Struct HostRolloutState 

pub struct HostRolloutState {

    pub rollout_id: RolloutId,
    pub hostname: String,
    pub channel: String,
    pub state: HostState,
    pub target_closure: ClosureHash,
    pub current_closure_at_dispatch: Option<ClosureHash>,
    pub current_closure: Option<ClosureHash>,
    pub reverted_to: Option<ClosureHash>,
    pub dispatched_at: DateTime<Utc>,
    pub dispatch_acked_at: Option<DateTime<Utc>>,
    pub activation_started_at: Option<DateTime<Utc>>,
    pub activation_completed_at: Option<DateTime<Utc>>,
    pub activation_failed_at: Option<DateTime<Utc>>,
    pub probe_observed_first_at: Option<DateTime<Utc>>,
    pub probe_failure_first_at: Option<DateTime<Utc>>,
    pub soak_due_at: Option<DateTime<Utc>>,
    pub converged_at: Option<DateTime<Utc>>,
    pub failed_at: Option<DateTime<Utc>>,
    pub policy_applied: Option<OnHealthFailure>,
    pub reverted_at: Option<DateTime<Utc>>,
    pub probes: HashMap<ProbeName, ProbeRecord>,
    pub last_event_seq: u64,
}

Per-(rollout, host) reducer state. Mirrors RFC-0005 §5 HostRolloutRecord; the persistence schema in Phase 4 serializes this struct.

Every transition timestamp is agent-supplied (received via the wire and stamped onto the corresponding field by the reducer). dispatched_at is the lone exception — it’s CP-issued, so CP wallclock is the source of truth there (RFC-0005 §5).

Fields

rollout_id: RolloutId

hostname: String

channel: String

state: HostState

target_closure: ClosureHash

current_closure_at_dispatch: Option<ClosureHash>

current_closure: Option<ClosureHash>

reverted_to: Option<ClosureHash>

dispatched_at: DateTime<Utc>

dispatch_acked_at: Option<DateTime<Utc>>

activation_started_at: Option<DateTime<Utc>>

activation_completed_at: Option<DateTime<Utc>>

activation_failed_at: Option<DateTime<Utc>>

probe_observed_first_at: Option<DateTime<Utc>>

probe_failure_first_at: Option<DateTime<Utc>>

soak_due_at: Option<DateTime<Utc>>

converged_at: Option<DateTime<Utc>>

failed_at: Option<DateTime<Utc>>

policy_applied: Option<OnHealthFailure>

reverted_at: Option<DateTime<Utc>>

probes: HashMap<ProbeName, ProbeRecord>

last_event_seq: u64

Monotonic per (hostname, rollout_id). Gaps signal lost events; out-of-order events are dropped with a warning at the runtime layer.

Implementations

impl HostRolloutState

pub fn new_pending( rollout_id: RolloutId, hostname: String, channel: String, target_closure: ClosureHash, dispatched_at: DateTime<Utc>, soak_due_at: DateTime<Utc>, ) -> Self

Construct the initial Pending state when CP queues a Dispatch (or the agent receives one via long-poll on /v1/agent/dispatch).

This is the only legitimate way to bring a (rollout_id, hostname) record into existence. Subsequent transitions go through [step].

Trait Implementations

impl Clone for HostRolloutState

fn clone(&self) -> HostRolloutState

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for HostRolloutState

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

Formats the value using the given formatter.