Idempotency, is it more than just a buzzword?

hey everyone, been thinking a lot about idempotency lately, especially with all the retries that happen in distributed systems. we all know what it is conceptually (applying an operation multiple times has the same effect as applying it once), but how do you *actually* implement it effectively? are you using unique request IDs and checking them on the server? what about managing the state for those checks? curious to hear your go-to strategies and any pitfalls youve run into.