Anybody have a good pattern for creating conditional task retries? I'm checking now but I would guess that

on_failure

is only called after all retries

Hey Dominic, what's the ideal pattern you're trying to achieve for conditional task retries? Just want to get a better understanding of your use case

Would mostly use for extraction tasks that can throw a variety of HTTP error codes, some of which require retries and others that should kill the flow there. For example, an authentication error would tell me the credentials expired and I should stop here, while a connection or timeout error is worth a retry after a delay.

Ah, understood, thank you for elaborating. Right now, I'm not aware of any out-of-the-box solution to achieve this pattern. However, I believe this open issue matches your desired behavior. Would you be willing to comment on the issue and provide the details you included here in our thread?