dynamo_runtime/pipeline/network/egress/

push_router.rs

// SPDX-FileCopyrightText: Copyright (c) 2024-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
// SPDX-License-Identifier: Apache-2.0

use super::{AsyncEngineContextProvider, ResponseStream};
use crate::error::{BackendError, DynamoError, ErrorType, match_error_chain};
use crate::pipeline::network::egress::route_span::{
    get_route_trace_context, record_route_error, record_route_span_start, wrap_route_span,
};
use crate::{
    component::{Client, DeviceType, Endpoint, Instance, RoutingInstances},
    discovery::EndpointInstanceId,
    dynamo_nvtx_range,
    engine::{AsyncEngine, AsyncEngineContext, Data},
    metrics::frontend_perf::{STAGE_DURATION_SECONDS, STAGE_ROUTE},
    pipeline::{
        AddressedPushRouter, AddressedRequest, Error, ManyIn, ManyOut, SingleIn, StreamingDispatch,
        error::{PipelineError, PipelineErrorExt},
    },
    protocols::{EndpointId, maybe_error::MaybeError},
    routing_policy::{
        CandidateView, OccupancyReservation, RouteCandidate, RouteContext, RouteDevice,
        RoutePicker, RoutePolicy, RouteTarget, RoutingOccupancyState,
        get_or_create_routing_occupancy_state,
    },
    traits::DistributedRuntimeProvider,
};
use async_trait::async_trait;
use futures::Stream;
use serde::{Deserialize, Serialize};
use std::{
    collections::{HashMap, HashSet},
    marker::PhantomData,
    pin::Pin,
    sync::{Arc, atomic::AtomicU64},
    task::Poll,
    time::Instant,
};
use tokio_stream::StreamExt;
use tracing::Instrument;

/// Check if an error chain indicates the worker should be reported as down.
fn is_inhibited(err: &(dyn std::error::Error + 'static)) -> bool {
    const INHIBITED: &[ErrorType] = &[
        ErrorType::CannotConnect,
        ErrorType::Disconnected,
        ErrorType::ConnectionTimeout,
        ErrorType::ResponseTimeout,
        ErrorType::Backend(BackendError::EngineShutdown),
        // A stream that ends mid-generation means this worker dropped the
        // request. Quarantine it, or a migration retry can reselect the same
        // worker before discovery removal catches up.
        ErrorType::Backend(BackendError::StreamIncomplete),
        // The addressed server has no handler for this instance: discovery is
        // stale or the worker is shutting down. Same reasoning as above.
        ErrorType::WorkerUnavailable,
    ];
    match_error_chain(err, INHIBITED, &[])
}

/// Read the backend response inactivity timeout from the environment.
/// Reuses `DYN_HTTP_BACKEND_STREAM_TIMEOUT_SECS` — the same env var
/// as the HTTP-layer safety net in `disconnect.rs`.
fn response_inactivity_timeout() -> Option<std::time::Duration> {
    use crate::config::environment_names::llm::DYN_HTTP_BACKEND_STREAM_TIMEOUT_SECS;
    std::env::var(DYN_HTTP_BACKEND_STREAM_TIMEOUT_SECS)
        .ok()
        .and_then(|s| s.parse::<u64>().ok())
        .filter(|&secs| secs > 0)
        .map(std::time::Duration::from_secs)
}

/// RAII handle for one in-flight unit of work charged against
/// [`RoutingOccupancyState`]. The counter is incremented at construction; the
/// matching decrement is emitted on drop (or by [`Self::into_tracked_stream`]).
struct OccupancyPermit {
    reservation: Option<OccupancyReservation>,
}

impl OccupancyPermit {
    fn acquire(state: Arc<RoutingOccupancyState>, instance_id: u64) -> Self {
        Self {
            reservation: Some(state.reserve(instance_id)),
        }
    }

    fn from_counter(
        state: Arc<RoutingOccupancyState>,
        instance_id: u64,
        counter: Arc<AtomicU64>,
    ) -> Self {
        Self {
            reservation: Some(OccupancyReservation::from_counter(
                state,
                instance_id,
                counter,
            )),
        }
    }

    fn retarget(&mut self, instance_id: u64) {
        self.reservation
            .as_mut()
            .expect("occupancy permit must be armed before stream tracking")
            .retarget(instance_id);
    }

    fn into_tracked_stream<U: Data + MaybeError>(mut self, stream: ManyOut<U>) -> ManyOut<U> {
        let engine_ctx = stream.context();
        ResponseStream::new(
            Box::pin(OccupancyTrackedStream {
                inner: stream,
                reservation: self.reservation.take(),
            }),
            engine_ctx,
        )
    }
}

/// Trait for monitoring worker load and determining overload state.
/// Implementations can define custom load metrics and overload thresholds.
#[async_trait]
pub trait WorkerLoadMonitor: Send + Sync {
    /// Start background monitoring of worker load.
    /// This should spawn background tasks that update the client's overloaded instances.
    async fn start_monitoring(&self) -> anyhow::Result<()>;
}

/// Query interface for routing against multimodal embedding cache state.
pub trait MultimodalCacheIndex: Send + Sync {
    fn workers_with_cache_key_hits(&self, cache_keys: &[String]) -> Vec<(u64, usize)>;
    fn remove_worker(&self, worker_id: u64);
}

pub type MultimodalCacheKeyExtractor<T> = Arc<dyn Fn(&T) -> Vec<String> + Send + Sync>;

#[derive(Clone)]
pub struct PushRouter<T, U>
where
    T: Data + Serialize,
    U: Data + for<'de> Deserialize<'de>,
{
    // TODO: This shouldn't be pub, but lib/bindings/python/rust/lib.rs exposes it.
    /// The Client is how we gather remote endpoint information from etcd.
    pub client: Client,

    /// How we choose which instance to send traffic to.
    ///
    /// Setting this to KV means we never intend to call `generate` on this PushRouter. We are
    /// not using it as an AsyncEngine.
    /// Instead we will decide whether to call random/round_robin/direct ourselves and call them directly.
    /// dynamo-llm's KV Routing does this.
    router_mode: RouterMode,

    /// Shared, scheduler-independent policy state. KV and Direct have no picker.
    picker: Option<Arc<RoutePicker>>,

    /// Policy-specific state for callers that explicitly request static routing,
    /// independently of the router's configured generate mode.
    round_robin_picker: Arc<RoutePicker>,
    random_picker: Arc<RoutePicker>,

    /// The final hop: after selecting an instance, `PushRouter` hands it to this
    /// `StreamingDispatch` (the request-plane `AddressedPushRouter` by default).
    /// A trait object so an alternate transport can swap it out.
    addressed: Arc<dyn StreamingDispatch<T, U>>,

    /// When false, `generate_with_fault_detection` skips fault detection logic:
    /// it won't call `report_instance_down` on errors, and it uses the raw discovery
    /// instance list instead of the filtered avail list. Use for recovery/query paths
    /// where transient failures are expected.
    fault_detection_enabled: bool,

    /// Cached response inactivity timeout. Read once at construction from
    /// [`environment_names::llm::DYN_HTTP_BACKEND_STREAM_TIMEOUT_SECS`](crate::config::environment_names::llm::DYN_HTTP_BACKEND_STREAM_TIMEOUT_SECS) to avoid a syscall per request.
    response_timeout: Option<std::time::Duration>,

    /// Shared request occupancy state for tracked routing modes.
    occupancy_state: Option<Arc<RoutingOccupancyState>>,

    /// Optional cache index for direct multimodal embedding cache lookups.
    /// Currently consumed by `RouterMode::DeviceAwareWeighted`.
    multimodal_cache_indexer: Option<Arc<dyn MultimodalCacheIndex>>,

    /// Optional typed request extractor for multimodal embedding cache keys.
    multimodal_cache_key_extractor: Option<MultimodalCacheKeyExtractor<T>>,

    /// An internal Rust type. This says that PushRouter is generic over the T and U types,
    /// which are the input and output types of it's `generate` function. It allows the
    /// compiler to specialize us at compile time.
    _phantom: PhantomData<(T, U)>,
}

#[derive(Default, Debug, Clone, Copy, PartialEq, Serialize, Deserialize)]
#[serde(rename_all = "snake_case")]
pub enum RouterMode {
    #[default]
    RoundRobin,
    Random,
    PowerOfTwoChoices,
    KV,
    Direct,
    LeastLoaded,
    /// Device-aware weighted routing for heterogeneous workers.
    DeviceAwareWeighted,
}

#[derive(Clone, Copy)]
enum TransportFallback<'a> {
    Allow,
    Deny,
    Within(&'a HashSet<u64>),
}

#[derive(Clone, Copy)]
enum OverloadCheck {
    Required,
    AlreadyAdmitted,
}

struct DeviceAwareCandidates {
    candidates: Vec<RouteCandidate>,
    context: RouteContext,
    embedding_cache_hit: bool,
    request_cache_keys: usize,
}

/// A DeviceAware policy decision whose optional occupancy booking is owned by
/// the caller's request lifecycle.
pub struct DeviceAwareSelection {
    worker_id: u64,
    candidate_count: usize,
    load: u64,
    is_cpu: bool,
    embedding_cache_hit: bool,
    request_cache_keys: usize,
    reservation: Option<OccupancyReservation>,
}

impl DeviceAwareSelection {
    pub fn worker_id(&self) -> u64 {
        self.worker_id
    }

    pub fn candidate_count(&self) -> usize {
        self.candidate_count
    }

    pub fn load(&self) -> u64 {
        self.load
    }

    pub fn is_cpu(&self) -> bool {
        self.is_cpu
    }

    pub fn embedding_cache_hit(&self) -> bool {
        self.embedding_cache_hit
    }

    pub fn request_cache_keys(&self) -> usize {
        self.request_cache_keys
    }

    pub fn into_reservation(self) -> Option<OccupancyReservation> {
        self.reservation
    }
}

impl RouterMode {
    pub const fn telemetry_label(self) -> &'static str {
        match self {
            Self::RoundRobin => "round-robin",
            Self::Random => "random",
            Self::PowerOfTwoChoices => "power-of-two-choices",
            Self::KV => "kv",
            Self::Direct => "direct",
            Self::LeastLoaded => "least-loaded",
            Self::DeviceAwareWeighted => "device-aware-weighted",
        }
    }

    pub fn is_kv_routing(&self) -> bool {
        *self == RouterMode::KV
    }

    pub fn is_direct_routing(&self) -> bool {
        *self == RouterMode::Direct
    }

    /// Whether this mode admits requests against host-owned occupancy counters.
    pub const fn requires_occupancy(self) -> bool {
        matches!(
            self,
            Self::PowerOfTwoChoices | Self::LeastLoaded | Self::DeviceAwareWeighted
        )
    }

    fn route_policy(self) -> Option<RoutePolicy> {
        match self {
            Self::RoundRobin => Some(RoutePolicy::RoundRobin),
            Self::Random => Some(RoutePolicy::Random),
            Self::PowerOfTwoChoices => Some(RoutePolicy::PowerOfTwoChoices),
            Self::LeastLoaded => Some(RoutePolicy::LeastLoaded),
            Self::DeviceAwareWeighted => Some(RoutePolicy::DeviceAwareWeighted),
            Self::KV | Self::Direct => None,
        }
    }
}

fn route_pickers(
    router_mode: RouterMode,
) -> (Arc<RoutePicker>, Arc<RoutePicker>, Option<Arc<RoutePicker>>) {
    let round_robin = Arc::new(RoutePicker::new(RoutePolicy::RoundRobin));
    let random = Arc::new(RoutePicker::new(RoutePolicy::Random));
    let configured = match router_mode {
        RouterMode::RoundRobin => Some(round_robin.clone()),
        RouterMode::Random => Some(random.clone()),
        mode => mode.route_policy().map(RoutePicker::new).map(Arc::new),
    };
    (round_robin, random, configured)
}

/// Pick the instance with lower in-flight count from two random candidates.
/// Returns the single instance if only one is available.
#[cfg(test)]
fn p2c_select_from(occupancy_state: &RoutingOccupancyState, instance_ids: &[u64]) -> u64 {
    RoutePicker::new(RoutePolicy::PowerOfTwoChoices)
        .peek(
            CandidateView::Workers(instance_ids),
            RouteContext::default(),
            |id| occupancy_state.load(id),
        )
        .expect("p2c selection requires at least one candidate")
        .target
        .worker_id
}

/// At most one `list_and_watch` per endpoint, across all `PushRouter`
/// instances. Entry removed on watcher exit so a later router can re-arm.
static ENDPOINT_WATCHER_ACTIVE: std::sync::OnceLock<dashmap::DashMap<EndpointId, ()>> =
    std::sync::OnceLock::new();

#[derive(Clone, Debug, Eq, Hash, PartialEq)]
struct RuntimeEndpointId {
    connection_id: u64,
    endpoint_id: EndpointId,
}

impl RuntimeEndpointId {
    fn for_endpoint(endpoint: &Endpoint) -> Self {
        Self {
            connection_id: endpoint.drt().connection_id(),
            endpoint_id: endpoint.id(),
        }
    }
}

/// At most one multimodal cache cleanup watcher per runtime endpoint.
static ENDPOINT_CACHE_INDEXER_WATCHER_ACTIVE: std::sync::OnceLock<
    dashmap::DashMap<RuntimeEndpointId, ()>,
> = std::sync::OnceLock::new();

/// Watch discovery for instance removals and cancel pending response-stream
/// registrations on the removed instance, unblocking queued requests with
/// a migratable `Disconnected` error. Uses raw `list_and_watch` events
/// (not a coalesced snapshot diff) so a rapid remove→re-add of the same
/// identity is not silently swallowed. Keyed by full `EndpointInstanceId`.
fn spawn_instance_removal_watcher<T, U>(
    endpoint: Endpoint,
    dispatch: Arc<dyn StreamingDispatch<T, U>>,
    cancel_token: tokio_util::sync::CancellationToken,
) where
    T: Data + Serialize + 'static,
    U: Data + for<'de> Deserialize<'de> + MaybeError + 'static,
{
    use crate::discovery::{
        DiscoveryEvent, DiscoveryInstance, DiscoveryInstanceId, DiscoveryQuery,
    };
    use tokio_stream::StreamExt as _;

    // One watcher per endpoint: if one is already running, skip.
    let guard = ENDPOINT_WATCHER_ACTIVE.get_or_init(dashmap::DashMap::new);
    let endpoint_id = endpoint.id();
    if guard.insert(endpoint_id.clone(), ()).is_some() {
        tracing::debug!(
            ?endpoint_id,
            "Instance removal watcher already running for this endpoint, skipping"
        );
        return;
    }

    let endpoint_name = endpoint.name().to_string();

    tokio::spawn(async move {
        // Release on every exit path (including panic); a leaked entry
        // silently disables removal cancellation until process restart.
        struct GuardRelease(EndpointId);
        impl Drop for GuardRelease {
            fn drop(&mut self) {
                if let Some(map) = ENDPOINT_WATCHER_ACTIVE.get() {
                    map.remove(&self.0);
                }
            }
        }
        let _release = GuardRelease(endpoint_id);

        let namespace = endpoint.component().namespace().name();
        let component = endpoint.component().name().to_string();

        // Reconnect on transient discovery failure; cancel-aware backoff.
        const RECONNECT_BACKOFF: std::time::Duration = std::time::Duration::from_secs(5);
        'reconnect: loop {
            let query = DiscoveryQuery::Endpoint {
                namespace: namespace.clone(),
                component: component.clone(),
                endpoint: endpoint_name.clone(),
            };

            let mut stream = match endpoint.drt().discovery().list_and_watch(query, None).await {
                Ok(s) => s,
                Err(e) => {
                    tracing::warn!(
                        endpoint = %endpoint_name,
                        "Failed to start instance removal watcher (will retry): {e}"
                    );
                    tokio::select! {
                        _ = tokio::time::sleep(RECONNECT_BACKOFF) => continue 'reconnect,
                        _ = cancel_token.cancelled() => break 'reconnect,
                    }
                }
            };

            loop {
                tokio::select! {
                    event = stream.next() => {
                        match event {
                            Some(Ok(DiscoveryEvent::Removed(id))) => {
                                if let DiscoveryInstanceId::Endpoint(eid) = &id {
                                    dispatch.on_instance_removed(eid).await;
                                }
                            }
                            Some(Ok(DiscoveryEvent::Added(DiscoveryInstance::Endpoint(inst)))) => {
                                let eid: EndpointInstanceId = inst.endpoint_instance_id();
                                dispatch.on_instance_added(&eid).await;
                            }
                            Some(Ok(DiscoveryEvent::Added(_)))
                            | Some(Ok(DiscoveryEvent::ModelTaintsUpdated(_)))
                            | Some(Ok(DiscoveryEvent::Resync(_))) => {}
                            Some(Err(e)) => {
                                tracing::warn!(
                                    endpoint = %endpoint_name,
                                    "Instance removal watcher stream error: {e}"
                                );
                            }
                            None => {
                                tracing::warn!(
                                    endpoint = %endpoint_name,
                                    "Instance removal watcher stream ended; reconnecting"
                                );
                                continue 'reconnect;
                            }
                        }
                    }
                    _ = cancel_token.cancelled() => {
                        break 'reconnect;
                    }
                }
            }
        }

        tracing::debug!(endpoint = %endpoint_name, "Instance removal watcher exiting");
    });
}

/// Watch discovery removals for cache-aware routers and drop stale worker cache entries.
fn spawn_multimodal_cache_cleanup_watcher(
    endpoint: Endpoint,
    indexer: Arc<dyn MultimodalCacheIndex>,
    cancel_token: tokio_util::sync::CancellationToken,
) {
    use crate::discovery::{DiscoveryEvent, DiscoveryInstanceId, DiscoveryQuery};
    use tokio_stream::StreamExt as _;

    let guard = ENDPOINT_CACHE_INDEXER_WATCHER_ACTIVE.get_or_init(dashmap::DashMap::new);
    let watcher_id = RuntimeEndpointId::for_endpoint(&endpoint);
    if guard.insert(watcher_id.clone(), ()).is_some() {
        tracing::debug!(
            connection_id = watcher_id.connection_id,
            ?watcher_id.endpoint_id,
            "Multimodal cache cleanup watcher already running for this runtime endpoint, skipping"
        );
        return;
    }

    let endpoint_name = endpoint.name().to_string();
    let namespace = endpoint.component().namespace().name();
    let component = endpoint.component().name().to_string();

    tokio::spawn(async move {
        struct GuardRelease(RuntimeEndpointId);
        impl Drop for GuardRelease {
            fn drop(&mut self) {
                if let Some(map) = ENDPOINT_CACHE_INDEXER_WATCHER_ACTIVE.get() {
                    map.remove(&self.0);
                }
            }
        }
        let _release = GuardRelease(watcher_id);

        const RECONNECT_BACKOFF: std::time::Duration = std::time::Duration::from_secs(5);
        'reconnect: loop {
            let query = DiscoveryQuery::Endpoint {
                namespace: namespace.clone(),
                component: component.clone(),
                endpoint: endpoint_name.clone(),
            };

            let mut stream = match endpoint.drt().discovery().list_and_watch(query, None).await {
                Ok(stream) => stream,
                Err(error) => {
                    tracing::warn!(
                        endpoint = %endpoint_name,
                        "Failed to start multimodal cache cleanup watcher (will retry): {error}"
                    );
                    tokio::select! {
                        _ = tokio::time::sleep(RECONNECT_BACKOFF) => continue 'reconnect,
                        _ = cancel_token.cancelled() => break 'reconnect,
                    }
                }
            };

            loop {
                tokio::select! {
                    event = stream.next() => {
                        match event {
                            Some(Ok(DiscoveryEvent::Removed(DiscoveryInstanceId::Endpoint(eid)))) => {
                                indexer.remove_worker(eid.instance_id);
                            }
                            Some(Ok(DiscoveryEvent::Added(_)))
                            | Some(Ok(DiscoveryEvent::ModelTaintsUpdated(_)))
                            | Some(Ok(DiscoveryEvent::Removed(_)))
                            | Some(Ok(DiscoveryEvent::Resync(_))) => {}
                            Some(Err(error)) => {
                                tracing::warn!(
                                    endpoint = %endpoint_name,
                                    "Multimodal cache cleanup watcher stream error: {error}"
                                );
                                continue 'reconnect;
                            }
                            None => {
                                tracing::warn!(
                                    endpoint = %endpoint_name,
                                    "Multimodal cache cleanup watcher stream ended; reconnecting"
                                );
                                continue 'reconnect;
                            }
                        }
                    }
                    _ = cancel_token.cancelled() => break 'reconnect,
                }
            }
        }

        tracing::debug!(endpoint = %endpoint_name, "Multimodal cache cleanup watcher exiting");
    });
}

async fn addressed_router(endpoint: &Endpoint) -> anyhow::Result<Arc<AddressedPushRouter>> {
    AddressedPushRouter::from_runtime_provider(endpoint).await
}

impl<T, U> PushRouter<T, U>
where
    T: Data + Serialize,
    U: Data + for<'de> Deserialize<'de> + MaybeError,
{
    pub fn router_mode(&self) -> RouterMode {
        self.router_mode
    }

    /// Create a new PushRouter without a worker load monitor (no overload detection)
    pub async fn from_client(client: Client, router_mode: RouterMode) -> anyhow::Result<Self> {
        Self::from_client_with_monitor(client, router_mode, None).await
    }

    /// Create a new PushRouter with fault detection disabled.
    ///
    /// Unlike `from_client`, this router will not call `report_instance_down` on
    /// transient errors, and `direct()` uses the raw discovery instance list instead
    /// of the filtered avail list. Use for recovery/query paths.
    pub async fn from_client_no_fault_detection(
        client: Client,
        router_mode: RouterMode,
    ) -> anyhow::Result<Self> {
        let addressed = addressed_router(&client.endpoint).await?;

        let occupancy_state = if router_mode.requires_occupancy() {
            Some(get_or_create_routing_occupancy_state(&client.endpoint).await)
        } else {
            None
        };

        // Type-erase to the seam so discovery-removal cleanup runs through it.
        let addressed: Arc<dyn StreamingDispatch<T, U>> = addressed;
        spawn_instance_removal_watcher(
            client.endpoint.clone(),
            addressed.clone(),
            client.endpoint.drt().primary_token(),
        );
        let (round_robin_picker, random_picker, picker) = route_pickers(router_mode);

        Ok(PushRouter {
            client,
            addressed,
            router_mode,
            picker,
            round_robin_picker,
            random_picker,
            fault_detection_enabled: false,
            response_timeout: response_inactivity_timeout(),
            occupancy_state,
            multimodal_cache_indexer: None,
            multimodal_cache_key_extractor: None,
            _phantom: PhantomData,
        })
    }

    /// Create a new PushRouter with an optional worker load monitor.
    ///
    /// The rejection path is gated by `fault_detection_enabled` (true here);
    /// durable overload detection is driven by the monitor through
    /// `client.set_overloaded_instances(...)`. Worker responses can also create
    /// bounded request-path overload leases.
    pub async fn from_client_with_monitor(
        client: Client,
        router_mode: RouterMode,
        worker_monitor: Option<Arc<dyn WorkerLoadMonitor>>,
    ) -> anyhow::Result<Self> {
        Self::from_client_with_state(client, router_mode, worker_monitor, None, None).await
    }

    /// Create a new PushRouter with optional load monitoring and multimodal cache indexing.
    pub async fn from_client_with_state(
        client: Client,
        router_mode: RouterMode,
        worker_monitor: Option<Arc<dyn WorkerLoadMonitor>>,
        multimodal_cache_indexer: Option<Arc<dyn MultimodalCacheIndex>>,
        multimodal_cache_key_extractor: Option<MultimodalCacheKeyExtractor<T>>,
    ) -> anyhow::Result<Self> {
        let addressed = addressed_router(&client.endpoint).await?;

        // Start worker monitor if provided and in dynamic mode
        if let Some(monitor) = worker_monitor.as_ref() {
            monitor.start_monitoring().await?;
        }

        let occupancy_state = if router_mode.requires_occupancy() {
            Some(get_or_create_routing_occupancy_state(&client.endpoint).await)
        } else {
            None
        };

        // Type-erase to the seam so discovery-removal cleanup runs through it.
        let addressed: Arc<dyn StreamingDispatch<T, U>> = addressed;
        spawn_instance_removal_watcher(
            client.endpoint.clone(),
            addressed.clone(),
            client.endpoint.drt().primary_token(),
        );

        // Drop stale cache-index entries when workers leave discovery.
        if let Some(indexer) = multimodal_cache_indexer.clone() {
            spawn_multimodal_cache_cleanup_watcher(
                client.endpoint.clone(),
                indexer,
                client.endpoint.drt().primary_token(),
            );
        }
        let (round_robin_picker, random_picker, picker) = route_pickers(router_mode);

        let router = PushRouter {
            client,
            addressed,
            router_mode,
            picker,
            round_robin_picker,
            random_picker,
            fault_detection_enabled: true,
            response_timeout: response_inactivity_timeout(),
            occupancy_state,
            multimodal_cache_indexer,
            multimodal_cache_key_extractor,
            _phantom: PhantomData,
        };

        Ok(router)
    }

    /// Like the other constructors but with a caller-supplied [`StreamingDispatch`]
    /// as the final hop. Fault detection is on, so the dispatch's `ErrorType`
    /// mapping drives report-down / overload / migration as usual.
    ///
    /// Wires frontend-local occupancy only — no `WorkerLoadMonitor` and no
    /// multimodal cache indexer, so `RouterMode::DeviceAwareWeighted` is
    /// non-functional; a caller needing those must extend it.
    pub async fn from_client_with_dispatch(
        client: Client,
        router_mode: RouterMode,
        dispatch: Arc<dyn StreamingDispatch<T, U>>,
    ) -> anyhow::Result<Self> {
        let occupancy_state = if router_mode.requires_occupancy() {
            Some(get_or_create_routing_occupancy_state(&client.endpoint).await)
        } else {
            None
        };

        spawn_instance_removal_watcher(
            client.endpoint.clone(),
            dispatch.clone(),
            client.endpoint.drt().primary_token(),
        );
        let (round_robin_picker, random_picker, picker) = route_pickers(router_mode);

        Ok(PushRouter {
            client,
            addressed: dispatch,
            router_mode,
            picker,
            round_robin_picker,
            random_picker,
            fault_detection_enabled: true,
            response_timeout: response_inactivity_timeout(),
            occupancy_state,
            multimodal_cache_indexer: None,
            multimodal_cache_key_extractor: None,
            _phantom: PhantomData,
        })
    }

    /// `ResourceExhausted` when workers are routable but all overloaded;
    /// `Unavailable` when no routable workers exist.
    fn empty_free_pool_error(&self, routing_instances: &RoutingInstances) -> anyhow::Error {
        if !routing_instances.routable_ids().is_empty() {
            let cause = PipelineError::ServiceOverloaded(
                "All workers are busy, please retry later".to_string(),
            );
            return DynamoError::builder()
                .error_type(ErrorType::ResourceExhausted)
                .message("All workers are busy, please retry later")
                .cause(cause)
                .build()
                .into();
        }
        DynamoError::builder()
            .error_type(ErrorType::Unavailable)
            .message(format!(
                "No workers available for endpoint {}",
                self.client.endpoint.id()
            ))
            .build()
            .into()
    }

    fn picker(&self) -> anyhow::Result<&RoutePicker> {
        self.picker.as_deref().ok_or_else(|| {
            anyhow::anyhow!(
                "{:?} routing does not use a worker picker",
                self.router_mode
            )
        })
    }

    fn select_untracked_worker(&self, picker: &RoutePicker) -> anyhow::Result<(u64, usize)> {
        let routing_instances = self.client.routing_instances();
        let candidates = routing_instances.free_ids();
        let decision = picker
            .select(
                CandidateView::Workers(candidates),
                RouteContext::default(),
                |_| 0,
            )
            .ok_or_else(|| self.empty_free_pool_error(&routing_instances))?;
        Ok((decision.target.worker_id, candidates.len()))
    }

    /// Snapshot workers currently eligible for a new routing decision.
    ///
    /// Selection stays outside `PushRouter`; this is the discovery/admission
    /// boundary used by routing hosts that own their policy lifecycle.
    pub fn selectable_worker_ids(&self) -> anyhow::Result<Vec<u64>> {
        self.with_selectable_worker_ids(<[u64]>::to_vec)
    }

    /// Borrow workers currently eligible for a new routing decision.
    pub fn with_selectable_worker_ids<R>(
        &self,
        select: impl FnOnce(&[u64]) -> R,
    ) -> anyhow::Result<R> {
        let routing_instances = self.client.routing_instances();
        if routing_instances.free_ids().is_empty() {
            return Err(self.empty_free_pool_error(&routing_instances));
        }
        Ok(select(routing_instances.free_ids()))
    }

    /// Shared O(1) occupancy capability for load-aware routing hosts.
    pub fn routing_occupancy_state(&self) -> Option<Arc<RoutingOccupancyState>> {
        self.occupancy_state.clone()
    }

    /// Select a DeviceAware worker while leaving request-lifecycle cleanup to
    /// the routing host. Request and device metadata is prepared before the
    /// occupancy admission lock is acquired.
    pub fn select_device_aware_and_reserve(
        &self,
        request: &T,
        pinned_worker: Option<u64>,
    ) -> anyhow::Result<DeviceAwareSelection> {
        anyhow::ensure!(
            self.router_mode == RouterMode::DeviceAwareWeighted,
            "{:?} routing is not DeviceAwareWeighted",
            self.router_mode
        );

        let state = self.occupancy_state()?;
        if let Some(worker_id) = pinned_worker {
            self.ensure_routable(worker_id)?;
            let selection = self.device_aware_candidates(request, &[worker_id]);
            let is_cpu = selection
                .candidates
                .first()
                .is_some_and(|candidate| candidate.device == RouteDevice::Cpu);
            let reservation = state.reserve(worker_id);
            let load = reservation.load();
            return Ok(DeviceAwareSelection {
                worker_id,
                candidate_count: 1,
                load,
                is_cpu,
                embedding_cache_hit: selection.embedding_cache_hit,
                request_cache_keys: selection.request_cache_keys,
                reservation: Some(reservation),
            });
        }

        let routing_instances = self.client.routing_instances();
        let instance_ids = routing_instances.free_ids();
        if instance_ids.is_empty() {
            return Err(self.empty_free_pool_error(&routing_instances));
        }
        let selection = self.device_aware_candidates(request, instance_ids);
        let (decision, counter) = state
            .select_and_admit(
                self.picker()?,
                CandidateView::DeviceAware(&selection.candidates),
                selection.context,
            )
            .ok_or_else(|| self.empty_free_pool_error(&routing_instances))?;
        let worker_id = decision.target.worker_id;
        let is_cpu = selection.candidates.iter().any(|candidate| {
            candidate.target.worker_id == worker_id && candidate.device == RouteDevice::Cpu
        });
        let reservation = counter
            .map(|counter| OccupancyReservation::from_counter(state.clone(), worker_id, counter));

        Ok(DeviceAwareSelection {
            worker_id,
            candidate_count: selection.candidates.len(),
            load: state.load(worker_id),
            is_cpu,
            embedding_cache_hit: selection.embedding_cache_hit,
            request_cache_keys: selection.request_cache_keys,
            reservation,
        })
    }

    /// Reject an exact target that local fault detection has removed from routing.
    pub fn ensure_routable(&self, instance_id: u64) -> anyhow::Result<()> {
        if self
            .client
            .routing_instances()
            .routable_ids()
            .contains(&instance_id)
        {
            return Ok(());
        }
        anyhow::bail!(
            "instance_id={instance_id} not found for endpoint {}",
            self.client.endpoint.id()
        )
    }

    fn ensure_discovered_for_dispatch(&self, instance_id: u64) -> anyhow::Result<()> {
        if self.client.is_instance_live(instance_id) {
            return Ok(());
        }
        Err(DynamoError::builder()
            .error_type(ErrorType::CannotConnect)
            .message(format!(
                "instance_id={instance_id} not found for endpoint {}",
                self.client.endpoint.id()
            ))
            .build()
            .into())
    }

    /// Issue a request to the next available instance in a round-robin fashion
    pub async fn round_robin(&self, request: SingleIn<T>) -> anyhow::Result<ManyOut<U>> {
        self.round_robin_prepared(request, |_, _| Ok(()))
            .await
            .map(|(_, stream)| stream)
    }

    async fn round_robin_prepared<M, F>(
        &self,
        request: SingleIn<T>,
        prepare: F,
    ) -> anyhow::Result<(M, ManyOut<U>)>
    where
        F: FnOnce(&mut T, u64) -> anyhow::Result<M>,
    {
        let (instance_id, candidate_count) =
            self.select_untracked_worker(self.round_robin_picker.as_ref())?;
        tracing::info!(
            router_mode = "round-robin",
            worker_id = instance_id,
            candidate_count,
            "Selected worker"
        );

        self.dispatch_selected(instance_id, request, None, prepare)
            .await
    }

    /// Issue a request to a random endpoint
    pub async fn random(&self, request: SingleIn<T>) -> anyhow::Result<ManyOut<U>> {
        self.random_prepared(request, |_, _| Ok(()))
            .await
            .map(|(_, stream)| stream)
    }

    async fn random_prepared<M, F>(
        &self,
        request: SingleIn<T>,
        prepare: F,
    ) -> anyhow::Result<(M, ManyOut<U>)>
    where
        F: FnOnce(&mut T, u64) -> anyhow::Result<M>,
    {
        let (instance_id, candidate_count) =
            self.select_untracked_worker(self.random_picker.as_ref())?;
        tracing::info!(
            router_mode = "random",
            worker_id = instance_id,
            candidate_count,
            "Selected worker"
        );

        self.dispatch_selected(instance_id, request, None, prepare)
            .await
    }

    /// Issue a request using power-of-two-choices: pick 2 random healthy workers,
    /// route to the one with fewer in-flight requests.
    pub async fn power_of_two_choices(&self, request: SingleIn<T>) -> anyhow::Result<ManyOut<U>> {
        self.power_of_two_choices_prepared(request, |_, _| Ok(()))
            .await
            .map(|(_, stream)| stream)
    }

    async fn power_of_two_choices_prepared<M, F>(
        &self,
        request: SingleIn<T>,
        prepare: F,
    ) -> anyhow::Result<(M, ManyOut<U>)>
    where
        F: FnOnce(&mut T, u64) -> anyhow::Result<M>,
    {
        let state = self.occupancy_state()?;
        let (instance_id, counter, candidate_count) = {
            let routing_instances = self.client.routing_instances();
            let candidates = routing_instances.free_ids();
            let (decision, counter) = state
                .select_and_admit(
                    self.picker()?,
                    CandidateView::Workers(candidates),
                    RouteContext::default(),
                )
                .ok_or_else(|| self.empty_free_pool_error(&routing_instances))?;
            (
                decision.target.worker_id,
                counter.expect("P2C selection always requests occupancy admission"),
                candidates.len(),
            )
        };
        tracing::info!(
            router_mode = "power-of-two-choices",
            worker_id = instance_id,
            candidate_count,
            load = state.load(instance_id),
            "Selected worker"
        );
        let permit = OccupancyPermit::from_counter(state, instance_id, counter);
        self.dispatch_selected(instance_id, request, Some(permit), prepare)
            .await
    }

    /// Issue a request to exactly one endpoint without transport fallback.
    pub async fn direct(
        &self,
        request: SingleIn<T>,
        instance_id: u64,
    ) -> anyhow::Result<ManyOut<U>> {
        tracing::info!(
            router_mode = "direct",
            worker_id = instance_id,
            "Selected worker"
        );
        self.generate_with_fault_detection(instance_id, request, TransportFallback::Deny)
            .await
    }

    /// Dispatch to a selected endpoint with transport fallback.
    ///
    /// Unlike [`Self::direct`], if the selected instance disappears between selection and
    /// dispatch, this method may reselect another worker. When `allowed_fallback` is `Some`,
    /// reselection is constrained to that set; callers that pre-narrowed the candidates (e.g.
    /// LoRA replica-set filtering) use it to prevent fallback to an arbitrary worker.
    pub async fn direct_within(
        &self,
        request: SingleIn<T>,
        instance_id: u64,
        allowed_fallback: Option<&HashSet<u64>>,
    ) -> anyhow::Result<ManyOut<U>> {
        self.direct_within_prepared(request, instance_id, allowed_fallback, |_, _| Ok(()))
            .await
            .map(|(_, stream)| stream)
    }

    /// Like [`Self::direct_within`], but prepares the request after transport resolution and
    /// returns the preparation metadata alongside the response stream.
    pub async fn direct_within_prepared<M, F>(
        &self,
        request: SingleIn<T>,
        instance_id: u64,
        allowed_fallback: Option<&HashSet<u64>>,
        prepare: F,
    ) -> anyhow::Result<(M, ManyOut<U>)>
    where
        F: FnOnce(&mut T, u64) -> anyhow::Result<M>,
    {
        let fallback = allowed_fallback
            .map(TransportFallback::Within)
            .unwrap_or(TransportFallback::Allow);
        self.direct_prepared_with_fallback(instance_id, request, fallback, prepare)
            .await
    }

    /// Dispatch a worker already selected by a routing host.
    ///
    /// `PushRouter` may still resolve transport fallback if the selected worker
    /// disappears before dispatch. The callback receives that final worker so
    /// the caller can retarget its own reservation and telemetry.
    pub async fn dispatch_preselected_prepared<M, F>(
        &self,
        request: SingleIn<T>,
        instance_id: u64,
        prepare: F,
    ) -> anyhow::Result<(M, ManyOut<U>)>
    where
        F: FnOnce(&mut T, u64) -> anyhow::Result<M>,
    {
        self.dispatch_preselected_with_fallback(
            instance_id,
            request,
            TransportFallback::Allow,
            prepare,
        )
        .await
    }

    async fn direct_prepared_with_fallback<M, F>(
        &self,
        instance_id: u64,
        request: SingleIn<T>,
        fallback: TransportFallback<'_>,
        prepare: F,
    ) -> anyhow::Result<(M, ManyOut<U>)>
    where
        F: FnOnce(&mut T, u64) -> anyhow::Result<M>,
    {
        // Fallback resolution owns the unavailable-target decision. Allowing it
        // to see a worker already absent from discovery preserves Direct's
        // historical standalone fallback; `Deny` remains exact-target behavior.
        let selected_is_discovered = self
            .client
            .instances()
            .iter()
            .any(|instance| instance.instance_id == instance_id);
        if !selected_is_discovered {
            let fallback_is_available = self
                .client
                .routing_instances()
                .free_ids()
                .iter()
                .copied()
                .any(|candidate| {
                    candidate != instance_id
                        && match fallback {
                            TransportFallback::Allow => true,
                            TransportFallback::Deny => false,
                            TransportFallback::Within(allowed) => allowed.contains(&candidate),
                        }
                });
            if !fallback_is_available {
                self.ensure_discovered_for_dispatch(instance_id)?;
            }
        }
        let ((metadata, resolved_instance_id), response_stream) = self
            .generate_with_fault_detection_prepared(
                instance_id,
                request,
                fallback,
                |request, resolved_instance_id| {
                    prepare(request, resolved_instance_id)
                        .map(|metadata| (metadata, resolved_instance_id))
                },
            )
            .await?;
        tracing::info!(
            router_mode = self.router_mode.telemetry_label(),
            worker_id = resolved_instance_id,
            "Selected worker"
        );
        Ok((metadata, response_stream))
    }

    async fn dispatch_preselected_with_fallback<M, F>(
        &self,
        instance_id: u64,
        request: SingleIn<T>,
        fallback: TransportFallback<'_>,
        prepare: F,
    ) -> anyhow::Result<(M, ManyOut<U>)>
    where
        F: FnOnce(&mut T, u64) -> anyhow::Result<M>,
    {
        // Fallback-enabled dispatch still honors a selected worker while it remains in
        // discovery. Local inhibition only filters worker selection owned by this router;
        // fallback is considered only if the selected worker disappears after this check.
        self.ensure_discovered_for_dispatch(instance_id)?;

        self.generate_with_fault_detection_prepared(instance_id, request, fallback, prepare)
            .await
    }

    /// Dispatch to exactly one worker without transport fallback.
    ///
    /// The worker is revalidated against the latest discovery and overload
    /// state immediately before dispatch.
    pub async fn dispatch_exact(
        &self,
        request: SingleIn<T>,
        instance_id: u64,
    ) -> anyhow::Result<ManyOut<U>> {
        self.generate_with_fault_detection(instance_id, request, TransportFallback::Deny)
            .await
    }

    /// Dispatch exactly to a worker whose KV selection step already performed
    /// overload admission.
    ///
    /// Discovery and fault detection are still enforced. The shared client
    /// overload state is not rechecked because admission may synchronously
    /// publish this request's own load before dispatch begins.
    pub async fn dispatch_kv_admitted(
        &self,
        request: SingleIn<T>,
        instance_id: u64,
    ) -> anyhow::Result<ManyOut<U>> {
        if !self.router_mode.is_kv_routing() {
            anyhow::bail!("admitted dispatch is only valid in KV routing mode");
        }
        self.generate_with_fault_detection_inner(
            instance_id,
            request,
            TransportFallback::Deny,
            OverloadCheck::AlreadyAdmitted,
        )
        .await
    }

    /// Select and book one worker, prepare the request for that exact worker,
    /// then dispatch without reselection or transport fallback.
    pub async fn select_and_dispatch_exact<M, F>(
        &self,
        mut request: SingleIn<T>,
        pinned_worker: Option<u64>,
        prepare: F,
    ) -> anyhow::Result<(M, ManyOut<U>)>
    where
        F: FnOnce(&mut T, u64) -> anyhow::Result<M>,
    {
        let (instance_id, permit) = self
            .select_exact_target(request.content(), pinned_worker)
            .await?;
        let metadata = prepare(&mut request, instance_id)?;
        let stream = self.dispatch_exact(request, instance_id).await?;
        let stream = match permit {
            Some(permit) => permit.into_tracked_stream(stream),
            None => stream,
        };
        Ok((metadata, stream))
    }

    /// Select a worker using the configured routing mode, prepare the request with the worker
    /// that survives transport resolution, then dispatch with normal fallback behavior.
    pub async fn select_and_dispatch<M, F>(
        &self,
        request: SingleIn<T>,
        prepare: F,
    ) -> anyhow::Result<(M, ManyOut<U>)>
    where
        F: FnOnce(&mut T, u64) -> anyhow::Result<M>,
    {
        match self.router_mode {
            RouterMode::Random => self.random_prepared(request, prepare).await,
            RouterMode::RoundRobin => self.round_robin_prepared(request, prepare).await,
            RouterMode::PowerOfTwoChoices => {
                self.power_of_two_choices_prepared(request, prepare).await
            }
            RouterMode::LeastLoaded => self.least_loaded_prepared(request, prepare).await,
            RouterMode::DeviceAwareWeighted => {
                self.device_aware_weighted_prepared(request, prepare).await
            }
            RouterMode::KV => anyhow::bail!("KV routing should not call select_and_dispatch"),
            RouterMode::Direct => anyhow::bail!(
                "Direct routing should use direct_within_prepared instead of select_and_dispatch"
            ),
        }
    }

    async fn dispatch_selected<M, F>(
        &self,
        instance_id: u64,
        request: SingleIn<T>,
        mut permit: Option<OccupancyPermit>,
        prepare: F,
    ) -> anyhow::Result<(M, ManyOut<U>)>
    where
        F: FnOnce(&mut T, u64) -> anyhow::Result<M>,
    {
        let (metadata, stream) = self
            .generate_with_fault_detection_prepared(
                instance_id,
                request,
                TransportFallback::Allow,
                |request, resolved_instance_id| {
                    if let Some(permit) = permit.as_mut() {
                        permit.retarget(resolved_instance_id);
                    }
                    prepare(request, resolved_instance_id)
                },
            )
            .await?;
        let stream = match permit {
            Some(permit) => permit.into_tracked_stream(stream),
            None => stream,
        };
        Ok((metadata, stream))
    }

    /// Issue a request using device-aware weighted routing.
    ///
    /// Instances are partitioned by device type (CPU vs non-CPU), then the router
    /// applies a budget policy and selects the least-loaded instance within the
    /// chosen group.
    ///
    /// If only one device class exists (all CPU or all non-CPU), this naturally
    /// degenerates to least-loaded routing over the available instances.
    pub async fn device_aware_weighted(&self, request: SingleIn<T>) -> anyhow::Result<ManyOut<U>> {
        self.device_aware_weighted_prepared(request, |_, _| Ok(()))
            .await
            .map(|(_, stream)| stream)
    }

    async fn device_aware_weighted_prepared<M, F>(
        &self,
        request: SingleIn<T>,
        prepare: F,
    ) -> anyhow::Result<(M, ManyOut<U>)>
    where
        F: FnOnce(&mut T, u64) -> anyhow::Result<M>,
    {
        let state = self.occupancy_state()?;
        let routing_instances = self.client.routing_instances();
        let instance_ids = routing_instances.free_ids();

        if instance_ids.is_empty() {
            return Err(self.empty_free_pool_error(&routing_instances));
        }

        // Apply a unified policy for all endpoints.
        let endpoint_id = self.client.endpoint.id();

        let selection = self.device_aware_candidates(request.content(), instance_ids);

        // Only full cache hits bypass weighted accounting; partial hits still follow the
        // device-aware ratio because some image encoding remains for this request.
        let (decision, counter) = state
            .select_and_admit(
                self.picker()?,
                CandidateView::DeviceAware(&selection.candidates),
                selection.context,
            )
            .ok_or_else(|| self.empty_free_pool_error(&routing_instances))?;
        let instance_id = decision.target.worker_id;
        let permit = counter
            .map(|counter| OccupancyPermit::from_counter(state.clone(), instance_id, counter));
        let is_cpu = selection.candidates.iter().any(|candidate| {
            candidate.target.worker_id == instance_id && candidate.device == RouteDevice::Cpu
        });
        tracing::info!(
            router_mode = "device-aware-weighted",
            worker_id = instance_id,
            candidate_count = selection.candidates.len(),
            load = state.load(instance_id),
            endpoint = %endpoint_id,
            is_cpu,
            embedding_cache_hit = selection.embedding_cache_hit,
            request_cache_keys = selection.request_cache_keys,
            "Selected worker"
        );

        self.dispatch_selected(instance_id, request, permit, prepare)
            .await
    }

    fn device_aware_candidates(&self, request: &T, instance_ids: &[u64]) -> DeviceAwareCandidates {
        let device_type_map = self
            .client
            .instances()
            .iter()
            .map(|instance| {
                let device = if matches!(instance.device_type, Some(DeviceType::Cpu)) {
                    RouteDevice::Cpu
                } else {
                    RouteDevice::Accelerator
                };
                (instance.instance_id, device)
            })
            .collect::<HashMap<_, _>>();
        let cuda_to_cpu_ratio = std::env::var("DYN_ENCODER_CUDA_TO_CPU_RATIO")
            .ok()
            .and_then(|value| value.parse::<usize>().ok())
            .filter(|value| *value >= 1)
            .unwrap_or(8);

        let (request_cache_keys, cache_matched_candidates) =
            if let (Some(indexer), Some(extractor)) = (
                self.multimodal_cache_indexer.as_ref(),
                self.multimodal_cache_key_extractor.as_ref(),
            ) {
                let request_cache_keys = extractor(request);
                let matched = if request_cache_keys.is_empty() {
                    Vec::new()
                } else {
                    let mut matched = indexer.workers_with_cache_key_hits(&request_cache_keys);
                    matched.retain(|(id, _)| instance_ids.contains(id));
                    matched
                };
                (request_cache_keys, matched)
            } else {
                (Vec::new(), Vec::new())
            };

        let embedding_cache_hit = !cache_matched_candidates.is_empty();
        let cache_hits = cache_matched_candidates
            .into_iter()
            .collect::<HashMap<_, _>>();
        let request_cache_key_count = request_cache_keys
            .iter()
            .collect::<std::collections::HashSet<_>>()
            .len();
        let candidates = instance_ids
            .iter()
            .map(|worker_id| RouteCandidate {
                target: RouteTarget::worker(*worker_id),
                device: device_type_map.get(worker_id).copied().unwrap_or_default(),
                cache_hits: cache_hits.get(worker_id).copied().unwrap_or_default(),
            })
            .collect::<Vec<_>>();

        DeviceAwareCandidates {
            candidates,
            context: RouteContext {
                required_cache_hits: request_cache_key_count,
                non_cpu_to_cpu_ratio: cuda_to_cpu_ratio,
            },
            embedding_cache_hit,
            request_cache_keys: request_cache_keys.len(),
        }
    }

    /// Issue a request to the instance with the fewest active connections.
    pub async fn least_loaded(&self, request: SingleIn<T>) -> anyhow::Result<ManyOut<U>> {
        self.least_loaded_prepared(request, |_, _| Ok(()))
            .await
            .map(|(_, stream)| stream)
    }

    async fn least_loaded_prepared<M, F>(
        &self,
        request: SingleIn<T>,
        prepare: F,
    ) -> anyhow::Result<(M, ManyOut<U>)>
    where
        F: FnOnce(&mut T, u64) -> anyhow::Result<M>,
    {
        let state = self.occupancy_state()?;
        let routing_instances = self.client.routing_instances();
        let instance_ids = routing_instances.free_ids();
        let (decision, counter) = state
            .select_and_admit(
                self.picker()?,
                CandidateView::Workers(instance_ids),
                RouteContext::default(),
            )
            .ok_or_else(|| self.empty_free_pool_error(&routing_instances))?;
        let instance_id = decision.target.worker_id;
        let permit = OccupancyPermit::from_counter(
            state.clone(),
            instance_id,
            counter.expect("least-loaded selection always requests occupancy admission"),
        );
        tracing::info!(
            router_mode = "least-loaded",
            worker_id = instance_id,
            candidate_count = instance_ids.len(),
            load = state.load(instance_id),
            "Selected worker"
        );

        self.dispatch_selected(instance_id, request, Some(permit), prepare)
            .await
    }

    /// Select the next worker according to the routing mode.
    /// Increments round-robin counter if applicable.
    /// Returns None for modes that require request lifecycle tracking or explicit routing hints.
    pub fn select_next_worker(&self) -> Option<u64> {
        let routing_instances = self.client.routing_instances();
        match self.router_mode {
            RouterMode::RoundRobin | RouterMode::Random => self
                .picker
                .as_deref()?
                .select(
                    CandidateView::Workers(routing_instances.free_ids()),
                    RouteContext::default(),
                    |_| 0,
                )
                .map(|decision| decision.target.worker_id),
            RouterMode::PowerOfTwoChoices
            | RouterMode::Direct
            | RouterMode::LeastLoaded
            | RouterMode::DeviceAwareWeighted => None,
            RouterMode::KV => {
                panic!(
                    "select_next_worker should not be called for {:?} routing mode",
                    self.router_mode
                )
            }
        }
    }

    /// Peek the next worker according to the routing mode without incrementing the counter.
    /// Useful for checking if a worker is suitable before committing to it.
    ///
    /// `None` for [`RouterMode::Direct`] (caller-supplied routing); panics for
    /// [`RouterMode::KV`], which selects via `kv_chooser::find_best_match`.
    pub fn peek_next_worker(&self) -> Option<u64> {
        // Select among free (admission-eligible) workers — see select_next_worker
        // for the per-mode selection rationale.
        let routing_instances = self.client.routing_instances();
        let instance_ids = routing_instances.free_ids();
        if instance_ids.is_empty() {
            return None;
        }

        match self.router_mode {
            RouterMode::RoundRobin | RouterMode::Random => self
                .picker
                .as_deref()?
                .peek(
                    CandidateView::Workers(instance_ids),
                    RouteContext::default(),
                    |_| 0,
                )
                .map(|decision| decision.target.worker_id),
            RouterMode::LeastLoaded | RouterMode::PowerOfTwoChoices => self
                .occupancy_state
                .as_deref()?
                .peek(
                    self.picker.as_deref()?,
                    CandidateView::Workers(instance_ids),
                    RouteContext::default(),
                )
                .map(|decision| decision.target.worker_id),
            RouterMode::DeviceAwareWeighted => {
                let state = self.occupancy_state.as_deref()?;
                let device_type_map: HashMap<u64, Option<DeviceType>> = self
                    .client
                    .instances()
                    .iter()
                    .map(|instance| (instance.instance_id, instance.device_type.clone()))
                    .collect();
                let cuda_to_cpu_ratio = std::env::var("DYN_ENCODER_CUDA_TO_CPU_RATIO")
                    .ok()
                    .and_then(|value| value.parse::<usize>().ok())
                    .filter(|value| *value >= 1)
                    .unwrap_or(8);
                let candidates = instance_ids
                    .iter()
                    .map(|worker_id| RouteCandidate {
                        target: RouteTarget::worker(*worker_id),
                        device: if matches!(
                            device_type_map.get(worker_id),
                            Some(Some(DeviceType::Cpu))
                        ) {
                            RouteDevice::Cpu
                        } else {
                            RouteDevice::Accelerator
                        },
                        cache_hits: 0,
                    })
                    .collect::<Vec<_>>();
                state
                    .peek(
                        self.picker.as_deref()?,
                        CandidateView::DeviceAware(&candidates),
                        RouteContext {
                            required_cache_hits: 0,
                            non_cpu_to_cpu_ratio: cuda_to_cpu_ratio,
                        },
                    )
                    .map(|decision| decision.target.worker_id)
            }
            RouterMode::Direct => None,
            RouterMode::KV => {
                panic!(
                    "peek_next_worker should not be called for {:?} routing mode",
                    self.router_mode
                )
            }
        }
    }

    #[cfg(any(test, feature = "testing"))]
    #[doc(hidden)]
    pub fn occupancy_for_test(&self, worker_id: u64) -> u64 {
        self.occupancy_state
            .as_deref()
            .map(|state| state.load(worker_id))
            .unwrap_or(0)
    }

    async fn select_exact_target(
        &self,
        request: &T,
        pinned_worker: Option<u64>,
    ) -> anyhow::Result<(u64, Option<OccupancyPermit>)> {
        if let Some(instance_id) = pinned_worker {
            let routing_instances = self.client.routing_instances();
            if !routing_instances.routable_ids().contains(&instance_id) {
                return Err(anyhow::anyhow!(
                    "instance_id={instance_id} not found for endpoint {}",
                    self.client.endpoint.id()
                ));
            }
            let permit = match self.router_mode {
                RouterMode::LeastLoaded
                | RouterMode::PowerOfTwoChoices
                | RouterMode::DeviceAwareWeighted => {
                    let state = self.occupancy_state()?;
                    Some(OccupancyPermit::acquire(state, instance_id))
                }
                RouterMode::RoundRobin
                | RouterMode::Random
                | RouterMode::Direct
                | RouterMode::KV => None,
            };
            return Ok((instance_id, permit));
        }

        match self.router_mode {
            RouterMode::LeastLoaded
            | RouterMode::PowerOfTwoChoices
            | RouterMode::DeviceAwareWeighted => {
                let state = self.occupancy_state()?;
                let routing_instances = self.client.routing_instances();
                let instance_ids = routing_instances.free_ids();
                if instance_ids.is_empty() {
                    return Err(self.empty_free_pool_error(&routing_instances));
                }

                let (decision, counter) = match self.router_mode {
                    RouterMode::LeastLoaded | RouterMode::PowerOfTwoChoices => state
                        .select_and_admit(
                            self.picker()?,
                            CandidateView::Workers(instance_ids),
                            RouteContext::default(),
                        )
                        .ok_or_else(|| self.empty_free_pool_error(&routing_instances))?,
                    RouterMode::DeviceAwareWeighted => {
                        let selection = self.device_aware_candidates(request, instance_ids);
                        state
                            .select_and_admit(
                                self.picker()?,
                                CandidateView::DeviceAware(&selection.candidates),
                                selection.context,
                            )
                            .ok_or_else(|| self.empty_free_pool_error(&routing_instances))?
                    }
                    _ => unreachable!(),
                };
                let instance_id = decision.target.worker_id;
                let permit = counter
                    .map(|counter| OccupancyPermit::from_counter(state, instance_id, counter));
                Ok((instance_id, permit))
            }
            RouterMode::RoundRobin => self
                .select_untracked_worker(self.round_robin_picker.as_ref())
                .map(|(instance_id, _)| (instance_id, None)),
            RouterMode::Random => self
                .select_untracked_worker(self.random_picker.as_ref())
                .map(|(instance_id, _)| (instance_id, None)),
            RouterMode::Direct => Err(anyhow::anyhow!(
                "Worker ID required for exact dispatch in Direct routing mode"
            )),
            RouterMode::KV => Err(anyhow::anyhow!(
                "select_and_dispatch_exact cannot select workers in KV routing mode"
            )),
        }
    }

    fn occupancy_state(&self) -> anyhow::Result<Arc<RoutingOccupancyState>> {
        self.occupancy_state.clone().ok_or_else(|| {
            anyhow::anyhow!(
                "routing occupancy state not initialized for endpoint {}",
                self.client.endpoint.id()
            )
        })
    }

    /*
    pub async fn r#static(&self, request: SingleIn<T>) -> anyhow::Result<ManyOut<U>> {
        let subject = self.client.endpoint.subject();
        tracing::debug!("static got subject: {subject}");
        let request = request.map(|req| AddressedRequest::new(req, subject));
        tracing::debug!("router generate");
        self.addressed.generate(request).await
    }
    */

    async fn generate_with_fault_detection(
        &self,
        instance_id: u64,
        request: SingleIn<T>,
        fallback: TransportFallback<'_>,
    ) -> anyhow::Result<ManyOut<U>> {
        self.generate_with_fault_detection_inner(
            instance_id,
            request,
            fallback,
            OverloadCheck::Required,
        )
        .await
    }

    async fn generate_with_fault_detection_inner(
        &self,
        instance_id: u64,
        request: SingleIn<T>,
        fallback: TransportFallback<'_>,
        overload_check: OverloadCheck,
    ) -> anyhow::Result<ManyOut<U>> {
        self.generate_with_fault_detection_prepared_inner(
            instance_id,
            request,
            fallback,
            overload_check,
            |_, _| Ok(()),
        )
        .await
        .map(|(_, stream)| stream)
    }

    async fn generate_with_fault_detection_prepared<M, F>(
        &self,
        instance_id: u64,
        request: SingleIn<T>,
        fallback: TransportFallback<'_>,
        prepare: F,
    ) -> anyhow::Result<(M, ManyOut<U>)>
    where
        F: FnOnce(&mut T, u64) -> anyhow::Result<M>,
    {
        self.generate_with_fault_detection_prepared_inner(
            instance_id,
            request,
            fallback,
            OverloadCheck::Required,
            prepare,
        )
        .await
    }

    async fn generate_with_fault_detection_prepared_inner<M, F>(
        &self,
        instance_id: u64,
        mut request: SingleIn<T>,
        fallback: TransportFallback<'_>,
        overload_check: OverloadCheck,
        prepare: F,
    ) -> anyhow::Result<(M, ManyOut<U>)>
    where
        F: FnOnce(&mut T, u64) -> anyhow::Result<M>,
    {
        let route_start = Instant::now();
        let request_id = request.id().to_string();
        let route_trace_context = get_route_trace_context(&request);
        let route_span = if matches!(self.router_mode, RouterMode::KV) {
            tracing::Span::none()
        } else {
            tracing::info_span!(
                target: "request_span",
                "router.route_request",
                otel.kind = "client",
                request_id = %request_id,
                worker_id = tracing::field::Empty,
                router_mode = ?self.router_mode,
                "request.attempt" = tracing::field::Empty,
                "request.outcome" = tracing::field::Empty,
                "migration.is_retry" = tracing::field::Empty,
                "migration.reason" = tracing::field::Empty,
                "migration.from_worker_id" = tracing::field::Empty,
                "migration.tokens_completed" = tracing::field::Empty,
                "cancellation.signal" = tracing::field::Empty,
                "error.type" = tracing::field::Empty,
                otel.status_code = tracing::field::Empty,
                otel.status_description = tracing::field::Empty,
            )
        };
        let (instance_id, address, transport_kind, instance) = match self
            .resolve_transport(instance_id, fallback)
        {
            Ok(resolved) => resolved,
            Err(error) => {
                record_route_span_start(&route_span, route_trace_context.as_deref(), instance_id);
                record_route_error(&route_span, error.as_ref());
                return Err(error);
            }
        };
        record_route_span_start(&route_span, route_trace_context.as_deref(), instance_id);
        if matches!(overload_check, OverloadCheck::Required)
            && let Err(error) = self.check_workers_available(instance_id, &request_id)
        {
            record_route_error(&route_span, error.as_ref());
            return Err(error);
        }

        let metadata = match prepare(&mut request, instance_id) {
            Ok(metadata) => metadata,
            Err(error) => {
                record_route_error(&route_span, error.as_ref());
                return Err(error);
            }
        };
        let request = request.map(|req| AddressedRequest::with_instance(req, address, instance));

        STAGE_DURATION_SECONDS
            .with_label_values(&[STAGE_ROUTE])
            .observe(route_start.elapsed().as_secs_f64());

        let _nvtx_transport = dynamo_nvtx_range!(transport_kind);
        let stream = self
            .addressed
            .generate(request)
            .instrument(route_span.clone())
            .await;
        let stream = self.wrap_with_fault_detection(stream, instance_id, route_span)?;
        Ok((metadata, stream))
    }

    /// Reject early if the selected worker is overloaded and fault detection
    /// is enabled. The request_id is only used for the debug-level "checked
    /// worker overload state" trace; pass an empty string from callers that
    /// don't have one handy.
    fn check_workers_available(&self, instance_id: u64, request_id: &str) -> anyhow::Result<()> {
        if !self.fault_detection_enabled {
            return Ok(());
        }
        let routing_instances = self.client.routing_instances();
        let selected_worker_overloaded = routing_instances.is_overloaded(instance_id);
        let counts = routing_instances.counts();
        if tracing::enabled!(tracing::Level::DEBUG) {
            tracing::debug!(
                request_id,
                instance_id,
                router_mode = ?self.router_mode,
                free_workers = counts.free,
                overloaded_workers = counts.overloaded,
                total_workers = counts.discovered,
                selected_worker_overloaded,
                "checked worker overload state"
            );
        }
        if !selected_worker_overloaded {
            return Ok(());
        }
        tracing::warn!(
            instance_id,
            overloaded_workers = counts.overloaded,
            total_workers = counts.discovered,
            "Rejecting request: selected worker is overloaded"
        );
        let cause = PipelineError::ServiceOverloaded(
            "Selected worker is overloaded, please retry later".into(),
        );
        Err(DynamoError::builder()
            .error_type(ErrorType::WorkerOverloaded)
            .message("Selected worker is overloaded, please retry later")
            .cause(cause)
            .build()
            .into())
    }

    /// Resolve `(instance_id, address, transport_kind_label, Instance)` for
    /// the selected worker. If that worker has disappeared, apply the caller's
    /// fallback policy. `CannotConnect` is returned when fallback is forbidden
    /// or when a selected fallback disappears before its transport can be
    /// resolved.
    fn resolve_transport(
        &self,
        instance_id: u64,
        fallback: TransportFallback<'_>,
    ) -> anyhow::Result<(u64, String, &'static str, Instance)> {
        use crate::component::TransportType;

        let lookup = |id: u64| {
            self.client.instance_by_id(id).map(|instance| {
                let (addr, kind) = match &instance.transport {
                    TransportType::Tcp(tcp_endpoint) => {
                        (tcp_endpoint.clone(), "transport.tcp.request")
                    }
                    TransportType::Nats(subject) => (subject.clone(), "transport.nats.request"),
                };
                (addr, kind, instance)
            })
        };

        if let Some((addr, kind, inst)) = lookup(instance_id) {
            return Ok((instance_id, addr, kind, inst));
        }
        let allowed_fallback = match fallback {
            TransportFallback::Allow => None,
            TransportFallback::Deny => {
                return Err(DynamoError::builder()
                    .error_type(ErrorType::CannotConnect)
                    .message(format!(
                        "instance_id={instance_id} not found for endpoint {}",
                        self.client.endpoint.id()
                    ))
                    .build()
                    .into());
            }
            TransportFallback::Within(allowed) => Some(allowed),
        };

        let routing_instances = self.client.routing_instances();
        let fallback_id = routing_instances.free_ids().iter().copied().find(|&id| {
            id != instance_id && allowed_fallback.is_none_or(|allowed| allowed.contains(&id))
        });
        match fallback_id {
            Some(id) => {
                tracing::warn!(
                    original_instance = instance_id,
                    fallback_instance = id,
                    "Instance disappeared during routing, reselecting"
                );
                let (addr, kind, inst) = lookup(id).ok_or_else(|| {
                    DynamoError::builder()
                        .error_type(ErrorType::CannotConnect)
                        .message(format!(
                            "Fallback instance {} also not found for endpoint {}",
                            id,
                            self.client.endpoint.id()
                        ))
                        .build()
                })?;
                Ok((id, addr, kind, inst))
            }
            // The selected instance vanished from discovery and no permitted
            // fallback is free. Typed rather than a bare `anyhow!` so
            // `error_type_from_chain` (route spans) and the frontend's status
            // mapping see a real category instead of `unknown`/500.
            //
            // Uniformly `Unavailable`, not a pool-state split: reaching here
            // means *this request's* worker is gone, which is a discovery fact,
            // and `transport_resolution_precedes_stale_overload_check` requires
            // that a vanished instance is never redressed as overload. Nor
            // `CannotConnect`, which stays the signature of exact dispatch
            // (`TransportFallback::Deny`) and must remain distinguishable from a
            // fallback-enabled failure.
            None => Err(DynamoError::builder()
                .error_type(ErrorType::Unavailable)
                .message(format!(
                    "Instance {} not found and no other instances available for endpoint {}",
                    instance_id,
                    self.client.endpoint.id()
                ))
                .build()
                .into()),
        }
    }

    /// Wrap a dispatched stream with fault detection + inactivity timeout.
    /// `is_inhibited` errors trigger `report_instance_down`; the timeout
    /// (driven by `DYN_HTTP_BACKEND_STREAM_TIMEOUT_SECS`) yields a synthetic
    /// `ResponseTimeout` and quarantines the worker.
    fn wrap_with_fault_detection(
        &self,
        stream: anyhow::Result<ManyOut<U>>,
        instance_id: u64,
        route_span: tracing::Span,
    ) -> anyhow::Result<ManyOut<U>> {
        let stream = match stream {
            Ok(stream) => stream,
            Err(err) => {
                record_route_error(&route_span, err.as_ref());
                if self.fault_detection_enabled {
                    if is_inhibited(err.as_ref()) {
                        tracing::debug!(
                            "Reporting instance {instance_id} down due to error: {err}"
                        );
                        self.client.report_instance_down(instance_id);
                    } else if match_error_chain(err.as_ref(), &[ErrorType::WorkerOverloaded], &[]) {
                        // Backpressure: the worker said "my queue is full,
                        // retry later". A bounded lease prevents an immediate
                        // retry loop. Fresh monitor data clears the lease early.
                        // Lease expiry permits a recovery probe when load events
                        // remain unchanged. This is not a fault quarantine.
                        tracing::debug!(
                            "Marking instance {instance_id} overloaded due to backpressure: {err}"
                        );
                        self.client.mark_overloaded_immediate(instance_id);
                    }
                }
                return Err(err);
            }
        };

        if !self.fault_detection_enabled {
            return Ok(wrap_route_span(stream, route_span));
        }

        let engine_ctx = stream.context();
        let client = self.client.clone();
        let client_for_timeout = self.client.clone();
        let stream = stream.map(move |res| {
            if let Some(err) = res.err()
                && is_inhibited(&err)
            {
                tracing::debug!(
                    "Reporting instance {instance_id} down due to migratable error: {err}"
                );
                client.report_instance_down(instance_id);
            }
            res
        });

        let stream: Pin<Box<dyn Stream<Item = U> + Send>> =
            if let Some(timeout) = self.response_timeout {
                Box::pin(async_stream::stream! {
                    let mut inner = Box::pin(stream);
                    loop {
                        tokio::select! {
                            biased;
                            item = inner.next() => {
                                match item {
                                    Some(item) => yield item,
                                    None => break,
                                }
                            }
                            _ = tokio::time::sleep(timeout) => {
                                tracing::warn!(
                                    instance_id,
                                    timeout_secs = timeout.as_secs(),
                                    "backend response inactivity timeout — quarantining worker"
                                );
                                client_for_timeout.report_instance_down(instance_id);
                                yield U::from_err(
                                    crate::error::DynamoError::builder()
                                        .error_type(crate::error::ErrorType::ResponseTimeout)
                                        .message("backend response inactivity timeout")
                                        .build()
                                );
                                break;
                            }
                        }
                    }
                })
            } else {
                Box::pin(stream)
            };

        Ok(wrap_route_span(
            ResponseStream::new(stream, engine_ctx),
            route_span,
        ))
    }
}

#[async_trait]
impl<T, U> AsyncEngine<SingleIn<T>, ManyOut<U>, Error> for PushRouter<T, U>
where
    T: Data + Serialize,
    U: Data + for<'de> Deserialize<'de> + MaybeError,
{
    async fn generate(&self, request: SingleIn<T>) -> Result<ManyOut<U>, Error> {
        match self.router_mode {
            RouterMode::Random => self.random(request).await,
            RouterMode::RoundRobin => self.round_robin(request).await,
            RouterMode::PowerOfTwoChoices => self.power_of_two_choices(request).await,
            RouterMode::KV => {
                anyhow::bail!("KV routing should not call generate on PushRouter");
            }
            RouterMode::Direct => {
                anyhow::bail!(
                    "Direct routing should not call generate on PushRouter directly; use RoutingHost"
                );
            }
            RouterMode::LeastLoaded => self.least_loaded(request).await,
            RouterMode::DeviceAwareWeighted => self.device_aware_weighted(request).await,
        }
    }
}

impl<T, U> PushRouter<T, U>
where
    T: Data + Serialize,
    U: Data + for<'de> Deserialize<'de> + MaybeError,
{
    /// Bidirectional sibling of [`Self::generate_with_fault_detection`].
    async fn bidirectional_dispatch(
        &self,
        instance_id: u64,
        input: ManyIn<T>,
    ) -> anyhow::Result<ManyOut<U>> {
        let route_start = Instant::now();
        let request_id = input.context().id().to_string();
        let route_trace_context = get_route_trace_context(&input);
        let route_span = tracing::info_span!(
            target: "request_span",
            "router.route_request_bidirectional",
            otel.kind = "client",
            request_id = %request_id,
            worker_id = tracing::field::Empty,
            router_mode = ?self.router_mode,
            "request.attempt" = tracing::field::Empty,
            "request.outcome" = tracing::field::Empty,
            "migration.is_retry" = tracing::field::Empty,
            "migration.reason" = tracing::field::Empty,
            "migration.from_worker_id" = tracing::field::Empty,
            "migration.tokens_completed" = tracing::field::Empty,
            "cancellation.signal" = tracing::field::Empty,
            "error.type" = tracing::field::Empty,
            otel.status_code = tracing::field::Empty,
            otel.status_description = tracing::field::Empty,
        );

        let (instance_id, address, transport_kind, instance) = match self
            .resolve_transport(instance_id, TransportFallback::Allow)
        {
            Ok(resolved) => resolved,
            Err(error) => {
                record_route_span_start(&route_span, route_trace_context.as_deref(), instance_id);
                record_route_error(&route_span, error.as_ref());
                return Err(error);
            }
        };
        record_route_span_start(&route_span, route_trace_context.as_deref(), instance_id);
        if let Err(error) = self.check_workers_available(instance_id, &request_id) {
            record_route_error(&route_span, error.as_ref());
            return Err(error);
        }

        STAGE_DURATION_SECONDS
            .with_label_values(&[STAGE_ROUTE])
            .observe(route_start.elapsed().as_secs_f64());

        let _nvtx_transport = dynamo_nvtx_range!(transport_kind);
        let stream: anyhow::Result<ManyOut<U>> = self
            .addressed
            .generate_bidirectional(instance, address, input)
            .instrument(route_span.clone())
            .await;
        self.wrap_with_fault_detection(stream, instance_id, route_span)
    }
}

/// Bidirectional `AsyncEngine` impl for streaming-input workloads (e.g. the
/// OpenAI Realtime API). Reserves a sticky worker up front — before any
/// inbound frame is observed — and binds the whole input stream to that
/// worker. KV and Direct modes inherit the same `bail!` invariants as the
/// unary impl.
///
/// **Reserve-before-observe rationale.** The router-mode strategies
/// (`RoundRobin`, `Random`, `PowerOfTwoChoices`, `LeastLoaded`,
/// `DeviceAwareWeighted`) don't depend on frame contents, so selection
/// runs immediately and connection setup proceeds in parallel with the
/// client producing its first frame. A client that connects but never
/// sends one still releases the slot via the response-stream-drop path;
/// the dispatch-side `cancel_both` cleanup covers the early-bail case.
#[async_trait]
impl<T, U> AsyncEngine<ManyIn<T>, ManyOut<U>, Error> for PushRouter<T, U>
where
    T: Data + Serialize,
    U: Data + for<'de> Deserialize<'de> + MaybeError,
{
    async fn generate(&self, input: ManyIn<T>) -> Result<ManyOut<U>, Error> {
        match self.router_mode {
            RouterMode::KV => {
                anyhow::bail!("KV routing should not call generate on PushRouter");
            }
            RouterMode::Direct => {
                anyhow::bail!(
                    "Direct routing should not call generate on PushRouter directly; use RoutingHost"
                );
            }
            // These modes drive `select_next_worker()` to `None` — they rely on
            // the occupancy/load-aware selection the bidirectional path does not
            // wire yet, which would otherwise surface as a misleading "no
            // instances available" error below. Reject them explicitly until
            // bidirectional support lands; tracked in
            // https://github.com/ai-dynamo/dynamo/issues/10320.
            RouterMode::PowerOfTwoChoices
            | RouterMode::LeastLoaded
            | RouterMode::DeviceAwareWeighted => {
                anyhow::bail!(
                    "{:?} routing is not yet supported for bidirectional dispatch",
                    self.router_mode
                );
            }
            RouterMode::RoundRobin | RouterMode::Random => {}
        }

        let instance_id = self
            .select_next_worker()
            .ok_or_else(|| anyhow::anyhow!("no instances available for bidirectional routing"))?;

        self.bidirectional_dispatch(instance_id, input).await
    }
}

struct OccupancyTrackedStream<U: Data + MaybeError> {
    inner: ManyOut<U>,
    reservation: Option<OccupancyReservation>,
}

impl<U: Data + MaybeError> OccupancyTrackedStream<U> {
    fn release(&mut self) {
        drop(self.reservation.take());
    }
}

impl<U: Data + MaybeError> Drop for OccupancyTrackedStream<U> {
    fn drop(&mut self) {
        self.release();
    }
}

impl<U: Data + MaybeError> std::fmt::Debug for OccupancyTrackedStream<U> {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        f.debug_struct("OccupancyTrackedStream")
            .field(
                "instance_id",
                &self
                    .reservation
                    .as_ref()
                    .map(OccupancyReservation::worker_id),
            )
            .finish()
    }
}

impl<U: Data + MaybeError> Stream for OccupancyTrackedStream<U> {
    type Item = U;

    fn poll_next(
        mut self: Pin<&mut Self>,
        cx: &mut std::task::Context<'_>,
    ) -> Poll<Option<Self::Item>> {
        let poll = self.inner.as_mut().poll_next(cx);
        if matches!(&poll, Poll::Ready(None))
            || matches!(&poll, Poll::Ready(Some(item)) if item.err().is_some())
        {
            self.release();
        }
        poll
    }
}

impl<U: Data + MaybeError> AsyncEngineContextProvider for OccupancyTrackedStream<U> {
    fn context(&self) -> Arc<dyn AsyncEngineContext> {
        self.inner.context()
    }
}

impl<U: Data + MaybeError> crate::engine::AsyncEngineStream<U> for OccupancyTrackedStream<U> {}

#[cfg(test)]
mod tests {
    use super::*;
    use std::sync::atomic::Ordering;

    use crate::{
        DistributedRuntime, Runtime,
        distributed::DistributedConfig,
        error::DynamoError,
        pipeline::{
            RequestStream, ResponseStream,
            context::{Context, Controller},
            network::egress::route_span,
        },
    };
    use serde::{Deserialize, Serialize};

    #[derive(Clone, Debug, Deserialize, Serialize)]
    struct TestResponse {
        error: Option<DynamoError>,
    }

    impl MaybeError for TestResponse {
        fn from_err(err: impl std::error::Error + 'static) -> Self {
            Self {
                error: Some(DynamoError::from(
                    Box::new(err) as Box<dyn std::error::Error + 'static>
                )),
            }
        }

        fn err(&self) -> Option<DynamoError> {
            self.error.clone()
        }
    }

    fn assert_cannot_connect(error: &anyhow::Error) {
        assert!(
            match_error_chain(error.as_ref(), &[ErrorType::CannotConnect], &[]),
            "expected CannotConnect error, got: {error}"
        );
        assert!(
            !match_error_chain(error.as_ref(), &[ErrorType::ResourceExhausted], &[]),
            "CannotConnect failure must not be masked as ResourceExhausted: {error}"
        );
    }

    fn assert_not_cannot_connect(error: &anyhow::Error) {
        assert!(
            !match_error_chain(error.as_ref(), &[ErrorType::CannotConnect], &[]),
            "fallback-enabled failure must preserve its existing error semantics: {error}"
        );
    }

    /// The no-permitted-fallback path must carry a typed `Unavailable`, not a
    /// bare `anyhow!`. Asserted through `error_type_from_chain` because that is
    /// what route spans and the frontend's 503 mapping actually call: an
    /// untyped error classifies as `Unknown` there and is exported as
    /// `error.type=unknown` / HTTP 500.
    fn assert_unavailable(error: &anyhow::Error) {
        assert_eq!(
            route_span::error_type_from_chain(error.as_ref()),
            ErrorType::Unavailable,
            "no-permitted-fallback failure must be typed Unavailable, got: {error}"
        );
        assert_not_cannot_connect(error);
    }

    struct StaticMultimodalCacheIndex {
        worker_id: u64,
    }

    impl MultimodalCacheIndex for StaticMultimodalCacheIndex {
        fn workers_with_cache_key_hits(&self, cache_keys: &[String]) -> Vec<(u64, usize)> {
            vec![(self.worker_id, cache_keys.len())]
        }

        fn remove_worker(&self, _worker_id: u64) {}
    }

    #[test]
    fn router_mode_telemetry_labels_are_stable() {
        assert_eq!(RouterMode::RoundRobin.telemetry_label(), "round-robin");
        assert_eq!(RouterMode::Random.telemetry_label(), "random");
        assert_eq!(
            RouterMode::PowerOfTwoChoices.telemetry_label(),
            "power-of-two-choices"
        );
        assert_eq!(RouterMode::KV.telemetry_label(), "kv");
        assert_eq!(RouterMode::Direct.telemetry_label(), "direct");
        assert_eq!(RouterMode::LeastLoaded.telemetry_label(), "least-loaded");
        assert_eq!(
            RouterMode::DeviceAwareWeighted.telemetry_label(),
            "device-aware-weighted"
        );
    }

    #[test]
    fn p2c_selects_lower_load_worker() {
        let state = RoutingOccupancyState::default();
        for _ in 0..10 {
            state.increment(1);
        }
        state.increment(2);

        // With only two workers, p2c_select_from must pick both and choose id=2 (lower load).
        let result = p2c_select_from(&state, &[1, 2]);
        assert_eq!(result, 2);
    }

    #[test]
    fn explicit_static_pickers_keep_policy_specific_state() {
        let (round_robin, random, configured) = route_pickers(RouterMode::KV);
        assert!(configured.is_none());
        assert_eq!(random.policy(), RoutePolicy::Random);

        let candidates = CandidateView::Workers(&[10, 20, 30, 40]);
        random
            .select(candidates, RouteContext::default(), |_| 0)
            .expect("random selection must have a candidate");
        let selected = (0..2)
            .map(|_| {
                round_robin
                    .select(candidates, RouteContext::default(), |_| 0)
                    .expect("round-robin selection must have a candidate")
                    .target
                    .worker_id
            })
            .collect::<Vec<_>>();
        assert_eq!(selected, [10, 20]);
    }

    #[test]
    fn p2c_selects_single_worker() {
        let state = RoutingOccupancyState::default();
        assert_eq!(p2c_select_from(&state, &[42]), 42);
    }

    #[test]
    fn p2c_treats_missing_counts_as_zero() {
        let state = RoutingOccupancyState::default();
        for _ in 0..5 {
            state.increment(1);
        }
        // Worker 2 has no entry — should be treated as 0, so it wins.
        let result = p2c_select_from(&state, &[1, 2]);
        assert_eq!(result, 2);
    }

    #[test]
    fn p2c_returns_valid_worker_on_tie() {
        let state = RoutingOccupancyState::default();
        for _ in 0..3 {
            state.increment(1);
            state.increment(2);
        }

        for _ in 0..100 {
            let result = p2c_select_from(&state, &[1, 2]);
            assert!(result == 1 || result == 2);
        }
    }

    #[test]
    fn occupancy_permit_decrements_before_stream_creation() {
        let state = Arc::new(RoutingOccupancyState::default());
        let counter = state.increment(42);
        let permit = OccupancyPermit::from_counter(state.clone(), 42, counter);
        assert_eq!(state.load(42), 1);
        drop(permit);
        assert_eq!(state.load(42), 0);
    }

    #[test]
    fn occupancy_tracked_stream_decrements_on_drop() {
        let state = Arc::new(RoutingOccupancyState::default());
        let counter = state.increment(7);
        let permit = OccupancyPermit::from_counter(state.clone(), 7, counter);
        let ctx: Arc<dyn AsyncEngineContext> = Arc::new(Controller::default());
        let stream = permit.into_tracked_stream(ResponseStream::new(
            Box::pin(tokio_stream::iter(vec![TestResponse { error: None }])),
            ctx,
        ));
        assert_eq!(state.load(7), 1);
        drop(stream);
        assert_eq!(state.load(7), 0);
    }

    #[tokio::test]
    async fn occupancy_tracked_stream_decrements_on_completion() {
        let state = Arc::new(RoutingOccupancyState::default());
        let counter = state.increment(7);
        let permit = OccupancyPermit::from_counter(state.clone(), 7, counter);
        let ctx: Arc<dyn AsyncEngineContext> = Arc::new(Controller::default());
        let mut stream = permit.into_tracked_stream(ResponseStream::new(
            Box::pin(tokio_stream::iter(vec![TestResponse { error: None }])),
            ctx,
        ));

        assert!(stream.next().await.unwrap().err().is_none());
        assert_eq!(state.load(7), 1);
        assert!(stream.next().await.is_none());
        assert_eq!(state.load(7), 0);
        drop(stream);
        assert_eq!(state.load(7), 0, "drop must not release twice after EOF");
    }

    #[tokio::test]
    async fn occupancy_tracked_stream_releases_before_yielding_error() {
        let state = Arc::new(RoutingOccupancyState::default());
        let counter = state.increment(7);
        let permit = OccupancyPermit::from_counter(state.clone(), 7, counter);
        let ctx: Arc<dyn AsyncEngineContext> = Arc::new(Controller::default());
        let error = DynamoError::builder()
            .error_type(ErrorType::WorkerOverloaded)
            .message("worker queue full")
            .build();
        let mut stream = permit.into_tracked_stream(ResponseStream::new(
            Box::pin(tokio_stream::iter(vec![TestResponse {
                error: Some(error),
            }])),
            ctx,
        ));

        let response = stream.next().await.expect("error response");
        assert!(response.err().is_some());
        assert_eq!(
            state.load(7),
            0,
            "occupancy must be released before retry observes the error"
        );
    }

    #[test]
    fn same_id_readd_keeps_existing_permit_accounting() {
        let state = Arc::new(RoutingOccupancyState::default());
        state.retain(&[7]);
        let old_counter = state.increment(7);
        let old_permit = OccupancyPermit::from_counter(state.clone(), 7, old_counter);

        state.retain(&[]);
        assert_eq!(state.load(7), 1);
        state.retain(&[7]);
        let new_permit = OccupancyPermit::acquire(state.clone(), 7);
        assert_eq!(state.load(7), 2);

        drop(old_permit);
        assert_eq!(
            state.load(7),
            1,
            "re-added worker must retain the still-live request"
        );
        drop(new_permit);
        assert_eq!(state.load(7), 0);
    }

    /// A mid-generation stream end means the worker dropped the request, so it
    /// must quarantine — otherwise a migration retry can reselect the same
    /// worker before discovery removal catches up.
    ///
    /// This pins `is_inhibited` against the migration layer's migratable set:
    /// a worker fault that is migratable must also inhibit, or migration
    /// bounces off the same dead worker.
    #[test]
    fn stream_incomplete_quarantines_the_worker() {
        let err = DynamoError::builder()
            .error_type(ErrorType::Backend(BackendError::StreamIncomplete))
            .message("stream ended before generation completed")
            .build();
        assert!(
            is_inhibited(&err),
            "StreamIncomplete must inhibit; it is migratable, so leaving it out \
             lets a retry reselect the failed worker"
        );

        let cancelled = DynamoError::builder()
            .error_type(ErrorType::Cancelled)
            .message("client went away")
            .build();
        assert!(
            !is_inhibited(&cancelled),
            "client cancellation is not a worker fault"
        );
    }

    #[test]
    fn worker_unavailable_quarantines_the_worker() {
        let err = DynamoError::builder()
            .error_type(ErrorType::WorkerUnavailable)
            .message("Server unavailable: unknown endpoint a/generate")
            .build();
        assert!(is_inhibited(&err));
    }

    #[test]
    fn p2c_lifecycle_tracks_inflight_counts_with_shared_tracker() {
        let state = Arc::new(RoutingOccupancyState::default());
        let mut permits = Vec::new();
        for _ in 0..5 {
            let selected = p2c_select_from(&state, &[1, 2]);
            permits.push(OccupancyPermit::acquire(state.clone(), selected));
        }

        let total = state.load(1) + state.load(2);
        assert_eq!(total, 5, "5 in-flight requests should be tracked");

        drop(permits);
        let total = state.load(1) + state.load(2);
        assert_eq!(total, 0, "All guards dropped, counts should be 0");
    }

    #[test]
    fn p2c_never_selects_dominated_worker() {
        let state = RoutingOccupancyState::default();
        for _ in 0..100 {
            state.increment(3);
        }

        let mut selected = [0u32; 3];
        for _ in 0..1000 {
            let result = p2c_select_from(&state, &[1, 2, 3]);
            match result {
                1 => selected[0] += 1,
                2 => selected[1] += 1,
                3 => selected[2] += 1,
                _ => panic!("unexpected worker id"),
            }
        }
        assert_eq!(
            selected[2], 0,
            "Worker 3 (load=100) should never be selected against load=0 workers, but got {} times",
            selected[2]
        );
    }

    #[tokio::test]
    async fn least_loaded_selects_exact_min_and_tracks_counts() {
        let state = Arc::new(RoutingOccupancyState::default());
        state.increment(1);
        state.increment(1);
        state.increment(2);

        let picker = RoutePicker::new(RoutePolicy::LeastLoaded);
        let (decision, counter) = state
            .select_and_admit(
                &picker,
                CandidateView::Workers(&[1, 2, 3]),
                RouteContext::default(),
            )
            .unwrap();
        let selected = decision.target.worker_id;
        assert_eq!(selected, 3);

        let permit = OccupancyPermit::from_counter(
            state.clone(),
            selected,
            counter.expect("least-loaded selection must acquire a counter"),
        );
        assert_eq!(state.load(selected), 1);
        drop(permit);
        assert_eq!(state.load(selected), 0);
    }

    #[tokio::test]
    async fn bidirectional_generate_bails_with_no_instances() {
        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let ns = drt.namespace("test_bidi_no_instances".to_string()).unwrap();
        let component = ns.component("test_component".to_string()).unwrap();
        let endpoint = component.endpoint("test_endpoint".to_string());
        let client = endpoint.client().await.unwrap();

        let router = PushRouter::<u64, TestResponse>::from_client(client, RouterMode::RoundRobin)
            .await
            .unwrap();

        let input: ManyIn<u64> =
            Context::new(RequestStream::new(Box::pin(tokio_stream::iter(vec![
                1u64, 2u64,
            ]))));
        let result = router.generate(input).await;
        assert!(
            result.is_err(),
            "bidirectional generate must bail when no instances are registered"
        );

        rt.shutdown();
    }

    #[tokio::test]
    async fn bidirectional_generate_bails_for_kv_router_mode() {
        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let ns = drt.namespace("test_bidi_kv_mode".to_string()).unwrap();
        let component = ns.component("test_component".to_string()).unwrap();
        let endpoint = component.endpoint("test_endpoint".to_string());
        let client = endpoint.client().await.unwrap();

        let router = PushRouter::<u64, TestResponse>::from_client(client, RouterMode::KV)
            .await
            .unwrap();

        let input: ManyIn<u64> =
            Context::new(RequestStream::new(Box::pin(tokio_stream::iter(vec![1u64]))));
        let result = router.generate(input).await;
        assert!(
            result.is_err(),
            "bidirectional generate must bail for RouterMode::KV"
        );
        let err_msg = format!("{:?}", result.unwrap_err());
        assert!(
            err_msg.contains("KV") || err_msg.contains("kv"),
            "error should mention KV: got {err_msg}"
        );

        rt.shutdown();
    }

    #[tokio::test]
    async fn bidirectional_generate_bails_for_direct_router_mode() {
        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let ns = drt.namespace("test_bidi_direct_mode".to_string()).unwrap();
        let component = ns.component("test_component".to_string()).unwrap();
        let endpoint = component.endpoint("test_endpoint".to_string());
        let client = endpoint.client().await.unwrap();

        let router = PushRouter::<u64, TestResponse>::from_client(client, RouterMode::Direct)
            .await
            .unwrap();

        let input: ManyIn<u64> =
            Context::new(RequestStream::new(Box::pin(tokio_stream::iter(vec![1u64]))));
        let result = router.generate(input).await;
        assert!(
            result.is_err(),
            "bidirectional generate must bail for RouterMode::Direct"
        );
        let err_msg = format!("{:?}", result.unwrap_err());
        assert!(
            err_msg.contains("Direct") || err_msg.contains("direct"),
            "error should mention Direct: got {err_msg}"
        );

        rt.shutdown();
    }

    #[tokio::test]
    async fn bidirectional_generate_rejects_unsupported_load_aware_modes() {
        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let ns = drt.namespace("test_bidi_load_aware".to_string()).unwrap();
        let component = ns.component("test_component".to_string()).unwrap();

        for mode in [
            RouterMode::PowerOfTwoChoices,
            RouterMode::LeastLoaded,
            RouterMode::DeviceAwareWeighted,
        ] {
            let endpoint = component.endpoint("test_endpoint".to_string());
            let client = endpoint.client().await.unwrap();
            let router = PushRouter::<u64, TestResponse>::from_client(client, mode)
                .await
                .unwrap();

            let input: ManyIn<u64> =
                Context::new(RequestStream::new(Box::pin(tokio_stream::iter(vec![1u64]))));
            let result = router.generate(input).await;
            assert!(
                result.is_err(),
                "bidirectional generate must reject {mode:?} (not yet supported)"
            );
            let err_msg = format!("{:?}", result.unwrap_err());
            assert!(
                err_msg.contains("not yet supported for bidirectional dispatch"),
                "error should explain the mode is unsupported, not 'no instances': got {err_msg}"
            );
        }

        rt.shutdown();
    }

    #[tokio::test]
    async fn least_loaded_peek_returns_available_worker_select_stays_none() {
        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let ns = drt
            .namespace("test_least_loaded_router".to_string())
            .unwrap();
        let component = ns.component("test_component".to_string()).unwrap();
        let endpoint = component.endpoint("test_endpoint".to_string());
        let client = endpoint.client().await.unwrap();

        endpoint.register_endpoint_instance().await.unwrap();
        client.wait_for_instances().await.unwrap();

        let router = PushRouter::<u64, TestResponse>::from_client(client, RouterMode::LeastLoaded)
            .await
            .unwrap();

        // LeastLoaded selection tracks request occupancy, so the advisory API is
        // separate from select_next_worker().
        assert_eq!(router.select_next_worker(), None);
        assert!(
            router.peek_next_worker().is_some(),
            "LeastLoaded peek must return the available worker for disagg bootstrap"
        );

        rt.shutdown();
    }

    #[tokio::test]
    async fn exact_selection_releases_occupancy_when_preparation_fails() {
        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let ns = drt
            .namespace("test_exact_prepare_failure".to_string())
            .unwrap();
        let component = ns.component("test_component".to_string()).unwrap();
        let endpoint = component.endpoint("test_endpoint".to_string());
        let client = endpoint.client().await.unwrap();
        endpoint.register_endpoint_instance().await.unwrap();
        let worker_id = client.wait_for_instances().await.unwrap()[0].id();

        let router = PushRouter::<u64, TestResponse>::from_client(client, RouterMode::LeastLoaded)
            .await
            .unwrap();
        let state = router.occupancy_state.clone().unwrap();
        let result = router
            .select_and_dispatch_exact(SingleIn::new(42), None, |_, _| {
                Err::<(), _>(anyhow::anyhow!("metadata preparation failed"))
            })
            .await;

        assert!(result.is_err());
        assert_eq!(
            state.load(worker_id),
            0,
            "preparation failure must release the selected worker"
        );
        rt.shutdown();
    }

    #[tokio::test]
    async fn exact_dispatch_revalidates_overload_after_preparation() {
        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let ns = drt
            .namespace("test_exact_overload_revalidation".to_string())
            .unwrap();
        let component = ns.component("test_component".to_string()).unwrap();
        let endpoint = component.endpoint("test_endpoint".to_string());
        let client = endpoint.client().await.unwrap();
        endpoint.register_endpoint_instance().await.unwrap();
        let worker_id = client.wait_for_instances().await.unwrap()[0].id();

        let router =
            PushRouter::<u64, TestResponse>::from_client(client.clone(), RouterMode::LeastLoaded)
                .await
                .unwrap();
        let state = router.occupancy_state.clone().unwrap();
        let result = router
            .select_and_dispatch_exact(SingleIn::new(42), Some(worker_id), |_, worker_id| {
                client.set_overloaded_instances(&[worker_id]);
                Ok(())
            })
            .await;

        assert!(result.is_err());
        assert_eq!(
            state.load(worker_id),
            0,
            "validation failure must release the selected worker"
        );
        rt.shutdown();
    }

    #[tokio::test]
    async fn transport_resolution_precedes_stale_overload_check() {
        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let endpoint = drt
            .namespace("test_transport_precedes_stale_overload".to_string())
            .unwrap()
            .component("test_component".to_string())
            .unwrap()
            .endpoint("test_endpoint".to_string());
        let client = endpoint.client().await.unwrap();
        let stale_id = 99999;
        client.override_instance_avail(vec![stale_id]);
        client.set_overloaded_instances(&[stale_id]);
        let router = PushRouter::<u64, TestResponse>::from_client(client, RouterMode::RoundRobin)
            .await
            .unwrap();

        let unary_error = router
            .direct(SingleIn::new(42), stale_id)
            .await
            .unwrap_err();
        assert_cannot_connect(&unary_error);

        let input: ManyIn<u64> =
            Context::new(RequestStream::new(Box::pin(tokio_stream::iter(vec![1u64]))));
        let bidirectional_error = router
            .bidirectional_dispatch(stale_id, input)
            .await
            .unwrap_err();
        assert_not_cannot_connect(&bidirectional_error);
        assert!(
            !match_error_chain(
                bidirectional_error.as_ref(),
                &[ErrorType::ResourceExhausted],
                &[]
            ),
            "transport resolution must precede the stale overload check: {bidirectional_error}"
        );

        rt.shutdown();
    }

    #[tokio::test]
    async fn selected_overloaded_worker_is_rejected_before_dispatch() {
        const TEST_RECONCILE_INTERVAL: std::time::Duration = std::time::Duration::from_millis(50);

        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let ns = drt
            .namespace("test_selected_overloaded_worker_rejected".to_string())
            .unwrap();
        let component = ns.component("test_component".to_string()).unwrap();
        let endpoint = component.endpoint("test_endpoint".to_string());
        let client = Client::with_reconcile_interval(endpoint.clone(), TEST_RECONCILE_INTERVAL)
            .await
            .unwrap();

        endpoint.register_endpoint_instance().await.unwrap();
        let instances = client.wait_for_instances().await.unwrap();
        let worker_id = instances[0].id();

        for _ in 0..10 {
            if client.instance_ids_avail().contains(&worker_id) {
                break;
            }
            tokio::time::sleep(TEST_RECONCILE_INTERVAL).await;
        }
        assert!(
            client.instance_ids_avail().contains(&worker_id),
            "worker should be routable before marking it overloaded"
        );

        client.set_overloaded_instances(&[worker_id]);
        let router = PushRouter::<u64, TestResponse>::from_client(client, RouterMode::RoundRobin)
            .await
            .unwrap();

        let result = router.generate(SingleIn::new(42u64)).await;
        assert!(result.is_err());
        let msg = format!("{}", result.unwrap_err());
        // With pre-selection filtering on free_ids, the single-overloaded-worker
        // case is now caught before selection rather than after — the chosen
        // worker is never overloaded because the candidate pool excludes it.
        // The post-selection check in route() remains as a race-condition
        // backstop.
        assert!(
            msg.contains("All workers are busy"),
            "expected empty-free-pool rejection, got: {msg}"
        );

        rt.shutdown();
    }

    #[tokio::test]
    async fn direct_within_rejects_overloaded_constrained_target() {
        const TEST_RECONCILE_INTERVAL: std::time::Duration = std::time::Duration::from_secs(3600);

        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let ns = drt
            .namespace("test_direct_within_overload_rejection".to_string())
            .unwrap();
        let component = ns.component("test_component".to_string()).unwrap();
        let endpoint = component.endpoint("test_endpoint".to_string());
        let client = Client::with_reconcile_interval(endpoint.clone(), TEST_RECONCILE_INTERVAL)
            .await
            .unwrap();

        endpoint.register_endpoint_instance().await.unwrap();
        let worker_id = client.wait_for_instances().await.unwrap()[0].id();
        client.set_overloaded_instances(&[worker_id]);

        let router = PushRouter::<u64, TestResponse>::from_client(client, RouterMode::RoundRobin)
            .await
            .unwrap();
        let allowed = HashSet::from([worker_id]);
        let error = router
            .direct_within(SingleIn::new(42), worker_id, Some(&allowed))
            .await
            .unwrap_err();

        // A *selected* worker being overloaded is single-worker overload, distinct
        // from pool-wide exhaustion: migration may retry elsewhere. Previously
        // both collapsed to ResourceExhausted, which blocked that retry.
        assert!(match_error_chain(
            error.as_ref(),
            &[ErrorType::WorkerOverloaded],
            &[]
        ));
        assert!(
            error.to_string().contains("Selected worker is overloaded"),
            "expected overload rejection, got: {error}"
        );

        rt.shutdown();
    }

    #[tokio::test]
    async fn no_workers_is_reported_as_unavailable() {
        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let ns = drt
            .namespace("test_no_workers_unavailable".to_string())
            .unwrap();
        let component = ns.component("test_component".to_string()).unwrap();
        let endpoint = component.endpoint("test_endpoint".to_string());
        let client = endpoint.client().await.unwrap();
        let router = PushRouter::<u64, TestResponse>::from_client(client, RouterMode::RoundRobin)
            .await
            .unwrap();

        let error = router.generate(SingleIn::new(42)).await.unwrap_err();
        assert!(match_error_chain(
            error.as_ref(),
            &[ErrorType::Unavailable],
            &[]
        ));

        rt.shutdown();
    }

    #[tokio::test]
    async fn round_robin_excludes_overloaded_workers_from_candidates() {
        // Long reconcile interval so the synthetic override below survives
        // the test. We still register a real endpoint instance up front so
        // the initial reconcile (which fires immediately when the monitor
        // task spawns) settles on a non-empty source — without that, the
        // first reconcile would clobber the override before it takes effect.
        const TEST_RECONCILE_INTERVAL: std::time::Duration = std::time::Duration::from_secs(3600);

        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let ns = drt
            .namespace("test_round_robin_excludes_overloaded".to_string())
            .unwrap();
        let component = ns.component("test_component".to_string()).unwrap();
        let endpoint = component.endpoint("test_endpoint".to_string());
        let client = Client::with_reconcile_interval(endpoint.clone(), TEST_RECONCILE_INTERVAL)
            .await
            .unwrap();

        endpoint.register_endpoint_instance().await.unwrap();
        let instances = client.wait_for_instances().await.unwrap();
        let real_id = instances[0].id();
        for _ in 0..50 {
            if client.instance_ids_avail().contains(&real_id) {
                break;
            }
            tokio::time::sleep(std::time::Duration::from_millis(20)).await;
        }

        // Now override with two synthetic IDs and mark one overloaded.
        // round_robin must never select the overloaded one — that's the
        // whole point of selecting from free_ids instead of routable_ids.
        // The post-selection overload check in route() would otherwise return 529
        // one of N requests on each pass, which is the bug this PR closes
        // for non-KV selectors.
        client.override_instance_avail(vec![1, 2]);
        client.set_overloaded_instances(&[1]);

        let router = PushRouter::<u64, TestResponse>::from_client(client, RouterMode::RoundRobin)
            .await
            .unwrap();

        // Round-robin over N requests should land on worker 2 every time.
        // We use peek_next_worker for a side-effect-free probe.
        for _ in 0..6 {
            let selected = router
                .peek_next_worker()
                .expect("peek should succeed with a free worker");
            assert_eq!(
                selected, 2,
                "overloaded worker 1 must not appear in the candidate set"
            );
        }

        rt.shutdown();
    }

    #[tokio::test]
    async fn device_aware_weighted_peek_returns_available_worker_select_stays_none() {
        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let ns = drt
            .namespace("test_device_aware_router".to_string())
            .unwrap();
        let component = ns.component("test_component".to_string()).unwrap();
        let endpoint = component.endpoint("test_endpoint".to_string());
        let client = endpoint.client().await.unwrap();

        endpoint.register_endpoint_instance().await.unwrap();
        client.wait_for_instances().await.unwrap();

        let router =
            PushRouter::<u64, TestResponse>::from_client(client, RouterMode::DeviceAwareWeighted)
                .await
                .unwrap();

        // DeviceAwareWeighted degenerates to least-loaded for peek (device-class
        // partitioning happens at dispatch); select_next_worker stays None.
        assert_eq!(router.select_next_worker(), None);
        assert!(
            router.peek_next_worker().is_some(),
            "DeviceAwareWeighted peek must return the available worker for disagg bootstrap"
        );

        rt.shutdown();
    }

    #[tokio::test]
    async fn device_aware_exact_selection_preserves_full_multimodal_cache_hit() {
        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let ns = drt
            .namespace("test_device_aware_affinity_cache".to_string())
            .unwrap();
        let component = ns.component("test_component".to_string()).unwrap();
        let endpoint = component.endpoint("test_endpoint".to_string());
        let client = endpoint.client().await.unwrap();
        endpoint.register_endpoint_instance().await.unwrap();
        let cache_worker = client.wait_for_instances().await.unwrap()[0].id();

        let router = PushRouter::<u64, TestResponse>::from_client_with_state(
            client,
            RouterMode::DeviceAwareWeighted,
            None,
            Some(Arc::new(StaticMultimodalCacheIndex {
                worker_id: cache_worker,
            })),
            Some(Arc::new(|_| vec!["image-key".to_string()])),
        )
        .await
        .unwrap();

        let selection = router.select_device_aware_and_reserve(&42, None).unwrap();
        assert_eq!(selection.worker_id(), cache_worker);
        assert!(
            selection.into_reservation().is_none(),
            "full cache hits bypass occupancy charging"
        );

        rt.shutdown();
    }

    /// Direct dispatch honors an upstream-selected worker even after local inhibition.
    #[tokio::test]
    async fn direct_dispatch_ignores_local_inhibition() {
        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let ns = drt
            .namespace("test_direct_bypasses_inhibition".to_string())
            .unwrap();
        let component = ns.component("test_component".to_string()).unwrap();
        let endpoint = component.endpoint("test_endpoint".to_string());
        let client = endpoint.client().await.unwrap();
        endpoint.register_endpoint_instance().await.unwrap();
        let instance_id = client.wait_for_instances().await.unwrap()[0].id();

        // KV routing selects upstream and dispatches through PushRouter::direct.
        let router = PushRouter::<u64, TestResponse>::from_client(client.clone(), RouterMode::KV)
            .await
            .unwrap();

        client.report_instance_down(instance_id);
        assert!(
            !client.instance_ids_avail().contains(&instance_id),
            "precondition: worker should be locally inhibited"
        );

        let result = router
            .direct_within_prepared(
                SingleIn::new(42),
                instance_id,
                None,
                |_, selected_instance_id| {
                    assert_eq!(selected_instance_id, instance_id);
                    Err::<(), _>(anyhow::anyhow!("direct prepare sentinel"))
                },
            )
            .await;
        let error = match result {
            Ok(_) => panic!("direct dispatch should reach request preparation"),
            Err(error) => error,
        };
        assert_eq!(error.to_string(), "direct prepare sentinel");

        let missing_instance_id = instance_id.wrapping_add(1);
        let result = router
            .direct_within_prepared(SingleIn::new(42), missing_instance_id, None, |_, _| {
                Ok::<(), anyhow::Error>(())
            })
            .await;
        let error = match result {
            Ok(_) => panic!("direct dispatch should reject a worker absent from discovery"),
            Err(error) => error,
        };
        assert!(
            error
                .to_string()
                .contains(&format!("instance_id={missing_instance_id} not found")),
            "unexpected missing-worker error: {error}"
        );

        rt.shutdown();
    }

    /// When the router selects an instance that has deregistered between selection
    /// and transport resolution, it should fall back to another available instance
    /// rather than returning a 500 error.
    #[tokio::test]
    async fn transport_resolution_falls_back_when_selected_instance_disappears() {
        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let ns = drt
            .namespace("test_transport_fallback".to_string())
            .unwrap();
        let component = ns.component("test_component".to_string()).unwrap();
        let endpoint = component.endpoint("test_endpoint".to_string());
        let client = endpoint.client().await.unwrap();

        // Register one real instance so it appears in instance_source.
        endpoint.register_endpoint_instance().await.unwrap();
        client.wait_for_instances().await.unwrap();

        let real_id = client.instance_ids()[0];

        // Inject a stale ID into instance_avail that does NOT exist in
        // instance_source. This simulates the race window where an instance
        // deregistered after selection but before transport resolution.
        let stale_id = real_id + 1000;
        client.override_instance_avail(vec![stale_id, real_id]);

        let router =
            PushRouter::<u64, TestResponse>::from_client(client.clone(), RouterMode::RoundRobin)
                .await
                .unwrap();

        // Exercise transport resolution directly. Sending a request to this
        // registration would wait forever because the test intentionally has
        // no worker handler.
        let (resolved_id, _, _, _) = router
            .resolve_transport(stale_id, TransportFallback::Allow)
            .expect("normal routing should fall back from a stale worker");
        assert_eq!(resolved_id, real_id);

        rt.shutdown();
    }

    #[tokio::test]
    async fn prepared_dispatch_observes_worker_after_transport_fallback() {
        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let endpoint = drt
            .namespace("test_prepared_transport_fallback".to_string())
            .unwrap()
            .component("test_component".to_string())
            .unwrap()
            .endpoint("test_endpoint".to_string());
        let client = endpoint.client().await.unwrap();
        endpoint.register_endpoint_instance().await.unwrap();
        let real_id = client.wait_for_instances().await.unwrap()[0].id();
        let stale_id = real_id.wrapping_add(1);
        client.override_instance_avail(vec![stale_id, real_id]);
        let router = PushRouter::<u64, TestResponse>::from_client(client, RouterMode::LeastLoaded)
            .await
            .unwrap();
        let state = router.occupancy_state.clone().unwrap();
        state.increment(real_id);
        let state_for_prepare = state.clone();
        let observed = Arc::new(AtomicU64::new(0));
        let observed_for_prepare = observed.clone();

        let _ = tokio::time::timeout(
            std::time::Duration::from_millis(100),
            router.select_and_dispatch(SingleIn::new(42), move |_, worker_id| {
                assert_eq!(state_for_prepare.load(stale_id), 0);
                assert_eq!(state_for_prepare.load(worker_id), 2);
                observed_for_prepare.store(worker_id, Ordering::Relaxed);
                Ok(())
            }),
        )
        .await;

        assert_eq!(observed.load(Ordering::Relaxed), real_id);
        assert_eq!(state.load(real_id), 1);
        state.decrement(real_id);
        rt.shutdown();
    }

    /// When no instances are available at all (both primary and fallback),
    /// the router should return a clear error.
    #[tokio::test]
    async fn transport_resolution_errors_when_no_instances_available() {
        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let ns = drt
            .namespace("test_transport_no_fallback".to_string())
            .unwrap();
        let component = ns.component("test_component".to_string()).unwrap();
        let endpoint = component.endpoint("test_endpoint".to_string());
        // Freeze `monitor_instance_source` before staging routing state. That
        // task reconciles `routable_ids` back to the real discovered set on
        // every discovery change *and* every `reconcile_interval` (5s by
        // default), so it silently undoes `override_instance_avail` mid-test.
        // Cancelling before registration means the only writer left is this
        // test. Safe because `wait_for_instances` reads `instance_source`
        // directly rather than the reconciled routing snapshot.
        let monitor = tokio_util::sync::CancellationToken::new();
        let client = endpoint
            .client_with_cancellation(monitor.clone())
            .await
            .unwrap();
        monitor.cancel();

        // Register an instance so we can create the router (needs transport setup).
        endpoint.register_endpoint_instance().await.unwrap();
        client.wait_for_instances().await.unwrap();

        let router =
            PushRouter::<u64, TestResponse>::from_client(client.clone(), RouterMode::RoundRobin)
                .await
                .unwrap();

        // Override avail to contain only a stale ID with no real backing
        // instance AND no other available fallback.
        let stale_id = 99999;
        client.override_instance_avail(vec![stale_id]);

        let request = SingleIn::new(42u64);
        let result = router.generate(request).await;

        assert!(result.is_err());
        let error = result.unwrap_err();
        assert_unavailable(&error);
        let msg = error.to_string();
        assert!(
            msg.contains("not found") && msg.contains("no other instances available"),
            "Expected clear error about missing instance with no fallback, got: {msg}"
        );

        rt.shutdown();
    }

    #[tokio::test]
    async fn transport_resolution_honors_fallback_policy() {
        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let ns = drt
            .namespace("test_exact_transport_no_fallback".to_string())
            .unwrap();
        let component = ns.component("test_component".to_string()).unwrap();
        let endpoint = component.endpoint("test_endpoint".to_string());
        // Freeze `monitor_instance_source` before staging routing state. That
        // task reconciles `routable_ids` back to the real discovered set on
        // every discovery change *and* every `reconcile_interval` (5s by
        // default), so it silently undoes `override_instance_avail` mid-test.
        // Cancelling before registration means the only writer left is this
        // test. Safe because `wait_for_instances` reads `instance_source`
        // directly rather than the reconciled routing snapshot.
        let monitor = tokio_util::sync::CancellationToken::new();
        let client = endpoint
            .client_with_cancellation(monitor.clone())
            .await
            .unwrap();
        monitor.cancel();
        endpoint.register_endpoint_instance().await.unwrap();
        let instances = client.wait_for_instances().await.unwrap();
        let real_id = instances[0].id();

        let router =
            PushRouter::<u64, TestResponse>::from_client(client.clone(), RouterMode::RoundRobin)
                .await
                .unwrap();
        let stale_id = real_id.wrapping_add(1);
        client.override_instance_avail(vec![stale_id, real_id]);

        assert!(
            router
                .resolve_transport(stale_id, TransportFallback::Allow)
                .is_ok(),
            "normal dispatch should preserve transport fallback"
        );
        let allowed = HashSet::from([real_id]);
        assert!(
            router
                .resolve_transport(stale_id, TransportFallback::Within(&allowed))
                .is_ok(),
            "constrained dispatch should fall back within the allowed worker set"
        );
        let disallowed = HashSet::new();
        let disallowed_error = router
            .resolve_transport(stale_id, TransportFallback::Within(&disallowed))
            .unwrap_err();
        assert_unavailable(&disallowed_error);

        let exact_error = router
            .resolve_transport(stale_id, TransportFallback::Deny)
            .unwrap_err();
        assert_cannot_connect(&exact_error);

        let second_stale_id = stale_id.wrapping_add(1);
        client.override_instance_avail(vec![stale_id, second_stale_id]);
        let stale_fallback_error = router
            .resolve_transport(stale_id, TransportFallback::Allow)
            .unwrap_err();
        assert_cannot_connect(&stale_fallback_error);
        assert!(
            stale_fallback_error
                .to_string()
                .contains("Fallback instance"),
            "expected fallback lookup failure, got: {stale_fallback_error}"
        );
        rt.shutdown();
    }

    /// The watcher dedup guard must be released even if the spawned task panics.
    /// Without this, a panic anywhere in the watcher body would leave a stale
    /// `ENDPOINT_WATCHER_ACTIVE` entry, silently disabling orphaned-pending-
    /// request cancellation for that endpoint until process restart.
    ///
    /// We exercise the Drop-guard pattern directly against the same static
    /// rather than driving `spawn_instance_removal_watcher` end-to-end (which
    /// would require staging a panicking discovery stream). The test mirrors
    /// the production code's GuardRelease shape; if the production code stops
    /// using a Drop guard, the integration would regress and the existing
    /// orphan-cancellation tests would fail.
    #[tokio::test]
    async fn watcher_dedup_guard_released_on_panic() {
        let endpoint_id = EndpointId {
            namespace: "panic-test-ns".to_string(),
            component: "panic-test-comp".to_string(),
            name: "panic-test-endpoint".to_string(),
        };

        // Mimic the production code's pre-spawn dedup insert.
        let map = ENDPOINT_WATCHER_ACTIVE.get_or_init(dashmap::DashMap::new);
        map.insert(endpoint_id.clone(), ());

        let endpoint_id_clone = endpoint_id.clone();
        let join = tokio::spawn(async move {
            // Same shape as in spawn_instance_removal_watcher.
            struct GuardRelease(EndpointId);
            impl Drop for GuardRelease {
                fn drop(&mut self) {
                    if let Some(map) = ENDPOINT_WATCHER_ACTIVE.get() {
                        map.remove(&self.0);
                    }
                }
            }
            let _release = GuardRelease(endpoint_id_clone);
            panic!("simulated watcher-task panic");
        });

        let result = join.await;
        assert!(result.is_err() && result.unwrap_err().is_panic());
        assert!(
            !map.contains_key(&endpoint_id),
            "Drop guard must release the dedup entry even on panic"
        );
    }

    /// Normal-exit path: the Drop guard releases the entry when the task
    /// finishes without panicking. This is the everyday case (cancel_token
    /// fires or discovery stream closes).
    #[tokio::test]
    async fn watcher_dedup_guard_released_on_normal_exit() {
        let endpoint_id = EndpointId {
            namespace: "normal-test-ns".to_string(),
            component: "normal-test-comp".to_string(),
            name: "normal-test-endpoint".to_string(),
        };

        let map = ENDPOINT_WATCHER_ACTIVE.get_or_init(dashmap::DashMap::new);
        map.insert(endpoint_id.clone(), ());

        let endpoint_id_clone = endpoint_id.clone();
        tokio::spawn(async move {
            struct GuardRelease(EndpointId);
            impl Drop for GuardRelease {
                fn drop(&mut self) {
                    if let Some(map) = ENDPOINT_WATCHER_ACTIVE.get() {
                        map.remove(&self.0);
                    }
                }
            }
            let _release = GuardRelease(endpoint_id_clone);
            // task body returns normally
        })
        .await
        .unwrap();

        assert!(!map.contains_key(&endpoint_id));
    }

    #[tokio::test]
    async fn cache_cleanup_watcher_identity_includes_runtime() {
        let runtime = Runtime::from_current().unwrap();
        let first = DistributedRuntime::new(runtime.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let second = DistributedRuntime::new(runtime.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let endpoint = |distributed: &DistributedRuntime| {
            distributed
                .namespace("cache-watcher-identity".to_string())
                .unwrap()
                .component("workers".to_string())
                .unwrap()
                .endpoint("generate".to_string())
        };
        let first = RuntimeEndpointId::for_endpoint(&endpoint(&first));
        let second = RuntimeEndpointId::for_endpoint(&endpoint(&second));

        assert_eq!(first.endpoint_id, second.endpoint_id);
        assert_ne!(first.connection_id, second.connection_id);
        assert_ne!(first, second);

        runtime.shutdown();
    }

    /// A `StreamingDispatch` that records what the router hands the seam, so the
    /// test can assert a *caller-supplied* dispatch (not just the default
    /// `AddressedPushRouter`) receives the selected address/instance and the
    /// discovery lifecycle events.
    #[derive(Default)]
    struct RecordingDispatch {
        unary: std::sync::Mutex<Vec<(u64, String, Option<u64>)>>,
        bidi: std::sync::Mutex<Vec<(String, u64)>>,
        added: std::sync::Mutex<Vec<u64>>,
        removed: std::sync::Mutex<Vec<u64>>,
    }

    impl RecordingDispatch {
        fn canned_stream() -> ManyOut<TestResponse> {
            let ctx: Arc<dyn AsyncEngineContext> = Arc::new(Controller::default());
            ResponseStream::new(
                Box::pin(tokio_stream::iter(vec![TestResponse { error: None }])),
                ctx,
            )
        }
    }

    #[async_trait::async_trait]
    impl StreamingDispatch<u64, TestResponse> for RecordingDispatch {
        async fn generate(
            &self,
            request: SingleIn<AddressedRequest<u64>>,
        ) -> Result<ManyOut<TestResponse>, Error> {
            let (addressed, _ctx) = request.transfer(());
            let (payload, address, instance) = addressed.into_parts();
            self.unary
                .lock()
                .unwrap()
                .push((payload, address, instance.map(|i| i.id())));
            Ok(Self::canned_stream())
        }

        async fn generate_bidirectional(
            &self,
            instance: Instance,
            address: String,
            _input: ManyIn<u64>,
        ) -> Result<ManyOut<TestResponse>, Error> {
            self.bidi.lock().unwrap().push((address, instance.id()));
            Ok(Self::canned_stream())
        }

        async fn on_instance_removed(&self, id: &EndpointInstanceId) {
            self.removed.lock().unwrap().push(id.instance_id);
        }

        async fn on_instance_added(&self, id: &EndpointInstanceId) {
            self.added.lock().unwrap().push(id.instance_id);
        }
    }

    /// The transport seam must deliver to a caller-supplied `StreamingDispatch`:
    /// unary and bidirectional requests arrive with the selected address and
    /// instance, and discovery removal/re-addition reach its lifecycle hooks.
    #[tokio::test]
    async fn from_client_with_dispatch_delivers_requests_and_lifecycle() {
        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let endpoint = drt
            .namespace("test_dispatch_seam".to_string())
            .unwrap()
            .component("test_component".to_string())
            .unwrap()
            .endpoint("test_endpoint".to_string());
        let client = endpoint.client().await.unwrap();

        endpoint.register_endpoint_instance().await.unwrap();
        let instance_id = client.wait_for_instances().await.unwrap()[0].id();

        let dispatch = Arc::new(RecordingDispatch::default());
        let router = PushRouter::<u64, TestResponse>::from_client_with_dispatch(
            client.clone(),
            RouterMode::RoundRobin,
            dispatch.clone(),
        )
        .await
        .unwrap();

        // Unary hop reaches the supplied dispatch with the selected worker.
        let mut stream = router.generate(SingleIn::new(42u64)).await.unwrap();
        while stream.next().await.is_some() {}
        {
            let unary = dispatch.unary.lock().unwrap();
            assert_eq!(unary.len(), 1, "one unary dispatch expected");
            let (payload, address, dispatched) = &unary[0];
            assert_eq!(*payload, 42);
            assert_eq!(*dispatched, Some(instance_id));
            assert!(!address.is_empty(), "selected transport address expected");
        }

        // Bidirectional hop reaches the supplied dispatch with the same worker.
        let input: ManyIn<u64> =
            Context::new(RequestStream::new(Box::pin(tokio_stream::iter(vec![
                1u64, 2u64,
            ]))));
        let mut stream = router.generate(input).await.unwrap();
        while stream.next().await.is_some() {}
        {
            let bidi = dispatch.bidi.lock().unwrap();
            assert_eq!(bidi.len(), 1, "one bidirectional dispatch expected");
            assert_eq!(bidi[0].1, instance_id);
            assert!(!bidi[0].0.is_empty());
        }

        // Gate on the initial-snapshot add before mutating discovery: this both
        // asserts on_instance_added is delivered and guarantees the watcher is
        // subscribed, so the removal broadcast can't race ahead of it.
        assert!(
            poll_until(|| dispatch.added.lock().unwrap().contains(&instance_id)).await,
            "on_instance_added (initial snapshot) not delivered to the supplied dispatch"
        );

        endpoint.unregister_endpoint_instance().await.unwrap();
        assert!(
            poll_until(|| dispatch.removed.lock().unwrap().contains(&instance_id)).await,
            "on_instance_removed not delivered to the supplied dispatch"
        );

        // A fresh add after re-registration must also reach the hook.
        let adds_before = dispatch.added.lock().unwrap().len();
        endpoint.register_endpoint_instance().await.unwrap();
        assert!(
            poll_until(|| dispatch.added.lock().unwrap().len() > adds_before).await,
            "on_instance_added (re-registration) not delivered to the supplied dispatch"
        );

        rt.shutdown();
    }

    #[tokio::test]
    async fn admitted_dispatch_does_not_reject_the_load_it_just_booked() {
        const TEST_RECONCILE_INTERVAL: std::time::Duration = std::time::Duration::from_secs(3600);

        let rt = Runtime::from_current().unwrap();
        let drt = DistributedRuntime::new(rt.clone(), DistributedConfig::process_local())
            .await
            .unwrap();
        let endpoint = drt
            .namespace("test_admitted_dispatch".to_string())
            .unwrap()
            .component("test_component".to_string())
            .unwrap()
            .endpoint("test_endpoint".to_string());
        let client = Client::with_reconcile_interval(endpoint.clone(), TEST_RECONCILE_INTERVAL)
            .await
            .unwrap();

        endpoint.register_endpoint_instance().await.unwrap();
        let instance_id = client.wait_for_instances().await.unwrap()[0].id();
        for _ in 0..50 {
            if client.instance_ids_avail().contains(&instance_id) {
                break;
            }
            tokio::time::sleep(std::time::Duration::from_millis(20)).await;
        }
        assert!(client.instance_ids_avail().contains(&instance_id));
        let dispatch = Arc::new(RecordingDispatch::default());
        let router = PushRouter::<u64, TestResponse>::from_client_with_dispatch(
            client.clone(),
            RouterMode::KV,
            dispatch.clone(),
        )
        .await
        .unwrap();

        client.set_overloaded_instances(&[instance_id]);
        let error = router
            .dispatch_exact(SingleIn::new(41), instance_id)
            .await
            .unwrap_err();
        assert!(match_error_chain(
            error.as_ref(),
            &[ErrorType::WorkerOverloaded],
            &[]
        ));
        assert!(dispatch.unary.lock().unwrap().is_empty());

        let mut stream = router
            .dispatch_kv_admitted(SingleIn::new(42), instance_id)
            .await
            .unwrap();
        while stream.next().await.is_some() {}
        let unary = dispatch.unary.lock().unwrap();
        assert_eq!(unary.len(), 1);
        assert_eq!(unary[0].0, 42);
        assert!(!unary[0].1.is_empty());
        assert_eq!(unary[0].2, Some(instance_id));
        drop(unary);

        rt.shutdown();
    }

    /// Poll a predicate until it holds or a short deadline elapses; discovery
    /// events reach the watcher's spawned task asynchronously.
    async fn poll_until(mut pred: impl FnMut() -> bool) -> bool {
        for _ in 0..200 {
            if pred() {
                return true;
            }
            tokio::time::sleep(std::time::Duration::from_millis(20)).await;
        }
        pred()
    }
}