[
https://issues.apache.org/jira/browse/FLINK-34348?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17815184#comment-17815184
]
Feng Jin commented on FLINK-34348:
----------------------------------
{color:#000000}I have completed the test, and overall it meets
expectations.{color}
h2. *The following is the main process of my test*
1. create a CDC source (here I use the paimon table),
2. Execute explain plan
3. Write relevant data to confirm if the results meet expectations.
h2. Create The CDC Source
!截屏2024-02-07 16.21.37.png!
!截屏2024-02-07 16.21.55.png!
h2. Execute explain plan
*TUMBLE*
!截屏2024-02-07 16.22.24.png!
*SESSION*
!截屏2024-02-07 16.23.12.png!
*HOP*
!截屏2024-02-07 16.23.27.png!
*CUMULATE*
!截屏2024-02-07 16.23.38.png!
h2. Verify the result
*TUMBLE*
Insert the data and the result as follows.
{code:sql}
SELECT window_start, window_end, SUM(cnt) FROM TABLE( TUMBLE(TABLE
test_source, DESCRIPTOR(`timestamp`), INTERVAL '10' MINUTES)) GROUP BY
window_start, window_end;
{code}
!截屏2024-02-07 16.29.09.png!
!截屏2024-02-07 16.29.21.png!
!截屏2024-02-07 16.29.34.png!
* SESSION *
Insert the data and the result as follows.
{code:sql}
SELECT window_start, window_end, SUM(cnt) FROM TABLE(SESSION(TABLE test_source,
DESCRIPTOR(`timestamp`), INTERVAL '10' MINUTES)) GROUP BY
window_start, window_end;
{code}
!截屏2024-02-07 16.46.12.png!
!截屏2024-02-07 16.46.23.png!
!截屏2024-02-07 16.46.37.png!
* HOP *
Insert the data and the result as follows.
{code:sql}
SELECT window_start, window_end, SUM(cnt) FROM TABLE( HOP(TABLE test_source,
DESCRIPTOR(`timestamp`), INTERVAL '10' MINUTES, INTERVAL '20'
MINUTES)) GROUP BY window_start, window_end;
{code}
!截屏2024-02-07 16.53.37.png!
!截屏2024-02-07 16.53.47.png!
!截屏2024-02-07 16.54.01.png!
* CUMULATE *
Insert the data and the result as follows.
{code:sql}
SELECT window_start, window_end, SUM(cnt) FROM TABLE(CUMULATE(TABLE
test_source, DESCRIPTOR(`timestamp`), INTERVAL '10' MINUTES, INTERVAL '20'
MINUTES)) GROUP BY window_start, window_end;
{code}
!截屏2024-02-07 16.59.22.png!
!截屏2024-02-07 16.59.33.png!
!截屏2024-02-07 16.59.42.png!
> Release Testing: Verify FLINK-20281 Window aggregation supports changelog
> stream input
> --------------------------------------------------------------------------------------
>
> Key: FLINK-34348
> URL: https://issues.apache.org/jira/browse/FLINK-34348
> Project: Flink
> Issue Type: Sub-task
> Components: Table SQL / API
> Affects Versions: 1.19.0
> Reporter: xuyang
> Assignee: Feng Jin
> Priority: Blocker
> Labels: release-testing
> Fix For: 1.19.0
>
> Attachments: 截屏2024-02-07 16.21.37.png, 截屏2024-02-07 16.21.55.png,
> 截屏2024-02-07 16.22.24.png, 截屏2024-02-07 16.23.12.png, 截屏2024-02-07
> 16.23.27.png, 截屏2024-02-07 16.23.38.png, 截屏2024-02-07 16.29.09.png,
> 截屏2024-02-07 16.29.21.png, 截屏2024-02-07 16.29.34.png, 截屏2024-02-07
> 16.46.12.png, 截屏2024-02-07 16.46.23.png, 截屏2024-02-07 16.46.37.png,
> 截屏2024-02-07 16.53.37.png, 截屏2024-02-07 16.53.47.png, 截屏2024-02-07
> 16.54.01.png, 截屏2024-02-07 16.59.22.png, 截屏2024-02-07 16.59.33.png,
> 截屏2024-02-07 16.59.42.png
>
>
> Window TVF aggregation supports changelog stream is ready for testing. User
> can add a window tvf aggregation as a down stream after CDC source or some
> nodes that will produce cdc records.
> Someone can verify this feature with:
> # Prepare a mysql table, and insert some data at first.
> # Start sql-client and prepare ddl for this mysql table as a cdc source.
> # You can verify the plan by `EXPLAIN PLAN_ADVICE` to check if there is a
> window aggregate node and the changelog contains "UA" or "UB" or "D" in its
> upstream.
> # Use different kinds of window tvf to test window tvf aggregation while
> updating the source data to check the data correctness.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
