In a Request Response pattern using MassTransit with RabbitMQ, I'm trying to create a request client. But when doing some research on the internet i saw two possibilities:
CreateRequestClient and CreatePublishRequestClient
Does someone know what's the difference between those two and when to use them?
see below for the methods:
public static class RequestClientExtensions
{
public static IRequestClient<TRequest, TResponse> CreateRequestClient<TRequest, TResponse>(this IBus bus, Uri address, TimeSpan timeout, TimeSpan? ttl = null, Action<SendContext<TRequest>> callback = null) where TRequest : class where TResponse : class
{
return (IRequestClient<TRequest, TResponse>) new MessageRequestClient<TRequest, TResponse>(bus, address, timeout, ttl, callback);
}
public static IRequestClient<TRequest, TResponse> CreatePublishRequestClient<TRequest, TResponse>(this IBus bus, TimeSpan timeout, TimeSpan? ttl = null, Action<SendContext<TRequest>> callback = null) where TRequest : class where TResponse : class
{
return (IRequestClient<TRequest, TResponse>) new PublishRequestClient<TRequest, TResponse>(bus, timeout, ttl, callback);
}
}
Well, the set of arguments explain the difference. This is the same difference as we have between Send and Publish. Publish uses fan-out exchange and Send delivers to a specific exchange only.
Normal RequestClient will do Send under the hood and needs the receiver address.
PublishRequestClient does not need any address since it will just publish the message and hope that someone will reply to it.
If you want to know more about Send vs Publish difference, you can check this blog post.
Related
I'm looking for a way to copy some headers from the request message to the response message when I use RabbitMq in RPC mode.
so far I have tried with setBeforeSendReplyPostProcessors but I can only access the response and add headers to it. but I don't have access to the request to get the values I need.
I have also tried with the advice chain, but the returnObject is null after proceeding so I can't modify it (I admit I don't understand why it is null... I thought I could get the object to modify it):
#Bean
public SimpleRabbitListenerContainerFactory simpleRabbitListenerContainerFactory(SimpleRabbitListenerContainerFactoryConfigurer simpleRabbitListenerContainerFactoryConfigurer, ConnectionFactory connectionFactory) {
SimpleRabbitListenerContainerFactory simpleRabbitListenerContainerFactory = new SimpleRabbitListenerContainerFactory();
simpleRabbitListenerContainerFactory.setAdviceChain(new MethodInterceptor() {
#Override
public Object invoke(MethodInvocation invocation) throws Throwable {
Object returnObject = invocation.proceed();
//returnObject is null here
return returnObject;
}
});
simpleRabbitListenerContainerFactoryConfigurer.configure(simpleRabbitListenerContainerFactory, connectionFactory);
return simpleRabbitListenerContainerFactory;
}
a working way is to change my method annotated with #RabbitListener so it returns a Message and there I can access both the requesting message (via arguments of the annotated method) and the response.
But I would like to do it automatically, since I need this feature at different places.
Basicaly I want to copy one header from the request message to the response.
this code do the job, but I want to do it through an aspect, or an interceptor.
#RabbitListener(queues = "myQueue"
, containerFactory = "simpleRabbitListenerContainerFactory")
public Message<MyResponseObject> execute(MyRequestObject myRequestObject, #Header("HEADER_TO_COPY") String headerToCopy) {
MyResponseObject myResponseObject = compute(myRequestObject);
return MessageBuilder.withPayload(myResponseObject)
.setHeader("HEADER_RESPONSE", headerToCopy)
.build();
}
The Message<?> return type support was added for this reason, but we could add an extension point to allow this, please open a GitHub issue.
Contributions are welcome.
I send a message from my client (implements IClientMessageInspector) using BeforeSendRequest() and receive the reply from the endpoint at AfterReceiveReply().
My question is what is the most effective way to "validate" the reply that I receive is in relation to the request I sent?
I found some article about using the correlationstate, but the examples were all too vauge.
Any help is greatly appreciated.
As far as I know, we could use the Correlationstate parameter to maintain the value in order to represent the relativity.
That’s why the BeofreSendRequest has the returned value.
public void AfterReceiveReply(ref Message reply, object correlationState)
{
Console.WriteLine(correlationState.ToString());
string displayText = $"the client has received the reply:\n{reply}\n";
Console.Write(displayText);
}
public object BeforeSendRequest(ref Message request, IClientChannel channel)
{
var correlationstate = Guid.NewGuid().ToString();
string displayText = $"the client send request message:\n{request}\n";
Console.WriteLine(displayText);
return correlationstate;
}
Here is a related discussion, wish it is useful to you.
https://social.msdn.microsoft.com/Forums/vstudio/en-US/c8de85bf-9ffe-478e-a23c-2514a6504bce/iclientmessageinspector-maintain-id-value-between-the-beforesendrequest-and-afterreceivereply?forum=wcf
Feel free to let me know if there is anything I can help with.
Class is defined as follows:
public class BizTalkRESTTransmitHandler : IClientMessageInspector
I'm a method with this signature:
public object BeforeSendRequest(ref Message request, IClientChannel channel)
So I think I need to manipulate the channel object.
The reason is this is being using in BizTalk 2010 SendPort to support JSON.
I tried this so far:
if (channel.RemoteAddress.Uri.AbsoluteUri == "http://api-stage2.mypartner.com/rest/events/2/"
|| channel.RemoteAddress.Uri.AbsoluteUri == "http://api.mypartner.com/rest/events/2/")
{
//TODO - "boxout" will become a variable obtained by parsing the message
Uri newUri = new Uri(channel.RemoteAddress.Uri.AbsoluteUri + "boxout");
channel.RemoteAddress.Uri = newUri;
}
Above gives compile error: "System.ServiceModel.EndpointAddress.Uri" cannot be assigned to - it is ready only" RemoteAddress seems to be read only as well.
I have referenced these questions but they don't use channel object.
Assign a URL to Url.AbsoluteUri in ASP.NET, and
WCF change endpoint address at runtime
But they don't seem to be dealing with channel object.
Update 1: I tried the following:
//try create new channel to change URL
WebHttpBinding myBinding = new WebHttpBinding();
EndpointAddress myEndpoint = new EndpointAddress(newURL);
ChannelFactory<IClientChannel> myChannelFactory = new ChannelFactory<IClientChannel>(myBinding, myEndpoint); //Change to you WCF interface
IClientChannel myNewChannel = myChannelFactory.CreateChannel();
channel = myNewChannel; //replace the channel parm passed to us
but it gave this error:
System.InvalidOperationException: Attempted to get contract type for IClientChannel, but that type is not a ServiceContract, nor does it inherit a ServiceContract.
IClientMessageInspector is not the right place the manipulate the Channel, you should use IEndpointBehavior instead:
From MSDN
Implements methods that can be used to extend run-time behavior for an
endpoint in either a service or client application.
Here is a simple example:
public void ApplyDispatchBehavior(ServiceEndpoint endpoint, EndpointDispatcher endpointDispatcher)
{
Uri endpointAddress = endpoint.Address.Uri;
string address = endpointAddress.ToString();
if (address == "http://api-stage2.mypartner.com/rest/events/2/"
|| address == "http://api.mypartner.com/rest/events/2/")
{
//TODO - "boxout" will become a variable obtained by parsing the message
Uri newUri = new Uri(address + "boxout");
ServiceHostBase host = endpointDispatcher.ChannelDispatcher.Host;
ChannelDispatcher newDispatcher = this.CreateChannelDispatcher(host, endpoint, newUri);
host.ChannelDispatchers.Add(newDispatcher);
}
}
Here you can read the excelent post of Carlos Figueira about IEndpointBehavior:
https://blogs.msdn.microsoft.com/carlosfigueira/2011/04/04/wcf-extensibility-iendpointbehavior/
Another alternative is to implement a simple Routing with WCF, here is link with an example:
WCF REST service url routing based on query parameters
Hope it helps.
Using the interface IEndpointBehavior, you'll have access to the ApplyClientBehavior method, which exposes the ServiceEndPoint instance.
Now you can change the value for the Address by defining a new EndpointAddress instance.
public class MyCustomEndpointBehavior : IEndpointBehavior
{
public void AddBindingParameters(ServiceEndpoint serviceEndpoint, System.ServiceModel.Channels.BindingParameterCollection bindingParameters)
{
}
public void ApplyClientBehavior(ServiceEndpoint serviceEndpoint, System.ServiceModel.Dispatcher.ClientRuntime behavior)
{
serviceEndpoint.Address = new System.ServiceModel.EndpointAddress("http://mynewaddress.com");
}
public void ApplyDispatchBehavior(ServiceEndpoint serviceEndpoint, System.ServiceModel.Dispatcher.EndpointDispatcher endpointDispatcher)
{
}
public void Validate(ServiceEndpoint serviceEndpoint)
{
}
}
I might be a bit too late but hoe it helps a bit.
I recently had a similar objective (also related to biztalk) where I needed to change the url based on some value sent on the message.
I tried using the ApplyDispatchBehavior method but it was never called and also, I couldn't see how to access the message from here so I started looking at method BeforeSendRequest (in the Inspector class).
Here is what i came up with:
object IClientMessageInspector.BeforeSendRequest(ref Message request, IClientChannel channel)
{
var queryDictionary = HttpUtility.ParseQueryString(request.Headers.To.Query);
string parameterValue = queryDictionary[this.BehaviourConfiguration.QueryParameter];
//Only change parameter value if it exists
if (parameterValue != null)
{
MessageBuffer buffer = request.CreateBufferedCopy(Int32.MaxValue);
request = buffer.CreateMessage();
//Necessary in order to read the message without having WCF throwing and error saying
//the messas was already read
var reqAux = buffer.CreateMessage();
//For some reason the message comes in binary inside tags <Binary>MESSAGE</Binary>
using (MemoryStream ms = new MemoryStream(Convert.FromBase64String(reqAux.ToString().Replace("<Binary>", "").Replace("</Binary>", ""))))
{
ms.Position = 0;
string val = ExtractNodeValueByXPath(ms, this.BehaviourConfiguration.FieldXpath);
queryDictionary.Set(this.BehaviourConfiguration.QueryParameter, DateTime.Now.ToString("yyyyMMddHHmmssfff") + "_" +
this.BehaviourConfiguration.Message + (string.IsNullOrWhiteSpace(val) ? string.Empty : "_" + val) + ".xml");
UriBuilder ub = new UriBuilder(request.Headers.To);
ub.Query = queryDictionary.ToString();
request.Headers.To = ub.Uri;
}
}
return null;
}
So, I discovered that, messing with the request.Headers.To I could change the endpoint.
I had several problems getting the message content and most examples on the internet (showing to use the MessageBuffer.CreateNavigator or Message.GetBody< string > which was always throwing an expcetion i couldn't get around) would not give me the biztalk message but rather the soap message?... not sure but it had a node header, body and inside the body there was some base64 string which was not my biztalk message.
Also, as you can see in Convert.FromBase64String(reqAux.ToString().Replace("<Binary>", "").Replace("</Binary>", "")), I had to do this ugly replaces. I don't don't why this comes in base64, probably some WCF configuration?, but by doing it, I could then look for my value.
NOTE: I haven't fully tested this, but so far it as worked for my examples.
By the way, any idea on what can i switch my MemoryStream with so it becomes a more streaming solution?
I have an ASP application which is client of WCF SERVICE1 , which is client of WCF SERVICE2.
I have added IDispatchMessageInspector and IClientMessageInspector to WCF SERVICE1.
Now I need to pass a custom value from ASP to WCF1 , then to WCF2.
from ASP to WCF1 it is trivial , via Message Headers.
The question is , how to pass a custom value from IDispatchMessageInspector.AfterReceiveRequest(request from ASP received by WCF1) to IClientMessageInspector.BeforeSendRequest(prepare to send request to WCF2) operation of WCF SERVICE 1 ?
Is there is some context which could be used ?
What does your code look like? Assuming that first Dispatch Message Inspector is the one making the request to WCF2, then simply using message properties would suffice.
However, if your dispatch message inspector does something; then the request continues processing and it is the service implementation that actually calls WCF2, then you'll need to jump through a few more hoops. In general, I'd say you'd need the inspector to put some data in the service request message properties that the service implementation would need to pick up and copy to the message to send to WCF2 so that the client inspector can pick them up.
That's ugly, and would kinda make the whole process more brittle.
Can you elaborate a bit more what you're trying to do? What kind of data are you hoping to pass around this way?
In my case, I had to identify and log nested service calls requested by client.
To do that, I stamp each service call by ThreadStatic property and add this property to the header of client call(service1 count as client for service2) than in AfterReceiveRequest method I have checked its existance. If exists,current method was requested by parent service.
public class GenericMessageInspector : IDispatchMessageInspector, IClientMessageInspector
{
[ThreadStatic]
private static string _masterServiceGUID;
public object AfterReceiveRequest(ref System.ServiceModel.Channels.Message request, System.ServiceModel.IClientChannel channel, System.ServiceModel.InstanceContext instanceContext)
{
if (request.Headers.Action == null)
return null;
//Control request header for nested call
string masterRequestId = string.Empty;
var IsMasterExist = request.Headers.FindHeader("MasterServiceGUID", "namespace");
if (IsMasterExist > -1)
{
//requested by internal service
masterRequestId = request.Headers.GetReaderAtHeader(IsMasterExist).ReadInnerXml();
}
}
public object BeforeSendRequest(ref Message request, IClientChannel channel)
{
if (!String.IsNullOrEmpty(_masterServiceGUID))
{
request.Headers.Add(MessageHeader.CreateHeader("MasterServiceGUID", "namespace", _masterServiceGUID));
}
return null;
}
}
}
We've got the following WCF Service Contracts:
[ServiceContract(Namespace = "http://example.com", Name = "Service1")]
public interface IService1
{
[OperationContract]
[FaultContract(typeof(Fault1))]
ValidateUserResult ValidateUser(
string username,
string password);
}
[ServiceContract(Namespace = "http://example.com", Name = "Service1")]
public interface IService1Async
{
[OperationContract(AsyncPattern = true)]
[FaultContract(typeof(Fault1))]
IAsyncResult BeginValidateUser(
string username,
string password,
AsyncCallback callback,
object userState);
ValidateUserResult EndValidateUser(IAsyncResult asyncResult);
}
[DataContract(Namespace = "http://example.com")]
public class Fault1
{
}
We are calling the async version of ValidateUser in the client side and we are throwing a FaultException<Fault1> on the server, but all the client receives is the base FaultException.
What can be the reason the contractually-specified fault is not being received?
We found now why. We are generating the fault from a Service Behaviour using the ProvideFault method. There we use code similar to the example at IErrorHandler.ProvideFault in msdn
The only difference was that we weren't passing the right action on the Message.CreateMessage overload. We copied exactly what gets generated in the case that we manually throw the fault and voila.
My excuses for not giving that last detail :-)
Can you show us the catch statements for your call? The contract and everything looks fine to me...
In which order do you check for faults?? You would have to check for FaultException<Fault1> before checking for FaultException or CommunicationException - any chance you might have that order mixed up somehow?
Does it work when you call the sync version of the method?