Getting 400 Bad Request when using Qdrant with N8N
I’m trying to set up a workflow that connects N8N with Qdrant vector database but keep running into this error:
Problem in node 'Qdrant Vector Store'
400 Bad Request: Wrong input: Not existing vector name error
I followed the basic setup from the official docs but something seems wrong with my collection configuration. The workflow includes these components:
- Manual trigger node
- OpenAI embeddings (using text-embedding-3-small model)
- GitHub document loader pointing to a demo repository
- Text splitter for processing documents
- Qdrant vector store node in insert mode
The main issues I’m facing:
- How do I properly create a Qdrant collection that works with N8N?
- What collection settings are needed for compatibility?
- Are there specific vector configurations required?
My current collection is named ‘documents2’ and I’ve set up the API credentials correctly. The error happens when the workflow tries to insert vectors into the collection.
Has anyone successfully integrated these two tools? What collection setup worked for you?
The vector name error occurs when the configuration of your collection does not align with what N8N anticipates. I faced a similar issue where the collection lacked the necessary vector field setup. It’s crucial to specify the vector field name explicitly when constructing the collection, rather than just mentioning the dimensions. Consider deleting your current collection entirely and recreating it via the Qdrant API, ensuring that you provide a specific vector field naming setup that aligns with N8N’s requirements. Additionally, make sure your Qdrant version is compatible with N8N’s vector store implementation, as some of the older versions have issues with vector field naming.
This error happens when your Qdrant collection setup doesn’t match what N8N expects. I’ve hit this before - you need to name your vectors properly when creating the collection. Instead of just setting the size directly, use a config like {“default”: {“size”: 1536, “distance”: “Cosine”}}. That “default” name is key. Also double-check your collection actually exists first. I’ve seen the API credentials work fine for other stuff even when collection creation failed silently. Recreate your collection with explicit vector naming and you should be good to go.
i had the same issue too, make sure your vector size matches the openai model’s size - it’s 1536 dimensions. also, double check that your collection is created in qdrant cuz n8n doesn’t create em automatically.