Struct ReconnectConfig 

pub struct ReconnectConfig {
    pub max_attempts: u32,
    pub base_delay: Duration,
    pub max_delay: Duration,
    pub min_healthy_uptime: Duration,
    pub max_total_reopens: u32,
}

SSE reconnect policy. The streaming client tracks the latest id: field seen on the wire and, on transient stream failures, re-issues the request with a Last-Event-ID: <stored> header so the server can resume the event sequence (per HTML5 EventSource spec).

Default is opt-in disabled (max_attempts: 0) so existing SDKs see no behaviour change.

Fields

max_attempts: u32

Maximum number of consecutive reconnect attempts with no healthy connection in between (the burst budget). 0 (default) disables reconnect entirely — stream errors bubble up. The budget is reset by a connection that both delivers an item and stays up at least min_healthy_uptime.

base_delay: Duration

Initial delay before the first reconnect attempt.

max_delay: Duration

Hard cap on delay between reconnect attempts.

min_healthy_uptime: Duration

Minimum time a connection must stay up — in addition to delivering at least one item — before its end resets the burst budget. Delivering a single item is too weak a health signal on its own: a peer that emits one item and immediately drops would reset the budget on every cycle and reopen forever, re-sending the auth token each time (#4740). A connection shorter than this counts against max_attempts like any other failed attempt.

max_total_reopens: u32

Absolute lifetime ceiling on reopens, independent of budget resets. It bounds the pathological peer that stays up just past min_healthy_uptime, delivers an item, and drops on a loop — which would otherwise reset the burst budget indefinitely. Set high enough that a genuinely healthy long-lived subscription (which reconnects rarely) never approaches it; 0 refuses reopen outright.

Implementations

impl ReconnectConfig

pub fn enabled(max_attempts: u32, base_delay: Duration) -> ReconnectConfig

Build a reconnect policy with up to max_attempts retries and the supplied initial delay (capped by max_delay, default 10s).

pub const fn disabled() -> ReconnectConfig

A policy that never reconnects: the stream’s first transport failure ends it.

The counterpart to ReconnectConfig::enabled. Default already yields max_attempts: 0, so this is behaviourally the same value — it exists so a call site that must not reconnect reads as a deliberate choice rather than an accepted default.

Generated clients pass this for a method whose open is fallible (#[streaming] async fn). Such an open carries domain semantics the client must not blindly repeat: any exclusion lease it acquired is owned by the returned stream’s lifetime, resume may be specified through the contract’s own cursor rather than Last-Event-ID, and a reconnect-time failure would arrive as a stream item — past the caller’s open-time error handling, which is the whole reason the fallible shape exists.

pub fn with_max_delay(self, max_delay: Duration) -> ReconnectConfig

Override the maximum delay between reconnect attempts.

pub fn with_min_healthy_uptime( self, min_healthy_uptime: Duration, ) -> ReconnectConfig

Override the minimum healthy connection uptime that resets the burst budget. See min_healthy_uptime.

pub fn with_max_total_reopens(self, max_total_reopens: u32) -> ReconnectConfig

Override the absolute lifetime cap on reopens. See max_total_reopens.

Trait Implementations

impl Clone for ReconnectConfig

fn clone(&self) -> ReconnectConfig

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ReconnectConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.

impl Default for ReconnectConfig

fn default() -> ReconnectConfig

Returns the “default value” for a type.