A proactive cache always has fresh tokens, except in two cases:
- when it's still starting up
- when an origin server is down
A token is the resource we're caching, with a time to live (TTL). An origin server is the server which provides fresh tokens. We don't want to overload it, hence the cache.
The logic can be generalized to cache arbitrary HTTP resources (e.g. with
appropriate Cache-Control directives), but the use case I had in mind was
OAuth 2.0 access tokens.
This repository has:
- a thread-based implementation in
sync - a coroutine-based implementation in
async