DeliveryManagerMultiNodeSpecs: node that restarts is actually a different node

[Aaronontheweb]

Akka.Cluster.Chunking/src/Akka.Cluster.Chunking.Tests/DeliveryManagerMultiNodeSpecs.cs

Lines 128 to 145 in 193e237

	await WithinAsync(TimeSpan.FromSeconds(10), async () =>
	{
	var cluster1 = Akka.Cluster.Cluster.Get(Sys);
	Shutdown(Sys2);
	await AwaitConditionAsync(() => cluster1.State.Members.Count == 2);
	});
	// have Sys message Sys3 (validate that we didn't break connections to healthy nodes)
	var msg2 = new ChunkedDelivery("hello2", probe3.Ref, probe.Ref);
	dm1.Tell(msg2);
	await probe3.ExpectMsgAsync("hello2");
	probe3.Reply("ok2");
	await probe.ExpectMsgAsync("ok2");
	// restart Sys2
	var newSys2 = ActorSystem.Create(Sys.Name, Sys.Settings.Config);
	var dm2New = CreateDeliveryManager(newSys2);
	var probe2New = CreateTestProbe(newSys2);

Node 2 restarts on port 0, making it effectively a different node - we need to grab the Cluster.SelfAddress for node2 and inject that into its akka.remote.dot-netty.tcp.port setting to ensure we're reusing the previous address.