hello! how can I access in a flow another flows’ metadata? let’s say I have a flow AFlow and BFlow, I need to have a step in BFlow which calls Flow(BFlow).latest_run.data object. I run flow in AWS sagemaker and save results in S3. It in BFlow returns Object not found error. Configuration file has METAFLOW_METASTORE_SYSROOT_S3 correctly specified

it may not be in the right namespace

namespace is the same, although I’ve tried to put namespace(None) in front of r = Run(AFlow/run_id) but getting the same error

could you paste the whole code (or the relevant section) and the full traceback.

Class Inference(FlowSpec): …… @step def load_last_model(self): “”" Loading last model from training flow “”" if self.is_prod: from metaflow import Run, namespace namespace(None) r = Run(self.run_id_actual_dict[‘IDCase’][‘training’]) self.model = r.data.lgbm_model self.last_model_file_name = r.data.model_file_name else: ……. Traceback:

Task is starting.
Bootstrapping virtual environment...
Environment bootstrapped
<flow DPITForecastInferenceIDCase step load_last_model> failed:
AWS Batch error:
Essential container in task exited This could be a transient error. Use @retry to retry.
Object not found:
Run('DPITForecastTrainIDCase/1704964756701873') does not exist

ok, based on the ID, that run ran without a metadata service (it was a “local” run) so you won’t be able to reference them.

it both flows I’m using the same configuration file and first flow was actually run on batch, as well as the second one. I can see metadata in S3 as well. You think this is still the case?

when you say you “see the metadata in S3”, what do you see in S3? S3 is used to store some things but there is also a metadata service. S3 is used during execution but the service is used for the client.

i mean I see directory that metaflow has created

hey sorry — went to bed

yes, you’re right. I have installed metaflow with simple pip install and didn’t configure metaflow service. Let me check the article and rebuild everything. Thank you so much for your hint!