19913d85-f662-4518-a9a1-7adc1793dfd4.png){kind=small}
Channels
#ask-community
Title
# ask-community
v
Vinnie
08/23/2022, 2:15 PMI have a job composed of two “ops”: a
multi_assetdefine_asset_job(selection=AssetSelection.keys().downstream())multi_assetSkipping step <dbt_op> due to skipped dependencies: [].For more context. here’s a minimal implementation of my
multi_asset Copy code
@multi_asset(
outs={
"asset1": AssetOut(
key_prefix="some_prefix",
is_required=False,
),
"asset2": AssetOut(
key_prefix="some_prefix",
is_required=False,
),
},
can_subset=True,
group_name="my_cool_group",
)
def my_assets(context):
elems = fetch()
for elem in elems:
if elem["dest_asset"] in context.selected_output_names:
processed_elem = process(elem)
yield Output(value=processed_elem["data"], output_name=elem["dest_asset"])c
chris
08/24/2022, 4:59 PMHow exactly are you declaring the dependency between the multi asset and the dbt models? It looks like there's some conditional branching happening here, but the fact that the dependency list is empty is concerning
v
Vinnie
08/24/2022, 5:34 PMHi Chris, I’ll PM you some more info, can’t talk much more publicly
c
chris
08/24/2022, 7:18 PMOkay so this is a known issue https://github.com/dagster-io/dagster/issues/9408 and represents a feature gap in dagster unfortunately. Basically, assets require all of their inputs to be resolved in order to run, and causes the dbt asset (which contains multiple models) to not run if only some inputs are provided.
As a workaround - I think what you can do is split up your call of
load_assets_from_dbt_projectLet me know if that makes sense, or if you need guidance on doing that.
v
Vinnie
08/24/2022, 7:23 PMShould be easy enough, I’ll try tomorrow and ping the thread :)
c
chris
08/24/2022, 7:30 PMSounds good!
72 Views