I am using OpenAI ada vectorizer for embeddings. The application I am working on sends requests to our hosted Weaviate instance through the python client (v3.11.0). We keep seeing a RequestsConnectionError Query not executed error in our logs - not for every request, but quite a few.
There is also a point in our code where we send requests asynchronously, hacked with
What are you getting back on the client side when that error happens?
Also your python client is very old (3.20 is current), don’t think anything significantly changed for the raw queries, but might still be worth to update.
Hi! Yes, the client is instantiated as a global singleton across the app. So there’s only one instance of weaviate.Client that gets used for all querying in the execution environment. Do you think that’s causing an issue?
yep! this is the code for instantiation: weaviate.Client( host, auth_client_secret=weaviate.AuthClientPassword(username=username, password=password), additional_headers={"X-OpenAI-Api-Key": get_openai_key()}, )
and then we have a class that uses this client and given a query calls this function: