Here is some context on how we are handling it currently, but have now started running into the issue outlined above. Is there better ways on handling this? Or any way to tell the consumer to forward any incoming response types sent via RespondAsync to the sender (alarm bells are probably ringing here)? Reasoning behind this is that I'd rather an exception being thrown immediately if the type is not handled, instead of the requester waiting 30s for a timeout exception due to the type not being specified (I'm also nervous as consumers evolve, developers not updating requests to match changed or additional types across a large system). I also could just throw exceptions, but I don't want to fill the error queue and telemetry with user input errors / continually need to filter out domain exceptions so they are ignored (plus having concrete response types makes it a lot easier to integrate with GraphQL and access the data). I know I can just add extension methods to support N amount of responses but it seems rather verbose, and something I'd rather not continue to tac onto if avoidable. This has left us in a bit of a pickle, since some consumers can have a fair few different negative responses and GetResponse accepts up to 3 response types.

The plan was to map any negative responses from the consumers into GraphQL types, where the client can make use of the data using Stage 6a: Error Union List + Interface Stage 6a: Error Union List + Interface. The api layer of the project uses GraphQL to ingest incoming requests, with MassTransit request / response to forward the request to the consumers. If I remove the call requestClient.GetResponse() the SUT times out without receiving the response from the stubbed consumer.I'm in search for a better or more elegant way to handle multiple response types (potentially half a dozen on more) when using request / response. The first message is dispatched by var response = await requestClient.GetResponse(startEvent) Īnd then the second happens in my system under test var result = await helper.SendPing().ConfigureAwait(false) Var result = await helper.SendPing().ConfigureAwait(false) Īssert.True(().Any()) Īssert.True(().Any()) Īssert.Equal(expectedResult.Time, result.Time) Īssert.Equal(expectedResult.Time, ) Var response = await requestClient.GetResponse(startEvent)

when both are called the StubbedConsumer is called twice this configures the response, these 2 are equivalent but without calling get response on the request client the tests time out Var requestClient = await harness.ConnectRequestClient().ConfigureAwait(false) Var helper = new PingHelper(harness.BusControl, _logger.Object) helper needs to be after harness is started Harness.Consumer(() => new StubbedConsumer(expectedResult)) the harness needs a consumer for request to work Var expectedResult = new EventsPingCompleted() My unit test is passing but when I debug it the stubbed consumer is being called in my unit test. Public async Task Consume(ConsumeContext context)Īwait context.RespondAsync(_response).ConfigureAwait(false) Public StubbedConsumer( TRt expectedResponse ) public class StubbedConsumer: IConsumer where T : class where TRt : class To do this I created a stubbed consumer because I don't want to unit test the consumer, I am doing that in other tests. I want to return a specific response as part of my unit test. Var response = await _bus.Request(message).ConfigureAwait(false) _logger.LogInformation($'Sending Events Ping') I am trying to test a method that calls Masstransit Request public async Task SendPing()

Masstransit request response