[jira] [Created] (FLINK-20374) Wrong result when shuffling changelog stream on non-primary-key columns

[Shang Yuanchun (Jira)]

Jark Wu created FLINK-20374:
-------------------------------

             Summary: Wrong result when shuffling changelog stream on non-primary-key columns
                 Key: FLINK-20374
                 URL: https://issues.apache.org/jira/browse/FLINK-20374
             Project: Flink
          Issue Type: Bug
          Components: Table SQL / Planner
            Reporter: Jark Wu


This is reported from user-zh ML: http://apache-flink.147419.n8.nabble.com/flink-1-11-2-cdc-cdc-sql-sink-save-point-job-sink-td8593.html

{code:sql}
CREATE TABLE test (
`id` INT,
`name` VARCHAR(255),
`time` TIMESTAMP(3),
`status` INT,
PRIMARY KEY(id) NOT ENFORCED
) WITH (
  'connector' = 'mysql-cdc',
  'hostname' = 'localhost',
  'port' = '3306',
  'username' = 'root',
  'password' = '1',
  'database-name' = 'ai_audio_lyric_task',
  'table-name' = 'test'
)

CREATE TABLE status (
`id` INT,
`name` VARCHAR(255),
PRIMARY KEY(id) NOT ENFORCED
) WITH (  
  'connector' = 'mysql-cdc',
  'hostname' = 'localhost',
  'port' = '3306',
  'username' = 'root',
  'password' = '1',
  'database-name' = 'ai_audio_lyric_task',
  'table-name' = 'status'
);

-- output
CREATE TABLE test_status (
`id` INT,
`name` VARCHAR(255),
`time` TIMESTAMP(3),
`status` INT,
`status_name` VARCHAR(255)
PRIMARY KEY(id) NOT ENFORCED
) WITH (
  'connector' = 'elasticsearch-7',
  'hosts' = 'xxx',
  'index' = 'xxx',
  'username' = 'xxx',
  'password' = 'xxx',
  'sink.bulk-flush.backoff.max-retries' = '100000',
  'sink.bulk-flush.backoff.strategy' = 'CONSTANT',
  'sink.bulk-flush.max-actions' = '5000',
  'sink.bulk-flush.max-size' = '10mb',
  'sink.bulk-flush.interval' = '1s'
);

INSERT into test_status
SELECT t.*, s.name
FROM test AS t
LEFT JOIN status AS s ON t.status = s.id;
{code}

Data in mysql table:

{code}
test:
0, name0, 2020-07-06 00:00:00 , 0
1, name1, 2020-07-06 00:00:00 , 1
2, name2, 2020-07-06 00:00:00 , 1
.....

status
0, status0
1, status1
2, status2
.....
{code}

Operations:
1. start job with paralleslim=40, result in test_status sink is correct:

{code}
    0, name0, 2020-07-06 00:00:00 , 0, status0
    1, name1, 2020-07-06 00:00:00 , 1, status1
    2, name2, 2020-07-06 00:00:00 , 1, status1
{code}

2. Update {{status}} of {{id=2}} record in table {{test}} from {{1}} to {{2}}.
3. Result is not correct because the {{id=2}} record is missing in the result.



The reason is that it shuffles the changelog {{test}} on {{status}} column which is not the primary key. Therefore, the ordering can't be guaranteed, and the result is wrong.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)