Expand description

Contains various failure accrual policies, which are used for the failure rate detection.

Structs

A policy based on a maximum number of consecutive failure
A combinator used for join two policies into new one.
A policy based on an exponentially-weighted moving average success rate over a time window. A moving average is used so the success rate calculation is biased towards more recent requests.

Traits

A FailurePolicy is used to determine whether or not the backend died.

Functions

A policy based on a maximum number of consecutive failures. If num_failures occur consecutively, mark_dead_on_failure will return a Some(Duration) to mark an endpoint dead for.
Returns a policy based on an exponentially-weighted moving average success rate over a time window. A moving average is used so the success rate calculation is biased towards more recent requests.