Bug when calling Lambda from Postgres RDS. Works for short function runs < 5 seconds, but not if the call lasts 10 seconds

I have found a bug when calling Lambda from Postgres RDS. I am using Postgres version 13. I have created a one minute video nailing down the bug: https://youtu.be/Y1ALUxEwzGk

I am using the following query to call the lambda: SELECT payload->>'answer' from aws_lambda.invoke(aws_commons.create_lambda_function_arn('arn:aws:lambda:eu-central-1:xxxxxx:function:ChatGPT', 'eu-central-1'), '{"prompt": ""}'::json)

Here is my working python code:

import time

def lambda_handler(event, context):
    time.sleep(1)
    return {
        'statusCode': 200,
        "answer": "Test"
    }

If I change this line:

time.sleep(1)

to

time.sleep(10)

Then the call will fail.

Postgres version: PostgreSQL 13.10 on aarch64-unknown-linux-gnu, compiled by gcc (GCC) 7.3.1 20180712 (Red Hat 7.3.1-6), 64-bit

Also, the timeout of the lambda is 3 minutes (far more than 10 seconds): https://drive.google.com/file/d/1iMEpZDM3PCA9b6glJnAh93tspW0V3BvD/view?usp=drivesdk

A related issue is that there seems to be a hard timeout of the aws_lambda.invoke of 100 seconds, and I would really like that to be adjustable. My use case is that I am querying chatgpt from inside postgres and it can really take a long time to return an answer, sometimes.

The error message after 103 seconds is: ERROR: invoke API failed DETAIL: AWS Lambda client returned 'Unable to connect to endpoint'. CONTEXT: SQL function "invoke" statement 1

So no timeout error message.