I'm getting this error when running a Langchain+Neo4j python api in an AWS lambda:
AttributeError: module 'neo4j' has no attribute 'GraphDatabase'
I've adapted the pattern for langchain and Neo4j vector embeddings app from a Graph Acadmey course and then split off the front end in a React js app, accessing the python components via an api (using Fastapi, and Mangum for AWS).
I can access the api fine on my local machine and return search results. I can run it in a docker container on my local machine. However when I run it in an AWS on the arm64 architecture, I get the above error. The logs show that the lambda function is running.
(I am using the arm64 architecture successfully to get round a dependency problem on the x86 architecture which gave me an error: No module named 'pydantic_core._pydantic_core'.)
(I have tried to deploy a docker container to google cloud run, I can run the docker container but not access it!)
Has anyone else seen this "module 'neo4j' has no attribute 'GraphDatabase'" error when running via python on AWS? (Importing the neo4j-driver.)
Thanks
Here is the error log, this appears to be happening in a langchain module:
neo4jvector = Neo4jVector.from_existing_index(
File "/opt/python/langchain_community/vectorstores/neo4j_vector.py", line 758, in from_existing_index
store = cls(
File "/opt/python/langchain_community/vectorstores/neo4j_vector.py", line 195, in init
self._driver = neo4j.GraphDatabase.driver(url, auth=(username, password))INIT_REPORT Init Duration: 6119.03 ms Phase: init Status: error Error Type: Runtime.ExitError