Struct RecoveryTimeouts 

pub struct RecoveryTimeouts {
    pub jit_handoff: Elapsed,
    pub startup: Elapsed,
    pub idle: Elapsed,
}

How long an attempt may sit in each pre-terminal state before recovery treats it as stuck.

No document in this taskflow states these durations. 03-control-flows.md flow 2.7 says a surplus runner “exits on its idle timeout” without giving one, and flow 4.4 says an expired JIT configuration is discarded without saying when it expires. RecoveryTimeouts::provisional therefore returns values chosen here, named so that a caller cannot mistake them for a product decision, and there is deliberately no Default impl — e1 and e3 should have to write the numbers down.

Fields

jit_handoff: Elapsed

How long allocated or jit_received may last before the JIT configuration is assumed lost.

startup: Elapsed

How long starting may last before the runner is assumed not to be coming up.

idle: Elapsed

How long idle may last, in both directions it is read.

For an attempt whose process is already gone it separates the two readings of that exit: past this, the surplus case; before it, a crash.

For one whose process is still alive and still registered it is a deadline rather than a reading — the point at which the agent stops the runner itself. Nothing else does: Runner.Listener run long-polls for an assignment indefinitely, so this value, and only this value, bounds how long a runner that never gets a job holds its capacity slot and its entry in the target’s runner settings.

Implementations

impl RecoveryTimeouts

pub const fn new(jit_handoff: Elapsed, startup: Elapsed, idle: Elapsed) -> Self

pub fn provisional() -> Self

Placeholder values, with one exception — see the type documentation.

idle is no longer a placeholder. It stopped being one when it became the only thing that bounds a live runner: five minutes is long enough that a runner GitHub is about to assign is not stopped out from under the assignment, and short enough that a machine does not carry an unusable slot, and a repository an unusable runner row, for hours. The other two still only separate readings of an event that already happened, and nothing here has had to decide what they should be.

Trait Implementations

impl Clone for RecoveryTimeouts

fn clone(&self) -> RecoveryTimeouts

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for RecoveryTimeouts

impl Debug for RecoveryTimeouts

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

Formats the value using the given formatter.

impl Eq for RecoveryTimeouts

impl PartialEq for RecoveryTimeouts

fn eq(&self, other: &RecoveryTimeouts) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for RecoveryTimeouts