Not receiving Terminated after Exception in Akka.net - akka.net

I have two actors (a parent and a child) and I use SupervisorStrategy.StoppingStrategy and Watch to handle errors. But the problem is that the parent is not receiving a Terminated-message when the child dies.
Below is a simple test to verify my problem.
[Fact]
public void ExceptionTest()
{
var child = Sys.ActorOf(Props.Create(() => new FooActor(), SupervisorStrategy.StoppingStrategy), "Name");
Watch(child);
child.Tell("Some message");
ExpectMsg<Terminated>(); // This will not occur.
}
public class FooActor : ReceiveActor
{
public FooActor()
{
ReceiveAny((e) =>
{
// Crashes on every message.
throw new Exception();
});
}
}
What I want is that the parent which is watching the child should get a Terminated whenever the child has an error which it doesn't handle. How can I accomplish it?
It works fine when I send a PoisonPill instead of throwing an Exception.

Actors in tests seems to be created with restart supervising strategy and thus will not kill the actor when the exception happens. I can't find a way to change the supervisor strategy of tests. So I wrote this to create actors in tests with stopping supervisor strategy with out having to write a "real" wrapper actor.
[Fact]
public async Task ExceptionTest()
{
var actor = await CreateActorWithStoppingStrategy(Props.Create(() => new AsGoodAsDead()));
Watch(actor);
actor.Tell("Death");
ExpectTerminated(actor);
}
private async Task<IActorRef> CreateActorWithStoppingStrategy(Props props)
{
IActorRef child = null;
Action<IActorDsl> actor = d =>
{
d.Strategy = SupervisorStrategy.StoppingStrategy;
d.OnPreStart = context => child = context.ActorOf(props);
d.ReceiveAny((o, context) => context.Sender.Tell("ready"));
};
var parent = Sys.ActorOf(Props.Create(() => new Act(actor)));
await parent.Ask("ready?");
return child;
}
public class AsGoodAsDead : UntypedActor
{
protected override void OnReceive(object message)
{
throw new Exception();
}
}
The following line you wrote will give the child the stopping supervisor strategy, meaning the child's children will be stopped when something happens.
var child = Sys.ActorOf(Props.Create(() => new FooActor(), SupervisorStrategy.StoppingStrategy), "Name");

Related

How to send Message to User id Saved in AspNetUsers Table Signal R

I am trying to send a message. I have tried to connection id
public Task SendMessaageToConnectionID(string ConnectionID,string Message)
{
return Clients.Clients(ConnectionID).SendAsync("RecieveMessage", Message);
}
it is successfully done
Now I am trying this
public Task SendMessageToUser(string userId,string Message)
{
return Clients.Clients(userId).SendAsync(Message);
}
I am sending the user id of user Saved in AspNetUser Table
How Can I send this to a User ID or is there any other way except connection id to send the message to user?
SignalR won't store the UserId-ConnectionId mappings for us. We need to do that by our own. For example, when some user sets up a connection to the Hub, it should trigger a ReJoinGroup() method.
In addition, in order to make sure the Groups property works fine, you need also :
invoke RemoveFromGroupAsync to remove the old <connectionId, groupName> mapping
invoke AddToGroupAsync to add a new <connectionId, groupName> mapping.
Typically, you might want to store these information in Redis or RDBMS. For testing purpose, I create a demo that stores these mappings in memory for your reference:
public class MyHub:Hub
{
/// a in-memory store that stores the <userId, connectionId> mappings
private Dictionary<string, string> _userConn = new Dictionary<string,string>();
private readonly SemaphoreSlim _semaphore = new SemaphoreSlim(1, 1);
public override async Task OnConnectedAsync()
{
// get the real group Name by userId,
// for testing purpose, I use the userId as the groupName,
// in your scenario, you could use the ChatRoom Id
var groupName = Context.UserIdentifier;
await this.ReJoinGroup(groupName);
}
// whenever a connection is setup, invoke this Hub method to update the store
public async Task<KeyValuePair<string,string>> ReJoinGroup(string groupName)
{
var newConnectionId = Context.ConnectionId;
var userId = Context.UserIdentifier;
await this._semaphore.WaitAsync();
try{
if(_userConn.TryGetValue(userId, out var oldConnectionId))
{
_userConn[userId]= newConnectionId;
// remove the old connectionId from the Group
if(!string.IsNullOrEmpty(groupName)){
await Groups.RemoveFromGroupAsync(oldConnectionId, groupName);
await Groups.AddToGroupAsync(newConnectionId, groupName);
}
} else {
_userConn[userId]= newConnectionId;
if(!string.IsNullOrEmpty(groupName)){
await Groups.AddToGroupAsync(newConnectionId, groupName);
}
}
} finally{
this._semaphore.Release();
}
return new KeyValuePair<string,string>(userId, newConnectionId);
}
/// your SendMessageToUser() method
public async Task SendMessageToUser(string userId,string Message)
{
// get the connectionId of target user
var userConn = await this.GetUserConnection(userId);
if( userConn.Equals(default(KeyValuePair<string,string>))) {
throw new Exception($"unknown user connection with userId={userId}");
}
await Clients.Clients(userConn.Value).SendAsync(Message);
}
/// a private helper that returns a pair of <UserId,ConnectionId>
private async Task<KeyValuePair<string,string>> GetUserConnection(string userId)
{
KeyValuePair<string,string> kvp = default;
string newConnectionId = default;
await this._semaphore.WaitAsync();
try{
if(this._userConn.TryGetValue(userId, out newConnectionId)){
kvp= new KeyValuePair<string, string>(userId, newConnectionId);
}
} finally{
this._semaphore.Release();
}
return kvp;
}
}

Unit testing for a publish method in mass transit using in memory bus

I am new to MassTransit and messaging. I am trying to write a unit test for a IBus.Publish and am not able to succeed with the result.
I am observing the fault from RabbitMQ and my observer looks like this:
public class FaultObserver : IReceiveObserver
{
public FaultObserver(IRequestUpdater statusUpdater,Lazy<IBus> bus)
{
this.statusUpdater = statusUpdater;
this.bus = bus;
}
public async Task ConsumeFault<T>(ConsumeContext<T> context, TimeSpan duration, string consumerType, Exception exception) where T : class
{
}
}
and my tests looks like the below code
var bus = fixture.Freeze<Mock<IBus>>();
bus.Setup(bu => bu.Publish<ReportFailedEvent>(It.IsAny<ReportFailedEvent>(),It.IsAny<CancellationToken>())).Verifiable();
var sut = fixture.Create<ReportRequestedFaultObserver>();
// Act
await sut.ConsumeFault(context.Object,new TimeSpan(0,0,1),string.Empty,exception);
// Assert
//bus.Verify(b => b.Publish<ReportFailedEvent>(It.IsAny<ReportFailedEvent>(), It.IsAny<CancellationToken>()), Times.Exactly(1));
bus.Verify(b =>b.Publish<ReportFailedEvent>(new ReportFailedEvent(request,exception.Message),It.IsAny<CancellationToken>()),Times.Once());
my setup looks like
[SetUp]
public void SetUp()
{
fixture.Customize(new AutoMoqCustomization());
var inMemoryTransportCache = new InMemoryTransportCache(Environment.ProcessorCount);
bus = Bus.Factory.CreateUsingInMemory(configure =>
{
configure.SetTransportProvider(inMemoryTransportCache);
configure.ReceiveEndpoint("input_queue", configurator =>
{
configurator.Handler<(cc =>
{
});
});
});
bus.Start().Ready.Wait();
}
I am not able to mock my Publish method. Does anyone knows what I am doing wrong?
Your mock for the bus is wrong. It should be:
bus.Setup(bu => bu.Publish<ReportFailedEvent>(
It.IsAny<object>(), It.IsAny<CancellationToken>()));
Publish takes an object and a CancellationToken, as per the interface definition in MassTransit:
Task Publish(object message, CancellationToken cancellationToken = default(CancellationToken));
Also, if you want to check the contents of the message, you can use the Moq Callback extension:
ReportFailedEvent message = null;
bus
.Setup(bu => bu.Publish<ReportFailedEvent>(It.IsAny<object>(), It.IsAny<CancellationToken>()))
.Callback<object, CancellationToken>((a, b) => { message = a as ReportFailedEvent; });
//.. your system under test code....
Assert.That(message.Property, Is.EqualTo(expectation));
If you are setting the priority, the MassTransit interface looks like this:
public static Task Publish<T>(
this IPublishEndpoint endpoint,
T message,
Action<PublishContext<T>> callback,
CancellationToken cancellationToken = default(CancellationToken))
where T : class;
And an example would be:
bus.Publish<ReportFailedEvent>(message, context =>
{
context.SetPriority(messagePriority.Priority);
});
And the accompanying Moq is:
ReportFailedEvent message = null;
bus
.Setup(bu => bu.Publish<ReportFailedEvent>(It.IsAny<object>(), It.IaAny<Action<PublishContext<ReportFailedEvent>>>(), It.IsAny<CancellationToken>()))
.Callback<object, Action<PublishContext<ReportFailedEvent>>, CancellationToken>((a, b, c) => { message = a as ReportFailedEvent; });
Just as a side note, ideally you should be publishing the interfaces rather than classes, so IReportFailedEvent rather than ReportFailedEvent.

My Akka.Net Demo is incredibly slow

I am trying to get a proof of concept running with akka.net. I am sure that I am doing something terribly wrong, but I can't figure out what it is.
I want my actors to form a graph of nodes. Later, this will be a complex graph of business objekts, but for now I want to try a simple linear structure like this:
I want to ask a node for a neighbour that is 9 steps away. I am trying to implement this in a recursive manner. I ask node #9 for a neighbour that is 9 steps away, then I ask node #8 for a neighbour that is 8 steps away and so on. Finally, this should return node #0 as an answer.
Well, my code works, but it takes more than 4 seconds to execute. Why is that?
This is my full code listing:
using System;
using System.Collections.Generic;
using System.Diagnostics;
using System.Linq;
using Akka;
using Akka.Actor;
namespace AkkaTest
{
class Program
{
public static Stopwatch stopwatch = new Stopwatch();
static void Main(string[] args)
{
var system = ActorSystem.Create("MySystem");
IActorRef[] current = new IActorRef[0];
Console.WriteLine("Initializing actors...");
for (int i = 0; i < 10; i++)
{
var current1 = current;
var props = Props.Create<Obj>(() => new Obj(current1, Guid.NewGuid()));
var actorRef = system.ActorOf(props, i.ToString());
current = new[] { actorRef };
}
Console.WriteLine("actors initialized.");
FindNeighboursRequest r = new FindNeighboursRequest(9);
stopwatch.Start();
var response = current[0].Ask(r);
FindNeighboursResponse result = (FindNeighboursResponse)response.Result;
stopwatch.Stop();
foreach (var d in result.FoundNeighbours)
{
Console.WriteLine(d);
}
Console.WriteLine("Search took " + stopwatch.ElapsedMilliseconds + "ms.");
Console.ReadLine();
}
}
public class FindNeighboursRequest
{
public FindNeighboursRequest(int distance)
{
this.Distance = distance;
}
public int Distance { get; private set; }
}
public class FindNeighboursResponse
{
private IActorRef[] foundNeighbours;
public FindNeighboursResponse(IEnumerable<IActorRef> descendants)
{
this.foundNeighbours = descendants.ToArray();
}
public IActorRef[] FoundNeighbours
{
get { return this.foundNeighbours; }
}
}
public class Obj : ReceiveActor
{
private Guid objGuid;
readonly List<IActorRef> neighbours = new List<IActorRef>();
public Obj(IEnumerable<IActorRef> otherObjs, Guid objGuid)
{
this.neighbours.AddRange(otherObjs);
this.objGuid = objGuid;
Receive<FindNeighboursRequest>(r => handleFindNeighbourRequest(r));
}
public Obj()
{
}
private async void handleFindNeighbourRequest (FindNeighboursRequest r)
{
if (r.Distance == 0)
{
FindNeighboursResponse response = new FindNeighboursResponse(new IActorRef[] { Self });
Sender.Tell(response, Self);
return;
}
List<FindNeighboursResponse> responses = new List<FindNeighboursResponse>();
foreach (var actorRef in neighbours)
{
FindNeighboursRequest req = new FindNeighboursRequest(r.Distance - 1);
var response2 = actorRef.Ask(req);
responses.Add((FindNeighboursResponse)response2.Result);
}
FindNeighboursResponse response3 = new FindNeighboursResponse(responses.SelectMany(rx => rx.FoundNeighbours));
Sender.Tell(response3, Self);
}
}
}
The reason of such slow behavior is the way you use Ask (an that you use it, but I'll cover this later). In your example, you're asking each neighbor in a loop, and then immediately executing response2.Result which is actively blocking current actor (and thread it resides on). So you're essentially making synchronous flow with blocking.
The easiest thing to fix that, is to collect all tasks returned from Ask and use Task.WhenAll to collect them all, without waiting for each one in a loop. Taking this example:
public class Obj : ReceiveActor
{
private readonly IActorRef[] _neighbours;
private readonly Guid _id;
public Obj(IActorRef[] neighbours, Guid id)
{
_neighbours = neighbours;
_id = id;
Receive<FindNeighboursRequest>(async r =>
{
if (r.Distance == 0) Sender.Tell(new FindNeighboursResponse(new[] {Self}));
else
{
var request = new FindNeighboursRequest(r.Distance - 1);
var replies = _neighbours.Select(neighbour => neighbour.Ask<FindNeighboursResponse>(request));
var ready = await Task.WhenAll(replies);
var responses = ready.SelectMany(x => x.FoundNeighbours);
Sender.Tell(new FindNeighboursResponse(responses.ToArray()));
}
});
}
}
This one is much faster.
NOTE: In general you shouldn't use Ask inside of an actor:
Each ask is allocating a listener inside current actor, so in general using Ask is A LOT heavier than passing messages with Tell.
When sending messages through chain of actors, cost of ask is additionally transporting message twice (one for request and one for reply) through each actor. One of the popular patterns is that, when you are sending request from A⇒B⇒C⇒D and respond from D back to A, you can reply directly D⇒A, without need of passing the message through whole chain back. Usually combination of Forward/Tell works better.
In general don't use async version of Receive if it's not necessary - at the moment, it's slower for an actor when compared to sync version.

How to do WaitAll with Akka.Net?

I have a hierarchy of actors in Akka.Net and am wondering whether I've chosen the right way to do something, or if there are better/simpler ways to achieve what I want.
My specific example is that I'm constructing a User actor in response to a user logging into the system, and when constructing this actor there are two pieces of data I need in order to complete the construction of the actor.
If this were regular .NET code I might have something like the following...
public Task<User> LoadUserAsync (string username)
{
IProfileService profileService = ...;
IMessageService messageService = ...;
var loadProfileTask = profileService.GetUserProfileAsync(username);
var loadMessagesTask = messageService.GetMessagesAsync(username);
Task.WaitAll(loadProfileTask, loadMessagesTask);
// Now construct the user from the result of both tasks
var user = new User
{
Profile = loadProfileTask.Result,
Messages = loadMessagesTask.Result
}
return Task.FromResult(user);
}
Here I use WaitAll to wait for the subordinate tasks to complete, and let them run concurrently.
My question is - if I wanted to do the same in Akka.Net, would the following be the most regular way to do this? Pictorially I've created the following...
When I create my User actor, I then construct a (temporary) User Loader Actor, whose job it is to get the full user details by calling to the Profile actor and the Messages actor. The leaf actors that get the data are as follows...
public class UserProfileLoader : ReceiveActor
{
public UserProfileLoader()
{
Receive<LoadUserRequest>(msg =>
{
// Load the user profile from somewhere
var profile = new UserProfile();
// And respond to the Sender
Sender.Tell(profile);
Self.Tell(PoisonPill.Instance);
});
}
}
public class UserMessagesLoader : ReceiveActor
{
public UserMessagesLoader()
{
Receive<LoadUserRequest>(msg =>
{
// Load the messages from somewhere
var messages = new List<Message>();
// And respond to the Sender
Sender.Tell(messages);
Self.Tell(PoisonPill.Instance);
});
}
}
It doesn't really matter where they get the data from for this discussion, but both simply respond to a request by returning some data.
Then I have the actor that coordinates the two data gathering actors...
public class UserLoaderActor : ReceiveActor
{
public UserLoaderActor()
{
Receive<LoadUserRequest>(msg => LoadProfileAndMessages(msg));
Receive<UserProfile>(msg =>
{
_profile = msg;
FinishIfPossible();
});
Receive<List<Message>>(msg =>
{
_messages = msg;
FinishIfPossible();
});
}
private void LoadProfileAndMessages(LoadUserRequest msg)
{
_originalSender = Sender;
Context.ActorOf<UserProfileLoader>().Tell(msg);
Context.ActorOf<UserMessagesLoader>().Tell(msg);
}
private void FinishIfPossible()
{
if ((null != _messages) && (null != _profile))
{
_originalSender.Tell(new LoadUserResponse(_profile, _messages));
Self.Tell(PoisonPill.Instance);
}
}
private IActorRef _originalSender;
private UserProfile _profile;
private List<Message> _messages;
}
This just creates the two subordinate actors, sends them a message to get cracking, and then waits for both to respond before sending back all the data that's been gathered to the original requestor.
So, does this seem like a reasonable way to coordinate two disparate responses, in order to combine them? Is there an easier way to do this than craft it up myself?
Thanks in advance for your responses!
Thanks folks, so I've now simplified the actor significantly into the following, based on both Roger and Jeff's suggestions...
public class TaskBasedUserLoader : ReceiveActor
{
public TaskBasedUserLoader()
{
Receive<LoadUserRequest>(msg => LoadProfileAndMessages(msg));
}
private void LoadProfileAndMessages(LoadUserRequest msg)
{
var originalSender = Sender;
var loadPreferences = this.LoadProfile(msg.UserId);
var loadMessages = this.LoadMessages(msg.UserId);
Task.WhenAll(loadPreferences, loadMessages)
.ContinueWith(t => new UserLoadedResponse(loadPreferences.Result, loadMessages.Result),
TaskContinuationOptions.AttachedToParent & TaskContinuationOptions.ExecuteSynchronously)
.PipeTo(originalSender);
}
private Task<UserProfile> LoadProfile(string userId)
{
return Task.FromResult(new UserProfile { UserId = userId });
}
private Task<List<Message>> LoadMessages(string userId)
{
return Task.FromResult(new List<Message>());
}
}
The LoadProfile and LoadMessages methods will ultimately call a repository to get the data, but for now I have a succinct way to do what I wanted.
Thanks again!
IMHO that's a valid process, as you fork action and then join it.
BTW you could use this.Self.GracefulStop(new TimeSpan(1)); instead of sending poison pill.
You could use a combination of Ask, WhenAll and PipeTo:
var task1 = actor1.Ask<Result1>(request1);
var task2 = actor2.Ask<Result2>(request2);
Task.WhenAll(task1, task2)
.ContinueWith(_ => new Result3(task1.Result, task2.Result))
.PipeTo(Self);
...
Receive<Result3>(msg => { ... });

Structuring tests (or property) for this reactive ui scenario

I'm not sure the correct way to structure this test. I've got a view model here:
public class ViewModel
{
public ReactiveCommand PerformSearchCommand { get; private set; }
private readonly ObservableAsPropertyHelper<bool> _IsBusy;
public bool IsBusy
{
get { return _IsBusy.Value; }
}
public ViewModel(IAdventureWorksRepository _awRepository)
{
PerformSearchCommand = new ReactiveCommand();
PerformSearchCommand.RegisterAsyncFunction((x) =>
{
return _awRepository.vIndividualCustomers.Take(1000).ToList();
}).Subscribe(rval =>
{
CustomerList = rval;
SelectedCustomer = CustomerList.FirstOrDefault();
});
PerformSearchCommand.IsExecuting.ToProperty(this, x => x.IsBusy, out _IsBusy);
PerformSearchCommand.Execute(null); // begin executing immediately
}
}
The dependency is a data access object to AdventureWorks
public interface IAdventureWorksRepository
{
IQueryable<vIndividualCustomer> vIndividualCustomers { get; }
}
Finally, my test looks something like this:
[TestMethod]
public void TestTiming()
{
new TestScheduler().With(sched =>
{
var repoMock = new Mock<IAdventureWorksRepository>();
repoMock.Setup(x => x.vIndividualCustomers).Returns(() =>
{
return new vIndividualCustomer[] {
new vIndividualCustomer { FirstName = "John", LastName = "Doe" }
};
});
var vm = new ViewModel(repoMock.Object);
Assert.AreEqual(true, vm.IsBusy); //fails?
Assert.AreEqual(1, vm.CustomerList.Count); //also fails, so it's not like the whole thing ran already
sched.AdvanceTo(2);
Assert.AreEqual(1, vm.CustomerList.Count); // success
// now the customer list is set at tick 2 (not at 1?)
// IsBusy was NEVER true.
});
}
So the viewmodel should immediately begin searching upon load
My immediate problem is that the IsBusy property doesn't seem to get set in the testing scheduler, even though it seems to work fine when I run the code normally. Am I using the ToProperty method correctly in the view model?
More generally, what is the proper way to do the full 'time travel' testing when my object under test has a dependency like this? The issue is that unlike most testing examples I'm seeing, the called interface is not an IObservable. It's just a synchronous query, used asynchronously in my view model. Of course in the view model test, I can mock the query to do whatever rx things I want. How would I set this up if I wanted the query to last 200 ticks, for example?
So, you've got a few things in your code that is stopping you from getting things to work correctly:
Don't invoke commands in ViewModel Constructors
First, calling Execute in the constructor means you'll never see the state change. The best pattern is to write that command but not execute it in the VM immediately, then in the View:
this.WhenAnyValue(x => x.ViewModel)
.InvokeCommand(this, x => x.ViewModel.PerformSearchCommand);
Move the clock after async actions
Ok, now that we can properly test the before and after state, we have to realize that after every time we do something that normally would be async, we have to advance the scheduler if we use TestScheduler. This means, that when we invoke the command, we should immediately advance the clock:
Assert.IsTrue(vm.PerformSearchCommand.CanExecute(null));
vm.PerformSearchCommand.Execute(null);
sched.AdvanceByMs(10);
Can't test Time Travel without IObservable
However, the trick is, your mock executes code immediately, there's no delay, so you'll never see it be busy. It just returns a canned value. Unfortunately, injecting the Repository makes this difficult to test if you want to see IsBusy toggle.
So, let's rig the constructor a little bit:
public ViewModel(IAdventureWorksRepository _awRepository, Func<IObservable<List<Customer>>> searchCommand = null)
{
PerformSearchCommand = new ReactiveCommand();
searchCommand = searchCommand ?? () => Observable.Start(() => {
return _awRepository.vIndividualCustomers.Take(1000).ToList();
}, RxApp.TaskPoolScheduler);
PerformSearchCommand.RegisterAsync(searchCommand)
.Subscribe(rval => {
CustomerList = rval;
SelectedCustomer = CustomerList.FirstOrDefault();
});
PerformSearchCommand.IsExecuting
.ToProperty(this, x => x.IsBusy, out _IsBusy);
}
Set up the test now
Now, we can set up the test, to replace PerformSearchCommand's action with something that has a delay on it:
new TestScheduler().With(sched =>
{
var repoMock = new Mock<IAdventureWorksRepository>();
var vm = new ViewModel(repoMock.Object, () =>
Observable.Return(new[] { new vIndividualCustomer(), })
.Delay(TimeSpan.FromSeconds(1.0), sched));
Assert.AreEqual(false, vm.IsBusy);
Assert.AreEqual(0, vm.CustomerList.Count);
vm.PerformSearchCommand.Execute(null);
sched.AdvanceByMs(10);
// We should be busy, we haven't finished yet - no customers
Assert.AreEqual(true, vm.IsBusy);
Assert.AreEqual(0, vm.CustomerList.Count);
// Skip ahead to after we've returned the customer
sched.AdvanceByMs(1000);
Assert.AreEqual(false, vm.IsBusy);
Assert.AreEqual(1, vm.CustomerList.Count);
});