Suddenly getting 'aiohttp.client_exceptions.ClientResponseError'

GPT builders
, ,
1

My GPT is now getting ‘aiohttp.client_exceptions.ClientResponseError’ on every attempt to call my API. Looking at my API logs, the API is never getting hit. I’ve verified my schema multiple times, validated the API side using Postman, etc. Are others seeing sudden issues with their GPT throwing ClientResponseError without actually making an API call?

1 Like
4

For those running into this in the future, through much head banging I was able to resolve this through the following steps:

  1. Delete the Action, save the GPT, interact with the new GPT, then recreate the Action
  2. When recreating the Action, make sure you increment the version of the schema in your info block
  3. Change your schema to be ultra protective - ‘oneOf’ qualifiers helped me even though I knew what I was returning.

From some research, it seems the GPT has a cache that it stores which is very hidden to the GPT Builder. That cache is then used to validate your schema and if that cache contains something that conflicts with your schema, even if the real API does not, the GPT Builder will throw this ClientResponseError on every attempt without ever checking the API. The above steps clear said cache and force the GPT Builder to re-initialize directly from the actual API.

Happy building…

3 Likes
5

Thanks for coming back to let us know.

Hope you stick around. We’ve got a great community!

1 Like
6

Thank you, that helped me.

But a month ago, calling any of the APIs took about a second, now it takes more than a minute.

My workflow used to take 2-3 minutes, now it takes several hours…

What could be the problem?