diff --git a/src/query/service/src/pipelines/builders/builder_mutation.rs b/src/query/service/src/pipelines/builders/builder_mutation.rs
index 18b64ffe59f0..3cd64c367fc5 100644
--- a/src/query/service/src/pipelines/builders/builder_mutation.rs
+++ b/src/query/service/src/pipelines/builders/builder_mutation.rs
@@ -20,6 +20,7 @@ use databend_common_exception::Result;
 use databend_common_expression::BlockThresholds;
 use databend_common_expression::DataSchema;
 use databend_common_expression::DataSchemaRef;
+use databend_common_pipeline_core::processors::create_resize_item;
 use databend_common_pipeline_core::processors::InputPort;
 use databend_common_pipeline_core::processors::OutputPort;
 use databend_common_pipeline_core::processors::ProcessorPtr;
@@ -239,6 +240,18 @@ impl PipelineBuilder {
     ) -> Result<()> {
         // we should avoid too much little block write, because for s3 write, there are too many
         // little blocks, it will cause high latency.
+        let mut origin_len = transform_len;
+        let mut resize_len = 1;
+        let mut pipe_items = Vec::with_capacity(2);
+        if need_match {
+            origin_len += 1;
+            resize_len += 1;
+            pipe_items.push(create_dummy_item());
+        }
+        pipe_items.push(create_resize_item(transform_len, 1));
+        self.main_pipeline
+            .add_pipe(Pipe::create(origin_len, resize_len, pipe_items));
+
         let mut builder = self.main_pipeline.add_transform_with_specified_len(
             |transform_input_port, transform_output_port| {
                 Ok(ProcessorPtr::create(AccumulatingTransformer::create(
@@ -247,13 +260,21 @@ impl PipelineBuilder {
                     BlockCompactBuilder::new(block_thresholds),
                 )))
             },
-            transform_len,
+            1,
         )?;
         if need_match {
             builder.add_items_prepend(vec![create_dummy_item()]);
         }
         self.main_pipeline.add_pipe(builder.finalize());
 
+        let mut pipe_items = Vec::with_capacity(2);
+        if need_match {
+            pipe_items.push(create_dummy_item());
+        }
+        pipe_items.push(create_resize_item(1, transform_len));
+        self.main_pipeline
+            .add_pipe(Pipe::create(resize_len, origin_len, pipe_items));
+
         let mut builder = self.main_pipeline.add_transform_with_specified_len(
             |transform_input_port, transform_output_port| {
                 Ok(ProcessorPtr::create(BlockMetaTransformer::create(
diff --git a/tests/sqllogictests/suites/base/09_fuse_engine/09_0011_change_tracking.test b/tests/sqllogictests/suites/base/09_fuse_engine/09_0011_change_tracking.test
index 2263a860cd04..6025b15c33f4 100644
--- a/tests/sqllogictests/suites/base/09_fuse_engine/09_0011_change_tracking.test
+++ b/tests/sqllogictests/suites/base/09_fuse_engine/09_0011_change_tracking.test
@@ -91,25 +91,25 @@ statement ok
 create table t2(a int)
 
 statement ok
-insert into t2 values(1),(2),(8)
+insert into t2 values(0),(2),(1)
 
 statement ok
 set enable_experimental_merge_into = 1
 
 query TTT
-merge into t using t2 on t.a = t2.a when matched and t2.a = 1 then update set t.a = 0 when matched and t2.a = 2 then delete when not matched then insert *
+merge into t using t2 on t.a = t2.a when matched and t2.a = 1 then update set t.a = 8 when matched and t2.a = 2 then delete when not matched then insert *
 ----
 1 1 1
 
 query IBBII
 select a, _origin_version is null, _origin_block_id is null, _origin_block_row_num, _row_version from t order by a
 ----
-0 0 0 0 1
+0 0 0 0 0
 3 0 0 1 1
 5 0 0 0 0
 6 0 0 0 0
 7 0 0 1 0
-8 0 0 0 0
+8 0 0 0 1
 
 statement ok
 create table t1(a int) change_tracking = true
@@ -131,12 +131,12 @@ merge into t using t1 on t.a = t1.a when matched and t1.a = 0 then update set t.
 query IBBII
 select a, _origin_version is null, _origin_block_id is null, _origin_block_row_num, _row_version from t order by a
 ----
-1 0 0 0 2
+1 0 0 0 1
 2 0 0 1 2
 5 0 0 0 0
 6 0 0 0 0
 7 0 0 1 0
-8 0 0 0 0
+8 0 0 0 1
 
 ###############
 # issue 14955 #