Struct FailoverController 

pub struct FailoverController { /* private fields */ }

A higher-level node health monitor that combines heartbeat-timeout detection with explicit mark/recover operations.

Unlike FailoverCoordinator (which is tied to the Raft election machinery), FailoverController is a self-contained monitor suitable for use by the cluster topology dashboard, the alerting subsystem, or external health-check drivers.

Thread safety

All methods use internal locks; the struct is Send + Sync.

Test injection

Tests may call set_last_seen to back-date the last heartbeat timestamp for a node without actually waiting.

Implementations

impl FailoverController

pub fn new(heartbeat_timeout: Duration) -> Self

Create a new controller.

heartbeat_timeout — how long a node can go without sending a heartbeat before it is considered failed.

pub fn record_heartbeat(&self, node_id: NodeId)

Record that a heartbeat was received from node_id.

If the node was previously marked as failed it is automatically recovered.

pub fn detect_failed_nodes(&self) -> Vec<NodeId>

Check all known nodes; return the IDs of nodes that have not sent a heartbeat within the timeout window.

Nodes detected here are also added to the internal failed set (visible via is_failed / failed_nodes).

pub fn mark_failed(&self, node_id: NodeId)

Explicitly mark a node as failed (e.g. after Raft loss-of-quorum).

pub fn mark_recovered(&self, node_id: NodeId)

Mark a node as recovered.

This removes it from the failed set. Call record_heartbeat as well to reset the timeout clock.

pub fn is_failed(&self, node_id: NodeId) -> bool

Return true if node_id is currently considered failed.

pub fn failed_nodes(&self) -> Vec<NodeId>

Return all currently failed node IDs as a Vec.

Trait Implementations

impl Debug for FailoverController

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

Formats the value using the given formatter.