GPT Module Error Handling

Questions
, ,
1

:bullseye: What is your goal?

I want to find the best error handling module for my use case

:thinking: What is the problem & what have you tried?

I have a scenario with multiple gpt modules. 9/10 times it works fine and gives good results. However, when it fails, the gpt modules just HANG . The circumference timer indicatoron the module just keeps going around. Sometimes the module takes up to 5 minutes to run and that’s fine. ButI know it has failedusually after 5 minutes. What kind of error handler should I be setting up?
The GPT api modules have system and user prompts, they receive webhook bundles and I use helper fields throughout, usual stuff. It’s the 10% I need to fix. I’d like to set error handler TRY AGAIN after 5 minutes. That possible? Would really appreciate a simple solution. THanks

:clipboard: Error messages or input/output bundles

My problem—there is never a specific error or timeout message, it just HANGS there until I need to FORCE STOP.

3

You can add a Break Module (Error Handler) to each ChatGPT Module. You can set like 2-3 attempts in a span of 5-6 minutes. (Enable Automatic completion)

Note: Scenario > Settings (Enable incomplete executions)

Let me know how it goes!

1 Like
4

Thank you Karmic. I just need to fully understand, will this set up allow the module to run for 4 minutes and if it hasn’t completed will run again? with 3 tries? Should I toggle on “map” too? Cheers!

image
image693Ă—605 46 KB

5

Those settings do not limit your ChatGPT Module to 4 minutes. That setting tells how often Make.com will retry and the interval of each retry if the ChatGPT Module fails. Leave the Map settings as is for now.

1 Like
6

Thanks Karmic, my real issue , hanging gpt modules are not technically “errors”, therefore Error Handling modules can’t manage that situation. I think it’s a general problem with API calls, especially when the prompts are getting complex and the data large.

7

Exactly! You can adjust your prompts. That’s fairly easy to do.