Insert into ClickHouse table which use ENGINE = Distributed causes incorrect result

[lupengfei-johnson]

clickhouse.testdb.test5 uses ENGINE = Distributed , there is 1 row in it.
trino> select * from clickhouse.testdb.test5;
eventdate | counterid | userid | appid
-------------------------+-----------+--------+-------
2021-04-15 19:00:00.000 | 123 | 123 | 123
(1 row)

Insert another row to it
trino> insert into clickhouse.testdb.test5 values(timestamp '2021-04-15 20:00:00',12345,123,123);
INSERT: 1 row

but there is 4 rows in the table.
trino> select * from clickhouse.testdb.test5;
eventdate | counterid | userid | appid
-------------------------+-----------+--------+-------
2021-04-15 19:00:00.000 | 123 | 123 | 123
2021-04-15 19:00:00.000 | 123 | 123 | 123
2021-04-15 20:00:00.000 | 12345 | 123 | 123
2021-04-15 20:00:00.000 | 12345 | 123 | 123
(4 rows)

[findepi]

cc @wgzhao

[wgzhao]

@lupengfei-johnson Similar to #7601, the Distributed Table Engine involves multi-node data sharding, and currently the clickhouse connector does not support distributed clickhouse writes

[nikita-sheremet-java-developer]

I have exact the same problem. But when inserttions are done by Spark there are no doubles. So it look like some trino settings are wrong.

@wgzhao
I also appologies if I'm bothering you, but what do think about this?

[wgzhao]

which version you are testing?

[nikita-sheremet-java-developer]

@wgzhao
Trino 464
ClickHouse 23.8.16.16

[wgzhao]

Could you describe the ClickHouse deployment environment in detail?
How many nodes are being used, and what is the data replication mode between the nodes?
Is it replica or shard?
Please provide the detailed table creation statements.