Postgres Vector Store fails with Supabase - Error 500 ENETUNREACH #5180
Description
Describe the bug
Hi, there,
I am unable to upload documents to Flowise Document Store when using Postgres vector store with Supabase as the database provider. Connection consistently fails with ENETUNREACH error despite correct credentials and table structure.
To Reproduce
Create new Supabase project with pgvector extension enabled
Create table with all required Flowise columns:
id, content, metadata, embedding, key, namespace, group_id, updated_at, created_at
Including unique constraint on (key, namespace)
In Flowise, create new Document Store with Postgres vector store
Configure with Supabase credentials (host, port, database, username, password)
Attempt to upload any document (.md, .txt, etc.)
ACTUAL BEHAVIOUR:
Upload fails with error:
Status: 500
Error: documentStoreServices.insertIntoVectorStoreMiddleware - Error:
documentStoreServices.insertIntoVectorStore - Error:
documentStoreServices._insertIntoVectorStoreWorkerThread - Error:
connect ENETUNREACH [IPv6 address]:6543 - Local (:::0)
What I've Tried
Using both Direct Connection and Pooler connections
Enabling IPv4 compatibility in Supabase
Using Shared Pooler (IPv4 compatible)
Creating fresh Supabase project
Disabling Row Level Security
Testing with simple, minimal content documents
Using different table names and structures
Additional Context
Other Flowise features work normally
Same documents upload fine to other vector stores, and were uploading well up to couple of weeks ago.
Supabase dashboard is accessible and SQL queries work
Issue appears to be IPv6/IPv4 network connectivity between Flowise and Supabase
Impact
Cannot use Supabase as vector store backend for document storage in Flowise, blocking development of agents that require document retrieval.
Expected behavior
Documents should upload successfully and be stored in the Supabase Postgres table.
Screenshots
No response
Flow
No response
Use Method
None
Flowise Version
version 3
Operating System
macOS
Browser
Chrome
Additional context
I use Flowise over Railway