Struct HealthMonitor 

pub struct HealthMonitor {
    block_id: String,
    inner: Mutex<HealthMonitorInner>,
    broker: Option<Arc<Broker>>,
}

Per-block agent health monitor.

Tracks output activity and error rates, computes health state, and emits WPS events on state transitions.

Fields

block_id: String

inner: Mutex<HealthMonitorInner>

broker: Option<Arc<Broker>>

Implementations

impl HealthMonitor

const STALL_SECS: u64 = 30

Stall threshold: no meaningful output for 30s during active turn.

const DEAD_SECS: u64 = 120

Dead threshold: no meaningful output for 120s during active turn.

const ERROR_WINDOW_SECS: u64 = 300

Error window duration.

const DEGRADED_TRANSIENT_THRESHOLD: usize = 5

Transient error count threshold for degraded.

pub fn new(block_id: String, broker: Option<Arc<Broker>>) -> Self

pub fn set_active_turn(&self, active: bool)

Called when a new turn starts (subprocess spawned).

pub fn set_exited(&self, exit_code: i32)

Called when the subprocess exits.

pub fn record_output(&self, meaningful: bool)

Called for each output line from stdout. meaningful is false for rate_limit_event and similar non-progress events.

pub fn record_error(&self, class: ErrorClass, message: String)

Called when an error is detected in the output stream.

pub fn is_active_turn(&self) -> bool

Whether there’s an active turn in progress.

pub fn check(&self)

Periodic health check — call this every ~5 seconds while a turn is active.

fn evaluate_and_transition(&self)

Compute current health and emit event if it changed.

fn compute_health(inner: &HealthMonitorInner) -> AgentHealth

Composite health computation.

fn make_detail(inner: &HealthMonitorInner, health: &AgentHealth) -> String

Generate human-readable detail string.

fn publish_health(&self, event: AgentHealthEvent)

Publish health event via WPS broker.