Struct ScaleManager 

pub struct ScaleManager { /* private fields */ }

Tracks all instances managed by this Box host.

Implementations

impl ScaleManager

pub fn new(max_instances: u32) -> Self

Create a new scale manager with the given capacity.

pub fn process_request(&mut self, request: &ScaleRequest) -> ScaleResponse

Process a scale request and return the response.

This determines how many instances to create or destroy but does not actually start/stop VMs — the caller is responsible for that based on the response.

pub fn register_instance( &mut self, service: &str, instance_id: &str, endpoint: Option<&str>, )

Register a new instance for a service.

pub fn update_state( &mut self, service: &str, instance_id: &str, new_state: InstanceState, ) -> Option<InstanceEvent>

Update an instance’s state and emit a transition event.

pub fn update_health( &mut self, service: &str, instance_id: &str, health: InstanceHealth, )

Update an instance’s health metrics.

pub fn update_endpoint( &mut self, service: &str, instance_id: &str, endpoint: &str, )

Update an instance’s endpoint.

pub fn deregister_instance(&mut self, service: &str, instance_id: &str) -> bool

Remove an instance from tracking.

pub fn instances_to_create(&self, service: &str) -> u32

Get instances that need to be created (target > current running).

pub fn instances_to_stop(&self, service: &str) -> Vec<String>

Get instances that should be stopped (current > target). Returns instance IDs to stop, preferring non-busy instances.

pub fn ready_instances(&self, service: &str) -> Vec<InstanceInfo>

Get all ready instances for a service (for traffic routing).

pub fn total_instances(&self) -> u32

Get the total number of instances across all services.

pub fn service_instance_count(&self, service: &str) -> u32

Get the number of instances for a specific service.

pub fn services(&self) -> Vec<String>

List all tracked services.

pub fn recent_events(&self, limit: usize) -> &[InstanceEvent]

Get recent events.

pub fn service_health(&self, service: &str) -> ServiceHealth

Aggregate health metrics for a service (for autoscaler decisions).

pub fn start_drain( &mut self, service: &str, instance_id: &str, ) -> Option<InstanceEvent>

Initiate graceful drain for an instance.

Transitions the instance to Draining state. The caller should:

1. Stop routing new requests to this instance
2. Wait for in-flight requests to complete (or timeout)
3. Call complete_drain() to transition to Stopping

pub fn complete_drain( &mut self, service: &str, instance_id: &str, ) -> Option<InstanceEvent>

Complete a drain and transition to Stopping.

Called after in-flight requests have completed or the drain timeout expired.

pub fn is_drain_complete(&self, service: &str, instance_id: &str) -> bool

Check if a draining instance has no in-flight requests.

pub fn draining_instances(&self, service: &str) -> Vec<String>

Get all instances currently draining.