I'm having the following problem: I'm using the

run_request_for_partition

function. It does return the error from the image attached. I did set the

name

attribute in my asset job. I cannot use

RunRequest

and pass

job_name

since it does not allow partitions. Similar code:

my_job = define_asset_job(
    name="my_job",
    selection=AssetSelection.groups("my_groups"),
    partitions_def=my_partitions,
)


@sensor(
    jobs=[my_job],
    default_status=DefaultSensorStatus.RUNNING,
    minimum_interval_seconds=30,
)
def my_sensor():

    ...
    for key in partitions:
        request =  my_job.run_request_for_partition(
            partition_key=key, run_key=key
        )

        yield request

Can you try using the parameter

job=my_job

instead in your sensor decorator ? Maybe it fails because it is expecting more than one job given you used the

jobs

parameter.

Indeed, I have two jobs (just tried to simplify the example above). This is more realistic:

my_job = define_asset_job(
    name="my_job",
    selection=AssetSelection.groups("my_groups"),
    partitions_def=my_partitions,
)

my_second_job = define_asset_job(
    name="my_second_job",
    selection=AssetSelection.groups("my_second_groups"),
    partitions_def=my_second_partitions,
)


@sensor(
    jobs=[my_job, my_second_job],
    default_status=DefaultSensorStatus.RUNNING,
    minimum_interval_seconds=30,
)
def my_sensor():

    ...
    for key in partitions:
        request =  my_job.run_request_for_partition(
            partition_key=key, run_key=key
        )

        yield request

    for key in second_partitions:
        second_request =  my_second_job.run_request_for_partition(
            partition_key=key, run_key=key
        )

        yield second_request

Ok, looked it up a bit, it seems all examples with multi-job sensors use

RunRequest

so my guess is that multi-job sensors is not yet compatible with

run_request_for_partition

. Looks like an oversight considering the

jobs

parameter is marked as experimental and the use case is not covered in the tests: https://github.com/dagster-io/dagster/blob/1.1.10/python_modules/dagster/dagster_tests/daemon_sensor_tests/test_sensor_run.py

Got it, thanks @Nicolas Parot Alvarez

I think it deserves a Github issue

Done: https://github.com/dagster-io/dagster/issues/11766

Light speed fix!