search

neo4j-labs / llm-graph-builder

Neo4j graph construction from unstructured data using LLMs
https://neo4j.com/labs/genai-ecosystem/llm-graph-builder/
Apache License 2.0
1.34k stars 184 forks source link

Authentication Error with Google Compute Engine #295

Closed cip22 closed 1 week ago

cip22 commented 2 months ago

When running the backend, uvicorn starts and the reloader process starts. Then: 2024-05-08 11:54:45,375 - Compute Engine Metadata server unavailable on attempt 1 of 3. Reason: timed out 2024-05-08 11:54:48,378 - Compute Engine Metadata server unavailable on attempt 2 of 3. Reason: timed out 2024-05-08 11:54:48,379 - Compute Engine Metadata server unavailable on attempt 3 of 3. Reason: [Errno 64] Host is down 2024-05-08 11:54:48,379 - Authentication failed using Compute Engine authentication due to unavailable metadata server.

Where is google Compute Engine used in this project?

ais-qianhao-liu commented 2 months ago

Its used here add_routes(app,ChatVertexAI(), path="/vertexai")

msenechal commented 2 months ago

Hi @cip22 Yes, in our app we do have the support for Gemini/VertexAI which is why you would need to setup GC if you want to run locally. If you don't want the Gemini integration, you can simply comment the line @ais-qianhao-liu posted above (which is in score.py ) and that should remove the error

wjiayis commented 2 months ago

If in your score.py there is

logging_client = gclogger.Client()
logger_name = "llm_experiments_metrics" # Saved in the google cloud logs
logger = logging_client.logger(logger_name)

You'll need to comment that out too.

jexp commented 2 months ago

It should be good enough to not configure Gemini as LLM, in the config, or @kartikpersistent

@kartikpersistent can we make the metrics logging also configurable via .env as it should be running for local or other deployments.