Vertex AI: model deployment to endpoint

Google Cloud Serverless Applications
1

I try to deploy a Vertex AI model using the following code:

from google.cloud import aiplatform # google-cloud-aiplatform 1.60.0

aiplatform.init(project=PROJECT_ID, location=LOCATION)

model = aiplatform.Model(model_name=f"projects/{PROJECT_ID}/locations/{LOCATION}/models/{MODEL_ID}")\endpoint_path = client.endpoint_path(project=PROJECT_ID, location=location, endpoint=endpoint_id)
model_endpoint = model.deploy()

But I get the following error:

File /opt/conda/envs/python311/lib/python3.11/site-packages/google/cloud/aiplatform/models.py:4876, in Model.deploy(self, endpoint, deployed_model_display_name, traffic_percentage, traffic_split, machine_type, min_replica_count, max_replica_count, accelerator_type, accelerator_count, tpu_topology, service_account, explanation_metadata, explanation_parameters, metadata, encryption_spec_key_name, network, sync, deploy_request_timeout, autoscaling_target_cpu_utilization, autoscaling_target_accelerator_duty_cycle, enable_access_logging, disable_container_logging, private_service_connect_config, deployment_resource_pool) 4865 raise ValueError( 4866 "Traffic splitting is not yet supported for PSA based PrivateEndpoint. " 4867 "Try calling deploy() without providing traffic_split. " 4868 “A maximum of one model can be deployed to each private Endpoint.” 4869 ) 4871 explanation_spec = _explanation_utils.create_and_validate_explanation_spec( 4872 explanation_metadata=explanation_metadata, 4873 explanation_parameters=explanation_parameters, 4874 )

2

Hi @george-kay ,

Welcome to Google Cloud Community!

Based on the error message, it is mentioned that traffic splitting is not yet supported for PSA based PrivateEndpoint. With this, please try to deploy without calling the traffic_split method and see if it works. You may also try other models to isolate if this is a model issue.

I hope this helps.