This document describes how to resolve errors that you might encounter when creating an agent.
Content generation errors
Issue:
You receive an error message similar to the following:
ValueError: Cannot get the Candidate text.
Response candidate content part has no text.
Possible cause:
This error might be caused by using a version of langchain-google-vertexai
that's not compatible with google-cloud-aiplatform. Version 1.0.2 or later
of langchain-google-vertexai is required. To check which version you're using,
run the following command in your terminal:
pip show langchain-google-vertexaiRecommended solution:
Install version 1.0.2 of langchain-google-vertexai. This version includes
the LangChain tool-calling
updates that are required to work with google-cloud-aiplatform. To update your
version of langchain-google-vertexai, run the following command in your
terminal:
pip install langchain-google-vertexai --upgradeAfter running the update command, verify that you're using version 1.0.2 or
later by running the following command in your terminal:
pip show langchain-google-vertexaiIf you're in a notebook instance (for example, Jupyter or Colab or Workbench), you might need to restart your runtime to use the updated packages.
Bring your own container (BYOC) first-time creation error
Issue:
You deploy agents through BYOC in your Cloud de Confiance by S3NS project for the first time in the region you are in, and you receive an error message similar to the following:
Reasoning Engine resource <reasoning-engine-id> failed to start and cannot serve traffic.
Possible cause:
The failure was likely due to Agent Runtime on Gemini Enterprise Agent Platform internal IAM provisioning. The changes take time to complete and propagate.
Recommended solution:
Retry the same agent deployment request, and it will succeed.