feat(grpc): site-side ICentralTransport seam + gRPC transport (T1A.3)

Introduce ICentralTransport as the site->central choke point inside
SiteCommunicationActor. The seven site->central sends (notification submit/
status, audit + cached-telemetry ingest, reconcile, health, heartbeat) now
delegate to an injected transport instead of owning ClusterClient.Send inline.

- AkkaCentralTransport: verbatim extraction of today's ClusterClient.Send path,
  including the exact sender-forwarding that routes central's reply straight
  back to the waiting Ask. Default when no transport is injected -> behaviour
  unchanged, existing SiteCommunicationActorTests pass as-is.
- GrpcCentralTransport + CentralChannelProvider: dial CentralControlService with
  sticky failover + background failback (1s-doubling-cap-60s), PSK +
  x-scadabridge-site via ControlPlaneCredentials, per-call deadlines mirroring
  today's Ask timeouts. Cross-node retry ONLY on provably-unsent
  connect failures; never on DeadlineExceeded. Heartbeat stays fire-and-forget.
- StaticSitePskProvider: site's single own-key provider (fail-closed).
- CommunicationOptions: CentralTransport flag (default Akka) + CentralGrpcEndpoints
  (validator: required when transport=Grpc). Host selects the impl; the
  ClusterClient is created only on the Akka path.

Tests: actor-with-fake-transport (7 delegations + fault routing + heartbeat
no-fault), AkkaCentralTransport sender-forwarding, GrpcCentralTransport over
in-process TestServer (failover flip, sticky, failback, PSK+header, deadline,
no-retry-on-deadline), validator. Communication.Tests 371 green, Host.Tests 391
green; the three above-seam suites pass unmodified.
This commit is contained in:
Joseph Doherty
2026-07-22 19:29:14 -04:00
parent 780bb9c369
commit 33b15f10a4
13 changed files with 1642 additions and 159 deletions
@@ -0,0 +1,373 @@
using System.Collections.Concurrent;
using Akka.Actor;
using Microsoft.AspNetCore.Builder;
using Microsoft.AspNetCore.Hosting;
using Microsoft.AspNetCore.TestHost;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Hosting;
using Microsoft.Extensions.Logging.Abstractions;
using Microsoft.Extensions.Options;
using ZB.MOM.WW.ScadaBridge.Commons.Messages.Notification;
using ZB.MOM.WW.ScadaBridge.Communication;
using ZB.MOM.WW.ScadaBridge.Communication.Grpc;
namespace ZB.MOM.WW.ScadaBridge.Host.Tests;
/// <summary>
/// T1A.3: <see cref="GrpcCentralTransport"/> + <see cref="CentralChannelProvider"/> over a real
/// gRPC stack (two in-process <see cref="TestServer"/> central nodes, the real
/// <see cref="CentralControlGrpcService"/> and <see cref="CentralControlAuthInterceptor"/>). Proves
/// the sticky failover/failback policy, the PSK + site-header attachment, the per-call deadline,
/// and — the hard rule — no cross-node retry on <c>DeadlineExceeded</c>.
/// </summary>
/// <remarks>
/// A "down" node is modelled by a <see cref="ToggleHandler"/> that throws before reaching the
/// TestServer, so BOTH the unary call and the failback <c>Heartbeat</c> probe see it as
/// <c>Unavailable</c> — the honest shape of a refused connection, and the only class the transport
/// fails over on. Readiness is always set, so a node that is "up" answers everything.
/// </remarks>
public class GrpcCentralTransportTests : IAsyncLifetime
{
private const string SiteA = "site-a";
private const string SiteAKey = "site-a-preshared-key";
private const string EndpointA = "http://central-a/";
private const string EndpointB = "http://central-b/";
private ActorSystem _system = null!;
private CentralNode _nodeA = null!;
private CentralNode _nodeB = null!;
/// <inheritdoc />
public async Task InitializeAsync()
{
_system = ActorSystem.Create("grpc-central-transport-test");
_nodeA = await CentralNode.StartAsync(_system, "A", SiteA, SiteAKey, repliesToSubmit: true);
_nodeB = await CentralNode.StartAsync(_system, "B", SiteA, SiteAKey, repliesToSubmit: true);
}
/// <inheritdoc />
public async Task DisposeAsync()
{
await _nodeA.DisposeAsync();
await _nodeB.DisposeAsync();
await _system.Terminate();
}
private CentralChannelProvider NewProvider(string? pskKey = SiteAKey) => new(
new[] { EndpointA, EndpointB },
new FixedPskProvider(pskKey),
SiteA,
new CommunicationOptions(),
NullLogger<CentralChannelProvider>.Instance,
handlerFactory: HandlerFor,
probeDeadline: TimeSpan.FromSeconds(2),
backoffBase: TimeSpan.FromMilliseconds(50),
backoffCap: TimeSpan.FromMilliseconds(200));
private HttpMessageHandler HandlerFor(string endpoint) => endpoint == EndpointA
? new ToggleHandler(_nodeA.Server.CreateHandler(), () => _nodeA.IsUp)
: new ToggleHandler(_nodeB.Server.CreateHandler(), () => _nodeB.IsUp);
private GrpcCentralTransport NewTransport(CentralChannelProvider provider, CommunicationOptions? options = null)
=> new(provider, options ?? new CommunicationOptions(), NullLogger<GrpcCentralTransport>.Instance);
[Fact]
public async Task HappyPath_ReachesThePreferredNode_AndRoutesTheAckBack()
{
using var provider = NewProvider();
var transport = NewTransport(provider);
var inbox = new Capture(_system);
transport.SubmitNotification(NewSubmit("n1"), inbox.Ref);
var ack = Assert.IsType<NotificationSubmitAck>(inbox.Receive(TimeSpan.FromSeconds(5)));
Assert.True(ack.Accepted);
Assert.Equal("n1", ack.NotificationId);
Assert.Equal(0, provider.CurrentIndex); // stayed on preferred
Assert.Equal(1, _nodeA.SubmitCount);
Assert.Equal(0, _nodeB.SubmitCount);
}
[Fact]
public async Task Sticky_StaysOnThePreferredNode_WhileHealthy()
{
using var provider = NewProvider();
var transport = NewTransport(provider);
for (var i = 0; i < 4; i++)
{
var inbox = new Capture(_system);
transport.SubmitNotification(NewSubmit($"n{i}"), inbox.Ref);
Assert.IsType<NotificationSubmitAck>(inbox.Receive(TimeSpan.FromSeconds(5)));
}
Assert.Equal(0, provider.CurrentIndex);
Assert.Equal(4, _nodeA.SubmitCount);
Assert.Equal(0, _nodeB.SubmitCount);
}
[Fact]
public async Task Failover_FlipsToThePeer_WhenThePreferredIsUnavailable()
{
using var provider = NewProvider();
var transport = NewTransport(provider);
_nodeA.IsUp = false; // preferred refuses connections
var inbox = new Capture(_system);
transport.SubmitNotification(NewSubmit("n1"), inbox.Ref);
var ack = Assert.IsType<NotificationSubmitAck>(inbox.Receive(TimeSpan.FromSeconds(5)));
Assert.True(ack.Accepted);
Assert.Equal(1, provider.CurrentIndex); // flipped to the peer
Assert.Equal(0, _nodeA.SubmitCount);
Assert.Equal(1, _nodeB.SubmitCount);
}
[Fact]
public async Task Failback_ReturnsToThePreferred_OnceItIsReachableAgain()
{
using var provider = NewProvider();
var transport = NewTransport(provider);
// Take the preferred down and drive one call so we flip to the peer + arm the failback probe.
_nodeA.IsUp = false;
var inbox = new Capture(_system);
transport.SubmitNotification(NewSubmit("n1"), inbox.Ref);
Assert.IsType<NotificationSubmitAck>(inbox.Receive(TimeSpan.FromSeconds(5)));
Assert.Equal(1, provider.CurrentIndex);
// Bring the preferred back; the background probe should fail us back within a few backoffs.
_nodeA.IsUp = true;
await WaitUntil(() => provider.CurrentIndex == 0, TimeSpan.FromSeconds(5));
Assert.Equal(0, provider.CurrentIndex);
// New calls resume on the preferred node.
var inbox2 = new Capture(_system);
transport.SubmitNotification(NewSubmit("n2"), inbox2.Ref);
Assert.IsType<NotificationSubmitAck>(inbox2.Receive(TimeSpan.FromSeconds(5)));
Assert.True(_nodeA.SubmitCount >= 1);
}
[Fact]
public async Task PskAndSiteHeader_AreAttached_SoTheGatedCallReachesTheService()
{
// The service is gated by CentralControlAuthInterceptor; a call that reaches it (and gets
// Accepted) proves both the bearer PSK and the x-scadabridge-site header were attached.
using var provider = NewProvider(pskKey: SiteAKey);
var transport = NewTransport(provider);
var inbox = new Capture(_system);
transport.SubmitNotification(NewSubmit("n1"), inbox.Ref);
var ack = Assert.IsType<NotificationSubmitAck>(inbox.Receive(TimeSpan.FromSeconds(5)));
Assert.True(ack.Accepted);
}
[Fact]
public async Task WrongPsk_IsRejected_AndNotRetriedOnThePeer()
{
// PermissionDenied is not a connect failure — the transport surfaces it as a transient
// Status.Failure without flipping to the peer.
using var provider = NewProvider(pskKey: "the-wrong-key");
var transport = NewTransport(provider);
var inbox = new Capture(_system);
transport.SubmitNotification(NewSubmit("n1"), inbox.Ref);
Assert.IsType<Status.Failure>(inbox.Receive(TimeSpan.FromSeconds(5)));
Assert.Equal(0, provider.CurrentIndex); // no flip
Assert.Equal(0, _nodeB.SubmitCount); // peer never tried
}
[Fact]
public async Task DeadlineExceeded_IsNotRetriedOnThePeer()
{
// THE hard rule. Node A is UP but never replies, so the call deadlines. The transport must
// surface Status.Failure and must NOT try node B (the call may already have executed).
_nodeA.SetBlackHole();
var shortDeadline = new CommunicationOptions { NotificationForwardTimeout = TimeSpan.FromMilliseconds(300) };
using var provider = NewProvider();
var transport = NewTransport(provider, shortDeadline);
var inbox = new Capture(_system);
transport.SubmitNotification(NewSubmit("n1"), inbox.Ref);
// A per-call deadline is applied (the call returns fast instead of hanging on the black hole).
Assert.IsType<Status.Failure>(inbox.Receive(TimeSpan.FromSeconds(5)));
Assert.Equal(0, provider.CurrentIndex); // no failover on a deadline
Assert.Equal(0, _nodeB.SubmitCount); // peer never tried
}
private static NotificationSubmit NewSubmit(string id) => new(
NotificationId: id,
ListName: "ops",
Subject: "s",
Body: "b",
SourceSiteId: SiteA,
SourceInstanceId: null,
SourceScript: null,
SiteEnqueuedAt: DateTimeOffset.UtcNow);
private static async Task WaitUntil(Func<bool> condition, TimeSpan timeout)
{
var deadline = DateTime.UtcNow + timeout;
while (DateTime.UtcNow < deadline)
{
if (condition())
{
return;
}
await Task.Delay(25);
}
}
/// <summary>
/// A raw message sink used as the transport's <c>replyTo</c>. Unlike Akka's <c>Inbox</c>, which
/// rethrows a <see cref="Status.Failure"/>'s cause on receive, this captures every message
/// verbatim so a test can assert on the <see cref="Status.Failure"/> itself.
/// </summary>
private sealed class Capture
{
private readonly BlockingCollection<object> _messages = new();
public Capture(ActorSystem system)
{
Ref = system.ActorOf(Props.Create(() => new CaptureActor(_messages)));
}
public IActorRef Ref { get; }
public object Receive(TimeSpan timeout)
=> _messages.TryTake(out var message, timeout)
? message
: throw new TimeoutException("No message captured within the timeout.");
private sealed class CaptureActor : ReceiveActor
{
public CaptureActor(BlockingCollection<object> messages) => ReceiveAny(messages.Add);
}
}
/// <summary>A gRPC channel handler that throws (a refused connection) while its node is "down".</summary>
private sealed class ToggleHandler : DelegatingHandler
{
private readonly Func<bool> _isUp;
public ToggleHandler(HttpMessageHandler inner, Func<bool> isUp)
{
InnerHandler = inner;
_isUp = isUp;
}
protected override async Task<HttpResponseMessage> SendAsync(HttpRequestMessage request, CancellationToken cancellationToken)
{
if (!_isUp())
{
throw new HttpRequestException("simulated central node down");
}
return await base.SendAsync(request, cancellationToken).ConfigureAwait(false);
}
}
private sealed class FixedPskProvider(string? key) : ISitePskProvider
{
public ValueTask<string> GetAsync(string siteId, CancellationToken ct)
=> key is null ? throw new InvalidOperationException("no key") : new ValueTask<string>(key);
public void Invalidate(string siteId) { }
}
/// <summary>One in-process central node: TestServer + real service/interceptor + a stub actor.</summary>
private sealed class CentralNode : IAsyncDisposable
{
private IHost _host = null!;
private IActorRef _stub = null!;
private readonly StubCounters _counters = new();
public TestServer Server { get; private set; } = null!;
public volatile bool IsUp = true;
public int SubmitCount => _counters.Submits;
public static async Task<CentralNode> StartAsync(
ActorSystem system, string label, string site, string key, bool repliesToSubmit)
{
var node = new CentralNode();
node._stub = system.ActorOf(
Props.Create(() => new StubCentralActor(node._counters, repliesToSubmit)), $"stub-{label}");
var service = new CentralControlGrpcService(
NullLogger<CentralControlGrpcService>.Instance,
Options.Create(new CommunicationOptions()));
service.SetReady(node._stub);
var psk = new MapPskProvider(new Dictionary<string, string> { [site] = key });
node._host = await new HostBuilder()
.ConfigureWebHost(web => web
.UseTestServer()
.ConfigureServices(services =>
{
services.AddGrpc(o => o.Interceptors.Add<CentralControlAuthInterceptor>());
services.AddSingleton<ISitePskProvider>(psk);
services.AddSingleton(service);
})
.Configure(app =>
{
app.UseRouting();
app.UseEndpoints(e => e.MapGrpcService<CentralControlGrpcService>());
}))
.StartAsync();
node.Server = node._host.GetTestServer();
return node;
}
/// <summary>Switches the node's actor to a black hole that counts but never replies.</summary>
public void SetBlackHole() => _counters.BlackHole = true;
public async ValueTask DisposeAsync()
{
await _host.StopAsync();
_host.Dispose();
}
private sealed class StubCounters
{
private int _submits;
public int Submits => Volatile.Read(ref _submits);
public void IncrementSubmits() => Interlocked.Increment(ref _submits);
public volatile bool BlackHole;
}
private sealed class StubCentralActor : ReceiveActor
{
public StubCentralActor(StubCounters counters, bool repliesToSubmit)
{
Receive<NotificationSubmit>(msg =>
{
counters.IncrementSubmits();
if (repliesToSubmit && !counters.BlackHole)
{
Sender.Tell(new NotificationSubmitAck(msg.NotificationId, Accepted: true, Error: null));
}
});
// Heartbeat lands here as a Tell (the failback probe); ignore it, no reply expected.
ReceiveAny(_ => { });
}
}
private sealed class MapPskProvider(IReadOnlyDictionary<string, string> keys) : ISitePskProvider
{
public ValueTask<string> GetAsync(string siteId, CancellationToken ct)
=> keys.TryGetValue(siteId, out var key)
? new ValueTask<string>(key)
: throw new InvalidOperationException($"no key for '{siteId}'");
public void Invalidate(string siteId) { }
}
}
}