Qdrant

1. Overview

Qdrant is a high-performance vector similarity search engine and vector database, specifically designed for handling high-dimensional vector data. It provides powerful vector search capabilities for large-scale machine learning applications, recommendation systems, semantic search, and similarity matching scenarios.

Through GoInsight's Qdrant node, you can fully integrate vector database operations into your automated workflows. You can achieve full lifecycle management of your vector data, including:

• Collection Management: Create, retrieve, update, delete, and list Qdrant vector collections.
• Point Operations: Upsert, query, update, delete, and retrieve vector points and their payload data.
• Advanced Search: Perform complex vector similarity searches, batch queries, and grouped searches.
• Index Management: Create and delete payload indices to optimize query performance.
• Payload Operations: Set, overwrite, delete, and clear the payload data of points.

2. Prerequisites

Before using this node, you need to have an accessible Qdrant server instance. You may need appropriate permissions to create and manage collections and their data.

3. Credentials

For a detailed guide on how to obtain and configure your credentials, please refer to our official documentation: Credential Configuration Guide.

4. Supported Operations

This node primarily operates on core resources such as Collections, Points, Payloads, Indices, and Vectors, and supports various search and analysis operations.

Summary

Operation Details

Create Collection

Creates a new vector collection in Qdrant with comprehensive configuration, including vector settings, sharding, replication, and indexing options.

Options:

• CollectionName: The name of the collection to be created.
• VectorsConfig: JSON-formatted vector configuration, including size and distance metric.
• ShardNumber: The number of shards for the collection.
• ReplicationFactor: The replication factor for the collection.
• WriteConsistencyFactor: The write consistency factor for the collection.
• OnDiskPayload: Whether to store the payload on disk.
• HnswConfig: HNSW index configuration in JSON format (optional).
• QuantizationConfig: Quantization configuration in JSON format (optional).
• Timeout: Operation timeout in seconds (optional).

Output:

• Result (bool): Indicates if the collection creation was successful.
• Status (string): Status message from the Qdrant API.
• Time (number): Execution time in seconds.
• StatusCode (number): HTTP status code.
• ErrorMessage (string): Error description if the operation fails.

Get Collection

Retrieves detailed information for a specific collection, including its configuration, status, and statistics.

Input Parameters:

• CollectionName: The name of the collection to retrieve.

Output:

• Result (bool): Indicates if the retrieval was successful.
• CollectionInfo (object): Detailed collection information object.
• VectorCount (number): The total number of vectors in the collection.
• PointsCount (number): The total number of points in the collection.
• StatusCode (number): HTTP status code.
• ErrorMessage (string): Error description if the operation fails.

Upsert Points

Inserts or updates vector points in a collection. If a point with the same ID already exists, it will be updated; otherwise, a new point will be created.

Input Parameters:

• CollectionName: The name of the target collection.
• Points: A JSON array of point objects. Each object must contain an id, vector, and an optional payload.

Options:

• Wait: If true, the operation will wait for completion before returning.
• Ordering: Write consistency level (weak, medium, or strong).

Output:

• Result (bool): Indicates if the upsert operation was successful.
• Status (string): Status message from the Qdrant API.
• PointsCount (number): The number of points successfully upserted.
• StatusCode (number): HTTP status code.
• ErrorMessage (string): Error description if the operation fails.

Find Points

Performs a complex vector similarity search in a collection, supporting various query types like nearest neighbor, recommendation, and discovery.

Input Parameters:

• CollectionName: The name of the target collection.
• Query: The query configuration in JSON format, which specifies the search vector and type.

Options:

• FilterCondition: A filter to apply during the search, in JSON format.
• Limit: The maximum number of results to return.
• WithPayload: If true, the payload of the resulting points will be included.
• WithVectors: If true, the vectors of the resulting points will be included.
• ScoreThreshold: A similarity score threshold to filter results.

Output:

• Result (bool): Indicates if the search operation was successful.
• Results (object-array): An array of search results, each containing the point's ID, score, and optional payload/vector.
• ResultsCount (number): The number of results returned.
• StatusCode (number): HTTP status code.
• ErrorMessage (string): Error description if the operation fails.

5. Example Usage

This section guides you through creating a simple workflow to create a new vector collection in your Qdrant database.

The workflow will look like this: Start -> Create Collection -> Answer.

Step-by-Step Guide

1. Add the Tool Node:
   • In the workflow canvas, click the + icon to add a new node.
   • Select the "Tool" tab in the pop-up panel.
   • Find and select Qdrant from the tool list.
   • In the list of supported operations for Qdrant, click Create Collection to add the node to the canvas.
2. Configure the Node:
   • Click the Create Collection node to open its configuration panel.
   • Credentials: Select your pre-configured Qdrant credential from the dropdown.
   • Parameters:
   • CollectionName: Enter a name for your collection, e.g., "my_text_embeddings".
   • VectorsConfig: Provide the vector configuration as a JSON string. For example, {"size": 384, "distance": "Cosine"} for 384-dimensional vectors using Cosine distance.
3. Run and Validate:
   • Once the required parameters are filled, click the "Run" button at the top-right of the canvas.
   • After execution, check the logs to see the output. A successful run will show Result: true and the details of the created collection.

Final Workflow

After completing these steps, your workflow is configured. Running it will create a new vector collection in your Qdrant database, ready for you to upsert points.

6. FAQs

Q: I received a 401 or 403 error. What's the cause?

A: This usually indicates an authentication or permission issue. Please check the following:

• Ensure your Qdrant credentials (e.g., API key) are correct and not expired.
• Verify that your API key has the necessary permissions for the operation.
• Check if your Qdrant server is accessible and firewall settings are correct.

Q: Why did my Create Collection operation fail with a parameter error?

A: Please check the following:

• Ensure VectorsConfig is a valid JSON string with size and distance fields.
• Verify that size is a positive integer.
• Confirm that distance is a valid value (e.g., "Cosine", "Euclid", "Dot").

Q: How should I format JSON configuration parameters?

A: Here are some examples:

• VectorsConfig: {"size": 768, "distance": "Cosine"}
• HnswConfig: {"m": 16, "ef_construct": 100}
• Points Data: [{"id": 1, "vector": [0.1, 0.2, 0.3], "payload": {"text": "sample text"}}]
• Filter Condition: {"must": [{"key": "category", "match": {"value": "document"}}]}

7. Official Documentation

For more detailed information, please refer to the Qdrant Official Documentation.