This is an automated email from the ASF dual-hosted git repository.
avamingli pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/cloudberry.git
The following commit(s) were added to refs/heads/main by this push:
new 0cd5c6067de Fix NULL locus of Shared Scan.
0cd5c6067de is described below
commit 0cd5c6067de85c50119a995a1477bd96aa1c4765
Author: Zhang Mingli <[email protected]>
AuthorDate: Mon Sep 29 17:38:32 2025 +0800
Fix NULL locus of Shared Scan.
The shared scan needs its locus type and parallel number to be
explicitly set when `gp_cte_sharing` is used, even without
parallelism. Failing to do so resulted in an incorrect NULL locus
appearing in the query plan.
Fix https://github.com/apache/cloudberry/issues/1376
Authored-by: Zhang Mingli [email protected]
---
src/backend/optimizer/plan/planshare.c | 3 ++
src/test/regress/expected/cbdb_parallel.out | 54 ++++++++++++++++++++++++++++-
src/test/regress/sql/cbdb_parallel.sql | 15 +++++++-
3 files changed, 70 insertions(+), 2 deletions(-)
diff --git a/src/backend/optimizer/plan/planshare.c
b/src/backend/optimizer/plan/planshare.c
index f60f45f933e..f3b716364ca 100644
--- a/src/backend/optimizer/plan/planshare.c
+++ b/src/backend/optimizer/plan/planshare.c
@@ -52,6 +52,9 @@ make_shareinputscan(PlannerInfo *root, Plan *inputplan)
sisc->scan.plan.plan_rows = inputplan->plan_rows;
sisc->scan.plan.plan_width = inputplan->plan_width;
+ sisc->scan.plan.locustype = inputplan->locustype;
+ sisc->scan.plan.parallel = 0; /* No parallel ShareInputScan */
+
return sisc;
}
diff --git a/src/test/regress/expected/cbdb_parallel.out
b/src/test/regress/expected/cbdb_parallel.out
index 942705e7471..da3216896ff 100644
--- a/src/test/regress/expected/cbdb_parallel.out
+++ b/src/test/regress/expected/cbdb_parallel.out
@@ -3516,7 +3516,59 @@ WHERE e.salary > (
David
(2 rows)
-
+--
+-- Test https://github.com/apache/cloudberry/issues/1376
+--
+create table t1(a int, b int);
+NOTICE: Table doesn't have 'DISTRIBUTED BY' clause -- Using column named 'a'
as the Apache Cloudberry data distribution key for this table.
+HINT: The 'DISTRIBUTED BY' clause determines the distribution of data. Make
sure column(s) chosen are the optimal data distribution key to minimize skew.
+create table t2 (like t1);
+NOTICE: table doesn't have 'DISTRIBUTED BY' clause, defaulting to
distribution columns from LIKE table
+set gp_cte_sharing = on;
+explain(locus, costs off) with x as
+ (select a, count(*) as b from t1 group by a union all
+ select a, count(*) as b from t2 group by a)
+ select count(*) from x a join x b on a.a = b.b;
+ QUERY PLAN
+------------------------------------------------------------------------
+ Finalize Aggregate
+ Locus: Entry
+ -> Gather Motion 3:1 (slice1; segments: 3)
+ Locus: Entry
+ -> Partial Aggregate
+ Locus: Hashed
+ -> Hash Join
+ Locus: Hashed
+ Hash Cond: (b.b = a.a)
+ -> Redistribute Motion 3:3 (slice2; segments: 3)
+ Locus: Hashed
+ Hash Key: b.b
+ -> Subquery Scan on b
+ Locus: Strewn
+ -> Shared Scan (share slice:id 2:0)
+ Locus: Hashed
+ -> Hash
+ Locus: Hashed
+ -> Subquery Scan on a
+ Locus: Hashed
+ -> Shared Scan (share slice:id 1:0)
+ Locus: Hashed
+ -> Append
+ Locus: Hashed
+ -> HashAggregate
+ Locus: Hashed
+ Group Key: t1.a
+ -> Seq Scan on t1
+ Locus: Hashed
+ -> HashAggregate
+ Locus: Hashed
+ Group Key: t2.a
+ -> Seq Scan on t2
+ Locus: Hashed
+ Optimizer: Postgres query optimizer
+(35 rows)
+
+reset gp_cte_sharing;
reset enable_parallel;
reset min_parallel_table_scan_size;
-- start_ignore
diff --git a/src/test/regress/sql/cbdb_parallel.sql
b/src/test/regress/sql/cbdb_parallel.sql
index 0ee6f72cb2a..f9d01dd8a00 100644
--- a/src/test/regress/sql/cbdb_parallel.sql
+++ b/src/test/regress/sql/cbdb_parallel.sql
@@ -1132,7 +1132,20 @@ WHERE e.salary > (
SELECT AVG(salary)
FROM employees
WHERE department_id = e.department_id);
-
+
+--
+-- Test https://github.com/apache/cloudberry/issues/1376
+--
+create table t1(a int, b int);
+create table t2 (like t1);
+set gp_cte_sharing = on;
+
+explain(locus, costs off) with x as
+ (select a, count(*) as b from t1 group by a union all
+ select a, count(*) as b from t2 group by a)
+ select count(*) from x a join x b on a.a = b.b;
+
+reset gp_cte_sharing;
reset enable_parallel;
reset min_parallel_table_scan_size;
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
