Wonjae Lee
04/11/2023, 6:56 PMclaire
04/11/2023, 10:41 PMload_input
/ handle_output
methods you could call context.asset_key
to determine which path you'd like the asset to write to.claire
04/11/2023, 10:48 PMWonjae Lee
04/12/2023, 12:32 AMWonjae Lee
04/12/2023, 12:58 AMWonjae Lee
04/12/2023, 2:40 AMclaire
04/12/2023, 5:13 PM.hardcoded_io_manager
function. I'm assuming the upstream asset is dim_company_events_active
with partition taging|2023-04-01
? Can you double check that that partition was materialized using the fs_io_manager
?
If so, I'm wondering if either of these files also exist for the upstream asset:
• taging/2023-04-01
• 2023-04-01/taging
We recently changed the path that multipartitioned asset partitions were stored in the fs_io_manager
, so wondering if that could also be interferingclaire
04/12/2023, 5:14 PMfs_io_manager
Wonjae Lee
04/12/2023, 6:11 PMWonjae Lee
04/12/2023, 7:06 PMclaire
04/12/2023, 9:21 PMfs_io_manager
, and your downstream asset loads the upstream asset using the fs_io_manager
.
I think if you want to load the upstream asset from bigquery instead, you'll need to swap out the IO manager for a bigquery IO manager. I think you should be able to customize the bigquery IO manager and provide the desired config following the docs here: https://docs.dagster.io/integrations/bigquery/reference#using-pandas-and-pyspark-dataframes-with-bigqueryWonjae Lee
04/12/2023, 9:56 PM@asset(
ins={"table": AssetIn(input_manager_key="bq_custom_manager")},
io_manager_key='different_io_manager')
will it load data through ‘bq_custom_manager’ and store output to ‘different_io_manager’??claire
04/12/2023, 10:08 PMWonjae Lee
04/13/2023, 2:41 AM