# Retries, limits and errors

> Page: https://docs.flung.dev/reliability. Index of every page: https://docs.flung.dev/llms.txt

How Flung keeps a retry from posting twice, how fast you may call it, and what a failure tells you.

## TL;DR

| Question | Answer |
|---|---|
| Is a retry safe? | Yes, with the same `Idempotency-Key`: you get the first answer back. The SDK sends one on every POST. |
| How fast may I call? | 600 requests a minute per API key or OAuth app. Past that the answer is `429` with `Retry-After`. |
| What does a failure look like? | One JSON shape, with a stable `code`. See [Errors](https://docs.flung.dev/reference/api/errors.md). |
| What does the SDK retry by itself? | Network failures, timeouts, 408, 409, 429 and 5xx answers, twice, waiting longer each time and honouring `Retry-After`. |

## Walk every page of a list

Lists return a page at a time with `hasMore`. The SDK walks them for you:

## Accounts that stop working

Platforms end access now and then: a token expires, a password changes, an app review lapses. Flung renews every token it can before it runs out. Accounts it cannot renew turn `needs_reconnect`, and `account.reconnect_required_soon` arrives a week before the ones with a fixed end. A reconnect needs the person, through the [connect panel](https://docs.flung.dev/connect-accounts.md) or a link from [Start reconnecting an account](https://docs.flung.dev/reference/api/post-v1-accounts-id-reconnect.md).