fix(communication): resolve Communication-012..015 — endpoint-aware gRPC client cache, address-change recreation, correlation-id validation, node-flip tests

tests/ScadaLink.Communication.Tests/Grpc/DebugStreamBridgeActorTests.cs

@@ -308,6 +308,46 @@ public class DebugStreamBridgeActorTests : TestKit
         Assert.True(ctx.TerminatedFlag[0]);
     }
 
+    [Fact]
+    public void On_GrpcError_Reconnects_To_Other_Node_Endpoint()
+    {
+        // Communication-015 regression: drive the bridge actor through a node flip
+        // with an endpoint-aware factory (one distinct mock client per endpoint).
+        // The first subscribe targets NodeA; after a gRPC error the bridge must
+        // reconnect via a client bound to the *NodeB* endpoint.
+        var commProbe = CreateTestProbe();
+        var factory = new EndpointTrackingGrpcClientFactory();
+        var events = new List<object>();
+        var terminated = new[] { false };
+
+        var props = Props.Create(typeof(DebugStreamBridgeActor),
+            SiteId, InstanceName, "corr-1", commProbe.Ref,
+            (Action<object>)(evt => { lock (events) { events.Add(evt); } }),
+            (Action)(() => terminated[0] = true),
+            factory, GrpcNodeA, GrpcNodeB);
+
+        var actor = Sys.ActorOf(props);
+        commProbe.ExpectMsg<SiteEnvelope>();
+
+        actor.Tell(new DebugViewSnapshot(
+            InstanceName,
+            new List<AttributeValueChanged>(),
+            new List<AlarmStateChanged>(),
+            DateTimeOffset.UtcNow));
+
+        // First subscribe goes to NodeA.
+        AwaitCondition(() => factory.ClientFor(GrpcNodeA).SubscribeCalls.Count == 1,
+            TimeSpan.FromSeconds(3));
+
+        // gRPC error → bridge flips to NodeB.
+        factory.ClientFor(GrpcNodeA).SubscribeCalls[0].OnError(new Exception("NodeA down"));
+
+        // The reconnect must reach a client bound to the NodeB endpoint.
+        AwaitCondition(() => factory.ClientFor(GrpcNodeB).SubscribeCalls.Count == 1,
+            TimeSpan.FromSeconds(5));
+        Assert.Equal("corr-1", factory.ClientFor(GrpcNodeB).SubscribeCalls[0].CorrelationId);
+    }
+
     [Fact]
     public void RetryCount_RecoveredOnlyAfterStreamStaysStableForStabilityWindow()
     {
@@ -415,3 +455,24 @@ internal class MockSiteStreamGrpcClientFactory : SiteStreamGrpcClientFactory
         return _mockClient;
     }
 }
+
+/// <summary>
+/// Endpoint-aware mock factory: hands out a distinct <see cref="MockSiteStreamGrpcClient"/>
+/// per endpoint, mirroring the real factory's corrected NodeA→NodeB failover behaviour
+/// so node-flip coverage is meaningful (Communication-015).
+/// </summary>
+internal class EndpointTrackingGrpcClientFactory : SiteStreamGrpcClientFactory
+{
+    private readonly System.Collections.Concurrent.ConcurrentDictionary<string, MockSiteStreamGrpcClient> _byEndpoint = new();
+
+    public EndpointTrackingGrpcClientFactory()
+        : base(Microsoft.Extensions.Logging.Abstractions.NullLoggerFactory.Instance)
+    {
+    }
+
+    public MockSiteStreamGrpcClient ClientFor(string endpoint) =>
+        _byEndpoint.GetOrAdd(endpoint, _ => new MockSiteStreamGrpcClient());
+
+    public override SiteStreamGrpcClient GetOrCreate(string siteIdentifier, string grpcEndpoint)
+        => ClientFor(grpcEndpoint);
+}