GoogleEncoder auth unclear

[FilippTrigub]

This may be a dumb question, but I struggle to get the GoogleEncoder to proprly authenticate.

I have tried with

• an API_KEY in teh env with th right permissions for Vertex AI and Compute Engine
• a service account
• initializing the aiplatform with eithre one bfore calling the GooglEncoder

Neverthless I get the constant error message:
2024-09-18 15:14:08 GMT - WARNING - Compute Engine Metadata server unavailable on attempt 1 of 3. Reason: [Errno 111] Connection refused
2024-09-18 15:14:09 GMT - WARNING - Compute Engine Metadata server unavailable on attempt 2 of 3. Reason: [Errno 111] Connection refused
2024-09-18 15:14:11 GMT - WARNING - Compute Engine Metadata server unavailable on attempt 3 of 3. Reason: [Errno 111] Connection refused
2024-09-18 15:14:11 GMT - INFO - PROCESSING: Error processing file dummy2.pdf: Google AI Platform client failed to initialize. Error:
Unable to authenticate your request.
Depending on your runtime environment, you can complete authentication by:

• if in local JupyterLab instance: !gcloud auth login
• if in Colab:
  -from google.colab import auth
  -auth.authenticate_user()
• if in service account or other: please follow guidance in https://cloud.google.com/docs/authentication

Would it not be simpler to give the GoogleEncoder an api_key (just as the OpenAIEncoder)?

[FilippTrigub]

Resolved. For clarity:
I am using a service account, which will not be stored in a json directly for CICD reasons. To properly auth, I had to give the SA the aiplatform user role, retrieve current credentials from that SA and initiate aiplatform with these credentials before initing the encoder.