<!-Hello All,
I am having a persistent issue with the Postgres node when trying to perform a semantic search with pgvector. The node fails with a TypeError: properties.string.split is not a function.
My Goal:
I want to run a SELECT query to find the cosine similarity between a stored vector and a new vector passed as a parameter.
My Setup:
- n8n Version: 1.113.3 (Self-hosted on Railway)
- Database: PostgreSQL with pgvector extension
- Embedding Model: OpenAI
text-embedding-3-small(1536 dimensions)
Node Configuration: - Node:
Postgres - Query:
SELECT summary, 1 - (embedding_vector <=> $1::vector) AS similarity FROM learned_facts ORDER BY similarity DESC LIMIT 1; - Query Parameters (Expression):
{{ [ $items("N2.2-Generar Embedding")[0].json.data[0].embedding ] }}
The Problem:
The execution fails on this Postgres node with the TypeError: properties.string.split is not a function. This happens on a clean, updated n8n instance. We have confirmed the database schema is correct (embedding_vector VECTOR(1536)), the extension is active, and the input data is a valid array of numbers.
It seems the Postgres node has a bug and cannot handle an array of numbers as a parameter for a vector type.
Is there a known workaround for this bug, or is this a confirmed issue?
Thank you.- Hey! The fastest way to find solutions is by using the 
search function at the upper right.
If your question hasn’t been asked before, please follow the template below. Skip the questions that are not relevant to you. →
Describe the problem/error/question
What is the error message (if any)?
Please share your workflow
(Select the nodes on your canvas and use the keyboard shortcuts CMD+C/CTRL+C and CMD+V/CTRL+V to copy and paste the workflow.)
Share the output returned by the last node
Information on your n8n setup
- n8n version:
- Database (default: SQLite):
- n8n EXECUTIONS_PROCESS setting (default: own, main):
- Running n8n via (Docker, npm, n8n cloud, desktop app):
- Operating system: