Struct ConsecutiveFailurePolicy 

pub struct ConsecutiveFailurePolicy {
    pub failure_threshold: u32,
    pub recovery_successes: u32,
    pub probe_interval: Duration,
}

Marks a node unhealthy after N consecutive failures, recovers after M successes.

When failure_threshold is 1, nodes jump directly from Healthy to Unhealthy — the Degraded state is skipped because there is no room for an intermediate warning.

Fields

failure_threshold: u32

Number of consecutive failures before a node becomes Unhealthy. At ⌈threshold/2⌉ failures the node first enters Degraded. Default: 3.

recovery_successes: u32

Number of consecutive successes an Unhealthy / Recovering node needs before returning to Healthy. Default: 2.

probe_interval: Duration

Minimum time after the last failure before an Unhealthy node is eligible for a probe request. Default: 10 s.

Implementations

impl ConsecutiveFailurePolicy

pub fn builder() -> ConsecutiveFailurePolicyBuilder

Create an instance of ConsecutiveFailurePolicy using the builder syntax

Trait Implementations

impl Default for ConsecutiveFailurePolicy

fn default() -> Self

Returns the “default value” for a type.

impl HealthPolicy for ConsecutiveFailurePolicy

fn on_success(&self, health: &mut NodeHealth)

Called after a successful request to the node.

fn on_failure(&self, health: &mut NodeHealth)

Called after a failed request to the node.

fn should_probe(&self, health: &NodeHealth) -> bool

Whether an Unhealthy node should be given a probe request to check if it has recovered.

fn on_outcome(&self, health: &mut NodeHealth, outcome: &Outcome)

Process a classified request outcome. The default implementation delegates to on_success / on_failure and ignores rate-limited outcomes (they don’t affect health state).