All incremental models started failing yesterday

Help
, ,
1

Yesterday around 9AM pacific. All of our Incremental models started failing with a message of “syntax error at or near “as” in context” Our incremental models use delete+insert (default). Since all of our models in prod, dev and local started seeing this issue. We are assuming it’s a global setting or bad push from dbt? We were running 1.9.1 when this happened, but upgraded to 1.9.3 just in case the issue was version specific. No change, it’s broken with both versions.

To get around the issue, we are forcing a full_refresh=true on all models. Obviously the system is struggling but our models are working again.

The dbt macro that builds the sql to execute against redshift. Started adding an alias to the delete statement. This alias doesn’t work in Redshift and we get the context error.

example of the code, now being generated by dbt. The extra code being added is the “as dbt_INTERNAL_DEST”.

delete from "database"."schema"."table" as DBT_INTERNAL_DEST
        where (table_id) in (
            select distinct table_id
            from "table__dbt_tmp154141418053" as DBT_INTERNAL_SOURCE
        );

Is anyone aware of a global setting that would start adding this alias to the delete statement in the redshift incremental code path? Looking back in the logs, it’s not there before 9AM and there every time after.

2

Hey! Please see the relevant PR. It’s not a bad idea to pin your dbt-core, dbt-adapters, and dbt-redshift versions for consistency! It’s somewhat rare for us to do a backport so you can be quite sure this won’t happen with version pins.

3

Thanks!!

We had core and redshift pinned, have added dbt-adapters to the list of pins.

Glad this was something simple.

Regards

1 Like