)) to a vector of boost ring type.
 */
@@ -208,6 +236,39 @@ private:
     ColumnFloat64::Container & second_container;
 };
 
+/// Serialize Point, LineString as LineString
+template <typename Point>
+class LineStringSerializer
+{
+public:
+    LineStringSerializer()
+        : offsets(ColumnUInt64::create())
+    {}
+
+    explicit LineStringSerializer(size_t n)
+        : offsets(ColumnUInt64::create(n))
+    {}
+
+    void add(const LineString<Point> & ring)
+    {
+        size += ring.size();
+        offsets->insertValue(size);
+        for (const auto & point : ring)
+            point_serializer.add(point);
+    }
+
+    ColumnPtr finalize()
+    {
+        return ColumnArray::create(point_serializer.finalize(), std::move(offsets));
+    }
+
+private:
+    size_t size = 0;
+    PointSerializer<Point> point_serializer;
+    ColumnUInt64::MutablePtr offsets;
+};
+
+/// Almost the same as LineStringSerializer
 /// Serialize Point, Ring as Ring
 template <typename Point>
 class RingSerializer
@@ -344,8 +405,16 @@ static void callOnGeometryDataType(DataTypePtr type, F && f)
     /// There is no Point type, because for most of geometry functions it is useless.
     if (factory.get("Point")->equals(*type))
         return f(ConverterType<ColumnToPointsConverter<Point>>());
+
+    /// We should take the name into consideration to avoid ambiguity.
+    /// Because for example both Ring and LineString are resolved to Array(Tuple(Point)).
+    else if (factory.get("LineString")->equals(*type) && type->getCustomName() && type->getCustomName()->getName() == "LineString")
+        return f(ConverterType<ColumnToLineStringsConverter<Point>>());
+
+    /// For backward compatibility if we call this function not on a custom type, we will consider Array(Tuple(Point)) as type Ring.
     else if (factory.get("Ring")->equals(*type))
         return f(ConverterType<ColumnToRingsConverter<Point>>());
+
     else if (factory.get("Polygon")->equals(*type))
         return f(ConverterType<ColumnToPolygonsConverter<Point>>());
     else if (factory.get("MultiPolygon")->equals(*type))

src/Functions/polygonsIntersection.cpp

@@ -73,6 +73,8 @@ public:
 
             if constexpr (std::is_same_v<ColumnToPointsConverter<Point>, LeftConverter> || std::is_same_v<ColumnToPointsConverter<Point>, RightConverter>)
                 throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Any argument of function {} must not be Point", getName());
+            else if constexpr (std::is_same_v<ColumnToLineStringsConverter<Point>, LeftConverter> || std::is_same_v<ColumnToLineStringsConverter<Point>, RightConverter>)
+                throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Any argument of function {} must not be LineString", getName());
             else
             {
                 auto first = LeftConverter::convert(arguments[0].column->convertToFullColumnIfConst());

src/Functions/polygonsSymDifference.cpp

@@ -71,6 +71,8 @@ public:
 
             if constexpr (std::is_same_v<ColumnToPointsConverter<Point>, LeftConverter> || std::is_same_v<ColumnToPointsConverter<Point>, RightConverter>)
                 throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Any argument of function {} must not be Point", getName());
+            else if constexpr (std::is_same_v<ColumnToLineStringsConverter<Point>, LeftConverter> || std::is_same_v<ColumnToLineStringsConverter<Point>, RightConverter>)
+                throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Any argument of function {} must not be LineString", getName());
             else
             {
                 auto first = LeftConverter::convert(arguments[0].column->convertToFullColumnIfConst());

src/Functions/polygonsUnion.cpp

@@ -71,6 +71,8 @@ public:
 
             if constexpr (std::is_same_v<ColumnToPointsConverter<Point>, LeftConverter> || std::is_same_v<ColumnToPointsConverter<Point>, RightConverter>)
                 throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Any argument of function {} must not be Point", getName());
+            else if constexpr (std::is_same_v<ColumnToLineStringsConverter<Point>, LeftConverter> || std::is_same_v<ColumnToLineStringsConverter<Point>, RightConverter>)
+                throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Any argument of function {} must not be LineString", getName());
             else
             {
                 auto first = LeftConverter::convert(arguments[0].column->convertToFullColumnIfConst());

src/Functions/polygonsWithin.cpp

@@ -75,6 +75,8 @@ public:
 
             if constexpr (std::is_same_v<ColumnToPointsConverter<Point>, LeftConverter> || std::is_same_v<ColumnToPointsConverter<Point>, RightConverter>)
                 throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Any argument of function {} must not be Point", getName());
+            else if constexpr (std::is_same_v<ColumnToLineStringsConverter<Point>, LeftConverter> || std::is_same_v<ColumnToLineStringsConverter<Point>, RightConverter>)
+                throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Any argument of function {} must not be LineString", getName());
             else
             {
                 auto first = LeftConverter::convert(arguments[0].column->convertToFullColumnIfConst());

src/Functions/readWkt.cpp

@@ -82,6 +82,11 @@ struct ReadWKTPointNameHolder
     static constexpr const char * name = "readWKTPoint";
 };
 
+struct ReadWKTLineStringNameHolder
+{
+    static constexpr const char * name = "readWKTLineString";
+};
+
 struct ReadWKTRingNameHolder
 {
     static constexpr const char * name = "readWKTRing";
@@ -102,6 +107,30 @@ struct ReadWKTMultiPolygonNameHolder
 REGISTER_FUNCTION(ReadWKT)
 {
     factory.registerFunction<FunctionReadWKT<DataTypePointName, CartesianPoint, PointSerializer<CartesianPoint>, ReadWKTPointNameHolder>>();
+    factory.registerFunction<FunctionReadWKT<DataTypeLineStringName, CartesianLineString, LineStringSerializer<CartesianPoint>, ReadWKTLineStringNameHolder>>(FunctionDocumentation
+    {
+        .description=R"(
+Parses a Well-Known Text (WKT) representation of a LineString geometry and returns it in the internal ClickHouse format.
+)",
+        .syntax = "readWKTLineString(wkt_string)",
+        .arguments{
+            {"wkt_string", "The input WKT string representing a LineString geometry."}
+        },
+        .returned_value = "The function returns a ClickHouse internal representation of the linestring geometry.",
+        .examples{
+            {"first call", "SELECT readWKTLineString('LINESTRING (1 1, 2 2, 3 3, 1 1)');", R"(
+┌─readWKTLineString('LINESTRING (1 1, 2 2, 3 3, 1 1)')─┐
+│ [(1,1),(2,2),(3,3),(1,1)]                            │
+└──────────────────────────────────────────────────────┘
+            )"},
+            {"second call", "SELECT toTypeName(readWKTLineString('LINESTRING (1 1, 2 2, 3 3, 1 1)'));", R"(
+┌─toTypeName(readWKTLineString('LINESTRING (1 1, 2 2, 3 3, 1 1)'))─┐
+│ LineString                                                       │
+└──────────────────────────────────────────────────────────────────┘
+            )"},
+        },
+        .categories{"Unique identifiers"}
+    });
     factory.registerFunction<FunctionReadWKT<DataTypeRingName, CartesianRing, RingSerializer<CartesianPoint>, ReadWKTRingNameHolder>>();
     factory.registerFunction<FunctionReadWKT<DataTypePolygonName, CartesianPolygon, PolygonSerializer<CartesianPoint>, ReadWKTPolygonNameHolder>>();
     factory.registerFunction<FunctionReadWKT<DataTypeMultiPolygonName, CartesianMultiPolygon, MultiPolygonSerializer<CartesianPoint>, ReadWKTMultiPolygonNameHolder>>();

src/Functions/tupleConcat.cpp

@@ -61,7 +61,7 @@ public:
         return std::make_shared<DataTypeTuple>(tuple_arg_types);
     }
 
-    ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t) const override
+    ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override
     {
         const size_t num_arguments = arguments.size();
         Columns columns;
@@ -92,6 +92,9 @@ public:
                 columns.push_back(inner_col);
         }
 
+        if (columns.empty())
+            return ColumnTuple::create(input_rows_count);
+
         return ColumnTuple::create(columns);
     }
 };

src/Functions/wkt.cpp

@@ -41,6 +41,14 @@ public:
 
     bool isSuitableForShortCircuitArgumentsExecution(const DataTypesWithConstInfo & /*arguments*/) const override { return true; }
 
+    /*
+    * Functions like recursiveRemoveLowCardinality don't pay enough attention to custom types and just erase
+    * the information about it during type conversions.
+    * While it is a big problem the quick solution would be just to disable default low cardinality implementation
+    * because it doesn't make a lot of sense for geo types.
+    */
+    bool useDefaultImplementationForLowCardinalityColumns() const override { return false; }
+
     ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr & /*result_type*/, size_t input_rows_count) const override
     {
         auto res_column = ColumnString::create();

src/IO/Archives/IArchiveReader.h

@@ -5,6 +5,7 @@
 #include <functional>
 #include <memory>
 
+#include <Poco/Timestamp.h>
 
 namespace DB
 {
@@ -25,6 +26,7 @@ public:
     {
         UInt64 uncompressed_size;
         UInt64 compressed_size;
+        Poco::Timestamp last_modified;
         bool is_encrypted;
     };
 

src/IO/Archives/LibArchiveReader.cpp

@@ -157,6 +157,7 @@ public:
             file_info.emplace();
             file_info->uncompressed_size = archive_entry_size(current_entry);
             file_info->compressed_size = archive_entry_size(current_entry);
+            file_info->last_modified = archive_entry_mtime(current_entry);
             file_info->is_encrypted = false;
         }
 

src/Interpreters/ActionsDAG.cpp

@@ -596,6 +596,34 @@ void ActionsDAG::removeUnusedActions(const std::unordered_set<const Node *> & us
     std::erase_if(inputs, [&](const Node * node) { return !visited_nodes.contains(node); });
 }
 
+
+void ActionsDAG::removeAliasesForFilter(const std::string & filter_name)
+{
+    const auto & filter_node = findInOutputs(filter_name);
+    std::stack<Node *> stack;
+    stack.push(const_cast<Node *>(&filter_node));
+
+    std::unordered_set<const Node *> visited;
+    visited.insert(stack.top());
+
+    while (!stack.empty())
+    {
+        auto * node = stack.top();
+        stack.pop();
+        for (auto & child : node->children)
+        {
+            while (child->type == ActionType::ALIAS)
+                child = child->children.front();
+
+            if (!visited.contains(child))
+            {
+                stack.push(const_cast<Node *>(child));
+                visited.insert(child);
+            }
+        }
+    }
+}
+
 ActionsDAGPtr ActionsDAG::cloneSubDAG(const NodeRawConstPtrs & outputs, bool remove_aliases)
 {
     auto actions = std::make_shared<ActionsDAG>();
@@ -1704,7 +1732,7 @@ void ActionsDAG::mergeNodes(ActionsDAG && second, NodeRawConstPtrs * out_outputs
     }
 }
 
-ActionsDAG::SplitResult ActionsDAG::split(std::unordered_set<const Node *> split_nodes, bool create_split_nodes_mapping) const
+ActionsDAG::SplitResult ActionsDAG::split(std::unordered_set<const Node *> split_nodes, bool create_split_nodes_mapping, bool avoid_duplicate_inputs) const
 {
     /// Split DAG into two parts.
     /// (first_nodes, first_outputs) is a part which will have split_list in result.
@@ -1718,6 +1746,14 @@ ActionsDAG::SplitResult ActionsDAG::split(std::unordered_set<const Node *> split
     /// List of nodes from current actions which are not inputs, but will be in second part.
     NodeRawConstPtrs new_inputs;
 
+    /// Avoid new inputs to have the same name as existing inputs.
+    /// It's allowed for DAG but may break Block invariant 'columns with identical name must have identical structure'.
+    std::unordered_set<std::string_view> duplicate_inputs;
+    size_t duplicate_counter = 0;
+    if (avoid_duplicate_inputs)
+        for (const auto * input : inputs)
+            duplicate_inputs.insert(input->result_name);
+
     struct Frame
     {
         const Node * node = nullptr;
@@ -1830,7 +1866,8 @@ ActionsDAG::SplitResult ActionsDAG::split(std::unordered_set<const Node *> split
                                 input_node.result_name = child->result_name;
                                 child_data.to_second = &second_nodes.emplace_back(std::move(input_node));
 
-                                new_inputs.push_back(child);
+                                if (child->type != ActionType::INPUT)
+                                    new_inputs.push_back(child);
                             }
                         }
 
@@ -1886,7 +1923,32 @@ ActionsDAG::SplitResult ActionsDAG::split(std::unordered_set<const Node *> split
 
     for (const auto * input : new_inputs)
     {
-        const auto & cur = data[input];
+        auto & cur = data[input];
+
+        if (avoid_duplicate_inputs)
+        {
+            bool is_name_updated = false;
+            while (!duplicate_inputs.insert(cur.to_first->result_name).second)
+            {
+                is_name_updated = true;
+                cur.to_first->result_name = fmt::format("{}_{}", input->result_name, duplicate_counter);
+                ++duplicate_counter;
+            }
+
+            if (is_name_updated)
+            {
+                Node input_node;
+                input_node.type = ActionType::INPUT;
+                input_node.result_type = cur.to_first->result_type;
+                input_node.result_name = cur.to_first->result_name;
+
+                auto * new_input = &second_nodes.emplace_back(std::move(input_node));
+                cur.to_second->type = ActionType::ALIAS;
+                cur.to_second->children = {new_input};
+                cur.to_second = new_input;
+            }
+        }
+
         second_inputs.push_back(cur.to_second);
         first_outputs.push_back(cur.to_first);
     }

src/Interpreters/ActionsDAG.h

@@ -195,6 +195,8 @@ public:
     /// Remove actions that are not needed to compute output nodes with required names
     void removeUnusedActions(const NameSet & required_names, bool allow_remove_inputs = true, bool allow_constant_folding = true);
 
+    void removeAliasesForFilter(const std::string & filter_name);
+
     /// Transform the current DAG in a way that leaf nodes get folded into their parents. It's done
     /// because each projection can provide some columns as inputs to substitute certain sub-DAGs
     /// (expressions). Consider the following example:
@@ -343,7 +345,7 @@ public:
     ///   initial DAG    : (a, b, c, d, e) -> (w, x, y, z)  | 1 a 2 b 3 c 4 d 5 e 6      ->  1 2 3 4 5 6 w x y z
     ///   split (first)  : (a, c, d) -> (i, j, k, w, y)     | 1 a 2 b 3 c 4 d 5 e 6      ->  1 2 b 3 4 5 e 6 i j k w y
     ///   split (second) : (i, j, k, y, b, e) -> (x, y, z)  | 1 2 b 3 4 5 e 6 i j k w y  ->  1 2 3 4 5 6 w x y z
-    SplitResult split(std::unordered_set<const Node *> split_nodes, bool create_split_nodes_mapping = false) const;
+    SplitResult split(std::unordered_set<const Node *> split_nodes, bool create_split_nodes_mapping = false, bool avoid_duplicate_inputs = false) const;
 
     /// Splits actions into two parts. Returned first half may be swapped with ARRAY JOIN.
     SplitResult splitActionsBeforeArrayJoin(const NameSet & array_joined_columns) const;

src/Interpreters/Context.cpp

@@ -740,12 +740,18 @@ struct ContextSharedPart : boost::noncopyable
 
     void initializeTraceCollector(std::shared_ptr<TraceLog> trace_log)
     {
-        if (!trace_log)
+        if (!trace_collector.has_value())
-            return;
+            throw Exception(ErrorCodes::LOGICAL_ERROR, "TraceCollector needs to be first created before initialization");
+
+        trace_collector->initialize(trace_log);
+    }
+
+    void createTraceCollector()
+    {
         if (hasTraceCollector())
             return;
 
-        trace_collector.emplace(std::move(trace_log));
+        trace_collector.emplace();
     }
 
     void addWarningMessage(const String & message) TSA_REQUIRES(mutex)
@@ -3891,6 +3897,11 @@ void Context::initializeSystemLogs()
     });
 }
 
+void Context::createTraceCollector()
+{
+    shared->createTraceCollector();
+}
+
 void Context::initializeTraceCollector()
 {
     shared->initializeTraceCollector(getTraceLog());

src/Interpreters/Context.h

@@ -1077,6 +1077,8 @@ public:
     void initializeSystemLogs();
 
     /// Call after initialization before using trace collector.
+    void createTraceCollector();
+
     void initializeTraceCollector();
 
     /// Call after unexpected crash happen.

src/Interpreters/DDLTask.cpp

@@ -568,8 +568,21 @@ void ZooKeeperMetadataTransaction::commit()
 
 ClusterPtr tryGetReplicatedDatabaseCluster(const String & cluster_name)
 {
-    if (const auto * replicated_db = dynamic_cast<const DatabaseReplicated *>(DatabaseCatalog::instance().tryGetDatabase(cluster_name).get()))
+    String name = cluster_name;
-        return replicated_db->tryGetCluster();
+    bool all_groups = false;
+    if (name.starts_with(DatabaseReplicated::ALL_GROUPS_CLUSTER_PREFIX))
+    {
+        name = name.substr(strlen(DatabaseReplicated::ALL_GROUPS_CLUSTER_PREFIX));
+        all_groups = true;
+    }
+
+    if (const auto * replicated_db = dynamic_cast<const DatabaseReplicated *>(DatabaseCatalog::instance().tryGetDatabase(name).get()))
+    {
+        if (all_groups)
+            return replicated_db->tryGetAllGroupsCluster();
+        else
+            return replicated_db->tryGetCluster();
+    }
     return {};
 }
 

src/Interpreters/InterpreterInsertQuery.cpp

@@ -279,6 +279,8 @@ Chain InterpreterInsertQuery::buildChain(
     std::atomic_uint64_t * elapsed_counter_ms,
     bool check_access)
 {
+    IInterpreter::checkStorageSupportsTransactionsIfNeeded(table, getContext());
+
     ProfileEvents::increment(ProfileEvents::InsertQueriesWithSubqueries);
     ProfileEvents::increment(ProfileEvents::QueriesWithSubqueries);
 

src/Interpreters/InterpreterSelectQuery.cpp

@@ -578,7 +578,9 @@ InterpreterSelectQuery::InterpreterSelectQuery(
                     settings.parallel_replicas_count,
                     settings.parallel_replica_offset,
                     std::move(custom_key_ast),
-                    settings.parallel_replicas_custom_key_filter_type,
+                    {settings.parallel_replicas_custom_key_filter_type,
+                     settings.parallel_replicas_custom_key_range_lower,
+                     settings.parallel_replicas_custom_key_range_upper},
                     storage->getInMemoryMetadataPtr()->columns,
                     context);
             }

src/Interpreters/TraceCollector.cpp

@@ -1,5 +1,4 @@
-#include "TraceCollector.h"
+#include <Interpreters/TraceCollector.h>
-
 #include <Core/Field.h>
 #include <IO/ReadBufferFromFileDescriptor.h>
 #include <IO/ReadHelpers.h>
@@ -14,8 +13,12 @@
 namespace DB
 {
 
-TraceCollector::TraceCollector(std::shared_ptr<TraceLog> trace_log_)
+namespace ErrorCodes
-    : trace_log(std::move(trace_log_))
+{
+    extern const int LOGICAL_ERROR;
+}
+
+TraceCollector::TraceCollector()
 {
     TraceSender::pipe.open();
 
@@ -28,6 +31,23 @@ TraceCollector::TraceCollector(std::shared_ptr<TraceLog> trace_log_)
     thread = ThreadFromGlobalPool(&TraceCollector::run, this);
 }
 
+void TraceCollector::initialize(std::shared_ptr<TraceLog> trace_log_)
+{
+    if (is_trace_log_initialized)
+        throw DB::Exception(ErrorCodes::LOGICAL_ERROR, "TraceCollector is already initialized");
+
+    trace_log_ptr = trace_log_;
+    is_trace_log_initialized.store(true, std::memory_order_release);
+}
+
+std::shared_ptr<TraceLog> TraceCollector::getTraceLog()
+{
+    if (!is_trace_log_initialized.load(std::memory_order_acquire))
+        return nullptr;
+
+    return trace_log_ptr;
+}
+
 void TraceCollector::tryClosePipe()
 {
     try
@@ -120,7 +140,7 @@ void TraceCollector::run()
             ProfileEvents::Count increment;
             readPODBinary(increment, in);
 
-            if (trace_log)
+            if (auto trace_log = getTraceLog())
             {
                 // time and time_in_microseconds are both being constructed from the same timespec so that the
                 // times will be equal up to the precision of a second.

src/Interpreters/TraceCollector.h

@@ -1,4 +1,5 @@
 #pragma once
+#include <atomic>
 #include <Common/ThreadPool.h>
 
 class StackTrace;
@@ -16,11 +17,17 @@ class TraceLog;
 class TraceCollector
 {
 public:
-    explicit TraceCollector(std::shared_ptr<TraceLog> trace_log_);
+    TraceCollector();
     ~TraceCollector();
 
+    void initialize(std::shared_ptr<TraceLog> trace_log_);
+
 private:
-    std::shared_ptr<TraceLog> trace_log;
+    std::shared_ptr<TraceLog> getTraceLog();
+
+    std::atomic<bool> is_trace_log_initialized = false;
+    std::shared_ptr<TraceLog> trace_log_ptr;
+
     ThreadFromGlobalPool thread;
 
     void tryClosePipe();

src/Interpreters/getCustomKeyFilterForParallelReplicas.cpp

@@ -7,7 +7,6 @@
 
 #include <Interpreters/Context.h>
 
-#include <DataTypes/DataTypesNumber.h>
 
 #include <boost/rational.hpp>
 
@@ -18,18 +17,19 @@ namespace DB
 namespace ErrorCodes
 {
     extern const int ILLEGAL_TYPE_OF_COLUMN_FOR_FILTER;
+    extern const int INVALID_SETTING_VALUE;
 }
 
 ASTPtr getCustomKeyFilterForParallelReplica(
     size_t replicas_count,
     size_t replica_num,
     ASTPtr custom_key_ast,
-    ParallelReplicasCustomKeyFilterType filter_type,
+    ParallelReplicasCustomKeyFilter filter,
     const ColumnsDescription & columns,
     const ContextPtr & context)
 {
     chassert(replicas_count > 1);
-    if (filter_type == ParallelReplicasCustomKeyFilterType::DEFAULT)
+    if (filter.filter_type == ParallelReplicasCustomKeyFilterType::DEFAULT)
     {
         // first we do modulo with replica count
         auto modulo_function = makeASTFunction("positiveModulo", custom_key_ast, std::make_shared<ASTLiteral>(replicas_count));
@@ -40,35 +40,80 @@ ASTPtr getCustomKeyFilterForParallelReplica(
         return equals_function;
     }
 
-    assert(filter_type == ParallelReplicasCustomKeyFilterType::RANGE);
+    chassert(filter.filter_type == ParallelReplicasCustomKeyFilterType::RANGE);
 
     KeyDescription custom_key_description
         = KeyDescription::getKeyFromAST(custom_key_ast, columns, context);
 
     using RelativeSize = boost::rational<ASTSampleRatio::BigNum>;
 
-    RelativeSize size_of_universum = 0;
+    RelativeSize range_upper = RelativeSize(0);
+    RelativeSize range_lower = RelativeSize(filter.range_lower);
     DataTypePtr custom_key_column_type = custom_key_description.data_types[0];
 
-    size_of_universum = RelativeSize(std::numeric_limits<UInt32>::max()) + RelativeSize(1);
     if (custom_key_description.data_types.size() == 1)
     {
         if (typeid_cast<const DataTypeUInt64 *>(custom_key_column_type.get()))
-            size_of_universum = RelativeSize(std::numeric_limits<UInt64>::max()) + RelativeSize(1);
+        {
+            range_upper = filter.range_upper > 0 ? RelativeSize(filter.range_upper) + RelativeSize(1)
+                                                 : RelativeSize(std::numeric_limits<UInt64>::max()) + RelativeSize(1);
+            if (range_upper > RelativeSize(std::numeric_limits<UInt64>::max()) + RelativeSize(1))
+                throw Exception(
+                    ErrorCodes::INVALID_SETTING_VALUE,
+                    "Invalid custom key range upper bound: {}. Value must be smaller than custom key column type (UInt64) max value",
+                    range_upper);
+        }
         else if (typeid_cast<const DataTypeUInt32 *>(custom_key_column_type.get()))
-            size_of_universum = RelativeSize(std::numeric_limits<UInt32>::max()) + RelativeSize(1);
+        {
+            range_upper = filter.range_upper > 0 ? RelativeSize(filter.range_upper) + RelativeSize(1)
+                                                 : RelativeSize(std::numeric_limits<UInt32>::max()) + RelativeSize(1);
+            if (range_upper > RelativeSize(std::numeric_limits<UInt32>::max()) + RelativeSize(1))
+                throw Exception(
+                    ErrorCodes::INVALID_SETTING_VALUE,
+                    "Invalid custom key range upper bound: {}. Value must be smaller than custom key column type (UInt32) max value",
+                    range_upper);
+        }
         else if (typeid_cast<const DataTypeUInt16 *>(custom_key_column_type.get()))
-            size_of_universum = RelativeSize(std::numeric_limits<UInt16>::max()) + RelativeSize(1);
+        {
+            range_upper = filter.range_upper > 0 ? RelativeSize(filter.range_upper) + RelativeSize(1)
+                                                 : RelativeSize(std::numeric_limits<UInt16>::max()) + RelativeSize(1);
+            if (range_upper > RelativeSize(std::numeric_limits<UInt16>::max()) + RelativeSize(1))
+                throw Exception(
+                    ErrorCodes::INVALID_SETTING_VALUE,
+                    "Invalid custom key range upper bound: {}. Value must be smaller than custom key column type (UInt16) max value",
+                    range_upper);
+        }
         else if (typeid_cast<const DataTypeUInt8 *>(custom_key_column_type.get()))
-            size_of_universum = RelativeSize(std::numeric_limits<UInt8>::max()) + RelativeSize(1);
+        {
+            range_upper = filter.range_upper > 0 ? RelativeSize(filter.range_upper) + RelativeSize(1)
+                                                 : RelativeSize(std::numeric_limits<UInt8>::max()) + RelativeSize(1);
+            if (range_upper > RelativeSize(std::numeric_limits<UInt8>::max()) + RelativeSize(1))
+                throw Exception(
+                    ErrorCodes::INVALID_SETTING_VALUE,
+                    "Invalid custom key range upper bound: {}. Value must be smaller than custom key column type (UInt8) max value",
+                    range_upper);
+        }
     }
 
-    if (size_of_universum == RelativeSize(0))
+    if (range_upper == RelativeSize(0))
         throw Exception(
             ErrorCodes::ILLEGAL_TYPE_OF_COLUMN_FOR_FILTER,
             "Invalid custom key column type: {}. Must be one unsigned integer type",
             custom_key_column_type->getName());
 
+    if (range_lower >= range_upper)
+        throw Exception(
+            ErrorCodes::INVALID_SETTING_VALUE,
+            "Invalid custom key filter range: Range upper bound {} must be larger than range lower bound {}",
+            range_lower,
+            range_upper);
+
+    RelativeSize size_of_universum = range_upper - range_lower;
+
+    if (size_of_universum <= RelativeSize(replicas_count))
+        throw Exception(
+            ErrorCodes::INVALID_SETTING_VALUE, "Invalid custom key filter range: Range must be larger than than the number of replicas");
+
     RelativeSize relative_range_size = RelativeSize(1) / replicas_count;
     RelativeSize relative_range_offset = relative_range_size * RelativeSize(replica_num);
 
@@ -76,19 +121,19 @@ ASTPtr getCustomKeyFilterForParallelReplica(
     bool has_lower_limit = false;
     bool has_upper_limit = false;
 
-    RelativeSize lower_limit_rational = relative_range_offset * size_of_universum;
+    RelativeSize lower_limit_rational = range_lower + relative_range_offset * size_of_universum;
-    RelativeSize upper_limit_rational = (relative_range_offset + relative_range_size) * size_of_universum;
+    RelativeSize upper_limit_rational = range_lower + (relative_range_offset + relative_range_size) * size_of_universum;
 
     UInt64 lower = boost::rational_cast<ASTSampleRatio::BigNum>(lower_limit_rational);
     UInt64 upper = boost::rational_cast<ASTSampleRatio::BigNum>(upper_limit_rational);
 
-    if (lower > 0)
+    if (lower_limit_rational > range_lower)
         has_lower_limit = true;
 
-    if (upper_limit_rational < size_of_universum)
+    if (upper_limit_rational < range_upper)
         has_upper_limit = true;
 
-    assert(has_lower_limit || has_upper_limit);
+    chassert(has_lower_limit || has_upper_limit);
 
     /// Let's add the conditions to cut off something else when the index is scanned again and when the request is processed.
     std::shared_ptr<ASTFunction> lower_function;
@@ -110,7 +155,7 @@ ASTPtr getCustomKeyFilterForParallelReplica(
             return upper_function;
     }
 
-    assert(upper_function && lower_function);
+    chassert(upper_function && lower_function);
 
     return makeASTFunction("and", std::move(lower_function), std::move(upper_function));
 }

src/Interpreters/getCustomKeyFilterForParallelReplicas.h

@@ -6,16 +6,24 @@
 #include <Storages/IStorage.h>
 #include <Core/SettingsEnums.h>
 #include <Interpreters/DatabaseAndTableWithAlias.h>
+#include <DataTypes/DataTypesNumber.h>
 
 namespace DB
 {
+struct ParallelReplicasCustomKeyFilter
+{
+    ParallelReplicasCustomKeyFilterType filter_type;
+    UInt64 range_lower;
+    UInt64 range_upper;
+};
+
 /// Get AST for filter created from custom_key
 /// replica_num is the number of the replica for which we are generating filter starting from 0
 ASTPtr getCustomKeyFilterForParallelReplica(
     size_t replicas_count,
     size_t replica_num,
     ASTPtr custom_key_ast,
-    ParallelReplicasCustomKeyFilterType filter_type,
+    ParallelReplicasCustomKeyFilter filter,
     const ColumnsDescription & columns,
     const ContextPtr & context);
 

src/Planner/PlannerJoinTree.cpp

@@ -501,12 +501,14 @@ FilterDAGInfo buildCustomKeyFilterIfNeeded(const StoragePtr & storage,
     LOG_TRACE(getLogger("Planner"), "Processing query on a replica using custom_key '{}'", settings.parallel_replicas_custom_key.value);
 
     auto parallel_replicas_custom_filter_ast = getCustomKeyFilterForParallelReplica(
-            settings.parallel_replicas_count,
+        settings.parallel_replicas_count,
-            settings.parallel_replica_offset,
+        settings.parallel_replica_offset,
-            std::move(custom_key_ast),
+        std::move(custom_key_ast),
-            settings.parallel_replicas_custom_key_filter_type,
+        {settings.parallel_replicas_custom_key_filter_type,
-            storage->getInMemoryMetadataPtr()->columns,
+         settings.parallel_replicas_custom_key_range_lower,
-            query_context);
+         settings.parallel_replicas_custom_key_range_upper},
+        storage->getInMemoryMetadataPtr()->columns,
+        query_context);
 
     return buildFilterInfo(parallel_replicas_custom_filter_ast, table_expression_query_info.table_expression, planner_context);
 }

src/Processors/Formats/Impl/NativeORCBlockInputFormat.cpp

@@ -269,7 +269,12 @@ convertFieldToORCLiteral(const orc::Type & orc_type, const Field & field, DataTy
             case orc::SHORT:
             case orc::INT:
             case orc::LONG: {
-                /// May throw exception
+                /// May throw exception.
+                ///
+                /// In particular, it'll throw if we request the column as unsigned, like this:
+                ///   SELECT * FROM file('t.orc', ORC, 'x UInt8') WHERE x > 10
+                /// We have to reject this, otherwise it would miss values > 127 (because
+                /// they're treated as negative by ORC).
                 auto val = field.get<Int64>();
                 return orc::Literal(val);
             }

src/Processors/Formats/Impl/ORCBlockOutputFormat.cpp

@@ -315,18 +315,20 @@ void ORCBlockOutputFormat::writeColumn(
     if (null_bytemap)
         orc_column.hasNulls = true;
 
+    /// ORC doesn't have unsigned types, so cast everything to signed and sign-extend to Int64 to
+    /// make the ORC library calculate min and max correctly.
     switch (type->getTypeId())
     {
         case TypeIndex::Enum8: [[fallthrough]];
         case TypeIndex::Int8:
         {
             /// Note: Explicit cast to avoid clang-tidy error: 'signed char' to 'long' conversion; consider casting to 'unsigned char' first.
-            writeNumbers<Int8, orc::LongVectorBatch>(orc_column, column, null_bytemap, [](const Int8 & value){ return static_cast<int64_t>(value); });
+            writeNumbers<Int8, orc::LongVectorBatch>(orc_column, column, null_bytemap, [](const Int8 & value){ return Int64(Int8(value)); });
             break;
         }
         case TypeIndex::UInt8:
         {
-            writeNumbers<UInt8, orc::LongVectorBatch>(orc_column, column, null_bytemap, [](const UInt8 & value){ return value; });
+            writeNumbers<UInt8, orc::LongVectorBatch>(orc_column, column, null_bytemap, [](const UInt8 & value){ return Int64(Int8(value)); });
             break;
         }
         case TypeIndex::Enum16: [[fallthrough]];
@@ -338,7 +340,7 @@ void ORCBlockOutputFormat::writeColumn(
         case TypeIndex::Date: [[fallthrough]];
         case TypeIndex::UInt16:
         {
-            writeNumbers<UInt16, orc::LongVectorBatch>(orc_column, column, null_bytemap, [](const UInt16 & value){ return value; });
+            writeNumbers<UInt16, orc::LongVectorBatch>(orc_column, column, null_bytemap, [](const UInt16 & value){ return Int64(Int16(value)); });
             break;
         }
         case TypeIndex::Date32: [[fallthrough]];
@@ -349,12 +351,12 @@ void ORCBlockOutputFormat::writeColumn(
         }
         case TypeIndex::UInt32:
         {
-            writeNumbers<UInt32, orc::LongVectorBatch>(orc_column, column, null_bytemap, [](const UInt32 & value){ return value; });
+            writeNumbers<UInt32, orc::LongVectorBatch>(orc_column, column, null_bytemap, [](const UInt32 & value){ return Int64(Int32(value)); });
             break;
         }
         case TypeIndex::IPv4:
         {
-            writeNumbers<IPv4, orc::LongVectorBatch>(orc_column, column, null_bytemap, [](const IPv4 & value){ return value.toUnderType(); });
+            writeNumbers<IPv4, orc::LongVectorBatch>(orc_column, column, null_bytemap, [](const IPv4 & value){ return Int64(Int32(value.toUnderType())); });
             break;
         }
         case TypeIndex::Int64:

src/Processors/QueryPlan/FilterStep.cpp

@@ -50,6 +50,8 @@ FilterStep::FilterStep(
     , filter_column_name(std::move(filter_column_name_))
     , remove_filter_column(remove_filter_column_)
 {
+    actions_dag = actions_dag->clone();
+    actions_dag->removeAliasesForFilter(filter_column_name);
 }
 
 void FilterStep::transformPipeline(QueryPipelineBuilder & pipeline, const BuildQueryPipelineSettings & settings)

src/Processors/QueryPlan/Optimizations/optimizePrewhere.cpp

@@ -118,7 +118,7 @@ void optimizePrewhere(Stack & stack, QueryPlan::Nodes &)
         outputs.resize(size);
     }
 
-    auto split_result = filter_step->getExpression()->split(optimize_result.prewhere_nodes, true);
+    auto split_result = filter_step->getExpression()->split(optimize_result.prewhere_nodes, true, true);
 
     /// This is the leak of abstraction.
     /// Splited actions may have inputs which are needed only for PREWHERE.

src/Storages/ObjectStorage/StorageObjectStorageSource.cpp

@@ -195,12 +195,14 @@ Chunk StorageObjectStorageSource::generate()
             const auto & object_info = reader.getObjectInfo();
             const auto & filename = object_info.getFileName();
             chassert(object_info.metadata);
-            VirtualColumnUtils::addRequestedPathFileAndSizeVirtualsToChunk(
+            VirtualColumnUtils::addRequestedFileLikeStorageVirtualsToChunk(
-                chunk,
+                chunk, read_from_format_info.requested_virtual_columns,
-                read_from_format_info.requested_virtual_columns,
+                {
-                getUniqueStoragePathIdentifier(*configuration, reader.getObjectInfo(), false),
+                    .path = getUniqueStoragePathIdentifier(*configuration, reader.getObjectInfo(), false),
-                object_info.metadata->size_bytes, &filename);
+                    .size = object_info.metadata->size_bytes,
-
+                    .filename = &filename,
+                    .last_modified = object_info.metadata->last_modified
+                });
             return chunk;
         }
 

src/Storages/RocksDB/StorageEmbeddedRocksDB.cpp

@@ -25,6 +25,7 @@
 
 #include <Poco/Logger.h>
 #include <Poco/Util/AbstractConfiguration.h>
+#include <Common/Logger.h>
 #include <Common/logger_useful.h>
 #include <Common/Exception.h>
 #include <Storages/AlterCommands.h>
@@ -42,6 +43,7 @@
 
 #include <cstddef>
 #include <filesystem>
+#include <memory>
 #include <utility>
 
 
@@ -185,11 +187,11 @@ StorageEmbeddedRocksDB::StorageEmbeddedRocksDB(const StorageID & table_id_,
         bool read_only_)
     : IStorage(table_id_)
     , WithContext(context_->getGlobalContext())
+    , log(getLogger(fmt::format("StorageEmbeddedRocksDB ({})", getStorageID().getNameForLogs())))
     , primary_key{primary_key_}
     , rocksdb_dir(std::move(rocksdb_dir_))
     , ttl(ttl_)
     , read_only(read_only_)
-    , log(getLogger(fmt::format("StorageEmbeddedRocksDB ({})", getStorageID().getNameForLogs())))
 {
     setInMemoryMetadata(metadata_);
     setSettings(std::move(settings_));
@@ -352,6 +354,72 @@ bool StorageEmbeddedRocksDB::optimize(
     return true;
 }
 
+static_assert(rocksdb::DEBUG_LEVEL == 0);
+static_assert(rocksdb::HEADER_LEVEL == 5);
+static constexpr std::array<std::pair<DB::LogsLevel, Poco::Message::Priority>, 6> rocksdb_logger_map = {
+    std::make_pair(DB::LogsLevel::debug, Poco::Message::Priority::PRIO_DEBUG),
+    std::make_pair(DB::LogsLevel::information, Poco::Message::Priority::PRIO_INFORMATION),
+    std::make_pair(DB::LogsLevel::warning, Poco::Message::Priority::PRIO_WARNING),
+    std::make_pair(DB::LogsLevel::error, Poco::Message::Priority::PRIO_ERROR),
+    std::make_pair(DB::LogsLevel::fatal, Poco::Message::Priority::PRIO_FATAL),
+    /// Same as default logger does for HEADER_LEVEL
+    std::make_pair(DB::LogsLevel::information, Poco::Message::Priority::PRIO_INFORMATION),
+};
+class StorageEmbeddedRocksDBLogger : public rocksdb::Logger
+{
+public:
+    explicit StorageEmbeddedRocksDBLogger(const rocksdb::InfoLogLevel log_level, LoggerRawPtr log_)
+        : rocksdb::Logger(log_level)
+        , log(log_)
+    {}
+
+    void Logv(const char * format, va_list ap) override
+        __attribute__((format(printf, 2, 0)))
+    {
+        Logv(rocksdb::InfoLogLevel::DEBUG_LEVEL, format, ap);
+    }
+
+    void Logv(const rocksdb::InfoLogLevel log_level, const char * format, va_list ap) override
+        __attribute__((format(printf, 3, 0)))
+    {
+        if (log_level < GetInfoLogLevel())
+            return;
+
+        auto level = rocksdb_logger_map[log_level];
+
+        /// stack buffer was enough
+        {
+            va_list backup_ap;
+            va_copy(backup_ap, ap);
+            std::array<char, 1024> stack;
+            if (vsnprintf(stack.data(), stack.size(), format, backup_ap) < static_cast<int>(stack.size()))
+            {
+                va_end(backup_ap);
+                LOG_IMPL(log, level.first, level.second, "{}", stack.data());
+                return;
+            }
+            va_end(backup_ap);
+        }
+
+        /// let's try with a bigger dynamic buffer (but not too huge, since
+        /// some of rocksdb internal code has also such a limitation, i..e
+        /// HdfsLogger)
+        {
+            va_list backup_ap;
+            va_copy(backup_ap, ap);
+            static constexpr int buffer_size = 30000;
+            std::unique_ptr<char[]> buffer(new char[buffer_size]);
+            if (vsnprintf(buffer.get(), buffer_size, format, backup_ap) >= buffer_size)
+                buffer[buffer_size - 1] = 0;
+            va_end(backup_ap);
+            LOG_IMPL(log, level.first, level.second, "{}", buffer.get());
+        }
+    }
+
+private:
+    LoggerRawPtr log;
+};
+
 void StorageEmbeddedRocksDB::initDB()
 {
     rocksdb::Status status;
@@ -448,6 +516,7 @@ void StorageEmbeddedRocksDB::initDB()
         }
     }
 
+    merged.info_log = std::make_shared<StorageEmbeddedRocksDBLogger>(merged.info_log_level, log.get());
     merged.table_factory.reset(rocksdb::NewBlockBasedTableFactory(table_options));
 
     if (ttl > 0)

src/Storages/RocksDB/StorageEmbeddedRocksDB.h

@@ -114,17 +114,19 @@ public:
 private:
     SinkToStoragePtr getSink(ContextPtr context, const StorageMetadataPtr & metadata_snapshot);
 
+    LoggerPtr log;
+
     MultiVersion<RocksDBSettings> storage_settings;
     const String primary_key;
+
     using RocksDBPtr = std::unique_ptr<rocksdb::DB>;
     RocksDBPtr rocksdb_ptr;
+
     mutable SharedMutex rocksdb_ptr_mx;
     String rocksdb_dir;
     Int32 ttl;
     bool read_only;
 
     void initDB();
-
-    LoggerPtr log;
 };
 }

src/Storages/S3Queue/S3QueueSource.cpp

@@ -421,8 +421,14 @@ Chunk StorageS3QueueSource::generate()
                 file_status->processed_rows += chunk.getNumRows();
                 processed_rows_from_file += chunk.getNumRows();
 
-                VirtualColumnUtils::addRequestedPathFileAndSizeVirtualsToChunk(
+                VirtualColumnUtils::addRequestedFileLikeStorageVirtualsToChunk(
-                    chunk, requested_virtual_columns, path, reader.getObjectInfo().metadata->size_bytes);
+                    chunk, requested_virtual_columns,
+                    {
+                        .path = path,
+                        .size = reader.getObjectInfo().metadata->size_bytes
+                    });
+
+
                 return chunk;
             }
         }

src/Storages/StorageDistributed.cpp

@@ -904,11 +904,13 @@ void StorageDistributed::read(
                 [my_custom_key_ast = std::move(custom_key_ast),
                  column_description = this->getInMemoryMetadataPtr()->columns,
                  custom_key_type = settings.parallel_replicas_custom_key_filter_type.value,
+                 custom_key_range_lower = settings.parallel_replicas_custom_key_range_lower.value,
+                 custom_key_range_upper = settings.parallel_replicas_custom_key_range_upper.value,
                  context = local_context,
                  replica_count = modified_query_info.getCluster()->getShardsInfo().front().per_replica_pools.size()](uint64_t replica_num) -> ASTPtr
             {
                 return getCustomKeyFilterForParallelReplica(
-                    replica_count, replica_num - 1, my_custom_key_ast, custom_key_type, column_description, context);
+                    replica_count, replica_num - 1, my_custom_key_ast, {custom_key_type, custom_key_range_lower, custom_key_range_upper}, column_description, context);
             };
         }
     }

src/Storages/StorageFile.cpp

@@ -1341,6 +1341,7 @@ Chunk StorageFileSource::generate()
                         chassert(file_enumerator);
                         current_path = fmt::format("{}::{}", archive_reader->getPath(), *filename_override);
                         current_file_size = file_enumerator->getFileInfo().uncompressed_size;
+                        current_file_last_modified = file_enumerator->getFileInfo().last_modified;
                         if (need_only_count && tryGetCountFromCache(current_archive_stat))
                             continue;
 
@@ -1370,6 +1371,7 @@ Chunk StorageFileSource::generate()
                 struct stat file_stat;
                 file_stat = getFileStat(current_path, storage->use_table_fd, storage->table_fd, storage->getName());
                 current_file_size = file_stat.st_size;
+                current_file_last_modified = Poco::Timestamp::fromEpochTime(file_stat.st_mtime);
 
                 if (getContext()->getSettingsRef().engine_file_skip_empty_files && file_stat.st_size == 0)
                     continue;
@@ -1436,8 +1438,15 @@ Chunk StorageFileSource::generate()
             progress(num_rows, chunk_size ? chunk_size : chunk.bytes());
 
             /// Enrich with virtual columns.
-            VirtualColumnUtils::addRequestedPathFileAndSizeVirtualsToChunk(
+            VirtualColumnUtils::addRequestedFileLikeStorageVirtualsToChunk(
-                chunk, requested_virtual_columns, current_path, current_file_size, filename_override.has_value() ? &filename_override.value() : nullptr);
+                chunk, requested_virtual_columns,
+                {
+                    .path = current_path,
+                    .size = current_file_size,
+                    .filename = (filename_override.has_value() ? &filename_override.value() : nullptr),
+                    .last_modified = current_file_last_modified
+                });
+
             return chunk;
         }
 

src/Storages/StorageFile.h

@@ -279,6 +279,7 @@ private:
     FilesIteratorPtr files_iterator;
     String current_path;
     std::optional<size_t> current_file_size;
+    std::optional<Poco::Timestamp> current_file_last_modified;
     struct stat current_archive_stat;
     std::optional<String> filename_override;
     Block sample_block;

src/Storages/StorageGenerateRandom.cpp

@@ -50,6 +50,12 @@ namespace ErrorCodes
 namespace
 {
 
+struct GenerateRandomState
+{
+    std::atomic<UInt64> add_total_rows = 0;
+};
+using GenerateRandomStatePtr = std::shared_ptr<GenerateRandomState>;
+
 void fillBufferWithRandomData(char * __restrict data, size_t limit, size_t size_of_type, pcg64 & rng, [[maybe_unused]] bool flip_bytes = false)
 {
     size_t size = limit * size_of_type;
@@ -532,10 +538,24 @@ ColumnPtr fillColumnWithRandomData(
 class GenerateSource : public ISource
 {
 public:
-    GenerateSource(UInt64 block_size_, UInt64 max_array_length_, UInt64 max_string_length_, UInt64 random_seed_, Block block_header_, ContextPtr context_)
+    GenerateSource(
+        UInt64 block_size_,
+        UInt64 max_array_length_,
+        UInt64 max_string_length_,
+        UInt64 random_seed_,
+        Block block_header_,
+        ContextPtr context_,
+        GenerateRandomStatePtr state_)
         : ISource(Nested::flattenNested(prepareBlockToFill(block_header_)))
-        , block_size(block_size_), max_array_length(max_array_length_), max_string_length(max_string_length_)
+        , block_size(block_size_)
-        , block_to_fill(std::move(block_header_)), rng(random_seed_), context(context_) {}
+        , max_array_length(max_array_length_)
+        , max_string_length(max_string_length_)
+        , block_to_fill(std::move(block_header_))
+        , rng(random_seed_)
+        , context(context_)
+        , shared_state(state_)
+    {
+    }
 
     String getName() const override { return "GenerateRandom"; }
 
@@ -549,7 +569,15 @@ protected:
             columns.emplace_back(fillColumnWithRandomData(elem.type, block_size, max_array_length, max_string_length, rng, context));
 
         columns = Nested::flattenNested(block_to_fill.cloneWithColumns(columns)).getColumns();
-        return {std::move(columns), block_size};
+
+        UInt64 total_rows = shared_state->add_total_rows.fetch_and(0);
+        if (total_rows)
+            addTotalRowsApprox(total_rows);
+
+        auto chunk = Chunk{std::move(columns), block_size};
+        progress(chunk.getNumRows(), chunk.bytes());
+
+        return chunk;
     }
 
 private:
@@ -561,6 +589,7 @@ private:
     pcg64 rng;
 
     ContextPtr context;
+    GenerateRandomStatePtr shared_state;
 
     static Block & prepareBlockToFill(Block & block)
     {
@@ -648,9 +677,6 @@ Pipe StorageGenerateRandom::read(
 {
     storage_snapshot->check(column_names);
 
-    Pipes pipes;
-    pipes.reserve(num_streams);
-
     const ColumnsDescription & our_columns = storage_snapshot->metadata->getColumns();
     Block block_header;
     for (const auto & name : column_names)
@@ -679,16 +705,24 @@ Pipe StorageGenerateRandom::read(
         }
     }
 
+    UInt64 query_limit = query_info.limit;
+    if (query_limit && num_streams * max_block_size > query_limit)
+    {
+        /// We want to avoid spawning more streams than necessary
+        num_streams = std::min(num_streams, static_cast<size_t>(((query_limit + max_block_size - 1) / max_block_size)));
+    }
+    Pipes pipes;
+    pipes.reserve(num_streams);
+
     /// Will create more seed values for each source from initial seed.
     pcg64 generate(random_seed);
 
+    auto shared_state = std::make_shared<GenerateRandomState>(query_info.limit);
+
     for (UInt64 i = 0; i < num_streams; ++i)
     {
-        auto source = std::make_shared<GenerateSource>(max_block_size, max_array_length, max_string_length, generate(), block_header, context);
+        auto source = std::make_shared<GenerateSource>(
-
+            max_block_size, max_array_length, max_string_length, generate(), block_header, context, shared_state);
-        if (i == 0 && query_info.limit)
-            source->addTotalRowsApprox(query_info.limit);
-
         pipes.emplace_back(std::move(source));
     }
 

src/Storages/StorageURL.cpp

@@ -411,7 +411,12 @@ Chunk StorageURLSource::generate()
             if (input_format)
                 chunk_size = input_format->getApproxBytesReadForChunk();
             progress(num_rows, chunk_size ? chunk_size : chunk.bytes());
-            VirtualColumnUtils::addRequestedPathFileAndSizeVirtualsToChunk(chunk, requested_virtual_columns, curr_uri.getPath(), current_file_size);
+            VirtualColumnUtils::addRequestedFileLikeStorageVirtualsToChunk(
+                chunk, requested_virtual_columns,
+                {
+                    .path = curr_uri.getPath(),
+                    .size = current_file_size
+                });
             return chunk;
         }
 

src/Storages/System/StorageSystemClusters.cpp

@@ -54,6 +54,10 @@ void StorageSystemClusters::fillData(MutableColumns & res_columns, ContextPtr co
             if (auto database_cluster = replicated->tryGetCluster())
                 writeCluster(res_columns, {name_and_database.first, database_cluster},
                              replicated->tryGetAreReplicasActive(database_cluster));
+
+            if (auto database_cluster = replicated->tryGetAllGroupsCluster())
+                writeCluster(res_columns, {DatabaseReplicated::ALL_GROUPS_CLUSTER_PREFIX + name_and_database.first, database_cluster},
+                             replicated->tryGetAreReplicasActive(database_cluster));
         }
     }
 }

src/Storages/System/StorageSystemZeros.cpp

@@ -16,7 +16,9 @@ namespace
 
 struct ZerosState
 {
+    explicit ZerosState(UInt64 limit) : add_total_rows(limit) { }
     std::atomic<UInt64> num_generated_rows = 0;
+    std::atomic<UInt64> add_total_rows = 0;
 };
 
 using ZerosStatePtr = std::shared_ptr<ZerosState>;
@@ -42,10 +44,13 @@ protected:
         auto column_ptr = column;
         size_t column_size = column_ptr->size();
 
-        if (state)
+        UInt64 total_rows = state->add_total_rows.fetch_and(0);
+        if (total_rows)
+            addTotalRowsApprox(total_rows);
+
+        if (limit)
         {
             auto generated_rows = state->num_generated_rows.fetch_add(column_size, std::memory_order_acquire);
-
             if (generated_rows >= limit)
                 return {};
 
@@ -103,36 +108,25 @@ Pipe StorageSystemZeros::read(
 {
     storage_snapshot->check(column_names);
 
-    bool use_multiple_streams = multithreaded;
+    UInt64 query_limit = limit ? *limit : 0;
+    if (query_info.limit)
+        query_limit = query_limit ? std::min(query_limit, query_info.limit) : query_info.limit;
 
-    if (limit && *limit < max_block_size)
+    if (query_limit && query_limit < max_block_size)
-    {
+        max_block_size = query_limit;
-        max_block_size = static_cast<size_t>(*limit);
-        use_multiple_streams = false;
-    }
 
-    if (!use_multiple_streams)
+    if (!multithreaded)
         num_streams = 1;
+    else if (query_limit && num_streams * max_block_size > query_limit)
+        /// We want to avoid spawning more streams than necessary
+        num_streams = std::min(num_streams, static_cast<size_t>(((query_limit + max_block_size - 1) / max_block_size)));
+
+    ZerosStatePtr state = std::make_shared<ZerosState>(query_limit);
 
     Pipe res;
-
-    ZerosStatePtr state;
-
-    if (limit)
-        state = std::make_shared<ZerosState>();
-
     for (size_t i = 0; i < num_streams; ++i)
     {
-        auto source = std::make_shared<ZerosSource>(max_block_size, limit ? *limit : 0, state);
+        auto source = std::make_shared<ZerosSource>(max_block_size, query_limit, state);
-
-        if (i == 0)
-        {
-            if (limit)
-                source->addTotalRowsApprox(*limit);
-            else if (query_info.limit)
-                source->addTotalRowsApprox(query_info.limit);
-        }
-
         res.addSource(std::move(source));
     }
 

src/Storages/TTLDescription.cpp

@@ -166,8 +166,14 @@ static ExpressionAndSets buildExpressionAndSets(ASTPtr & ast, const NamesAndType
 {
     ExpressionAndSets result;
     auto ttl_string = queryToString(ast);
-    auto syntax_analyzer_result = TreeRewriter(context).analyze(ast, columns);
+    auto context_copy = Context::createCopy(context);
-    ExpressionAnalyzer analyzer(ast, syntax_analyzer_result, context);
+    /// FIXME All code here will work with old analyzer, however for TTL
+    /// with subqueries it's possible that new analyzer will be enabled in ::read method
+    /// of underlying storage when all other parts of infra are not ready for it
+    /// (built with old analyzer).
+    context_copy->setSetting("allow_experimental_analyzer", Field{0});
+    auto syntax_analyzer_result = TreeRewriter(context_copy).analyze(ast, columns);
+    ExpressionAnalyzer analyzer(ast, syntax_analyzer_result, context_copy);
     auto dag = analyzer.getActionsDAG(false);
 
     const auto * col = &dag->findInOutputs(ast->getColumnName());
@@ -177,7 +183,7 @@ static ExpressionAndSets buildExpressionAndSets(ASTPtr & ast, const NamesAndType
     dag->getOutputs() = {col};
     dag->removeUnusedActions();
 
-    result.expression = std::make_shared<ExpressionActions>(dag, ExpressionActionsSettings::fromContext(context));
+    result.expression = std::make_shared<ExpressionActions>(dag, ExpressionActionsSettings::fromContext(context_copy));
     result.sets = analyzer.getPreparedSets();
 
     return result;

src/Storages/VirtualColumnUtils.cpp

@@ -26,6 +26,7 @@
 #include <DataTypes/DataTypesNumber.h>
 #include <DataTypes/DataTypeString.h>
 #include <DataTypes/DataTypeLowCardinality.h>
+#include <DataTypes/DataTypeDateTime.h>
 
 #include <Processors/QueryPlan/QueryPlan.h>
 #include <Processors/QueryPlan/BuildQueryPipelineSettings.h>
@@ -111,7 +112,7 @@ void filterBlockWithDAG(ActionsDAGPtr dag, Block & block, ContextPtr context)
 
 NameSet getVirtualNamesForFileLikeStorage()
 {
-    return {"_path", "_file", "_size"};
+    return {"_path", "_file", "_size", "_time"};
 }
 
 VirtualColumnsDescription getVirtualsForFileLikeStorage(const ColumnsDescription & storage_columns)
@@ -129,6 +130,7 @@ VirtualColumnsDescription getVirtualsForFileLikeStorage(const ColumnsDescription
     add_virtual("_path", std::make_shared<DataTypeLowCardinality>(std::make_shared<DataTypeString>()));
     add_virtual("_file", std::make_shared<DataTypeLowCardinality>(std::make_shared<DataTypeString>()));
     add_virtual("_size", makeNullable(std::make_shared<DataTypeUInt64>()));
+    add_virtual("_time", makeNullable(std::make_shared<DataTypeDateTime>()));
 
     return desc;
 }
@@ -187,32 +189,40 @@ ColumnPtr getFilterByPathAndFileIndexes(const std::vector<String> & paths, const
     return block.getByName("_idx").column;
 }
 
-void addRequestedPathFileAndSizeVirtualsToChunk(
+void addRequestedFileLikeStorageVirtualsToChunk(
-    Chunk & chunk, const NamesAndTypesList & requested_virtual_columns, const String & path, std::optional<size_t> size, const String * filename)
+    Chunk & chunk, const NamesAndTypesList & requested_virtual_columns,
+    VirtualsForFileLikeStorage virtual_values)
 {
     for (const auto & virtual_column : requested_virtual_columns)
     {
         if (virtual_column.name == "_path")
         {
-            chunk.addColumn(virtual_column.type->createColumnConst(chunk.getNumRows(), path)->convertToFullColumnIfConst());
+            chunk.addColumn(virtual_column.type->createColumnConst(chunk.getNumRows(), virtual_values.path)->convertToFullColumnIfConst());
         }
         else if (virtual_column.name == "_file")
         {
-            if (filename)
+            if (virtual_values.filename)
             {
-                chunk.addColumn(virtual_column.type->createColumnConst(chunk.getNumRows(), *filename)->convertToFullColumnIfConst());
+                chunk.addColumn(virtual_column.type->createColumnConst(chunk.getNumRows(), (*virtual_values.filename))->convertToFullColumnIfConst());
             }
             else
             {
-                size_t last_slash_pos = path.find_last_of('/');
+                size_t last_slash_pos = virtual_values.path.find_last_of('/');
-                auto filename_from_path = path.substr(last_slash_pos + 1);
+                auto filename_from_path = virtual_values.path.substr(last_slash_pos + 1);
                 chunk.addColumn(virtual_column.type->createColumnConst(chunk.getNumRows(), filename_from_path)->convertToFullColumnIfConst());
             }
         }
         else if (virtual_column.name == "_size")
         {
-            if (size)
+            if (virtual_values.size)
-                chunk.addColumn(virtual_column.type->createColumnConst(chunk.getNumRows(), *size)->convertToFullColumnIfConst());
+                chunk.addColumn(virtual_column.type->createColumnConst(chunk.getNumRows(), *virtual_values.size)->convertToFullColumnIfConst());
+            else
+                chunk.addColumn(virtual_column.type->createColumnConstWithDefaultValue(chunk.getNumRows())->convertToFullColumnIfConst());
+        }
+        else if (virtual_column.name == "_time")
+        {
+            if (virtual_values.last_modified)
+                chunk.addColumn(virtual_column.type->createColumnConst(chunk.getNumRows(), virtual_values.last_modified->epochTime())->convertToFullColumnIfConst());
             else
                 chunk.addColumn(virtual_column.type->createColumnConstWithDefaultValue(chunk.getNumRows())->convertToFullColumnIfConst());
         }

src/Storages/VirtualColumnUtils.h

@@ -68,8 +68,18 @@ void filterByPathOrFile(std::vector<T> & sources, const std::vector<String> & pa
     sources = std::move(filtered_sources);
 }
 
-void addRequestedPathFileAndSizeVirtualsToChunk(
+struct VirtualsForFileLikeStorage
-    Chunk & chunk, const NamesAndTypesList & requested_virtual_columns, const String & path, std::optional<size_t> size, const String * filename = nullptr);
+{
+    const String & path;
+    std::optional<size_t> size { std::nullopt };
+    const String * filename { nullptr };
+    std::optional<Poco::Timestamp> last_modified { std::nullopt };
+
+};
+
+void addRequestedFileLikeStorageVirtualsToChunk(
+    Chunk & chunk, const NamesAndTypesList & requested_virtual_columns,
+    VirtualsForFileLikeStorage virtual_values);
 }
 
 }

tests/ci/build_download_helper.py

@@ -27,6 +27,8 @@ except ImportError:
 
 DOWNLOAD_RETRIES_COUNT = 5
 
+logger = logging.getLogger(__name__)
+
 
 class DownloadException(Exception):
     pass
@@ -42,7 +44,7 @@ def get_with_retries(
     sleep: int = 3,
     **kwargs: Any,
 ) -> requests.Response:
-    logging.info(
+    logger.info(
         "Getting URL with %i tries and sleep %i in between: %s", retries, sleep, url
     )
     exc = Exception("A placeholder to satisfy typing and avoid nesting")
@@ -54,7 +56,7 @@ def get_with_retries(
             return response
         except Exception as e:
             if i + 1 < retries:
-                logging.info("Exception '%s' while getting, retry %i", e, i + 1)
+                logger.info("Exception '%s' while getting, retry %i", e, i + 1)
                 time.sleep(sleep)
 
             exc = e
@@ -103,7 +105,7 @@ def get_gh_api(
             )
             try_auth = e.response.status_code == 404
             if (ratelimit_exceeded or try_auth) and not token_is_set:
-                logging.warning(
+                logger.warning(
                     "Received rate limit exception, setting the auth header and retry"
                 )
                 set_auth_header()
@@ -114,10 +116,10 @@ def get_gh_api(
             exc = e
 
         if try_cnt < retries:
-            logging.info("Exception '%s' while getting, retry %i", exc, try_cnt)
+            logger.info("Exception '%s' while getting, retry %i", exc, try_cnt)
             time.sleep(sleep)
 
-    raise APIException("Unable to request data from GH API") from exc
+    raise APIException(f"Unable to request data from GH API: {url}") from exc
 
 
 def get_build_name_for_check(check_name: str) -> str:
@@ -128,25 +130,25 @@ def read_build_urls(build_name: str, reports_path: Union[Path, str]) -> List[str
     for root, _, files in os.walk(reports_path):
         for file in files:
             if file.endswith(f"_{build_name}.json"):
-                logging.info("Found build report json %s for %s", file, build_name)
+                logger.info("Found build report json %s for %s", file, build_name)
                 with open(
                     os.path.join(root, file), "r", encoding="utf-8"
                 ) as file_handler:
                     build_report = json.load(file_handler)
                     return build_report["build_urls"]  # type: ignore
 
-    logging.info("A build report is not found for %s", build_name)
+    logger.info("A build report is not found for %s", build_name)
     return []
 
 
 def download_build_with_progress(url: str, path: Path) -> None:
-    logging.info("Downloading from %s to temp path %s", url, path)
+    logger.info("Downloading from %s to temp path %s", url, path)
     for i in range(DOWNLOAD_RETRIES_COUNT):
         try:
             response = get_with_retries(url, retries=1, stream=True)
             total_length = int(response.headers.get("content-length", 0))
             if path.is_file() and total_length and path.stat().st_size == total_length:
-                logging.info(
+                logger.info(
                     "The file %s already exists and have a proper size %s",
                     path,
                     total_length,
@@ -155,14 +157,14 @@ def download_build_with_progress(url: str, path: Path) -> None:
 
             with open(path, "wb") as f:
                 if total_length == 0:
-                    logging.info(
+                    logger.info(
                         "No content-length, will download file without progress"
                     )
                     f.write(response.content)
                 else:
                     dl = 0
 
-                    logging.info("Content length is %ld bytes", total_length)
+                    logger.info("Content length is %ld bytes", total_length)
                     for data in response.iter_content(chunk_size=4096):
                         dl += len(data)
                         f.write(data)
@@ -177,8 +179,8 @@ def download_build_with_progress(url: str, path: Path) -> None:
         except Exception as e:
             if sys.stdout.isatty():
                 sys.stdout.write("\n")
-            if os.path.exists(path):
+            if path.exists():
-                os.remove(path)
+                path.unlink()
 
             if i + 1 < DOWNLOAD_RETRIES_COUNT:
                 time.sleep(3)
@@ -189,7 +191,7 @@ def download_build_with_progress(url: str, path: Path) -> None:
 
     if sys.stdout.isatty():
         sys.stdout.write("\n")
-    logging.info("Downloading finished")
+    logger.info("Downloading finished")
 
 
 def download_builds(
@@ -198,7 +200,7 @@ def download_builds(
     for url in build_urls:
         if filter_fn(url):
             fname = os.path.basename(url.replace("%2B", "+").replace("%20", " "))
-            logging.info("Will download %s to %s", fname, result_path)
+            logger.info("Will download %s to %s", fname, result_path)
             download_build_with_progress(url, result_path / fname)
 
 
@@ -210,7 +212,7 @@ def download_builds_filter(
 ) -> None:
     build_name = get_build_name_for_check(check_name)
     urls = read_build_urls(build_name, reports_path)
-    logging.info("The build report for %s contains the next URLs: %s", build_name, urls)
+    logger.info("The build report for %s contains the next URLs: %s", build_name, urls)
 
     if not urls:
         raise DownloadException("No build URLs found")
@@ -247,7 +249,7 @@ def get_clickhouse_binary_url(
 ) -> Optional[str]:
     build_name = get_build_name_for_check(check_name)
     urls = read_build_urls(build_name, reports_path)
-    logging.info("The build report for %s contains the next URLs: %s", build_name, urls)
+    logger.info("The build report for %s contains the next URLs: %s", build_name, urls)
     for url in urls:
         check_url = url
         if "?" in check_url:

tests/ci/cherry_pick.py

@@ -127,13 +127,11 @@ close it.
                 to_pop.append(i)
             elif pr.head.ref.startswith(f"backport/{self.name}"):
                 self.backport_pr = pr
+                self._backported = True
                 to_pop.append(i)
             else:
                 assert False, f"BUG! Invalid PR's branch [{pr.head.ref}]"
 
-            # Cherry-pick or backport PR found, set @backported flag for current release branch
-            self._backported = True
-
         for i in reversed(to_pop):
             # Going from the tail to keep the order and pop greater index first
             prs.pop(i)
@@ -218,6 +216,7 @@ close it.
                     self.name,
                     self.pr.number,
                 )
+                self._backported = True
                 return
         except CalledProcessError:
             # There are most probably conflicts, they'll be resolved in PR
@@ -247,7 +246,6 @@ close it.
             self.cherrypick_pr.add_to_labels(Labels.PR_CRITICAL_BUGFIX)
         elif Labels.PR_BUGFIX in [label.name for label in self.pr.labels]:
             self.cherrypick_pr.add_to_labels(Labels.PR_BUGFIX)
-        self._backported = True
         self._assign_new_pr(self.cherrypick_pr)
         # update cherrypick PR to get the state for PR.mergable
         self.cherrypick_pr.update()
@@ -359,10 +357,10 @@ class Backport:
         self._fetch_from = fetch_from
         self.dry_run = dry_run
 
-        self.must_create_backport_label = (
+        self.must_create_backport_labels = (
-            Labels.MUST_BACKPORT
+            [Labels.MUST_BACKPORT]
             if self._repo_name == self._fetch_from
-            else Labels.MUST_BACKPORT_CLOUD
+            else [Labels.MUST_BACKPORT_CLOUD, Labels.MUST_BACKPORT]
         )
         self.backport_created_label = (
             Labels.PR_BACKPORTS_CREATED
@@ -468,7 +466,7 @@ class Backport:
         query_args = {
             "query": f"type:pr repo:{self._fetch_from} -label:{self.backport_created_label}",
             "label": ",".join(
-                self.labels_to_backport + [self.must_create_backport_label]
+                self.labels_to_backport + self.must_create_backport_labels
             ),
             "merged": [since_date, tomorrow],
         }
@@ -492,7 +490,10 @@ class Backport:
     def process_pr(self, pr: PullRequest) -> None:
         pr_labels = [label.name for label in pr.labels]
 
-        if self.must_create_backport_label in pr_labels:
+        if (
+            any(label in pr_labels for label in self.must_create_backport_labels)
+            or self._repo_name != self._fetch_from
+        ):
             branches = [
                 ReleaseBranch(br, pr, self.repo, self.backport_created_label)
                 for br in self.release_branches
@@ -532,8 +533,8 @@ class Backport:
             br.process(self.dry_run)
 
         for br in branches:
-            assert br.backported, f"BUG! backport to branch [{br}] failed"
+            if br.backported:
-        self.mark_pr_backported(pr)
+                self.mark_pr_backported(pr)
 
     def mark_pr_backported(self, pr: PullRequest) -> None:
         if self.dry_run:

tests/ci/ci.py

@@ -424,6 +424,7 @@ def _configure_jobs(
     s3: S3Helper,
     pr_info: PRInfo,
     ci_settings: CiSettings,
+    skip_jobs: bool,
 ) -> CiCache:
     """
     returns CICache instance with configured job's data
@@ -434,20 +435,29 @@ def _configure_jobs(
     """
 
     # get all jobs
-    job_configs = CI_CONFIG.get_workflow_jobs_with_configs(
+    if not skip_jobs:
-        is_mq=pr_info.is_merge_queue,
+        job_configs = CI_CONFIG.get_workflow_jobs_with_configs(
-        is_docs_only=pr_info.has_changes_in_documentation_only(),
+            is_mq=pr_info.is_merge_queue,
-        is_master=pr_info.is_master,
+            is_docs_only=pr_info.has_changes_in_documentation_only(),
-    )
+            is_master=pr_info.is_master,
+        )
+    else:
+        job_configs = {}
 
     # filter jobs in accordance with ci settings
     job_configs = ci_settings.apply(
-        job_configs, pr_info.is_release, is_pr=pr_info.is_pr, labels=pr_info.labels
+        job_configs,
+        pr_info.is_release,
+        is_pr=pr_info.is_pr,
+        is_mq=pr_info.is_merge_queue,
+        labels=pr_info.labels,
     )
 
     # check jobs in ci cache
     ci_cache = CiCache.calc_digests_and_create(
-        s3, job_configs, cache_enabled=not ci_settings.no_ci_cache and CI
+        s3,
+        job_configs,
+        cache_enabled=not ci_settings.no_ci_cache and not skip_jobs and CI,
     )
     ci_cache.update()
     ci_cache.apply(job_configs, is_release=pr_info.is_release)
@@ -971,6 +981,7 @@ def main() -> int:
             s3,
             pr_info,
             ci_settings,
+            args.skip_jobs,
         )
         ci_cache.print_status()
 
@@ -989,15 +1000,15 @@ def main() -> int:
         result["ci_settings"] = ci_settings.as_dict()
         if not args.skip_jobs:
             result["stages_data"] = _generate_ci_stage_config(ci_cache.jobs_to_do)
-        result["jobs_data"] = {
+            result["jobs_data"] = {
-            "jobs_to_do": list(ci_cache.jobs_to_do),
+                "jobs_to_do": list(ci_cache.jobs_to_do),
-            "jobs_to_skip": ci_cache.jobs_to_skip,
+                "jobs_to_skip": ci_cache.jobs_to_skip,
-            "digests": ci_cache.job_digests,
+                "digests": ci_cache.job_digests,
-            "jobs_params": {
+                "jobs_params": {
-                job: {"batches": config.batches, "num_batches": config.num_batches}
+                    job: {"batches": config.batches, "num_batches": config.num_batches}
-                for job, config in ci_cache.jobs_to_do.items()
+                    for job, config in ci_cache.jobs_to_do.items()
-            },
+                },
-        }
+            }
         result["docker_data"] = docker_data
     ### CONFIGURE action: end
 

tests/ci/ci_config.py

@@ -53,6 +53,8 @@ class CILabels(metaclass=WithIter):
     CI_SET_SYNC = "ci_set_sync"
     CI_SET_ARM = "ci_set_arm"
     CI_SET_REQUIRED = "ci_set_required"
+    CI_SET_NORMAL_BUILDS = "ci_set_normal_builds"
+    CI_SET_SPECIAL_BUILDS = "ci_set_special_builds"
     CI_SET_NON_REQUIRED = "ci_set_non_required"
     CI_SET_OLD_ANALYZER = "ci_set_old_analyzer"
 
@@ -685,10 +687,8 @@ class CIConfig:
     def get_job_parents(self, check_name: str) -> List[str]:
         res = []
         check_name = normalize_string(check_name)
-
         for config in (
             self.build_config,
-            self.builds_report_config,
             self.test_configs,
             self.other_jobs_configs,
         ):
@@ -899,6 +899,40 @@ CI_CONFIG = CIConfig(
             ]
         ),
         CILabels.CI_SET_REQUIRED: LabelConfig(run_jobs=REQUIRED_CHECKS),
+        CILabels.CI_SET_NORMAL_BUILDS: LabelConfig(
+            run_jobs=[
+                JobNames.STYLE_CHECK,
+                JobNames.BUILD_CHECK,
+                Build.PACKAGE_RELEASE,
+                Build.PACKAGE_AARCH64,
+                Build.PACKAGE_ASAN,
+                Build.PACKAGE_UBSAN,
+                Build.PACKAGE_TSAN,
+                Build.PACKAGE_MSAN,
+                Build.PACKAGE_DEBUG,
+                Build.BINARY_RELEASE,
+                Build.PACKAGE_RELEASE_COVERAGE,
+                Build.FUZZERS,
+            ]
+        ),
+        CILabels.CI_SET_SPECIAL_BUILDS: LabelConfig(
+            run_jobs=[
+                JobNames.STYLE_CHECK,
+                JobNames.BUILD_CHECK_SPECIAL,
+                Build.BINARY_TIDY,
+                Build.BINARY_DARWIN,
+                Build.BINARY_AARCH64,
+                Build.BINARY_AARCH64_V80COMPAT,
+                Build.BINARY_FREEBSD,
+                Build.BINARY_DARWIN_AARCH64,
+                Build.BINARY_PPC64LE,
+                Build.BINARY_RISCV64,
+                Build.BINARY_S390X,
+                Build.BINARY_LOONGARCH64,
+                Build.BINARY_AMD64_COMPAT,
+                Build.BINARY_AMD64_MUSL,
+            ]
+        ),
         CILabels.CI_SET_NON_REQUIRED: LabelConfig(
             run_jobs=[job for job in JobNames if job not in REQUIRED_CHECKS]
         ),

tests/ci/ci_settings.py

@@ -134,6 +134,7 @@ class CiSettings:
         job_config: JobConfig,
         is_release: bool,
         is_pr: bool,
+        is_mq: bool,
         labels: Iterable[str],
     ) -> bool:  # type: ignore #too-many-return-statements
         if self.do_not_test:
@@ -189,7 +190,7 @@ class CiSettings:
 
         if job_config.release_only and not is_release:
             return False
-        elif job_config.pr_only and not is_pr:
+        elif job_config.pr_only and not is_pr and not is_mq:
             return False
 
         return not to_deny
@@ -199,6 +200,7 @@ class CiSettings:
         job_configs: Dict[str, JobConfig],
         is_release: bool,
         is_pr: bool,
+        is_mq: bool,
         labels: Iterable[str],
     ) -> Dict[str, JobConfig]:
         """
@@ -207,16 +209,24 @@ class CiSettings:
         res = {}
         for job, job_config in job_configs.items():
             if self._check_if_selected(
-                job, job_config, is_release=is_release, is_pr=is_pr, labels=labels
+                job,
+                job_config,
+                is_release=is_release,
+                is_pr=is_pr,
+                is_mq=is_mq,
+                labels=labels,
             ):
                 res[job] = job_config
 
+        add_parents = []
         for job in list(res):
             parent_jobs = CI_CONFIG.get_job_parents(job)
             for parent_job in parent_jobs:
                 if parent_job not in res:
+                    add_parents.append(parent_job)
                     print(f"Job [{job}] requires [{parent_job}] - add")
-                    res[parent_job] = job_configs[parent_job]
+        for job in add_parents:
+            res[job] = job_configs[job]
 
         for job, job_config in res.items():
             batches = []

tests/ci/functional_test_check.py

@@ -309,9 +309,6 @@ def main():
         state, description, test_results, additional_logs = process_results(
             result_path, server_log_path
         )
-        # FIXME (alesapin)
-        if "azure" in check_name:
-            state = "success"
     else:
         print(
             "This is validate bugfix or flaky check run, but no changes test to run - skip with success"

tests/ci/pr_info.py

@@ -59,7 +59,7 @@ def get_pr_for_commit(sha, ref):
         data = response.json()
         our_prs = []  # type: List[Dict]
         if len(data) > 1:
-            print("Got more than one pr for commit", sha)
+            logging.warning("Got more than one pr for commit %s", sha)
         for pr in data:
             # We need to check if the PR is created in our repo, because
             # https://github.com/kaynewu/ClickHouse/pull/2
@@ -71,13 +71,20 @@ def get_pr_for_commit(sha, ref):
             if pr["head"]["ref"] in ref:
                 return pr
             our_prs.append(pr)
-        print(
+        logging.warning(
-            f"Cannot find PR with required ref {ref}, sha {sha} - returning first one"
+            "Cannot find PR with required ref %s, sha %s - returning first one",
+            ref,
+            sha,
         )
         first_pr = our_prs[0]
         return first_pr
     except Exception as ex:
-        print(f"Cannot fetch PR info from commit {ref}, {sha}", ex)
+        logging.error(
+            "Cannot fetch PR info from commit ref %s, sha %s, exception: %s",
+            ref,
+            sha,
+            ex,
+        )
     return None
 
 
@@ -259,12 +266,12 @@ class PRInfo:
                     self.diff_urls.append(
                         self.compare_url(
                             pull_request["base"]["repo"]["default_branch"],
-                            pull_request["head"]["label"],
+                            pull_request["head"]["sha"],
                         )
                     )
                     self.diff_urls.append(
                         self.compare_url(
-                            pull_request["head"]["label"],
+                            pull_request["head"]["sha"],
                             pull_request["base"]["repo"]["default_branch"],
                         )
                     )
@@ -279,7 +286,7 @@ class PRInfo:
                     # itself, but as well files changed since we branched out
                     self.diff_urls.append(
                         self.compare_url(
-                            pull_request["head"]["label"],
+                            pull_request["head"]["sha"],
                             pull_request["base"]["repo"]["default_branch"],
                         )
                     )
@@ -289,8 +296,10 @@ class PRInfo:
             else:
                 # assume this is a dispatch
                 self.event_type = EventType.DISPATCH
-            print("event.json does not match pull_request or push:")
+            logging.warning(
-            print(json.dumps(github_event, sort_keys=True, indent=4))
+                "event.json does not match pull_request or push:\n%s",
+                json.dumps(github_event, sort_keys=True, indent=4),
+            )
             self.sha = os.getenv(
                 "GITHUB_SHA", "0000000000000000000000000000000000000000"
             )
@@ -330,7 +339,7 @@ class PRInfo:
         return self.event_type == EventType.DISPATCH
 
     def compare_pr_url(self, pr_object: dict) -> str:
-        return self.compare_url(pr_object["base"]["label"], pr_object["head"]["label"])
+        return self.compare_url(pr_object["base"]["sha"], pr_object["head"]["sha"])
 
     @staticmethod
     def compare_url(first: str, second: str) -> str:
@@ -357,7 +366,7 @@ class PRInfo:
             diff_object = PatchSet(response.text)
             self.changed_files.update({f.path for f in diff_object})
         self.changed_files_requested = True
-        print(f"Fetched info about {len(self.changed_files)} changed files")
+        logging.info("Fetched info about %s changed files", len(self.changed_files))
 
     def get_dict(self):
         return {

tests/ci/test_ci_options.py

@@ -179,7 +179,11 @@ class TestCIOptions(unittest.TestCase):
         )
         filtered_jobs = list(
             ci_options.apply(
-                jobs_configs, is_release=False, is_pr=True, labels=["TEST_LABEL"]
+                jobs_configs,
+                is_release=False,
+                is_pr=True,
+                is_mq=False,
+                labels=["TEST_LABEL"],
             )
         )
         self.assertCountEqual(
@@ -212,7 +216,9 @@ class TestCIOptions(unittest.TestCase):
         jobs_configs["fuzzers"].run_by_label = "TEST_LABEL"
         # no settings are set
         filtered_jobs = list(
-            CiSettings().apply(jobs_configs, is_release=False, is_pr=True, labels=[])
+            CiSettings().apply(
+                jobs_configs, is_release=False, is_pr=False, is_mq=True, labels=[]
+            )
         )
         self.assertCountEqual(
             filtered_jobs,
@@ -220,9 +226,21 @@ class TestCIOptions(unittest.TestCase):
                 "Fast test",
             ],
         )
-
         filtered_jobs = list(
-            CiSettings().apply(jobs_configs, is_release=True, is_pr=False, labels=[])
+            CiSettings().apply(
+                jobs_configs, is_release=False, is_pr=True, is_mq=False, labels=[]
+            )
+        )
+        self.assertCountEqual(
+            filtered_jobs,
+            [
+                "Fast test",
+            ],
+        )
+        filtered_jobs = list(
+            CiSettings().apply(
+                jobs_configs, is_release=True, is_pr=False, is_mq=False, labels=[]
+            )
         )
         self.assertCountEqual(
             filtered_jobs,
@@ -240,7 +258,11 @@ class TestCIOptions(unittest.TestCase):
         # no settings are set
         filtered_jobs = list(
             ci_settings.apply(
-                jobs_configs, is_release=False, is_pr=True, labels=["TEST_LABEL"]
+                jobs_configs,
+                is_release=False,
+                is_pr=True,
+                is_mq=False,
+                labels=["TEST_LABEL"],
             )
         )
         self.assertCountEqual(
@@ -253,7 +275,11 @@ class TestCIOptions(unittest.TestCase):
         ci_settings.include_keywords = ["Fast"]
         filtered_jobs = list(
             ci_settings.apply(
-                jobs_configs, is_release=True, is_pr=False, labels=["TEST_LABEL"]
+                jobs_configs,
+                is_release=True,
+                is_pr=False,
+                is_mq=False,
+                labels=["TEST_LABEL"],
             )
         )
         self.assertCountEqual(
@@ -277,7 +303,11 @@ class TestCIOptions(unittest.TestCase):
         jobs_configs["Integration tests (asan)"].release_only = True
         filtered_jobs = list(
             ci_options.apply(
-                jobs_configs, is_release=False, is_pr=True, labels=["TEST_LABEL"]
+                jobs_configs,
+                is_release=False,
+                is_pr=True,
+                is_mq=False,
+                labels=["TEST_LABEL"],
             )
         )
         self.assertCountEqual(

tests/clickhouse-test

@@ -808,10 +808,10 @@ class SettingsRandomizer:
         "merge_tree_coarse_index_granularity": lambda: random.randint(2, 32),
         "optimize_distinct_in_order": lambda: random.randint(0, 1),
         "max_bytes_before_external_sort": threshold_generator(
-            0.3, 0.5, 1, 10 * 1024 * 1024 * 1024
+            0.3, 0.5, 0, 10 * 1024 * 1024 * 1024
         ),
         "max_bytes_before_external_group_by": threshold_generator(
-            0.3, 0.5, 1, 10 * 1024 * 1024 * 1024
+            0.3, 0.5, 0, 10 * 1024 * 1024 * 1024
         ),
         "max_bytes_before_remerge_sort": lambda: random.randint(1, 3000000000),
         "min_compress_block_size": lambda: random.randint(1, 1048576 * 3),
@@ -850,6 +850,11 @@ class SettingsRandomizer:
         "merge_tree_read_split_ranges_into_intersecting_and_non_intersecting_injection_probability": lambda: round(
             random.random(), 2
         ),
+        "prefer_external_sort_block_bytes": lambda: random.choice([0, 1, 100000000]),
+        "cross_join_min_rows_to_compress": lambda: random.choice([0, 1, 100000000]),
+        "cross_join_min_bytes_to_compress": lambda: random.choice([0, 1, 100000000]),
+        "min_external_table_block_size_bytes": lambda: random.choice([0, 1, 100000000]),
+        "max_parsing_threads": lambda: random.choice([0, 1, 10]),
     }
 
     @staticmethod

tests/config/config.d/rocksdb.xml

@@ -0,0 +1,7 @@
+<clickhouse>
+    <rocksdb>
+        <options>
+            <info_log_level>DEBUG_LEVEL</info_log_level>
+        </options>
+    </rocksdb>
+</clickhouse>