You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Title: Flowise Freezing with RAG Setup Using Pinecone & SQLite Record Manager
Hi everyone,
I'm facing an issue where Flowise Cloud freezes and turns white whenever I query my RAG agent. I've checked multiple possible causes, but I still can't identify the root problem.
My Flow Setup
Conversational Q&A Retriever → OpenAI GPT-4 → Document Store (Pinecone)
Document Loader: PDF file with Recursive Character Text Splitter
Chunking: 1000 characters per chunk with 200 overlap (~180 chunks)
Embeddings: OpenAI text-embedding-3-small (1536 dimensions)
Vector Store: Pinecone (cosine similarity, april2025 index, Top K = 4)
Record Manager: SQLite
Issue
Flowise freezes when I test the agent, and the console logs show:
🚨 TypeError: Cannot read properties of null (reading '4')
Additionally, I previously had Redis configured, but after facing connection issues, I switched to Pinecone and SQLite Record Manager. The issue persists even after upserting documents successfully.
Troubleshooting Steps Tried
✅ Recreated the vector store and re-upserted documents successfully
✅ Verified SQLite is properly configured (no permission errors)
✅ Cleared browser cache, tried incognito mode, and different browsers
✅ Checked CPU/memory usage during query (no major spikes)
✅ Tested with smaller document size and different chunking settings
Possible Causes & Questions
Could the SQLite Record Manager be causing the issue? Should I try without it?
Could Pinecone retrieval be causing memory overload?
Any way to debug why the UI crashes (logs, error traces, etc.)?
I'd appreciate any insights or suggestions! If anyone has experience with Pinecone + SQLite Record Manager on Flowise Cloud, I’d love to hear your setup.
Thanks in advance! 🚀
The text was updated successfully, but these errors were encountered:
Dear Henry,
Thank you very much for your answer.
I atatch the record here.
6a4e51df-6f27-45ce-b818-0abda9221394.mp4
It does happen only to the agent that is using the specific document store.
For your info, I have created 3 doc stores (200 chunks), 3 RAG agent linked to each document store, and all of them are experiencing the same issue
Title: Flowise Freezing with RAG Setup Using Pinecone & SQLite Record Manager
Hi everyone,
I'm facing an issue where Flowise Cloud freezes and turns white whenever I query my RAG agent. I've checked multiple possible causes, but I still can't identify the root problem.
My Flow Setup
Conversational Q&A Retriever → OpenAI GPT-4 → Document Store (Pinecone)
Document Loader: PDF file with Recursive Character Text Splitter
Chunking: 1000 characters per chunk with 200 overlap (~180 chunks)
Embeddings: OpenAI text-embedding-3-small (1536 dimensions)
Vector Store: Pinecone (cosine similarity, april2025 index, Top K = 4)
Record Manager: SQLite
Issue
Flowise freezes when I test the agent, and the console logs show:
🚨 TypeError: Cannot read properties of null (reading '4')
Additionally, I previously had Redis configured, but after facing connection issues, I switched to Pinecone and SQLite Record Manager. The issue persists even after upserting documents successfully.
Troubleshooting Steps Tried
✅ Recreated the vector store and re-upserted documents successfully
✅ Verified SQLite is properly configured (no permission errors)
✅ Cleared browser cache, tried incognito mode, and different browsers
✅ Checked CPU/memory usage during query (no major spikes)
✅ Tested with smaller document size and different chunking settings
Possible Causes & Questions
Could the SQLite Record Manager be causing the issue? Should I try without it?
Could Pinecone retrieval be causing memory overload?
Any way to debug why the UI crashes (logs, error traces, etc.)?
I'd appreciate any insights or suggestions! If anyone has experience with Pinecone + SQLite Record Manager on Flowise Cloud, I’d love to hear your setup.
Thanks in advance! 🚀
The text was updated successfully, but these errors were encountered: