Qdrant (Advanced)

Full Qdrant API — collections & points, with JSON expressions everywhere

Overview

This node provides advanced integration with the Qdrant vector search engine, allowing users to perform a variety of operations on Qdrant collections and points. The "Search Points" operation enables searching for vectors similar to a given embedding within a specified collection, optionally filtered by metadata.

Common scenarios include:

  • Finding nearest neighbors or similar items in a vector database based on an embedding vector.
  • Filtering search results by specific criteria using Qdrant's filtering capabilities.
  • Limiting the number of returned results to optimize performance or focus on top matches.

Practical example:

  • A user has a collection of product embeddings and wants to find the top 10 products most similar to a customer's query embedding, possibly filtering out discontinued products.

Properties

Name Meaning
Collection Name The name of the Qdrant collection to operate on. Supports expressions for dynamic values.
Search Vector (JSON Array) The embedding vector used to search against the collection. If left blank, the node uses incoming data as the vector. Must be a JSON array of numbers.
Filter (JSON) A Qdrant filter object to narrow down search results based on payload conditions. Supports expressions.
Limit Maximum number of search results to return. Must be at least 1. Defaults to 50.

Output

The output is a JSON array where each item contains the search results from Qdrant. The structure includes:

  • json: An object representing the response from Qdrant's search API, typically containing:
    • Matched points with their IDs.
    • Distances or similarity scores.
    • Payload data associated with each point.

No binary data output is produced by this operation.

Dependencies

  • Requires a Qdrant API endpoint URL and an API key credential configured in n8n.
  • Uses the official Qdrant JavaScript client library (@qdrant/js-client-rest) bundled within the node.
  • The node expects valid JSON input for vector and filter parameters, supporting expressions for dynamic content.

Troubleshooting

  • Missing Collection Name: The node will throw an error if the collection name is not provided. Ensure this property is set.
  • Invalid Search Vector: If the search vector is empty or improperly formatted, an error occurs. Provide a valid JSON array of numbers or ensure incoming data is suitable.
  • Limit Less Than 1: The limit must be at least 1; otherwise, an error is thrown.
  • Filter Parsing Errors: Filters must be valid JSON objects. Malformed filters cause errors.
  • API Connection Issues: Verify that the Qdrant API URL and API key are correctly configured and that the Qdrant service is reachable.
  • Unsupported Operation: Using an operation other than those supported by the node will result in an error.

Links and References

Discussion