Files
ScadaBridge/tests/ZB.MOM.WW.ScadaBridge.Host.Tests/GrpcSiteTransportFailoverTests.cs
Joseph Doherty 86ad4d5c8e feat(comm): T1B.3 — central-side gRPC site-command transport seam (default Akka)
Extract the central→site send path in CentralCommunicationActor behind a new
ISiteCommandTransport, selected by ScadaBridge:Communication:SiteTransport
(Akka | Grpc, default Akka — rollback = flip the flag). CommunicationService's
27 commands, SiteCallAuditActor's 2 parked relays and DebugStreamBridgeActor's
subscribe/unsubscribe are untouched; the seam sits below SiteEnvelope.

- AkkaSiteTransport: today's per-site ClusterClient path extracted verbatim
  (the _siteClients lookup + ClusterClient.Send with the reply-to sender
  preserved, and the "no client ⇒ warn + drop, caller's Ask times out" path).
- GrpcSiteTransport: dials the site SiteCommandService (T1B.1 proto client) via
  SiteCommandDtoMapper, PSK + x-scadabridge-site on the channel through
  ControlPlaneCredentials, per-command deadlines set EQUAL to today's
  CommunicationService Ask timeouts (per-command, not per-group: DeploymentState
  query and TriggerSiteFailover use QueryTimeout; the two parked relays map to
  QueryTimeout so SiteCallAudit's inner RelayTimeout 10s < 30s ordering holds;
  WaitForAttribute keeps its dynamic Timeout + IntegrationTimeout).
- SitePairChannelProvider: per-site A/B channel pair with sticky failover
  (flip only on Unavailable — NEVER on DeadlineExceeded, a write/deploy/failover
  may have run), background failback probe to the preferred node with 1s→60s
  doubling backoff, PSK invalidation on site removal. Fed by the SAME DB refresh
  loop (extended to carry GrpcNodeA/GrpcNodeBAddress) — no second poll.

Tests: actor-with-substitute-transport (routing, Ask-reply plumbing, per-site
lifecycle across refreshes), ResolveDeadline pinned to each command's current
Ask timeout, and GrpcSiteTransport/SitePairChannelProvider over dual in-process
TestServers (PSK+header+deadline attached, Unavailable failover + stickiness,
failback to preferred, no-retry-on-DeadlineExceeded). Proto csproj untouched
(no active <Protobuf> item). Full solution builds 0 warnings; Communication.Tests
607 green with Akka default.
2026-07-22 20:09:43 -04:00

219 lines
8.6 KiB
C#

using Grpc.Core;
using Grpc.Net.Client;
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.Lifecycle;
using ZB.MOM.WW.ScadaBridge.Communication;
using ZB.MOM.WW.ScadaBridge.Communication.Grpc;
namespace ZB.MOM.WW.ScadaBridge.Host.Tests;
/// <summary>
/// T1B.3: <see cref="SitePairChannelProvider"/> failover/failback + credential/deadline proof over
/// two in-process gRPC <see cref="TestServer"/>s (NodeA preferred, NodeB standby). Exercises the
/// real client stack — PSK call credentials, per-call deadline, sticky failover, no-retry on
/// <see cref="StatusCode.DeadlineExceeded"/>, and failback to the preferred node.
/// </summary>
public sealed class GrpcSiteTransportFailoverTests : IAsyncLifetime
{
private const string SiteId = "site-1";
private const string SiteKey = "the-site-1-key";
private const string EndpointA = "http://node-a";
private const string EndpointB = "http://node-b";
private IHost _hostA = null!;
private IHost _hostB = null!;
private StubSiteCommandService _stubA = null!;
private StubSiteCommandService _stubB = null!;
private SitePairChannelProvider _provider = null!;
private volatile bool _preferredReachable;
/// <inheritdoc />
public async Task InitializeAsync()
{
(_hostA, _stubA) = await StartServerAsync();
(_hostB, _stubB) = await StartServerAsync();
var handlerA = _hostA.GetTestServer().CreateHandler();
var handlerB = _hostB.GetTestServer().CreateHandler();
_provider = new SitePairChannelProvider(
new FixedPskProvider(SiteKey),
Options.Create(new CommunicationOptions()),
NullLogger<SitePairChannelProvider>.Instance,
handlerFactory: endpoint => endpoint == EndpointA ? handlerA : handlerB,
reachabilityProbe: (_, _) => Task.FromResult(_preferredReachable));
_provider.UpdateSite(SiteId, EndpointA, EndpointB);
}
/// <inheritdoc />
public async Task DisposeAsync()
{
_provider.Dispose();
await _hostA.StopAsync();
await _hostB.StopAsync();
_hostA.Dispose();
_hostB.Dispose();
}
private static async Task<(IHost, StubSiteCommandService)> StartServerAsync()
{
var stub = new StubSiteCommandService();
var host = await new HostBuilder()
.ConfigureWebHost(web => web
.UseTestServer()
.ConfigureServices(services =>
{
services.AddGrpc();
services.AddSingleton(stub);
})
.Configure(app =>
{
app.UseRouting();
app.UseEndpoints(e => e.MapGrpcService<StubSiteCommandService>());
}))
.StartAsync();
return (host, stub);
}
private static LifecycleRequest EnableReq() =>
SiteCommandDtoMapper.ToLifecycleRequest(
new EnableInstanceCommand("cmd-1", "Site1.Pump1", DateTimeOffset.UtcNow));
private Task<LifecycleReply> CallAsync(TimeSpan? deadline = null) =>
_provider.ExecuteAsync(
SiteId,
(channel, ct) =>
{
var client = new SiteCommandService.SiteCommandServiceClient(channel);
return client.ExecuteLifecycleAsync(
EnableReq(),
deadline: DateTime.UtcNow + (deadline ?? TimeSpan.FromSeconds(10)),
cancellationToken: ct).ResponseAsync;
},
CancellationToken.None);
[Fact]
public async Task HealthyCall_HitsPreferredNodeA_WithPskAndSiteHeaderAndDeadline()
{
var reply = await CallAsync();
Assert.Equal(LifecycleReply.ReplyOneofCase.InstanceLifecycle, reply.ReplyCase);
Assert.Equal(1, _stubA.LifecycleCalls);
Assert.Equal(0, _stubB.LifecycleCalls);
Assert.Equal($"Bearer {SiteKey}", _stubA.LastAuthHeader);
Assert.Equal(SiteId, _stubA.LastSiteHeader);
Assert.True(_stubA.DeadlineWasSet, "the client must set a per-call deadline");
Assert.True(_provider.IsOnPreferredNode(SiteId));
}
[Fact]
public async Task Unavailable_OnNodeA_FailsOverToNodeB_AndStaysSticky()
{
_stubA.ThrowUnavailable = true;
var reply = await CallAsync();
// Failed over: A was tried (and threw), B answered.
Assert.Equal(LifecycleReply.ReplyOneofCase.InstanceLifecycle, reply.ReplyCase);
Assert.Equal(1, _stubA.LifecycleCalls);
Assert.Equal(1, _stubB.LifecycleCalls);
Assert.False(_provider.IsOnPreferredNode(SiteId));
// Sticky: the next call goes straight to B without re-touching A.
await CallAsync();
Assert.Equal(1, _stubA.LifecycleCalls);
Assert.Equal(2, _stubB.LifecycleCalls);
}
[Fact]
public async Task Failback_ReturnsToPreferredNodeA_OncePreferredIsReachableAgain()
{
_stubA.ThrowUnavailable = true;
await CallAsync(); // flips to B
Assert.False(_provider.IsOnPreferredNode(SiteId));
// NodeA recovers; the failback probe reports it reachable.
_stubA.ThrowUnavailable = false;
_preferredReachable = true;
var back = await _provider.TryFailbackAsync(SiteId, CancellationToken.None);
Assert.True(back);
Assert.True(_provider.IsOnPreferredNode(SiteId));
var beforeA = _stubA.LifecycleCalls;
await CallAsync();
Assert.Equal(beforeA + 1, _stubA.LifecycleCalls); // next call back on A
}
[Fact]
public async Task DeadlineExceeded_OnNodeA_IsNotRetriedOnNodeB()
{
// A DeadlineExceeded is ambiguous (a WriteTag/Deploy/Failover may already have executed), so
// — unlike Unavailable — it must NOT fail over to B. Modelled by A returning the status
// directly, isolating the retry-decision from TestServer's own timeout mechanics.
_stubA.StatusToThrow = StatusCode.DeadlineExceeded;
var ex = await Assert.ThrowsAsync<RpcException>(() => CallAsync());
Assert.Equal(StatusCode.DeadlineExceeded, ex.StatusCode);
Assert.Equal(1, _stubA.LifecycleCalls);
Assert.Equal(0, _stubB.LifecycleCalls); // B was never tried
Assert.True(_provider.IsOnPreferredNode(SiteId), "a deadline must not flip stickiness");
}
[Fact]
public async Task UnknownSite_Throws_SiteChannelUnavailable()
{
await Assert.ThrowsAsync<SiteChannelUnavailableException>(
() => _provider.ExecuteAsync<LifecycleReply>(
"not-configured", (_, _) => Task.FromResult(new LifecycleReply()), CancellationToken.None));
}
private sealed class FixedPskProvider(string key) : ISitePskProvider
{
public ValueTask<string> GetAsync(string siteId, CancellationToken ct) => new(key);
public void Invalidate(string siteId) { }
}
/// <summary>Stub SiteCommandService recording what the client sent and letting a test steer faults.</summary>
private sealed class StubSiteCommandService : SiteCommandService.SiteCommandServiceBase
{
private int _lifecycleCalls;
public int LifecycleCalls => Volatile.Read(ref _lifecycleCalls);
public string? LastAuthHeader { get; private set; }
public string? LastSiteHeader { get; private set; }
public bool DeadlineWasSet { get; private set; }
public volatile bool ThrowUnavailable;
public StatusCode? StatusToThrow;
public override Task<LifecycleReply> ExecuteLifecycle(LifecycleRequest request, ServerCallContext context)
{
Interlocked.Increment(ref _lifecycleCalls);
LastAuthHeader = context.RequestHeaders
.FirstOrDefault(h => h.Key == ControlPlaneCredentials.AuthorizationHeader)?.Value;
LastSiteHeader = context.RequestHeaders
.FirstOrDefault(h => h.Key == ControlPlaneCredentials.SiteHeader)?.Value;
DeadlineWasSet = context.Deadline != DateTime.MaxValue;
if (ThrowUnavailable)
{
throw new RpcException(new Status(StatusCode.Unavailable, "node down"));
}
if (StatusToThrow is { } status)
{
throw new RpcException(new Status(status, "modelled fault"));
}
return Task.FromResult(SiteCommandDtoMapper.ToLifecycleReply(
new InstanceLifecycleResponse("cmd-1", "Site1.Pump1", true, null, DateTimeOffset.UtcNow)));
}
}
}