ScadaBridge/src/ZB.MOM.WW.ScadaBridge.StoreAndForward/StoreAndForwardService.cs

using System.Threading.Channels;
using Microsoft.Extensions.Logging;
using ZB.MOM.WW.ScadaBridge.Commons.Interfaces.Services;
using ZB.MOM.WW.ScadaBridge.Commons.Observability;
using ZB.MOM.WW.ScadaBridge.Commons.Types;
using ZB.MOM.WW.ScadaBridge.Commons.Types.Enums;
using ZB.MOM.WW.ScadaBridge.SiteEventLogging;

namespace ZB.MOM.WW.ScadaBridge.StoreAndForward;

/// <summary>
/// Core store-and-forward service.
///
/// Lifecycle:
///   1. Caller attempts immediate delivery via IDeliveryHandler
///   2. On transient failure → buffer in SQLite → retry loop
///   3. On success → remove from buffer
///   4. On reaching MaxRetries → park (a MaxRetries of 0 means "no limit" — the
///      message is retried until delivered and is never parked for retry exhaustion)
///   5. Permanent failures are returned to caller immediately (never buffered)
///
/// Fixed retry interval (not exponential). Per-source-entity retry settings.
///        Background timer-based retry sweep.
///
/// Parked messages queryable, retryable, and discardable.
///
/// Buffer depth reported as health metric. Activity logged to site event log.
///
/// CachedCall idempotency is the caller's responsibility.
///        This service does not deduplicate — if the same message is enqueued twice,
///        it will be delivered twice. Callers using ExternalSystem.CachedCall() must
///        design their payloads to be idempotent (e.g., include unique request IDs
///        and handle duplicate detection on the remote end).
/// </summary>
public class StoreAndForwardService
{
    private readonly StoreAndForwardStorage _storage;
    private readonly StoreAndForwardOptions _options;
    private readonly ReplicationService? _replication;
    private readonly ILogger<StoreAndForwardService> _logger;
    /// <summary>
    /// Site-side observer notified
    /// after every cached-call delivery attempt. Optional — when null no
    /// telemetry is emitted; the legacy retry loop behaviour is
    /// preserved exactly.
    /// </summary>
    private readonly ICachedCallLifecycleObserver? _cachedCallObserver;
    /// <summary>
    /// Optional site operational-event log. When non-null the service maps
    /// its own buffer/retry/park activity (the same activity that drives
    /// <see cref="OnActivity"/>) onto site events — <c>store_and_forward</c> for the
    /// cached-call categories and <c>notification</c> for the site's
    /// forward-to-central notification path. Best-effort and fire-and-forget so a
    /// failing logger never affects delivery bookkeeping.
    /// </summary>
    private readonly ISiteEventLogger? _siteEventLogger;
    /// <summary>
    /// Site id stamped onto the
    /// cached-call attempt context so the audit bridge can build the
    /// <see cref="SiteCallOperational"/> half of the telemetry packet.
    /// </summary>
    private readonly string _siteId;

    /// <summary>
    /// Distinctive marker stamped onto cached-call audit
    /// telemetry when the host has not registered an
    /// <see cref="IStoreAndForwardSiteContext"/>. Chosen with a leading <c>$</c>
    /// so it cannot collide with a real site id (which is a configuration
    /// identifier and never starts with <c>$</c>). Surfacing this in the
    /// central audit log makes a missing site-context binding immediately
    /// recognisable instead of an unattributable empty string.
    /// </summary>
    public const string UnknownSiteSentinel = "$unknown-site";

    /// <summary>
    /// The detail-string prefix written by <see cref="EnqueueAsync"/>
    /// when an immediate forward attempt throws and the message is buffered for
    /// the retry sweep. <see cref="EmitSiteEvent"/> matches on this same prefix
    /// to distinguish a forward <i>failure</i> (logged) from a routine
    /// no-handler enqueue (not logged), so both the construction site and the
    /// check reference this single constant rather than duplicating the
    /// literal — keeping the two ends from drifting apart.
    /// </summary>
    private const string BufferedForRetryDetailPrefix = "Buffered for retry";

    private Timer? _retryTimer;
    private int _retryInProgress;

    /// <summary>
    /// Active-node delivery gate. When set, the retry sweep runs ONLY when the
    /// gate reports true — the standby node applies replicated buffer operations
    /// but must never deliver (Component-StoreAndForward.md "standby-passive"
    /// model). Re-evaluated on every sweep tick so a failover resumes delivery
    /// within one RetryTimerInterval without re-wiring. Null (tests, central
    /// hosts) preserves the ungated legacy behaviour. A throwing gate is treated
    /// as standby — safe-by-default, mirroring SiteCommunicationActor's
    /// DefaultIsActiveCheck fallback.
    /// </summary>
    private Func<bool>? _deliveryGate;

    /// <summary>Installs the active-node delivery gate (see <see cref="_deliveryGate"/>).</summary>
    /// <param name="gate">Predicate returning <c>true</c> when this node should run the retry sweep.</param>
    public void SetDeliveryGate(Func<bool> gate) => _deliveryGate = gate;

    /// <summary>
    /// The in-flight retry sweep <see cref="Task"/>, or
    /// <c>null</c> when no sweep is currently running. Captured when the timer
    /// callback starts a sweep so <see cref="StopAsync"/> can wait for it to
    /// finish before the host disposes downstream dependencies
    /// (<see cref="_storage"/>, <see cref="_replication"/>) that the sweep is
    /// still touching. Written from the timer thread and from
    /// <see cref="StopAsync"/>, so reads are synchronised via the
    /// <see cref="Volatile"/> APIs.
    /// </summary>
    private Task? _sweepTask;

    /// <summary>
    /// Ordered, single-reader queue of cached-call audit-observer notifications.
    /// The retry sweep <b>posts</b> to this channel instead of awaiting the
    /// observer inline, so a slow observer (a SQLite audit write) cannot stretch
    /// the sweep; the single reader (<see cref="_observerPump"/>) preserves
    /// per-operation event order. Re-created in <see cref="StartAsync"/> because
    /// <see cref="StopAsync"/> completes it (a restarted instance needs a fresh
    /// channel). Before <see cref="StartAsync"/> starts the pump, posts fall back
    /// to inline processing (see <see cref="PostObserverNotification"/>).
    /// </summary>
    private Channel<Func<Task>> _observerQueue =
        Channel.CreateUnbounded<Func<Task>>(new UnboundedChannelOptions { SingleReader = true });

    /// <summary>
    /// The single-reader pump draining <see cref="_observerQueue"/>, or
    /// <c>null</c> when the service is not started. Started in
    /// <see cref="StartAsync"/>, completed + awaited in <see cref="StopAsync"/>.
    /// </summary>
    private Task? _observerPump;

    /// <summary>
    /// How long <see cref="StopAsync"/> waits for an
    /// in-flight retry sweep to finish before returning. The default — 10 s —
    /// is generous enough to let a typical sweep over the buffered queue drain,
    /// but bounded so a hung downstream call (a stuck SQLite write, a
    /// long-running delivery handler) cannot block host shutdown indefinitely.
    /// On timeout the wait is abandoned and the timer is still disposed; the
    /// sweep keeps running but will throw on the next call into a disposed
    /// dependency — preferred to blocking shutdown forever.
    /// </summary>
    private static readonly TimeSpan SweepShutdownWaitTimeout = TimeSpan.FromSeconds(10);

    /// <summary>
    /// Cached count of messages currently buffered for
    /// forwarding — i.e. rows in <see cref="StoreAndForwardMessageStatus.Pending"/>,
    /// the live store-and-forward queue waiting to be delivered. This backs the
    /// <c>scadabridge.store_and_forward.queue.depth</c> observable gauge.
    /// <para>
    /// The gauge's collection callback is synchronous and is invoked frequently by
    /// the OpenTelemetry/Prometheus collector, so it must never run an async SQLite
    /// <c>COUNT(*)</c>. Instead this <see cref="long"/> is seeded once from storage
    /// in <see cref="StartAsync"/> and then adjusted in-process on the existing
    /// paths that change the Pending population: <see cref="BufferAsync"/> (+1),
    /// successful-retry removal and Pending→Parked transitions in
    /// <see cref="RetryMessageAsync"/> (-1), and operator requeue in
    /// <see cref="RetryParkedMessageAsync"/> (+1). The provider registered with
    /// <see cref="ScadaBridgeTelemetry.SetQueueDepthProvider"/> reads it via
    /// <see cref="Interlocked.Read"/> — non-blocking and sync-safe. It is an
    /// approximate, eventually-consistent gauge (concurrent failover replication
    /// applies to the standby's own store, not this counter), which is exactly
    /// what a queue-depth metric needs.
    /// </para>
    /// </summary>
    private long _bufferedCount;

    /// <summary>
    /// Test seam: simulates a concurrent pre-seed
    /// <see cref="BufferAsync"/> increment landing on <see cref="_bufferedCount"/>
    /// before <see cref="StartAsync"/> seeds it, so a test can prove the seed uses
    /// <see cref="Interlocked.Add"/> (additive) rather than Exchange (clobbering).
    /// </summary>
    internal void TestOnly_IncrementBufferedCount() =>
        Interlocked.Increment(ref _bufferedCount);

    /// <summary>
    /// An instance field that guards against a single instance
    /// registering the queue-depth provider (and re-seeding the counter) more than
    /// once — e.g. a second <see cref="StartAsync"/> on the same instance. It does NOT
    /// coordinate across instances: the gauge slot in <see cref="ScadaBridgeTelemetry"/>
    /// is process-global, so in a multi-instance process the last <see cref="StartAsync"/>
    /// wins the global slot. 0 = not yet registered, 1 = done.
    /// </summary>
    private int _queueDepthProviderRegistered;

    /// <summary>
    /// The exact provider delegate this instance registered with
    /// the process-global <see cref="ScadaBridgeTelemetry"/> gauge slot, retained so
    /// <see cref="StopAsync"/> can deregister it by identity (compare-and-clear). Holding
    /// the same reference both registers and clears the slot; the identity check ensures a
    /// late stop of this instance cannot clear a newer instance's provider. Null until
    /// <see cref="StartAsync"/> registers.
    /// </summary>
    private Func<long>? _queueDepthProvider;

    /// <summary>
    /// Delivery handler delegate. The return value / exception is interpreted
    /// the same way on both the immediate-delivery path (<see cref="EnqueueAsync"/>)
    /// and the background retry path (<c>RetryMessageAsync</c>):
    /// <list type="bullet">
    ///   <item><description><c>true</c> — delivered successfully. The message is
    ///   removed from the buffer (or, on the immediate path, never buffered).</description></item>
    ///   <item><description><c>false</c> — permanent failure. On the immediate path
    ///   the message is NOT buffered; on a retry the message is already buffered and
    ///   is parked immediately (no further retries).</description></item>
    ///   <item><description>throws — transient failure. On the immediate path the
    ///   message is buffered for retry; on a retry the retry count is incremented and
    ///   the message is parked once <see cref="StoreAndForwardMessage.MaxRetries"/> is
    ///   reached.</description></item>
    /// </list>
    /// </summary>
    private readonly Dictionary<StoreAndForwardCategory, Func<StoreAndForwardMessage, Task<bool>>> _deliveryHandlers = new();

    /// <summary>
    /// Event callback for logging S&amp;F activity to site event log.
    /// </summary>
    public event Action<string, StoreAndForwardCategory, string>? OnActivity;

    /// <summary>
    /// Initializes a new instance of the StoreAndForwardService.
    /// </summary>
    /// <param name="storage">The storage backend for buffered messages.</param>
    /// <param name="options">Configuration options.</param>
    /// <param name="logger">Logger instance.</param>
    /// <param name="replication">Optional replication service for standby synchronization.</param>
    /// <param name="cachedCallObserver">Optional observer for cached call lifecycle events.</param>
    /// <param name="siteId">The site identifier this service belongs to.</param>
    /// <param name="siteEventLogger">
    /// Optional site operational-event log. When non-null, buffer/retry/park
    /// activity is mirrored to site events (<c>store_and_forward</c> /
    /// <c>notification</c> by category). Optional with a <c>null</c> default so the
    /// many direct-construction tests still compile unchanged.
    /// </param>
    public StoreAndForwardService(
        StoreAndForwardStorage storage,
        StoreAndForwardOptions options,
        ILogger<StoreAndForwardService> logger,
        ReplicationService? replication = null,
        ICachedCallLifecycleObserver? cachedCallObserver = null,
        string siteId = "",
        ISiteEventLogger? siteEventLogger = null)
    {
        _storage = storage;
        _options = options;
        _logger = logger;
        _replication = replication;
        _cachedCallObserver = cachedCallObserver;
        _siteId = string.IsNullOrWhiteSpace(siteId) ? UnknownSiteSentinel : siteId;
        _siteEventLogger = siteEventLogger;

        // Ride the existing activity hook to emit site operational events.
        // RaiseActivity already isolates a throwing subscriber, so a failing
        // event log can never be misclassified as a transient delivery failure.
        // Only subscribe when a logger is wired so the
        // legacy (test/central) construction path stays a no-op.
        if (_siteEventLogger != null)
        {
            OnActivity += EmitSiteEvent;
        }
    }

    /// <summary>
    /// Maps one store-and-forward activity to a site operational event,
    /// following the Site Event Logging spec's per-category scope
    /// (Component-SiteEventLogging.md §"Events Logged"):
    /// <list type="bullet">
    ///   <item><description>Cached-call categories
    ///   (<see cref="StoreAndForwardCategory.ExternalSystem"/> /
    ///   <see cref="StoreAndForwardCategory.CachedDbWrite"/>) log under
    ///   <c>store_and_forward</c> for queued / retried / parked / retry-delivered
    ///   activity.</description></item>
    ///   <item><description>The site's notification forward-to-central path
    ///   (<see cref="StoreAndForwardCategory.Notification"/>) logs under
    ///   <c>notification</c> ONLY on a forward FAILURE (buffered after the
    ///   immediate forward threw) or a park (long-buffered / retries exhausted).
    ///   Routine enqueue and forward-success are deliberately NOT logged — central's
    ///   <c>Notifications</c> table is the record of audit; the site only fills the
    ///   in-transit blind spot when central is unreachable.</description></item>
    /// </list>
    /// A successful immediate cached-call <c>Delivered</c> is the normal hot path and
    /// is not logged.
    /// </summary>
    private void EmitSiteEvent(string action, StoreAndForwardCategory category, string detail)
    {
        var logger = _siteEventLogger;
        if (logger == null)
        {
            return;
        }

        // An immediate-delivery success is the normal hot path, not an
        // operational event. A retry-loop success (detail "Delivered to … after
        // N retries") IS logged for cached calls — it records a recovery.
        if (action == "Delivered" && detail.StartsWith("Immediate", StringComparison.Ordinal))
        {
            return;
        }

        if (category == StoreAndForwardCategory.Notification)
        {
            // Spec: log only forward-failure (the immediate forward threw and the
            // notification was buffered for retry — detail prefixed
            // BufferedForRetryDetailPrefix) and park. A routine "No handler
            // registered, buffered" enqueue and a forward-success "Delivered"
            // are deliberately NOT logged.
            var isForwardFailure = action == "Queued"
                && detail.StartsWith(BufferedForRetryDetailPrefix, StringComparison.Ordinal);
            if (!isForwardFailure && action != "Parked")
            {
                return;
            }

            var notifSeverity = action == "Parked" ? "Error" : "Warning";
            _ = logger.LogEventAsync(
                "notification", notifSeverity, instanceId: null,
                source: "StoreAndForwardService",
                message: $"Notification {action.ToLowerInvariant()}: {detail}");
            return;
        }

        // Cached-call categories: queued / retried / parked / retry-delivered.
        // Severity: parking is an Error (delivery abandoned for retry purposes);
        // queue/retry/requeue are Warning; a retry-loop Delivered is Info.
        var severity = action switch
        {
            "Parked" => "Error",
            "Delivered" => "Info",
            _ => "Warning",
        };

        _ = logger.LogEventAsync(
            "store_and_forward", severity, instanceId: null,
            source: "StoreAndForwardService",
            message: $"Operation {action.ToLowerInvariant()}: {detail}");
    }

    /// <summary>
    /// Registers a delivery handler for a given message category. See the
    /// <c>_deliveryHandlers</c> field documentation for the true/false/throws contract,
    /// which applies identically on the immediate and retry paths.
    /// </summary>
    /// <param name="category">The message category to handle.</param>
    /// <param name="handler">The delivery handler function.</param>
    public void RegisterDeliveryHandler(
        StoreAndForwardCategory category,
        Func<StoreAndForwardMessage, Task<bool>> handler)
    {
        _deliveryHandlers[category] = handler;
    }

    /// <summary>
    /// Initializes storage and starts the background retry timer.
    /// </summary>
    /// <returns>A task representing the asynchronous start operation.</returns>
    public async Task StartAsync()
    {
        await _storage.InitializeAsync();

        var pending = await _storage.GetMessageCountByStatusAsync(
            StoreAndForwardMessageStatus.Pending);
        if (Interlocked.CompareExchange(ref _queueDepthProviderRegistered, 1, 0) == 0)
        {
            Interlocked.Add(ref _bufferedCount, pending);
            var provider = (Func<long>)(() => Interlocked.Read(ref _bufferedCount));
            _queueDepthProvider = provider;
            ScadaBridgeTelemetry.SetQueueDepthProvider(provider);
        }

        // (Re)create the observer queue + start its single-reader pump. A prior
        // StopAsync completes the channel, so a restarted instance needs a fresh
        // one. The pump is best-effort: an observer that throws is logged and
        // swallowed so a failing audit pipeline never corrupts retry bookkeeping.
        _observerQueue = Channel.CreateUnbounded<Func<Task>>(
            new UnboundedChannelOptions { SingleReader = true });
        _observerPump = Task.Run(async () =>
        {
            await foreach (var work in _observerQueue.Reader.ReadAllAsync())
            {
                try
                {
                    await work();
                }
                catch (Exception ex)
                {
                    _logger.LogWarning(ex,
                        "Cached-call audit observer threw; ignored (best-effort per alog.md §7)");
                }
            }
        });

        _retryTimer = new Timer(
            _ => KickSweep(),
            null,
            _options.RetryTimerInterval,
            _options.RetryTimerInterval);

        _logger.LogInformation(
            "Store-and-forward service started. Retry interval: {Interval}s",
            _options.DefaultRetryInterval.TotalSeconds);
    }

    /// <summary>
    /// Stops the background retry timer and waits (bounded) for any in-flight
    /// retry sweep to finish before returning.
    /// </summary>
    /// <returns>A task representing the asynchronous stop operation.</returns>
    public async Task StopAsync()
    {
        if (_retryTimer != null)
        {
            // Stop the periodic callback first so no new sweep starts while we
            // are waiting for the in-flight one to drain.
            await _retryTimer.DisposeAsync();
            _retryTimer = null;
        }

        var inflight = Volatile.Read(ref _sweepTask);
        if (inflight is not null && !inflight.IsCompleted)
        {
            try
            {
                // WaitAsync with a finite timeout: a hung delivery handler /
                // storage call cannot block host shutdown indefinitely. On timeout
                // the sweep keeps running but the host is free to proceed with
                // disposal — preferred to never returning.
                await inflight.WaitAsync(SweepShutdownWaitTimeout).ConfigureAwait(false);
            }
            catch (TimeoutException)
            {
                _logger.LogWarning(
                    "Store-and-forward retry sweep did not finish within {Timeout}; " +
                    "shutdown is proceeding while the sweep is still in-flight",
                    SweepShutdownWaitTimeout);
            }
            catch (Exception ex)
            {
                // The sweep itself already logs at Error on failure (see
                // RetryPendingMessagesAsync's catch); we only log here so a
                // surprise fault during shutdown is still visible. Swallow so the
                // host's shutdown sequence can continue regardless.
                _logger.LogWarning(ex,
                    "Store-and-forward retry sweep faulted during shutdown wait");
            }
        }

        // Complete the observer queue and wait (bounded) for the pump to drain any
        // posted-but-not-yet-processed notifications, so shutdown does not silently
        // drop audit events that the sweep already handed off.
        _observerQueue.Writer.TryComplete();
        var pump = _observerPump;
        if (pump is not null)
        {
            try
            {
                await pump.WaitAsync(SweepShutdownWaitTimeout).ConfigureAwait(false);
            }
            catch (TimeoutException)
            {
                _logger.LogWarning(
                    "Audit-observer pump did not drain within {Timeout}", SweepShutdownWaitTimeout);
            }
            _observerPump = null;
        }

        var provider = _queueDepthProvider;
        if (provider is not null)
        {
            ScadaBridgeTelemetry.ClearQueueDepthProvider(provider);
            _queueDepthProvider = null;
        }
        Interlocked.Exchange(ref _bufferedCount, 0);
        Interlocked.Exchange(ref _queueDepthProviderRegistered, 0);
    }

    /// <summary>
    /// Enqueues a message for store-and-forward delivery.
    /// Attempts immediate delivery first. On transient failure, buffers for retry.
    /// On permanent failure (handler returns false), returns false immediately.
    ///
    /// Retry-count lifecycle — the immediate (or caller-made) delivery attempt
    /// is attempt 0 and is not counted; the background retry sweep increments
    /// <see cref="StoreAndForwardMessage.RetryCount"/> on each retry. A buffered
    /// message is parked once <c>RetryCount</c> reaches <paramref name="maxRetries"/>
    /// — <b>but only when <paramref name="maxRetries"/> is greater than 0</b>. A
    /// <paramref name="maxRetries"/> of <c>0</c> means <b>no limit</b>: the message is
    /// retried on every sweep until it is delivered and is <b>never parked</b> on a
    /// retry-count basis. It is therefore <i>not</i> a "do not retry" value — callers
    /// that want delivery abandoned after a bounded number of attempts must pass a
    /// positive <paramref name="maxRetries"/>.
    ///
    /// CachedCall idempotency note — this method does not deduplicate.
    /// The caller (e.g., ExternalSystem.CachedCall()) is responsible for ensuring
    /// that the remote system can handle duplicate deliveries safely.
    /// </summary>
    /// <param name="category">Message category (selects the delivery handler).</param>
    /// <param name="target">Target system name (external system / notification list / DB connection).</param>
    /// <param name="payloadJson">JSON-serialized call payload, treated opaquely.</param>
    /// <param name="originInstanceName">Instance that originated the message (survives instance deletion).</param>
    /// <param name="maxRetries">
    /// Maximum background retry-sweep attempts before the message is parked.
    /// <b><c>0</c> = no limit</b> — the message is retried on every sweep until
    /// delivered and is never parked for exhausting retries; it is <b>not</b> a
    /// "never retry" value. <c>null</c> uses <see cref="StoreAndForwardOptions.DefaultMaxRetries"/>.
    /// Must be positive to bound delivery attempts. Mirrors the
    /// <see cref="StoreAndForwardMessage.MaxRetries"/> contract.
    /// </param>
    /// <param name="retryInterval">Fixed interval between retry sweeps for this message; <c>null</c> uses the configured default.</param>
    /// <param name="attemptImmediateDelivery">
    /// When <c>false</c>, the caller has already made its own delivery attempt and the
    /// message is buffered directly for the retry sweep (the handler is not invoked here).
    /// </param>
    /// <param name="deferToSweep">
    /// When <c>true</c>, the handler is never invoked inline — the message is buffered
    /// due-immediately (<see cref="StoreAndForwardMessage.LastAttemptAt"/> stays null)
    /// and a background sweep is kicked. Use for callers on latency-sensitive threads
    /// (e.g. script dispatchers) whose delivery involves a long Ask timeout. Unlike
    /// <paramref name="attemptImmediateDelivery"/><c>: false</c>, no delivery attempt is
    /// presumed to have already been made, so the row is due on the very next sweep.
    /// </param>
    /// <param name="messageId">
    /// An explicit, caller-supplied message id. <c>null</c> (the default) makes the
    /// service mint a fresh GUID. The Notification Outbox enqueue path supplies its own
    /// id so the script-generated <c>NotificationId</c> is the single idempotency key —
    /// it is the buffered row's <see cref="StoreAndForwardMessage.Id"/>, it is carried
    /// inside the payload, and it is the id the forwarder submits to central.
    /// </param>
    /// <param name="executionId">
    /// The originating script execution's
    /// per-run correlation id. Threaded onto the buffered row so the retry-loop
    /// cached-call audit rows carry it. <c>null</c> for callers (e.g. notifications)
    /// that do not supply one.
    /// </param>
    /// <param name="sourceScript">
    /// The originating script identifier,
    /// threaded onto the buffered row alongside <paramref name="executionId"/>
    /// so the retry-loop audit rows carry the same provenance the script-side
    /// cached rows do. <c>null</c> when not known.
    /// </param>
    /// <param name="parentExecutionId">
    /// The <c>ExecutionId</c> of the
    /// inbound-API request that spawned the originating script execution.
    /// Threaded onto the buffered row alongside <paramref name="executionId"/>
    /// so the retry-loop cached-call audit rows carry it. <c>null</c> for a
    /// non-routed run and for callers (e.g. notifications) that
    /// do not supply one.
    /// </param>
    /// <returns>A task that resolves to a result indicating whether the message was delivered or buffered.</returns>
    public async Task<StoreAndForwardResult> EnqueueAsync(
        StoreAndForwardCategory category,
        string target,
        string payloadJson,
        string? originInstanceName = null,
        int? maxRetries = null,
        TimeSpan? retryInterval = null,
        bool attemptImmediateDelivery = true,
        bool deferToSweep = false,
        string? messageId = null,
        Guid? executionId = null,
        string? sourceScript = null,
        Guid? parentExecutionId = null)
    {
        var message = new StoreAndForwardMessage
        {
            Id = messageId ?? Guid.NewGuid().ToString("N"),
            Category = category,
            Target = target,
            PayloadJson = payloadJson,
            RetryCount = 0,
            MaxRetries = maxRetries ?? _options.DefaultMaxRetries,
            RetryIntervalMs = (long)(retryInterval ?? _options.DefaultRetryInterval).TotalMilliseconds,
            CreatedAt = DateTimeOffset.UtcNow,
            Status = StoreAndForwardMessageStatus.Pending,
            OriginInstanceName = originInstanceName,
            ExecutionId = executionId,
            SourceScript = sourceScript,
            ParentExecutionId = parentExecutionId
        };

        // Latency-sensitive caller: never run the handler inline. Buffer
        // due-immediately (LastAttemptAt intentionally left null so the row is due
        // on the very next sweep, not held back by RetryInterval) and kick a
        // background sweep so the healthy-path latency is milliseconds.
        if (deferToSweep)
        {
            await BufferAsync(message);
            RaiseActivity("Queued", category, $"Deferred to sweep: {target}");
            TriggerSweep();
            return new StoreAndForwardResult(true, message.Id, true);
        }

        // Attempt immediate delivery — unless the caller has already made a
        // delivery attempt of its own (attemptImmediateDelivery: false). In that
        // case re-invoking the handler here would dispatch the request twice.
        if (attemptImmediateDelivery && _deliveryHandlers.TryGetValue(category, out var handler))
        {
            try
            {
                var success = await handler(message);
                if (success)
                {
                    RaiseActivity("Delivered", category, $"Immediate delivery to {target}");
                    return new StoreAndForwardResult(true, message.Id, false);
                }

                // Permanent failure — do not buffer
                return new StoreAndForwardResult(false, message.Id, false);
            }
            catch (Exception ex)
            {
                // Transient failure — buffer for retry. The immediate attempt is
                // attempt 0; RetryCount tracks only sweep retries, so it stays 0
                // here.
                _logger.LogWarning(ex,
                    "Immediate delivery to {Target} failed (transient), buffering for retry",
                    target);

                message.LastAttemptAt = DateTimeOffset.UtcNow;
                message.LastError = ex.Message;
                await BufferAsync(message);

                RaiseActivity("Queued", category, $"{BufferedForRetryDetailPrefix}: {target} ({ex.Message})");
                return new StoreAndForwardResult(true, message.Id, true);
            }
        }

        // Either no handler is registered yet, or the caller already attempted
        // delivery itself — buffer for the background retry sweep to deliver.
        // The initial attempt (caller-made, or skipped because no handler is
        // registered) is attempt 0; RetryCount tracks only sweep retries and
        // therefore stays 0 here.
        if (!attemptImmediateDelivery)
        {
            message.LastAttemptAt = DateTimeOffset.UtcNow;
        }
        await BufferAsync(message);
        RaiseActivity("Queued", category, attemptImmediateDelivery
            ? $"No handler registered, buffered: {target}"
            : $"{BufferedForRetryDetailPrefix}: {target}");
        return new StoreAndForwardResult(true, message.Id, true);
    }

    /// <summary>
    /// Persists a message to the local SQLite buffer and replicates the
    /// add to the standby node so a failover does not lose the buffered message.
    /// </summary>
    private async Task BufferAsync(StoreAndForwardMessage message)
    {
        await _storage.EnqueueAsync(message);
        _replication?.ReplicateEnqueue(message);
        Interlocked.Increment(ref _bufferedCount);
    }

    /// <summary>
    /// Kicks a background retry sweep now (fire-and-forget). No-op before
    /// <see cref="StartAsync"/> (storage may be uninitialized and the timer, whose
    /// presence gates this, is not yet set). Overlap-safe: publishes into
    /// <see cref="_sweepTask"/> only when the CAS is won — see <see cref="KickSweep"/>.
    /// </summary>
    public void TriggerSweep()
    {
        if (_retryTimer == null) return;
        KickSweep();
    }

    /// <summary>The current drain handle — test seam for the N3 clobber regression.</summary>
    internal Task? CurrentSweepTaskForTest => Volatile.Read(ref _sweepTask);

    /// <summary>
    /// Starts a sweep IF none is in flight, publishing the new task into
    /// <see cref="_sweepTask"/> only when this call wins the <see cref="_retryInProgress"/>
    /// CAS. A kick that loses the CAS returns without touching <see cref="_sweepTask"/> —
    /// pre-fix it overwrote the drain handle with an instantly-completed no-op, so
    /// <see cref="StopAsync"/> proceeded with disposal under a still-running sweep
    /// (review 02 round 2, N3) — defeating the drain exactly when a sweep outlives a tick.
    /// </summary>
    private void KickSweep()
    {
        if (Interlocked.CompareExchange(ref _retryInProgress, 1, 0) != 0)
            return;
        Volatile.Write(ref _sweepTask, RunSweepOwnedAsync());
    }

    /// <summary>
    /// Background retry sweep. Processes all pending messages that are due for retry.
    /// Self-CASes for direct callers (existing tests); the entry CAS is skipped when
    /// ownership is already held by <see cref="KickSweep"/>.
    /// </summary>
    /// <returns>A task representing the asynchronous retry sweep.</returns>
    internal async Task RetryPendingMessagesAsync()
    {
        // Prevent overlapping retry sweeps
        if (Interlocked.CompareExchange(ref _retryInProgress, 1, 0) != 0)
            return;
        await RunSweepOwnedAsync();
    }

    /// <summary>
    /// The actual retry sweep body. The caller MUST already own the
    /// <see cref="_retryInProgress"/> flag (won the CAS); this method releases it in its
    /// finally. Never call directly without holding ownership.
    /// </summary>
    private async Task RunSweepOwnedAsync()
    {
        try
        {
            var gate = _deliveryGate;
            if (gate != null)
            {
                bool isActive;
                try { isActive = gate(); }
                catch (Exception ex)
                {
                    _logger.LogWarning(ex,
                        "S&F delivery gate threw; treating this node as standby for this sweep");
                    isActive = false;
                }
                if (!isActive)
                {
                    _logger.LogDebug("S&F retry sweep skipped: this node is not the active site node");
                    return;
                }
            }

            var messages = await _storage.GetMessagesForRetryAsync(_options.SweepBatchLimit);
            if (messages.Count == 0) return;

            _logger.LogDebug("Retry sweep: {Count} messages due for retry", messages.Count);

            // Group due rows into per-(category,target) lanes. GroupBy is stable, so
            // each lane preserves created_at-ASC (oldest-first) order. Lanes run
            // concurrently up to SweepTargetParallelism, so one slow/dead target no
            // longer blocks delivery to healthy ones; within a lane delivery stays
            // strictly sequential (per-target FIFO).
            var lanes = messages
                .GroupBy(m => (m.Category, m.Target))
                .Select(g => g.ToList())
                .ToList();

            using var laneCap = new SemaphoreSlim(Math.Max(1, _options.SweepTargetParallelism));
            var laneTasks = lanes.Select(async lane =>
            {
                await laneCap.WaitAsync();
                try
                {
                    foreach (var message in lane)
                    {
                        // Short-circuit, per lane: one transient failure means
                        // the target is down — skip the rest of this lane this sweep
                        // rather than burning a full timeout per remaining message.
                        // Skipped rows keep their RetryCount/LastAttemptAt untouched.
                        var outcome = await RetryMessageAsync(message);
                        if (outcome == RetryOutcome.TransientFailure)
                            return;
                    }
                }
                finally { laneCap.Release(); }
            }).ToList();
            await Task.WhenAll(laneTasks);
        }
        catch (Exception ex)
        {
            _logger.LogError(ex, "Error during retry sweep");
        }
        finally
        {
            Interlocked.Exchange(ref _retryInProgress, 0);
        }
    }

    /// <summary>
    /// Outcome of a single message's retry attempt, used by the sweep to
    /// short-circuit a <c>(category, target)</c> lane after its first transient
    /// failure. <see cref="Skipped"/> = no attempt was made (no handler, or the
    /// row's status changed under a concurrent apply); its RetryCount is untouched.
    /// </summary>
    internal enum RetryOutcome { Delivered, Parked, TransientFailure, Skipped }

    private async Task<RetryOutcome> RetryMessageAsync(StoreAndForwardMessage message)
    {
        if (!_deliveryHandlers.TryGetValue(message.Category, out var handler))
        {
            _logger.LogWarning("No delivery handler for category {Category}", message.Category);
            return RetryOutcome.Skipped;
        }

        // Measure per-attempt
        // duration so the audit row carries a meaningful DurationMs. Captured
        // around the handler invocation only — storage / replication overhead
        // is excluded.
        var attemptStartUtc = DateTime.UtcNow;
        var attemptStopwatch = System.Diagnostics.Stopwatch.StartNew();

        try
        {
            var success = await handler(message);
            attemptStopwatch.Stop();
            if (success)
            {
                await _storage.RemoveMessageAsync(message.Id);
                _replication?.ReplicateRemove(message.Id);
                Interlocked.Decrement(ref _bufferedCount);
                RaiseActivity("Delivered", message.Category,
                    $"Delivered to {message.Target} after {message.RetryCount} retries");

                // Terminal Delivered observer notification — the audit
                // bridge maps this to Attempted + CachedResolve(Delivered).
                await PostObserverNotification(
                    message,
                    CachedCallAttemptOutcome.Delivered,
                    lastError: null,
                    httpStatus: null,
                    occurredAtUtc: attemptStartUtc,
                    durationMs: (int)attemptStopwatch.ElapsedMilliseconds);
                return RetryOutcome.Delivered;
            }

            // Permanent failure on retry — park immediately.
            message.Status = StoreAndForwardMessageStatus.Parked;
            message.LastAttemptAt = DateTimeOffset.UtcNow;
            message.LastError = "Permanent failure (handler returned false)";
            var parked = await _storage.UpdateMessageIfStatusAsync(
                message, StoreAndForwardMessageStatus.Pending);
            if (!parked)
            {
                _logger.LogDebug(
                    "Message {MessageId} changed status during delivery; sweep park skipped",
                    message.Id);
                return RetryOutcome.Skipped;
            }
            Interlocked.Decrement(ref _bufferedCount);
            _replication?.ReplicatePark(message);
            RaiseActivity("Parked", message.Category,
                $"Permanent failure for {message.Target}: handler returned false");

            // Terminal PermanentFailure observer notification — the
            // audit bridge maps this to Attempted(Failed) + CachedResolve(Parked).
            await PostObserverNotification(
                message,
                CachedCallAttemptOutcome.PermanentFailure,
                lastError: message.LastError,
                httpStatus: null,
                occurredAtUtc: attemptStartUtc,
                durationMs: (int)attemptStopwatch.ElapsedMilliseconds);
            return RetryOutcome.Parked;
        }
        catch (Exception ex)
        {
            attemptStopwatch.Stop();
            // Transient failure — increment retry, check max
            message.RetryCount++;
            message.LastAttemptAt = DateTimeOffset.UtcNow;
            message.LastError = ex.Message;

            if (message.MaxRetries > 0 && message.RetryCount >= message.MaxRetries)
            {
                message.Status = StoreAndForwardMessageStatus.Parked;
                var parked = await _storage.UpdateMessageIfStatusAsync(
                    message, StoreAndForwardMessageStatus.Pending);
                if (!parked)
                {
                    _logger.LogDebug(
                        "Message {MessageId} changed status during delivery; sweep park skipped",
                        message.Id);
                    return RetryOutcome.Skipped;
                }
                Interlocked.Decrement(ref _bufferedCount);
                _replication?.ReplicatePark(message);
                RaiseActivity("Parked", message.Category,
                    $"Max retries ({message.MaxRetries}) reached for {message.Target}");
                _logger.LogWarning(
                    "Message {MessageId} parked after {MaxRetries} retries to {Target}",
                    message.Id, message.MaxRetries, message.Target);

                // Terminal ParkedMaxRetries observer notification — the
                // audit bridge maps this to Attempted(Failed) + CachedResolve(Parked).
                await PostObserverNotification(
                    message,
                    CachedCallAttemptOutcome.ParkedMaxRetries,
                    lastError: ex.Message,
                    httpStatus: null,
                    occurredAtUtc: attemptStartUtc,
                    durationMs: (int)attemptStopwatch.ElapsedMilliseconds);
                return RetryOutcome.Parked;
            }
            else
            {
                if (!await _storage.UpdateMessageIfStatusAsync(
                        message, StoreAndForwardMessageStatus.Pending))
                {
                    _logger.LogDebug(
                        "Message {MessageId} changed status during delivery; sweep retry-count update skipped",
                        message.Id);
                    return RetryOutcome.Skipped;
                }
                RaiseActivity("Retried", message.Category,
                    $"Retry {message.RetryCount}/{message.MaxRetries} for {message.Target}: {ex.Message}");

                // Per-attempt TransientFailure observer notification —
                // the audit bridge maps this to Attempted(Failed).
                await PostObserverNotification(
                    message,
                    CachedCallAttemptOutcome.TransientFailure,
                    lastError: ex.Message,
                    httpStatus: null,
                    occurredAtUtc: attemptStartUtc,
                    durationMs: (int)attemptStopwatch.ElapsedMilliseconds);
                return RetryOutcome.TransientFailure;
            }
        }
    }

    /// <summary>
    /// Hands a cached-call observer notification to the ordered single-reader pump
    /// (see <see cref="_observerQueue"/>) and returns immediately, so a slow
    /// observer cannot stretch the sweep. When the service is not started (no pump —
    /// e.g. a unit test driving <see cref="RetryPendingMessagesAsync"/> directly),
    /// the notification is processed inline so it is still delivered synchronously.
    /// The buffered <paramref name="message"/> is not mutated again after this call
    /// within a sweep (the next tick loads fresh row instances), so the deferred
    /// read of its fields by the pump is safe.
    /// </summary>
    private ValueTask PostObserverNotification(
        StoreAndForwardMessage message,
        CachedCallAttemptOutcome outcome,
        string? lastError,
        int? httpStatus,
        DateTime occurredAtUtc,
        int? durationMs)
    {
        if (_cachedCallObserver == null)
        {
            return ValueTask.CompletedTask;
        }

        Func<Task> work = () => NotifyCachedCallObserverAsync(
            message, outcome, lastError, httpStatus, occurredAtUtc, durationMs);

        if (_observerPump is null)
        {
            // Not started: process inline (preserves prior test behavior).
            return new ValueTask(work());
        }

        _observerQueue.Writer.TryWrite(work);
        return ValueTask.CompletedTask;
    }

    /// <summary>
    /// Notify the registered
    /// <see cref="ICachedCallLifecycleObserver"/> of the just-completed
    /// attempt. Invoked from the single-reader observer pump (latency-isolated
    /// from the sweep; order preserved per posting sequence) — or inline when the
    /// service is not started. Only fires for cached-call categories
    /// (<see cref="StoreAndForwardCategory.ExternalSystem"/> and
    /// <see cref="StoreAndForwardCategory.CachedDbWrite"/>); the
    /// <see cref="StoreAndForwardCategory.Notification"/> category has its
    /// own central-side audit pipeline (Notification Outbox) and must
    /// not surface on this hook.
    /// </summary>
    /// <remarks>
    /// Best-effort: an observer that throws is logged and swallowed so a
    /// failing audit pipeline cannot corrupt S&amp;F retry bookkeeping
    /// (alog.md §7 contract). Messages whose ids are not valid GUIDs (callers
    /// that didn't thread a TrackedOperationId in) are silently
    /// skipped — the observer requires a parseable id by contract.
    /// </remarks>
    private async Task NotifyCachedCallObserverAsync(
        StoreAndForwardMessage message,
        CachedCallAttemptOutcome outcome,
        string? lastError,
        int? httpStatus,
        DateTime occurredAtUtc,
        int? durationMs)
    {
        if (_cachedCallObserver == null)
        {
            return;
        }

        // Only cached-call categories generate audit telemetry on this hook —
        // notifications have their own outbox-side audit pipeline.
        var channel = message.Category switch
        {
            StoreAndForwardCategory.ExternalSystem => "ApiOutbound",
            StoreAndForwardCategory.CachedDbWrite => "DbOutbound",
            _ => null,
        };
        if (channel is null)
        {
            return;
        }

        if (!TrackedOperationId.TryParse(message.Id, out var trackedId))
        {
            _logger.LogWarning(
                "Cached-call audit observer skipped: message id {MessageId} is not a parseable TrackedOperationId (category {Category}, outcome {Outcome}). " +
                "Audit lifecycle for this operation will have no rows.",
                message.Id, message.Category, outcome);
            return;
        }

        CachedCallAttemptContext context;
        try
        {
            context = new CachedCallAttemptContext(
                TrackedOperationId: trackedId,
                Channel: channel,
                Target: message.Target,
                SourceSite: _siteId,
                Outcome: outcome,
                RetryCount: message.RetryCount,
                LastError: lastError,
                HttpStatus: httpStatus,
                CreatedAtUtc: message.CreatedAt.UtcDateTime,
                OccurredAtUtc: DateTime.SpecifyKind(occurredAtUtc, DateTimeKind.Utc),
                DurationMs: durationMs,
                SourceInstanceId: message.OriginInstanceName,
                // The buffered message carries the originating script
                // execution's ExecutionId + SourceScript; surface them on the
                // context so the bridge can stamp the retry-loop cached audit
                // rows. Null on rows buffered before this field existed
                // (back-compat).
                ExecutionId: message.ExecutionId,
                SourceScript: message.SourceScript,
                // The buffered message also carries the spawning inbound-API
                // request's ExecutionId; surface it so the bridge stamps it
                // onto the retry-loop cached rows. Null for a non-routed run
                // and on rows buffered before this field existed (back-compat).
                ParentExecutionId: message.ParentExecutionId);
        }
        catch (Exception buildEx)
        {
            // Defensive — record construction shouldn't throw, but the alog.md
            // §7 contract requires this path be exception-safe regardless.
            _logger.LogWarning(buildEx,
                "Failed to build cached-call attempt context for {MessageId}; observer skipped",
                message.Id);
            return;
        }

        try
        {
            await _cachedCallObserver.OnAttemptCompletedAsync(context, CancellationToken.None)
                .ConfigureAwait(false);
        }
        catch (Exception ex)
        {
            // alog.md §7 best-effort: an audit observer outage must NEVER be
            // misclassified as a transient delivery failure or corrupt the
            // S&F retry bookkeeping.
            _logger.LogWarning(ex,
                "ICachedCallLifecycleObserver threw for {MessageId} (Outcome {Outcome}); ignored",
                message.Id, outcome);
        }
    }

    /// <summary>
    /// Gets parked messages for central query (Pattern 8).
    /// </summary>
    /// <param name="category">Optional category filter, or null for all categories.</param>
    /// <param name="pageNumber">The page number (1-based).</param>
    /// <param name="pageSize">The page size.</param>
    /// <returns>A tuple of parked messages and the total count.</returns>
    public async Task<(List<StoreAndForwardMessage> Messages, int TotalCount)> GetParkedMessagesAsync(
        StoreAndForwardCategory? category = null,
        int pageNumber = 1,
        int pageSize = 50)
    {
        return await _storage.GetParkedMessagesAsync(category, pageNumber, pageSize);
    }

    /// <summary>
    /// Retries a parked message (moves back to pending queue).
    ///
    /// An operator requeue is a buffer state change and is
    /// replicated to the standby (as a <see cref="ReplicationOperationType.Requeue"/>)
    /// so a failover preserves the operator's retry intent.
    /// The activity-log entry carries the message's true
    /// category rather than a hard-coded one.
    /// The parked row is captured <i>before</i> the local
    /// requeue write rather than re-read after it, so a concurrent
    /// <c>RemoveMessageAsync</c> or <c>DiscardParkedMessageAsync</c> running
    /// between the two storage calls cannot leave the standby in <c>Parked</c>
    /// while the active node has already requeued — we always have the row in
    /// hand for the <c>Requeue</c> replication.
    /// </summary>
    /// <param name="messageId">The identifier of the message to retry.</param>
    /// <returns>True if successfully retried, false otherwise.</returns>
    public async Task<bool> RetryParkedMessageAsync(string messageId)
    {
        var captured = await _storage.GetMessageByIdAsync(messageId);
        if (captured is null || captured.Status != StoreAndForwardMessageStatus.Parked)
        {
            return false;
        }

        var success = await _storage.RetryParkedMessageAsync(messageId);
        if (!success)
        {
            return false;
        }

        Interlocked.Increment(ref _bufferedCount);

        // The active node just rewrote this row to Pending with retry_count = 0
        // and cleared last_error / last_attempt_at (see
        // StoreAndForwardStorage.RetryParkedMessageAsync). Reconstruct the
        // post-requeue state on the captured POCO so the standby applies the
        // same mutations even if a concurrent writer has already deleted the
        // row underneath us.
        captured.Status = StoreAndForwardMessageStatus.Pending;
        captured.RetryCount = 0;
        captured.LastError = null;
        captured.LastAttemptAt = null;
        _replication?.ReplicateRequeue(captured);

        RaiseActivity("Retry", captured.Category,
            $"Parked message {messageId} moved back to queue");
        return true;
    }

    /// <summary>
    /// Permanently discards a parked message.
    ///
    /// An operator discard is a buffer removal and is replicated
    /// to the standby (as a <see cref="ReplicationOperationType.Remove"/>) so the
    /// discarded message does not reappear after a failover.
    /// The activity-log entry carries the message's true
    /// category rather than a hard-coded one.
    /// </summary>
    /// <param name="messageId">The identifier of the message to discard.</param>
    /// <returns>True if successfully discarded, false otherwise.</returns>
    public async Task<bool> DiscardParkedMessageAsync(string messageId)
    {
        // Capture the category before the row is deleted so the activity log is
        // labelled correctly.
        var message = await _storage.GetMessageByIdAsync(messageId);
        var success = await _storage.DiscardParkedMessageAsync(messageId);
        if (success)
        {
            _replication?.ReplicateRemove(messageId);
            RaiseActivity("Discard", message?.Category ?? StoreAndForwardCategory.ExternalSystem,
                $"Parked message {messageId} discarded");
        }
        return success;
    }

    /// <summary>
    /// Gets buffer depth by category for health reporting.
    /// </summary>
    /// <returns>A dictionary of buffer depths by category.</returns>
    public async Task<Dictionary<StoreAndForwardCategory, int>> GetBufferDepthAsync()
    {
        return await _storage.GetBufferDepthByCategoryAsync();
    }

    /// <summary>
    /// Gets count of S&amp;F messages for a given instance (for verifying survival on deletion).
    /// </summary>
    /// <param name="instanceName">The instance name to query.</param>
    /// <returns>The number of messages originating from the instance.</returns>
    public async Task<int> GetMessageCountForInstanceAsync(string instanceName)
    {
        return await _storage.GetMessageCountByOriginInstanceAsync(instanceName);
    }

    /// <summary>
    /// Notification Outbox: looks up a buffered message by its id, or <c>null</c> if it
    /// is not (or no longer) in the buffer. <c>Notify.Status</c> uses this to detect a
    /// notification still in transit at the site — central reports it not-found while
    /// the S&amp;F buffer still holds it, which is the site-local <c>Forwarding</c> state.
    /// </summary>
    /// <param name="messageId">The message identifier.</param>
    /// <returns>The message, or null if not found.</returns>
    public async Task<StoreAndForwardMessage?> GetMessageByIdAsync(string messageId)
    {
        return await _storage.GetMessageByIdAsync(messageId);
    }

    /// <summary>
    /// Raises the S&amp;F activity notification. The
    /// delegate is snapshotted (so a concurrent unsubscribe cannot NRE) and every
    /// subscriber invocation is wrapped so a slow/throwing subscriber (e.g. the site
    /// event log) cannot abort the caller. Crucially, a subscriber exception raised
    /// from <see cref="EnqueueAsync"/> or <c>RetryMessageAsync</c> must NOT be
    /// misclassified as a transient delivery failure — pre-fix it escaped into the
    /// delivery try/catch and caused a successfully delivered message to be buffered
    /// (or its retry count to be bumped). Activity logging is best-effort.
    /// </summary>
    private void RaiseActivity(string action, StoreAndForwardCategory category, string detail)
    {
        var handlers = OnActivity;
        if (handlers == null) return;

        foreach (var handler in handlers.GetInvocationList().Cast<Action<string, StoreAndForwardCategory, string>>())
        {
            try
            {
                handler(action, category, detail);
            }
            catch (Exception ex)
            {
                _logger.LogWarning(ex,
                    "Store-and-forward activity subscriber threw for action {Action}; ignored",
                    action);
            }
        }
    }
}

/// <summary>
/// Result of an enqueue operation.
/// </summary>
public record StoreAndForwardResult(
    /// <summary>True if the message was accepted (either delivered immediately or buffered).</summary>
    bool Accepted,
    /// <summary>Unique message ID for tracking.</summary>
    string MessageId,
    /// <summary>True if the message was buffered (not delivered immediately).</summary>
    bool WasBuffered);