Pin max_rows_to_group_by and prefer_localhost_replica in the regression test

groeneai · claude · groeneai · commit fea96be84f58 · 2026-06-24T18:32:10.000Z
The CI test profile (tests/config/users.d/limits.yaml) sets max_rows_to_group_by
to 10G, and make_distributed_plan rejects aggregation with a non-zero limit
(Code 344, SUPPORT_IS_DISABLED). That fired before the query could reach the
ScatterExchangeStep path, so the Fast test for this regression failed without
exercising the fix.

clickhouse-test also randomizes prefer_localhost_replica. With it set to 0 the
Distributed engine serializes the per-child plan to the localhost replica, where
the experimental BlocksMarshalling step is not deserializable (Code 47,
UNKNOWN_IDENTIFIER). That path is unrelated to the crash, which happens earlier
while building the per-child plan.

Pin both settings so the query deterministically reaches the multi-bucket
ScatterExchangeStep. EXPLAIN distributed=1 confirms the per-child plan still
carries a ShuffleExchange over a bucketed ReadFromMergeTree (the step that threw
'should have one source shard, got 8' before the fix), so crash-path coverage is
preserved. 20/20 runs pass under full CI randomization.

Co-Authored-By: Claude Opus 4.8 &lt;noreply@anthropic.com&gt;
diff --git a/tests/queries/0_stateless/04367_distributed_plan_merge_scatter_multishard.sql b/tests/queries/0_stateless/04367_distributed_plan_merge_scatter_multishard.sql
@@ -23,8 +23,14 @@ CREATE TABLE d107946_1 AS base107946_1 ENGINE = Distributed(test_shard_localhost
 CREATE TABLE d107946_4 AS base107946_4 ENGINE = Distributed(test_shard_localhost, currentDatabase(), base107946_4);
 CREATE TABLE m107946 ENGINE = Merge(currentDatabase(), '^d107946_(1|4)$');
 
+-- max_rows_to_group_by must be 0: the CI test profile sets it to 10G, and make_distributed_plan
+-- rejects aggregation with a non-zero limit (Code 344), which would mask the crash path.
+-- prefer_localhost_replica must be 1: with 0 the Distributed engine serializes the plan to the
+-- localhost replica, where the experimental BlocksMarshalling step is not deserializable (Code 47);
+-- that path is unrelated to this crash, which happens earlier while building the per-child plan.
 SET make_distributed_plan = 1, enable_parallel_replicas = 0, distributed_plan_execute_locally = 1,
-    use_statistics = 1, distributed_plan_optimize_exchanges = 1, enable_join_runtime_filters = 0;
+    use_statistics = 1, distributed_plan_optimize_exchanges = 1, enable_join_runtime_filters = 0,
+    max_rows_to_group_by = 0, prefer_localhost_replica = 1;
 
 -- The exact reproducer from the issue. The Merge child plans are distributed, so the scatter built for
 -- the per-child aggregation sees the bucketed distributed read as its source. No rows match (the
