You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hello Drill community,
Recently, we have encountered with the following issue( UNSUPPORTED_OPERATION ERROR: HashJoinBatch does not support schema changes in build side. Prior schema), when joining two CTEs.
There two CTEs which having filter conditions from list of items. and need to join based on given condition.
it seems, after join operation, schema changes for field filtered__items1
[filtered__items1 (NULL:OPTIONAL)]]
[filtered__items1 (MAP:REPEATED)
Executed queries:
WITH cte_1 AS (
SELECT t.customer_id as customer_id1, t.customer_name as customer_name1,
COLLECT_LIST('val', t.p_item) filtered__items1
FROM (
SELECT d.customer_id, d.customer_name, flatten(d.purchased_items) AS p_item
FROM dfs.root.`/datas3/customers/*` d
) t
WHERE t.p_item['product_id'] IN (777, 888)
GROUP BY t.customer_id, t.customer_name
),
cte_2 AS (
SELECT t.customer_id as customer_id2, t.customer_name as customer_name2,
COLLECT_LIST('val', t.p_item) filtered__items2
FROM (
SELECT d.customer_id, d.customer_name, flatten(d.purchased_items) AS p_item
FROM dfs.root.`/datas3/customers/*` d
) t
WHERE t.p_item['product_id'] IN (999)
GROUP BY t.customer_id, t.customer_name
)
SELECT cte_1.*
FROM cte_1
JOIN cte_2 ON cte_1.customer_id1 = cte_2.customer_id2;
Error details:
-------------------------------------------------------------------------------------------------------
org.apache.drill.common.exceptions.UserRemoteException: UNSUPPORTED_OPERATION ERROR: HashJoinBatch does not support schema changes in build side.
Prior schema :
BatchSchema [fields=[[`customer_id1` (BIGINT:OPTIONAL)], [`customer_name1` (VARCHAR:OPTIONAL)], [`filtered__items1` (NULL:OPTIONAL)]], selectionVector=NONE]
New schema :
BatchSchema [fields=[[`customer_id1` (BIGINT:OPTIONAL)], [`customer_name1` (VARCHAR:OPTIONAL)], [`filtered__items1` (MAP:REPEATED), children=([`val` (MAP:REQUIRED), children=([`item_id` (BIGINT:OPTIONAL)], [`item_class` (VARCHAR:OPTIONAL)], [`product_id` (BIGINT:OPTIONAL)], [`created_at` (VARCHAR:OPTIONAL)])])]], selectionVector=NONE]
Unsupported schema change
Fragment: 0:0
[Error Id: 6b857b70-583e-4fb0-ab02-d1e6b536c569 on 8a288c59e4a7:31010]
------------------------------------------------------------------------
Drill version
Apache drill version: 1.21.2
Additional context
You may download as parquet file from here
Test data
Hello Drill community,
Recently, we have encountered with the following issue(
UNSUPPORTED_OPERATION ERROR: HashJoinBatch does not support schema changes in build side. Prior schema
), when joining two CTEs.There two CTEs which having filter conditions from list of items. and need to join based on given condition.
it seems, after join operation, schema changes for field
filtered__items1
[
filtered__items1
(NULL:OPTIONAL)]][
filtered__items1
(MAP:REPEATED)Executed queries:
Error details:
Drill version
Apache drill version: 1.21.2
Additional context
You may download as parquet file from here
Test data
The text was updated successfully, but these errors were encountered: