Caching / rate limiting for API

[Daniel15]

I just had an issue where my server got a huge number of hits to the /api/v1/status endpoint, which basically killed my PHP-FPM server (it hit the maximum number of workers, preventing people from connecting to any other sites in the same pool).

Does Cachet cache the response to this API call? It would be good to cache its response (even if only for a short period of time, like 30 seconds), so that a large number of hits don't cause any perf issues. If a lot of hits come in at the same time, they could all receive the cached response rather than having to recompute it + hit the DB.

[jbrooksuk]

Hey @Daniel15! The API endpoints aren't actually cached at the moment, but they definitely should.

Out of interest, where are you using Cachet? I see you all over the place! 😄

[Daniel15]

Thanks for the reply! In theory I could use Nginx FastCGI caching to cache the endpoint, but it seems like Cachet is sending a Cache-Control: no-cache header, making the content uncacheable. Is there an easy way to modify /api/v1/status to send a caching header? Then I could just configure Nginx to cache it.

@jbrooksuk Cachet is being used for the status page for Yarn: https://status.yarnpkg.com/

[jbrooksuk]

I've addressed this is a Draft PR, so that we can talk about it there and then get this fixed for you.

Cachet is being used for the status page for Yarn: https://status.yarnpkg.com/

That's where I know you from! 😄