A polite wait signal
When a server is overloaded or rate limiting, it can return a Retry After header telling the client how long to wait before retrying. It commonly pairs with 429 Too Many Requests or 503 Service Unavailable. The value is either a number of seconds or an absolute date.
Why clients should honor it
- It prevents a retry storm where many clients hammer a struggling server and keep it down.
- It lets the server schedule recovery, for example during a planned maintenance window with a date value.
- It replaces blind guessing with the server's own estimate of when capacity returns.
Combining with backoff
A robust client treats Retry After as a floor, then adds exponential backoff with jitter so retries spread out rather than all firing at the same instant. Ignoring the header and retrying instantly is a common cause of cascading failure under load.
Key idea
The Retry After header gives a client a server supplied wait time, usually with 429 or 503, and honoring it with added jitter prevents retry storms.