New RunLLM response generated: >>>*# Question* Hey, how does

retries

work? Our spark tasks sometimes have transient failures but they are not retried after a "user error". Then I tested with a python task, just raising an Exception, it's not retried either. Is the intended behaviour? # Answer Flyte's retry mechanism differentiates between recoverable and non-recoverable errors. Recoverable errors are retried, while non-recoverable errors are not. By default, all user exceptions are considered non-recoverable unless they are a subclass of

FlyteRecoverableException

(source, source). For transient failures in Spark tasks, ensure that the exceptions raised are subclasses of

FlyteRecoverableException

to trigger retries. For Python tasks, raising a generic

Exception

will not trigger retries unless it is wrapped in a

FlyteRecoverableException

. Would you like more details on how to handle specific exceptions in Flyte? Sources: - Build Indestructible Pipelines with Flyte - Flyte Task System Retry Slack Link https://.slack.com/archives/CP2HDHKE1/p1727791110.858489 👍 0 button 👎 0 button 💡 Hint: Mention @runllm in the thread for followups.