Exactly my thought! I was also thinking that Octo sounds way too much like a monolith. I wouldn't let retries tax the same instance. I bet they could use some

Exactly my thought! I was also thinking that Octo sounds way too much like a monolith. I wouldn't let retries tax the same instance. I bet they could use something like RabbitMQ to sort work distribution and retries between instances without relying on nested callbacks.