Hi, can we run the dbt test when running the dbt-asset?

Hey Julius! I set up dbt test assets using an asset factory like below:

def test_asset_factory(
    model: str, non_argument_deps: set[AssetKey] | set[str] | None
) -> AssetsDefinition:
    @asset(
        name=f"test_{model}",
        non_argument_deps=non_argument_deps,
        required_resource_keys={"dbt"},
        group_name="dbt_tests",
        key_prefix="dbt_project",
        compute_kind="dbt",
    )
    def _asset(context: OpExecutionContext) -> DbtCliOutput:
        res = context.resources.dbt.test(models=[model])
        return res

    return _asset


test_assets = [
    test_asset_factory(**kwargs)
    for kwargs in [
        ...
    ]
]

Some commands have a parameter

use_build_command

to use

dbt build

instead of

dbt run

ah TIL thanks, Benoit daggy love

@Benoit Perigaud I'm using the

use_build_command

, but for some reason, the jobs don't catch the dbt-test if it is a failure test. I also want whole job fails when dbt-test fail.

Hi! Thanks for the information! I have a question. Do the sensors interfere in any way with use_build_command? we use a sensor for executions

On my side, the project I have behave as expected with

build

, if a test fails, the dbt command will skip the downstream model and Dagster will report an error.

@Benoit Perigaud ok, thanks for the information, let me try more with

use_build_command

. Did you use the lastest version of dbt-dagster?

I use it on a personal project (yes, dbt and dagster in my spare time 🙈), that doesn’t fail often but I got it failing as expected when I was on dagster 1.2 and dbt 1.4

Thanks @Julius ! And the

use_build_command

do you include it in the job.py?

@Nicolas Luchetti yes, include it in the job.py. The dbt-run-job does run dbt-test when all the test pass, but when I set the test fail (on purpose), it skips the dbt-test and the jobs still mark as success. let me investigate more with the

use_build_command

Thank @Julius! If I manage to solve it I will let you know too.

thanks a lot @Nicolas Luchetti!