I have a selfhosted qdrant running on a cloud, it is properly deployed with API key, but n8n cannot access any collections. Here is the strange part:
When I provide the credentials (API-Key and URL) the connection is always tested successfully, no matter what API key is provided, even the correct one.
When I use a qdrant vector store through the ‘official’ qdrant page, it works.
Anyone experience the same issues?
n8n version: 1.68.0
Database: default
EXECUTIONS_PROCESS default
Running n8n via Docker (compose)
Operating system: Linux
Are you running both in containers and are you trying to use 127.0.0.1 or localhost as the address for qdrant? If you are can you try 172.17.0.1 instead or the ip of the machine running qdrant
both running on my Hetzner Cloud, accessible through a URL. Both run in a docker container managed by Coolify. If I use tools like langflow or flowise, this problem does not occur, only in n8n.
Having the same issue on n8n on coolify. qdrant node goes green to say connectin successful then when i try, it wont work. Says unable to fetch. Log files on the containers show connections rejected, I dont know why
Hey there,
seems that nobody has a solution for this… I tried everything, even hardcode the API keys within the files in the container, no luck. Meanwhile I switched to Supabase, since QDrant is useless in n8n.
I have qdrant running behind a proxy, can call the UI and N8N connection says “successful”, but on insert i get “fetch failed”.
If i try the same in qdrant cloud (just different credentials and url) it works out of the box.
Both running on Docker next to each other. The API Key settings gives successful message all the time but the Vector Store can’t connect. The logs show fetch failed only.
I tried using both service address with plain HTTP (such as http://n8n-qdrant/) or HTTPS with a reverse proxy (such as https://qdrant.example.com/) but it doesn’t help. No matter I do, the node fails to connect. Regular request execution works. Also, the Qdrant dashboard works. All with the same API key. I don’t understand what’s the problem.
I don’t understand, you expose 6334, which (as far as I experienced) is always exposed if you deploy it, and then you use http://xxxxxxxx:6333 in the Credentials in n8n?
Unfortunately this doesnt work for me. Anything else you set up? How does the docker compose file look like? How (or on which platform (Coolify/Dokploy etc.)) did you host it?
This is exactly what I did. I exposed the gRPC port on the container and on the service, and it started working with http://n8n-qdrant:6333/ as the URL along with the API key. Previously, only the HTTP port was exposed, and I mapped it to port 80. Keeping it on 6333 and exposing 6334 solved it for me.
I’m on mobile now, so forgive me the screenshots form the code.