github-actions[bot] commented on code in PR #66443:
URL: https://github.com/apache/doris/pull/66443#discussion_r3713395076


##########
be/test/exec/operator/materialization_shared_state_test.cpp:
##########
@@ -473,8 +508,8 @@ TEST_F(MaterializationSharedStateTest, 
TestMergeMultiResponseStaleBlockMaps) {
         ASSERT_TRUE(rel0_block.serialize(0, pb0, &us, &cs, &ct, 
CompressionTypePB::LZ4).ok());
         _shared_state->response_blocks[0] = rel0_block.clone_empty();
 
-        // blocks[1]: empty (BE_1 has no data for relation 1)
-        response.add_blocks();
+        // BE_1 has no data for relation 1. This also models an ignored 
transport failure for

Review Comment:
   **[P2] Add coverage for the actual reused-response failure path**
   
   This fixture cannot arise from the failure it claims to model: BE 1 has a 
non-empty relation-0 request and a valid relation-0 response, while only 
relation 1's block is omitted. A transport failure applies to the whole RPC and 
`validate_rpc_results()` rejects this request because its aggregate row count 
is nonzero; on success, `RowIdStorageReader::read_by_rowids()` appends one 
positional block for every relation, including empty ones.
   
   As a result, none of the new tests would fail if the newly added 
`response.Clear()` or consecutive-batch cleanup were removed. Please add a 
two-batch reuse test (or extract dispatch preparation into a testable helper) 
that seeds a stale non-empty response, then runs an all-empty failed backend 
alongside a data-bearing successful backend and verifies controller reset, 
stable relation indexing, no stale block/profile use, and cleared row/file 
arrays.



-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to