To deal with SQL timeouts I'm trying to use SqlAzureExecutionStrategy (https://msdn.microsoft.com/en-us/data/dn456835.aspx)
The problem I am running into is it prevents "user initiated transactions" which seem to be the recommended way to implement "with (nolock)" in EF (http://www.hanselman.com/blog/GettingLINQToSQLAndLINQToEntitiesToUseNOLOCK.aspx, NOLOCK with Linq to SQL).
example code
public AspnetUser GetAspnetUserByUserName(string userName)
{
using (var tx = new TransactionScope(TransactionScopeOption.Required, new TransactionOptions() { IsolationLevel = IsolationLevel.ReadUncommitted }))
{
return context.AspnetUsers.Where(x => x.UserName == userName).FirstOrDefault();
}
}
throws error
The configured execution strategy 'SqlAzureExecutionStrategy' does not support user initiated transactions. See http://go.microsoft.com/fwlink/?LinkId=309381 for additional information.
I've seen the answers that say to turn off the SqlAzureExecutionStrategy on a per call basis, but that would defeat the purpose of using it, if all my reads ignored the strategy. It is possible to have both "NoLock" and SqlAzureExecutionStrategy
SqlAzureExecutionStrategy doesn't support transactions initiated outside the action to be retried. To work around this restriction you would need to suspend the strategy, create the transaction scope and do the work as an action that you manually pass to the execution strategy to be retried:
public AspnetUser GetAspnetUserByUserName(string userName)
{
new SuspendableSqlAzureExecutionStrategy().Execute(() =>
{
using (var tx = new TransactionScope(
TransactionScopeOption.Required,
new TransactionOptions() { IsolationLevel = IsolationLevel.ReadUncommitted }))
{
return context.AspnetUsers.Where(x => x.UserName == userName).FirstOrDefault();
}
});
}
Here I am using an alternative to the suspendable strategy from https://msdn.microsoft.com/en-us/data/dn307226 that will suspend any nested invocations automatically:
using System.Data.Entity.Infrastructure;
using System.Data.Entity.SqlServer;
using System.Data.Entity.Utilities;
using System.Runtime.Remoting.Messaging;
using System.Threading;
using System.Threading.Tasks;
public class SuspendableSqlAzureExecutionStrategy : IDbExecutionStrategy
{
private readonly IDbExecutionStrategy _azureExecutionStrategy;
public SuspendableSqlAzureExecutionStrategy()
{
_azureExecutionStrategy = new SqlAzureExecutionStrategy();
}
private static bool Suspend
{
get { return (bool?)CallContext.LogicalGetData("SuspendExecutionStrategy") ?? false; }
set { CallContext.LogicalSetData("SuspendExecutionStrategy", value); }
}
public bool RetriesOnFailure
{
get { return !Suspend; }
}
public virtual void Execute(Action operation)
{
if (!RetriesOnFailure)
{
operation();
return;
}
try
{
Suspend = true;
_azureExecutionStrategy.Execute(operation);
}
finally
{
Suspend = false;
}
}
public virtual TResult Execute<TResult>(Func<TResult> operation)
{
if (!RetriesOnFailure)
{
return operation();
}
try
{
Suspend = true;
return _azureExecutionStrategy.Execute(operation);
}
finally
{
Suspend = false;
}
}
public virtual async Task ExecuteAsync(Func<Task> operation, CancellationToken cancellationToken)
{
if (!RetriesOnFailure)
{
await operation();
return;
}
try
{
Suspend = true;
await _azureExecutionStrategy.ExecuteAsync(operation, cancellationToken);
}
finally
{
Suspend = false;
}
}
public virtual async Task<TResult> ExecuteAsync<TResult>(Func<Task<TResult>> operation, CancellationToken cancellationToken)
{
if (!RetriesOnFailure)
{
return await operation();
}
try
{
Suspend = true;
return await _azureExecutionStrategy.ExecuteAsync(operation, cancellationToken);
}
finally
{
Suspend = false;
}
}
}
public class MyConfiguration : DbConfiguration
{
public MyConfiguration()
{
SetExecutionStrategy("System.Data.SqlClient", () => new SuspendableSqlAzureExecutionStrategy());
}
}
Related
I have scenario to pass async function as callback to my own resource manager(which implements IEnlistmentNotification interface), and need to invoke asynchronously in prepare method, but it works when invoke as synchronous way, is there any way to make it without wait or asynchronous, the wait producing the AggregatorException rather than my custom exception?
Resource Manager
public class ResourceManager : IEnlistmentNotification
{
private Func<Task>? _doWorkCallback;
public async Task EnlistAsync(Func<Task> doWorkCallback)
{
_doWorkCallback = doWorkCallback;
var transaction = Transaction.Current;
if (transaction != null)
{
await transaction.EnlistVolatileAsync(this, EnlistmentOptions.None).ConfigureAwait(false);
}
}
public void Prepare(PreparingEnlistment preparingEnlistment)
{
try
{
_doWorkCallback?.Invoke().Wait();
preparingEnlistment.Prepared();
}
catch
{
preparingEnlistment.ForceRollback();
}
}
public void Commit(Enlistment enlistment)
{
enlistment.Done();
}
public void Rollback(Enlistment enlistment)
{
enlistment.Done();
}
public void InDoubt(Enlistment enlistment)
{
enlistment.Done();
}
}
public static class TranscationExtensions
{
public static Task EnlistVolatileAsync(this Transaction transaction,
IEnlistmentNotification
enlistmentNotification,
EnlistmentOptions enlistmentOptions)
{
return Task.FromResult(transaction.EnlistVolatile
(enlistmentNotification,
enlistmentOptions));
}
}
Usage Code
public class Test
{
private async Task DoWork()
{
Thread.Sleep(1000);// considerer here my custom exception
await Task.CompletedTask;
}
public async Task TestMethod()
{
ResourceManager rm = new ResourceManager();
await rm.EnlistAsync(async () => await DoWork().ConfigureAwait(false)).ConfigureAwait(false);
}
}
Bot Info
SDK Platform: .NET
Active Channels: Direct Line
Deployment Environment: Azure Bot Service
Question
How to pass user context details from from bot Controller to FormDialog?
Code Example
public virtual async Task < HttpResponseMessage > Post([FromBody] Activity activity) {
if (activity != null && activity.GetActivityType() == ActivityTypes.Message) {
await Conversation.SendAsync(activity, () => {
return Chain.From(() => FormDialog.FromForm(RequestOrder.BuildEnquiryForm));
});
} else {
HandleSystemMessage(activity);
}
return new HttpResponseMessage(System.Net.HttpStatusCode.OK);
}
public static IForm < RequestOrder > BuildEnquiryForm() {
return new FormBuilder < RequestOrder > ()
.Message("Hello {***Pass current user name?????****}Welcome to request bot!")
.Field(nameof(IsTermsAgreed))
.Field(nameof(ServiceRequired))
.AddRemainingFields()
.OnCompletion(ProcessParkingPermitRequest)
.Message("Thank you, I have submitted your request.")
.Build();
}
Fei Han answer is correct but using a static variable might lead to some unexpected error since all instances are sharing the same value. A better approach would be using the state of the form.
Request Order From
In your RequestOrder class you need to add a new member variable username.
public class RequestOrder
{
public string username;
/* Rest of your member variables */
}
The .Message method allows you to access the state of the form. You can get the username from the state of the form as below:
public static IForm < RequestOrder > BuildForm()
{
return new FormBuilder < RequestOrder > ()
.Message(async (state) => {
return new PromptAttribute($"Hi {state.username}, Welcome to request bot! ");
})
.Field(nameof(IsTermsAgreed))
.Field(nameof(ServiceRequired))
.AddRemainingFields()
.OnCompletion(ProcessParkingPermitRequest)
.Message("Thank you, I have submitted your request.")
.Build();
}
Root Dialog
In your root Dialog, before calling the BuildForm you need to create a new instance of your RequestOrder class and initialize username as the current user's name. Then pass your form to the BuildForm with option FormOptions.PromptInStart.
[Serializable]
public class RootDialog : IDialog<object>
{
public Task StartAsync(IDialogContext context)
{
context.Wait(MessageReceivedAsync);
return Task.CompletedTask;
}
private async Task MessageReceivedAsync(IDialogContext context, IAwaitable<object> result)
{
var form = new RequestOrder()
{
username = context.Activity.From.Id
};
var requestOrderform = new FormDialog<RequestOrder>(form, RequestOrder.BuildForm, FormOptions.PromptInStart);
context.Call<RequestOrder>(requestOrderform, SampleFormSubmitted);
}
private async Task SampleFormSubmitted(IDialogContext context, IAwaitable<SampleForm> result)
{
try
{
var query = await result;
context.Done(true);
}
catch (FormCanceledException<SampleForm> e)
{
string reply;
if (e.InnerException == null)
{
reply = $"You quit. Maybe you can fill some other time.";
}
else
{
reply = $"Something went wrong. Please try again.";
}
context.Done(true);
await context.PostAsync(reply);
}
}
}
This is what you get:
In following sample code, I define a constructor of SandwichOrder class with a string type parameter, then I call FormDialog as a child dialog from root dialog (not from Messages Controller directly) and pass user name as parameter, which works for me, you can refer to it.
In RootDialog:
[Serializable]
public class RootDialog : IDialog<object>
{
public Task StartAsync(IDialogContext context)
{
context.Wait(MessageReceivedAsync);
return Task.CompletedTask;
}
private async Task MessageReceivedAsync(IDialogContext context, IAwaitable<object> result)
{
var username = "Fei Han";
var myform = new Microsoft.Bot.Builder.FormFlow.FormDialog<SandwichOrder>(new SandwichOrder($"{username}"), SandwichOrder.BuildForm, Microsoft.Bot.Builder.FormFlow.FormOptions.PromptInStart, null);
context.Call<SandwichOrder>(myform, FormCompleteCallback);
}
private async Task FormCompleteCallback(IDialogContext context, IAwaitable<SandwichOrder> result)
{
await context.PostAsync($"The form is completed!");
context.Done(this);
}
}
In SandwichOrder class:
namespace BotFormFlowTest
{
public enum SandwichOptions
{
BLT, BlackForestHam, BuffaloChicken, ChickenAndBaconRanchMelt, ColdCutCombo, MeatballMarinara,
OvenRoastedChicken, RoastBeef, RotisserieStyleChicken, SpicyItalian, SteakAndCheese, SweetOnionTeriyaki, Tuna,
TurkeyBreast, Veggie
};
public enum LengthOptions { SixInch, FootLong };
public enum BreadOptions { NineGrainWheat, NineGrainHoneyOat, Italian, ItalianHerbsAndCheese, Flatbread };
public enum CheeseOptions { American, MontereyCheddar, Pepperjack };
public enum ToppingOptions
{
Avocado, BananaPeppers, Cucumbers, GreenBellPeppers, Jalapenos,
Lettuce, Olives, Pickles, RedOnion, Spinach, Tomatoes
};
public enum SauceOptions
{
ChipotleSouthwest, HoneyMustard, LightMayonnaise, RegularMayonnaise,
Mustard, Oil, Pepper, Ranch, SweetOnion, Vinegar
};
[Serializable]
public class SandwichOrder
{
public static string username = "User";
public SandwichOrder(string uname)
{
username = uname;
}
public SandwichOptions? Sandwich;
public LengthOptions? Length;
public BreadOptions? Bread;
public CheeseOptions? Cheese;
public List<ToppingOptions> Toppings;
public List<SauceOptions> Sauce;
public static IForm<SandwichOrder> BuildForm()
{
return new FormBuilder<SandwichOrder>()
.Message($"Hello {username}, Welcome to the simple sandwich order bot!")
.Build();
}
};
}
Test Result:
I am working on a Chat app. After the messages of a chat are loaded and the messages were visible for 5 seconds, I want to send a read confirmation to the server. This is what I've come up with so far:
public async void RefreshLocalData()
{
// some async code to load the messages
if (_selectedChat.countNewMessages > 0)
{
Device.StartTimer(TimeSpan.FromSeconds(5), SendReadConfirmation);
}
}
When RefreshLocalData() is called, I know that either another chat was selected by the user or new messages came in for the current chat. So when RefreshLocalData() is called, I have to cancel the current timer to start a new one.
Another situation where I have to cancel the timer is when I navigate to another Page. This is no problem, because the whole ViewModel is disposed when this happens.
With the code above, if RefreshLocalData() is called again but the stated TimeSpan of 5 seconds is not over yet, the method is still executing.
Is there a way to cancel the timer (if RefreshLocalData() is called again)?
I have found this answer in the Xamarin forum: https://forums.xamarin.com/discussion/comment/149877/#Comment_149877
I have changed it a little bit to meet my needs and this solution is working:
public class StoppableTimer
{
private readonly TimeSpan timespan;
private readonly Action callback;
private CancellationTokenSource cancellation;
public StoppableTimer(TimeSpan timespan, Action callback)
{
this.timespan = timespan;
this.callback = callback;
this.cancellation = new CancellationTokenSource();
}
public void Start()
{
CancellationTokenSource cts = this.cancellation; // safe copy
Device.StartTimer(this.timespan,
() => {
if (cts.IsCancellationRequested) return false;
this.callback.Invoke();
return false; // or true for periodic behavior
});
}
public void Stop()
{
Interlocked.Exchange(ref this.cancellation, new CancellationTokenSource()).Cancel();
}
public void Dispose()
{
}
}
And this is how I use it in the RefreshLocalData() method:
private StoppableTimer stoppableTimer;
public async void RefreshLocalData()
{
if (stoppableTimer != null)
{
stoppableTimer.Stop();
}
// some async code to load the messages
if (_selectedChat.countNewMessages > 0)
{
if (stoppableTimer == null)
{
stoppableTimer = new StoppableTimer(TimeSpan.FromSeconds(5), SendReadConfirmation);
stoppableTimer.Start();
}
else
{
stoppableTimer.Start();
}
}
}
You can try using this class I found, it covers some of the limits to the DeviceTimer:
public class MySystemDeviceTimer
{
private readonly TimeSpan timespan;
private readonly Action callback;
private CancellationTokenSource cancellation;
public bool running { get; private set; }
public MySystemDeviceTimer(TimeSpan timespan, Action callback)
{
this.timespan = timespan;
this.callback = callback;
this.cancellation = new CancellationTokenSource();
}
public void Start()
{
running = true;
start(true);
}
private void start(bool continuous)
{
CancellationTokenSource cts = this.cancellation; // safe copy
Device.StartTimer(this.timespan,
() =>
{
if (cts.IsCancellationRequested)
{
running = false;
return false;
}
this.callback.Invoke();
return continuous;
});
}
public void FireOnce()
{
running = true;
start(false);
running = false;
}
public void Stop()
{
Interlocked.Exchange(ref this.cancellation, new CancellationTokenSource()).Cancel();
}
}
Then for your purpose:
MySystemDeviceTimer timer;
if (timer == null)
{
timer = new MySystemDeviceTimer(TimeSpan.FromSeconds(5), SendReadConfirmation);
timer.FireOnce();
}
else if (timer.running)
timer.Stop();
Yes you can with Device.StartTimer() as long as you return true to have the function repeat. I typically handle this through a Boolean variable that I can control in my ViewModel. Something like below:
bool shouldRun = true;
public async void RefreshLocalData()
{
// some async code to load the messages
if (_selectedChat.countNewMessages > 0)
{
Device.StartTimer(TimeSpan.FromSeconds(5), async() =>
{
await SendReadConfirmationAsync()
return shouldRun;
});
}
}
public async Task SendReadConfirmationAsync()
{
//Do some stuff
if(we want to stop call)
shouldRun = false;
}
I have a consumer that is also publishing a response back to the bus. I can get an IReceiveObserver wired up and working on the bus, but I haven't been able to get either an ISendObserver or IPublishObserver running. I have confirmed with RabbitMQ management console that the messages are being published correctly.
class Program
{
static BusHandle _BusHandle;
static void Main(string[] args)
{
InitLogging();
InitStructureMap();
InitBus();
System.Console.WriteLine("Starting processing, ENTER to stop...");
System.Console.ReadLine();
System.Console.WriteLine("See you later, alligator!");
StopBus();
}
static void InitBus()
{
var busCtrl = ObjectFactory.Container.GetInstance<IBusControl>();
var recObserver = ObjectFactory.Container.GetInstance<IReceiveObserver>();
var sendObserver = ObjectFactory.Container.GetInstance<ISendObserver>();
busCtrl.ConnectReceiveObserver(recObserver);
busCtrl.ConnectSendObserver(sendObserver);
_BusHandle = busCtrl.Start();
}
static void StopBus()
{
_BusHandle.Stop();
}
static void InitLogging()
{
XmlConfigurator.Configure();
Log4NetLogger.Use();
}
static void InitStructureMap()
{
ObjectFactory.Initialize(x => {
x.AddRegistry<MyTestConsoleRegistry>();
x.AddRegistry<MyTestRegistry>();
});
}
}
public class MyTestConsoleRegistry : Registry
{
public MyTestConsoleRegistry()
{
var rabbitURI = ConfigurationManager.AppSettings["rabbitMQHostUri"];
var queueName = ConfigurationManager.AppSettings["massTransitQueue"];
For<IBusControl>(new SingletonLifecycle())
.Use("Configure IBusControl for MassTransit consumers with RabbitMQ transport",
ctx => Bus.Factory.CreateUsingRabbitMq(cfg => {
cfg.UseJsonSerializer();
cfg.PublisherConfirmation = true;
var host = cfg.Host(new Uri(rabbitURI), rabbitCfg => { });
cfg.ReceiveEndpoint(host, queueName, endpointCfg => {
endpointCfg.LoadFrom(ctx);
});
})
);
For<IReceiveObserver>().Use<MassTransitObserver>();
For<ISendObserver>().Use<MassTransitObserver>();
// ...snip...
}
}
public class MyTestRegistry : Registry
{
public MyTestRegistry()
{
ForConcreteType<MyTestConsumer>();
// ...snip...
}
}
public class MassTransitObserver : IReceiveObserver, ISendObserver
{
// Does nothing for now, just trying to wire it up...
public Task ConsumeFault<T>(ConsumeContext<T> context, TimeSpan duration, string consumerType, Exception exception) where T : class
{
return Task.CompletedTask;
}
public Task PostConsume<T>(ConsumeContext<T> context, TimeSpan duration, string consumerType) where T : class
{
return Task.CompletedTask;
}
public Task PostReceive(ReceiveContext context)
{
return Task.CompletedTask;
}
public Task PreReceive(ReceiveContext context)
{
return Task.CompletedTask;
}
public Task ReceiveFault(ReceiveContext context, Exception exception)
{
return Task.CompletedTask;
}
public Task PreSend<T>(SendContext<T> context) where T : class
{
return Task.CompletedTask;
}
public Task PostSend<T>(SendContext<T> context) where T : class
{
return Task.CompletedTask;
}
public Task SendFault<T>(SendContext<T> context, Exception exception) where T : class
{
return Task.CompletedTask;
}
}
public class MyTestConsumer : IConsumer<MyTestMessage>,
// for testing only:
IConsumer<MyTestResponse>
{
readonly IDoSomething _DoSomething;
public TestConsumer(IDoSomething doSomething)
{
_DoSomething = doSomething;
}
public Task Consume(ConsumeContext<MyTestResponse> context)
{
// For testing only...
return Task.CompletedTask;
}
public async Task Consume(ConsumeContext<MyTestMessage> context)
{
var result = await _DoSomething(context.Message.Id);
var resp = new MyTestResponseMessage(result);
await context
.Publish<MyTestResponse>(resp);
}
}
Given this code, the IReceiveObserver methods are getting called, but the ISendObserver methods are not.
I'm new to MassTransit, I expect this is probably a straightforward issue.
EDIT: A unit test using NUnit and Moq, doesn't use StructureMap. I believe this properly illustrates what I'm seeing.
[Test]
public void TestSendObserver()
{
var bus = CreateBus();
var busHandle = bus.Start();
var sendObs = new Mock<ISendObserver>();
sendObs.Setup(x => x.PreSend<TestMessage>(It.IsAny<SendContext<TestMessage>>()))
.Returns(Task.FromResult(0))
.Verifiable();
sendObs.Setup(x => x.PostSend<TestMessage>(It.IsAny<SendContext<TestMessage>>()))
.Returns(Task.FromResult(0))
.Verifiable();
using (bus.ConnectSendObserver(sendObs.Object)) {
var pubTask = bus.Publish(new TestMessage { Message = "Some test message" });
pubTask.Wait();
}
busHandle.Stop();
// Fails, neither PreSend nor PostSend have been called
sendObs.Verify(x => x.PreSend<TestMessage>(It.IsAny<SendContext<TestMessage>>()), Times.Once());
sendObs.Verify(x => x.PostSend<TestMessage>(It.IsAny<SendContext<TestMessage>>()), Times.Once());
}
IBusControl CreateBus()
{
return MassTransit.Bus.Factory.CreateUsingRabbitMq(x => {
var host = x.Host(new Uri("rabbitmq://localhost/"), h => {
h.Username("guest");
h.Password("guest");
});
});
}
public class TestMessage
{
public String Message { get; set; }
}
Up till now, we've implemented synchronous WebApi methods. Each action has a transaction created prior to the action call and that transaction is committed at the end of the call if all successful. If I now need to implement an async action, how would I need to change the action filter below to still have the transaction commit correctly? or would I need to implement a DelegatingHandler? Thank you in advance.
[AttributeUsage(AttributeTargets.Method, AllowMultiple = false)]
public class TransactionAttribute : ActionFilterAttribute
{
private IWindsorContainer Container
{
get { return IocContainer.Container; }
}
private WebRequestContext WebRequestContext
{
get
{
var webRequestContext = Container.Resolve<WebRequestContext>();
return webRequestContext;
}
}
private TransactionScope TransactionScope
{
get { return WebRequestContext.TransactionScope; }
set { WebRequestContext.TransactionScope = value; }
}
public override void OnActionExecuting(HttpActionContext actionContext)
{
var transactionOptions = new TransactionOptions
{
IsolationLevel = IsolationLevel.ReadCommitted
};
TransactionScope = new TransactionScope(TransactionScopeOption.Required,
transactionOptions);
base.OnActionExecuting(actionContext);
}
public override void OnActionExecuted(HttpActionExecutedContext actionExecutedContext)
{
using (TransactionScope)
{
base.OnActionExecuted(actionExecutedContext);
if (IsSuccessfulInvocation(actionExecutedContext))
{
TransactionScope.Complete();
}
}
}
private static bool IsSuccessfulInvocation(HttpActionExecutedContext actionExecutedContext)
{
return actionExecutedContext.Exception == null;
}