How is a CloudFront origin deemed unavailable?

Question

If an origin is unavailable, CloudFront will serve the requested object from cache even if it is stale. I have observed that when the origin server is offline, the stale object is served very quickly and it's obvious that the origin isn't contacted during the viewer request.

How does CloudFront know that the origin is unavailable? Is there a heuristic for this, ie. how many requests have to fail before the origin is marked unavailable? At what point will the origin be contacted again?

Thank you!

Answer

According to the documentations, [Controlling origin timeouts and attempts](https://docs.aws.amazon.com/AmazonCloudFront/latest/DeveloperGuide/high_availability_origin_failover.html#controlling-attempts-and-timeouts):

"By default, CloudFront tries to connect to the primary origin in an origin group for as long as 30 seconds (3 connection attempts of 10 seconds each) before failing over to the secondary origin. "

You can customize these.

How is a CloudFront origin deemed unavailable?

Relevant content