add setting remove_empty_parts

src/Storages/MergeTree/MergeTreeData.cpp

@@ -1217,6 +1217,9 @@ void MergeTreeData::clearOldWriteAheadLogs()
 
 void MergeTreeData::clearEmptyParts()
 {
+    if (!getSettings()->remove_empty_parts)
+        return;
+
     auto parts = getDataPartsVector();
     for (const auto & part : parts)
     {

src/Storages/MergeTree/MergeTreeSettings.h

@@ -105,6 +105,7 @@ struct Settings;
     M(UInt64, concurrent_part_removal_threshold, 100, "Activate concurrent part removal (see 'max_part_removal_threads') only if the number of inactive data parts is at least this.", 0) \
     M(String, storage_policy, "default", "Name of storage disk policy", 0) \
     M(Bool, allow_nullable_key, false, "Allow Nullable types as primary keys.", 0) \
+    M(Bool, remove_empty_parts, true, "Remove empty parts after they were pruned by TTL, mutation, or collapsing merge algorithm", 0) \
     \
     /** Settings for testing purposes */ \
     M(Bool, randomize_part_type, false, "For testing purposes only. Randomizes part type between wide and compact", 0) \

tests/queries/0_stateless/00933_alter_ttl.reference

-CREATE TABLE default.ttl\n(\n    `d` Date,\n    `a` Int32\n)\nENGINE = MergeTree\nPARTITION BY toDayOfMonth(d)\nORDER BY a\nTTL d + toIntervalDay(1)\nSETTINGS index_granularity = 8192
+CREATE TABLE default.ttl\n(\n    `d` Date,\n    `a` Int32\n)\nENGINE = MergeTree\nPARTITION BY toDayOfMonth(d)\nORDER BY a\nTTL d + toIntervalDay(1)\nSETTINGS remove_empty_parts = 0, index_granularity = 8192
 2100-10-10	3
 2100-10-10	4
 d	Date					

tests/queries/0_stateless/00933_alter_ttl.sql

@@ -2,14 +2,13 @@ set send_logs_level = 'fatal';
 
 drop table if exists ttl;
 
-create table ttl (d Date, a Int) engine = MergeTree order by a partition by toDayOfMonth(d);
+create table ttl (d Date, a Int) engine = MergeTree order by a partition by toDayOfMonth(d) settings remove_empty_parts = 0;
 alter table ttl modify ttl d + interval 1 day;
 show create table ttl;
 insert into ttl values (toDateTime('2000-10-10 00:00:00'), 1);
 insert into ttl values (toDateTime('2000-10-10 00:00:00'), 2);
 insert into ttl values (toDateTime('2100-10-10 00:00:00'), 3);
 insert into ttl values (toDateTime('2100-10-10 00:00:00'), 4);
-select sleep(1) format Null; -- wait if very fast merge happen
 optimize table ttl partition 10 final;
 
 select * from ttl order by d;
@@ -18,7 +17,7 @@ alter table ttl modify ttl a; -- { serverError 450 }
 
 drop table if exists ttl;
 
-create table ttl (d Date, a Int) engine = MergeTree order by tuple() partition by toDayOfMonth(d);
+create table ttl (d Date, a Int) engine = MergeTree order by tuple() partition by toDayOfMonth(d) settings remove_empty_parts = 0;
 alter table ttl modify column a Int ttl d + interval 1 day;
 desc table ttl;
 alter table ttl modify column d Int ttl d + interval 1 day; -- { serverError 43 }

tests/queries/0_stateless/00933_ttl_simple.sql

@@ -11,7 +11,9 @@ select a, b from ttl_00933_1;
 
 drop table if exists ttl_00933_1;
 
-create table ttl_00933_1 (d DateTime, a Int, b Int) engine = MergeTree order by toDate(d) partition by tuple() ttl d + interval 1 second;
+create table ttl_00933_1 (d DateTime, a Int, b Int)
+    engine = MergeTree order by toDate(d) partition by tuple() ttl d + interval 1 second
+    settings remove_empty_parts = 0;
 insert into ttl_00933_1 values (now(), 1, 2);
 insert into ttl_00933_1 values (now(), 3, 4);
 insert into ttl_00933_1 values (now() + 1000, 5, 6);
@@ -30,7 +32,9 @@ select * from ttl_00933_1 order by d;
 
 drop table if exists ttl_00933_1;
 
-create table ttl_00933_1 (d DateTime, a Int) engine = MergeTree order by tuple() partition by tuple() ttl d + interval 1 day;
+create table ttl_00933_1 (d DateTime, a Int)
+    engine = MergeTree order by tuple() partition by tuple() ttl d + interval 1 day
+    settings remove_empty_parts = 0;
 insert into ttl_00933_1 values (toDateTime('2000-10-10 00:00:00'), 1);
 insert into ttl_00933_1 values (toDateTime('2000-10-10 00:00:00'), 2);
 insert into ttl_00933_1 values (toDateTime('2100-10-10 00:00:00'), 3);
@@ -39,7 +43,9 @@ select * from ttl_00933_1 order by d;
 
 drop table if exists ttl_00933_1;
 
-create table ttl_00933_1 (d Date, a Int) engine = MergeTree order by a partition by toDayOfMonth(d) ttl d + interval 1 day;
+create table ttl_00933_1 (d Date, a Int)
+    engine = MergeTree order by a partition by toDayOfMonth(d) ttl d + interval 1 day
+    settings remove_empty_parts = 0;
 insert into ttl_00933_1 values (toDate('2000-10-10'), 1);
 insert into ttl_00933_1 values (toDate('2100-10-10'), 2);
 optimize table ttl_00933_1 final;

tests/queries/0_stateless/00976_ttl_with_old_parts.sql

 drop table if exists ttl;
 
-create table ttl (d Date, a Int) engine = MergeTree order by a partition by toDayOfMonth(d);
+create table ttl (d Date, a Int) engine = MergeTree order by a partition by toDayOfMonth(d) settings remove_empty_parts = 0;
 insert into ttl values (toDateTime('2000-10-10 00:00:00'), 1);
 insert into ttl values (toDateTime('2000-10-10 00:00:00'), 2);
 insert into ttl values (toDateTime('2100-10-10 00:00:00'), 3);

tests/queries/0_stateless/01006_simpod_empty_part_single_column_write.sh

@@ -16,7 +16,7 @@ ${CLICKHOUSE_CLIENT} --query="CREATE TABLE table_with_empty_part
 ENGINE = MergeTree()
 ORDER BY id
 PARTITION BY id
-SETTINGS vertical_merge_algorithm_min_rows_to_activate=0, vertical_merge_algorithm_min_columns_to_activate=0
+SETTINGS vertical_merge_algorithm_min_rows_to_activate=0, vertical_merge_algorithm_min_columns_to_activate=0, remove_empty_parts = 0
 "
 
 

tests/queries/0_stateless/01047_simple_aggregate_sizes_of_columns_bug.sql

 DROP TABLE IF EXISTS column_size_bug;
 
-CREATE TABLE column_size_bug (date_time DateTime, value SimpleAggregateFunction(sum,UInt64)) ENGINE = AggregatingMergeTree PARTITION BY toStartOfInterval(date_time, INTERVAL 1 DAY) ORDER BY (date_time);
+CREATE TABLE column_size_bug (date_time DateTime, value SimpleAggregateFunction(sum,UInt64)) ENGINE = AggregatingMergeTree PARTITION BY toStartOfInterval(date_time, INTERVAL 1 DAY) ORDER BY (date_time) SETTINGS remove_empty_parts = 0;
 
 INSERT INTO column_size_bug VALUES(now(),1);
 INSERT INTO column_size_bug VALUES(now(),1);