Re: Insert on conflict do update fails with duplicate rows - there are no duplicates

On Mon, 2025-01-27 at 10:18 -0300, Pedro Moraes wrote:
>
> I managed to replicate the issue with the following data & tables attached download link.
>
> I am trying to insert from bugtest.temp_on_conflict_test into bugtest.history, both tables have a primary key on history_id so there cannot be duplicates
>
> The insert query also uses distinct on (history_id)
>
> INSERT INTO bugtest.history (account_id,appo_id,account_name,resource_owner_name,resource_owner_id,resource_owner_user_id,resource_id,resource_type,preview,resource_date,history_id)
> select distinct on (history_id) * from bugtest.temp_on_conflict_test limit 2
> ON CONFLICT (history_id) DO UPDATE SET
> account_id = excluded.account_id,
> resource_owner_name = excluded.resource_owner_name,
> resource_owner_user_id = excluded.resource_owner_user_id,
> resource_owner_id = excluded.resource_owner_id,
> preview = excluded.preview,
> resource_date = excluded.resource_date,
> account_name = excluded.account_name
> RETURNING *
>
> Reproducible dump:
> https://limewire.com/d/2d598eff-f7c9-47d2-80d6-7e522ee1688c#ru2TRVilIqQdqmEjr2-_H4z67PHYe2JZAshDErA8umw

I get this error message:

ERROR: ON CONFLICT DO UPDATE command cannot affect row a second time
HINT: Ensure that no rows proposed for insertion within the same command have duplicate constrained values.

The problem is that you are using "SELECT *", which means that you will get the columns
in the order they were defined in the table. So you end up inserting
"temp_on_conflict_test.account_name" into "history.history_id", which causes the error,
because all rows in "temp_on_conflict_test" have the same value in that column.

Lessons to learn:
- don't use "SELECT *"
- use more appropriate data types
In the case at hand, using type "uuid" for "history_id" would have got you a more
understandable error message.

Yours,
Laurenz Albe