qdrant and rest api with 0.5.2?

[LStromann]

=Hey team,

I'm trying to go live with user management enabled in a production environment (postgre + neo4j + qdrant + gemini) with the REST api.

At the moment, I am receiving the following message: "Connection failed: Unsupported vector database provider: qdrant. Supported providers are: LanceDB, PGVector, neptune_analytics, ChromaDB"

Is it possible to use qdrant with the REST API?

I can't find a clear answer in the documentation ...

This discussion was automatically pulled from Discord.

[LStromann]

QDrant is part of our community repo, here is how you can add QDrant support to Cognee:

https://github.com/topoteretes/cognee-community/tree/main/packages/vector/qdrant

Check the README.md there along with the examples

[LStromann]

To enable Neo4j with Cognee in a multi user setting you can check these docs:

https://docs.cognee.ai/core-concepts/multi-user-mode/dataset-database-handlers/existing-dataset-database-handlers/neo4j-aura-dev#neo4j-aura-dataset-database-handler

As currently it is only officially supported through Neo4js Aura service

[LStromann]

Thank you for your reply, but my question is more focused on the REST API of Cognee (the HTTP server embedded in Cognee) and how to make it use Qdrant.

[LStromann]

It should be the same, making this import in code:
# NOTE: Importing the register module we let cognee know it can use the Qdrant vector adapter from cognee_community_vector_adapter_qdrant import register

Before starting the Cognee server so it will be aware of QDrant adapter

[LStromann]

Okay, so that means we would need to patch the Cognee REST server to add this line of code? Is there no provision for this in the configuration?

[LStromann]

Not currently, you would also need to add the dependency to pyproject.toml

you don't have to patch the Cognee REST server you can import it in a new project/script and run something like:

from cognee_community_vector_adapter_qdrant import register
from cognee.api.client import start_api_server
import os

if name == "main":
start_api_server(host="0.0.0.0", port=8000)

[LStromann]

We also support PGVector which is the vector store in PostgreSQL on the Cognee core side, for that one you could just run it with Cognee out of the box

[LStromann]

<@508682252595626015> we should document this better!