Merge pull request ClickHouse#88147 from ClickHouse/backport/25.8/88095

Backport ClickHouse#88095 to 25.8: Fix data corruption with expired column TTL and sparse serialization

Author: clickhouse-gh[bot]

src/Storages/MergeTree/IMergeTreeDataPart.cpp

@@ -558,6 +558,7 @@ void IMergeTreeDataPart::setColumns(const NamesAndTypesList & new_columns, const
     serialization_infos = new_infos;
     metadata_version = new_metadata_version;
 
+    serializations.clear();
     column_name_to_position.clear();
     column_name_to_position.reserve(new_columns.size());
     size_t pos = 0;

src/Storages/MergeTree/MergeTask.cpp

@@ -509,17 +509,9 @@ bool MergeTask::ExecuteAndFinalizeHorizontalPart::prepare() const
     const auto & expired_columns = global_ctx->new_data_part->expired_columns;
     if (!expired_columns.empty())
     {
-        auto part_serialization_infos = global_ctx->new_data_part->getSerializationInfos();
-        for (const auto & expired_column : expired_columns)
-            part_serialization_infos.erase(expired_column);
-
         global_ctx->gathering_columns = global_ctx->gathering_columns.eraseNames(expired_columns);
         global_ctx->merging_columns = global_ctx->merging_columns.eraseNames(expired_columns);
         global_ctx->storage_columns = global_ctx->storage_columns.eraseNames(expired_columns);
-        global_ctx->new_data_part->setColumns(
-            global_ctx->storage_columns,
-            part_serialization_infos,
-            global_ctx->metadata_snapshot->getMetadataVersion());
     }
 
     global_ctx->new_data_part->uuid = global_ctx->future_part->uuid;

tests/queries/0_stateless/03642_column_ttl_sparse.reference

@@ -0,0 +1,4 @@
+435
+435
+435
+435

tests/queries/0_stateless/03642_column_ttl_sparse.sql

@@ -0,0 +1,26 @@
+DROP TABLE IF EXISTS ttl_sparse_repro;
+
+CREATE TABLE ttl_sparse_repro
+(
+    a UInt64,
+    dt DateTime,
+    b UInt64 TTL dt + INTERVAL 2 SECOND, c UInt64
+)
+ENGINE = MergeTree ORDER BY a SETTINGS ratio_of_defaults_for_sparse_serialization = 0.9;
+
+INSERT INTO ttl_sparse_repro SELECT number, now(), 1, if (number % 23 = 0, 1, 0) FROM numbers(10000);
+
+SELECT sum(c) FROM ttl_sparse_repro;
+
+SELECT sleep(3) FORMAT Null;
+
+OPTIMIZE TABLE ttl_sparse_repro FINAL;
+SELECT sum(c) FROM ttl_sparse_repro;
+
+OPTIMIZE TABLE ttl_sparse_repro FINAL;
+SELECT sum(c) FROM ttl_sparse_repro;
+
+OPTIMIZE TABLE ttl_sparse_repro FINAL;
+SELECT sum(c) FROM ttl_sparse_repro;
+
+DROP TABLE IF EXISTS ttl_sparse_repro;