[Bug]: Deploying Langgraph service in GCP VertexAI Reasoning Engine give this error InternalServerError: 500

[puranjay123]

File Name

https://colab.research.google.com/github/GoogleCloudPlatform/generative-ai/blob/main/gemini/reasoning-engine/tutorial_langgraph.ipynb#scrollTo=dced52d8c524

What happened?

Similar to this issue
I am encountering this issue while following the [tutorial_langgraph]
I have tried removing all locks and testing with them enabled, but the issue persists. The logs consistently show the following error:

Invalid filter resource.type=
InternalServerError: 500
Additionally, it gave me this error

InternalServerError: 500 
 Please navigate to the Cloud Console Log Explorer page (https://console.cloud.google.com/logs/query;query=resource.type="aiplatform.googleapis.com%2FReasoningEngine"%0Aresource.labels.reasoning_engine_id=~"xxxxxxx"?project=xxxxx

Below is the code block that seems to be causing the issue:

remote_agent = reasoning_engines.ReasoningEngine.create(
    SimpleLangGraphApp(project=PROJECT_ID, location=LOCATION),
    requirements=[
        "google-cloud-aiplatform[langchain,reasoningengine]",
        "cloudpickle==3.0.0",
        "pydantic>=2.10",
        "langgraph",
        "httpx",
    ],
    display_name="Reasoning Engine with LangGraph",
    description="This is a sample custom application in Reasoning Engine that uses LangGraph",
    extra_packages=[],
)

I believe its the issue with dependencies I have tried all the possible combinations but still getting the same issue
Has anyone else encountered this issue or found a workaround? Any insights would be greatly appreciated!

Code of Conduct

• I agree to follow this project's Code of Conduct

[koverholt]

Thanks for the report. I ran the notebook just now and can confirm that it's working as expected (see below screenshot).

• Did you make changes to the agent class or add new imports for dependencies that need to be installed?
• Are you running in Colab or a different environment?
• Which version of Python are you using?
• Can you print the library versions that you are using? Specifically the ones at the top of the notebook that get installed by pip.
• Is there any additional information in the Console logs linked from the error message?

[puranjay123]

Hey @koverholt

"Thank you for your quick response! I was following the default notebook from the GCP documentation and the official [Google Cloud Platform GitHub]. It turns out I don’t have access to this API as it’s part of the pre-GA offerings."

Closing this issue:) Thanks for helping

[koverholt]

Thanks for the update! Reasoning Engine is pre-GA but is in Public Preview stage, so you should be able to use it if you have the Vertex AI APIs enabled in your Google Cloud account without the need to do additional allowlisting. If you post the full text of the 500 error from the logs in the Console, we can take a deeper look and see why you are getting the 500 error.