Hi everyone,
I encountered an issue with a workflow that has been configured and working perfectly fine for a long time.
### The Problem:
The issue occurs specifically in the **Gemini “Analyze an Image”** node. The node gets stuck on “loading” for a very long time and eventually returns nothing (times out).
This behavior persisted continuously for about 2 hours, during which no images could be processed. After 2 hours, the workflow started working normally again without any changes from my side.
I would like to understand what could cause this temporary “hanging” of the Gemini node and how to prevent it or handle it properly in the future so it doesn’t break the automation.
### My setup:
- **n8n version:** [ n8n@2.21.5]
Any advice on why the Gemini node might freeze on image processing like this? Thanks in advance for your help!
Welcome to the n8n community @Product
This looks like a service intermittency issue.
Was this problem a one-time occurrence or is it recurring?
The Gemini documentation treats failures and timeouts as issues that can occur both on the API backend and on the client/SDK side. I would add retry/backoff and error handling to reprocess the image when there’s a timeout, in addition to checking image size, quota/rate limits, and Gemini API status when the problem occurs.
Thank you! The issue is recurring. I already have retry logic implemented — 3 attempts with a 5-second interval between each — but the problem still persists. I’m on a paid API plan and I’m not exceeding any rate limits or quotas. Could this be related to scheduled maintenance on Gemini’s side? Is there a way to check upcoming maintenance windows in advance so we can plan around them?
@Product needs to speak with them about maintenance.
Normally the monitoring is done through the status dashboards and incident reports from Google Cloud/AI when instabilities occur.
Also, which Gemini model is the best to choose in terms of stability and rarely breaking down (for photo recognition)? I am currently using ‘gemini-flash-latest’
