A debatch step followed by another request-response step using the same send port fails

[refurbs]

Firstly, great work on this project!

But, I've hit a potential issue...

I have request-response step which is configured to debatch 3 messages on send port xyz. This works as expected and returns 3 responses.

But when I add another single request-response step afterwards which also uses send port xyz, the first debatch step returns 1 response, then fails to return the other 2 and fails with a timeout?

If I configure the second request-response step to use a different send port abc then all is well again???

Is this a bug or am I doing something wrong? Any thoughts appreciated.

[refurbs]

Further analysis suggests the asynchronous setup/teardown behaviour combined with the 1-1 pipe connection are the cause of this issue. I wonder if introducing a 1-* pipe connection is the potential solution? I will look to test this theory as and when time allows.

[svetvasilev]

Hi, thanks for the feedback and taking the time to register, and digging deeper into this issue. I think you are right when saying that this behavior is caused by the fact that i try to maintain a 1-1 relationship of client-server connection. One of the earlier versions of the adapter was 1 to many implementation but there were many issue keeping things in sync, so I ought to go the 1-1 way. In this particular case however the actual cause is, i suspect as i haven't had the time to re-create it, caused by the fact that the steps performing receive operation (MockReceive and MockRequestResponse) start the named pipe server during step validation. In BizUnit life cycle this happens at the start of the test. Hence you end up in having 2 instances of named pipe servers listening on the same address. While this is totally legitimate scenario it kind of undermines the purpose of what you try to achieve with this test. This is one of the caveats of not only TransMock but the programming model of BizUnit as well.
I suggest you try to set up the debatch step to expect one extra message and configure the verification to be of cascading type. This way you have the freedom of verifying each received message separately. The last one although not produced by the debatch operation will still be received as part of this steps instance. In case you have another step executing between the debatch and non-debatched messages on this port please set the debatch step for concurrent execution.

[refurbs]

Hi Svet,
Thanks for giving some insight into the history, cause and suggesting a workaround. Admittedly, I would have preferred a more elegant solution, but I accept that may not be possible based on what you have said. As I don't want to abandon TransMock, I will try and use the workaround for now. Thanks again.
Rob

[svetvasilev]

Hi Rob, were you able to progress with your test case after following my suggestions? Wanted to check with you so that I find out what to do with this issue.
Thanks, Svet

[refurbs]

Hi Svet,

Thanks for asking. I did attempt to implement your suggestion, but was unable to make it work. Here's what I did:

• I combined the debatch and single MockRequestResponseStep into one.
• I increased the DebatchedMessageCount from 3 to 4.
• Set the ValidationMode to Cascading.
• Created and attached a collection of 4 CascadingSubSteps of type LambdaValidationStep.

It worked in the sense I was able to trigger a validation event for each individual message in the batch. However, by combining steps I needed the flexibility to set a different ResponsePath for the final "non batch" message. I could not find any way to do this in the current object model? The parent MockRequestResponseStep is not accessible from within the LambdaValidationStep?

Rob

[svetvasilev]

Hi Rob, were you able to get this running with the latest version? In 1.4 you have the ability to govern the response content for each and every request in a debatch step, which should be sufficient to allow you to only use a single step instance in your scenario. Have a look at my blog post here on how you can achieve this if you havent done so yet. Give a sound in case you need some help.
Cheers, Svet

[Shille88]

I have the same issue but with different setup.
I have two [TestMethod] which runs sequencially (succeed and soap fault). Both are configured to hit the same endpoint on BizTalk. The first will always succeed while the second one will always fail, caused by timeout, this also applies when I change the order of the [TestMethod]'s. If I run them seperateley by manually activating them 1 by 1, both will succeed.

    [TestMethod]
    [DeploymentItem(@"TestData\LabTrans\Success\HoursServiceCreateResponse.xml")]
    [DeploymentItem(@"TestData\LabTrans\Success\PublishAXLABTRANS.xml")]
    public void LabTrans()
    {
        var serviceMock = new EndpointsMock<Quant_IntMockAddresses>();
        serviceMock.SetupReceive(ep => ep.Rcv_Int_MX_Hours_Http)
            .SetupSendRequestAndReceiveResponse(ep => ep.Snd_Int_AX_Hours_Http);

        var testClient = serviceMock.CreateMessagingClient();
        testClient.Send(
                r => r.Rcv_Int_MX_Hours_Http,
                "PublishAXLABTRANS.xml",
                Encoding.UTF8,
                10,
                beforeSendAction: ctx => ctx.DebugInfo("Sending message in to Rcv_MX_Hours_Http")
            )
           .ReceiveRequestAndSendResponse(rp => rp.Snd_Int_AX_Hours_Http,
           responseSelector: rs => new StaticFileResponseSelector()
           {
               FilePath = "HoursServiceCreateResponse.xml"
           },
           requestValidator: x => ValidateOutMessage(x));
    }

    [TestMethod]
    [DeploymentItem(@"TestData\LabTrans\ZeroHours\PublishAXLABTRANS.xml")]
    [DeploymentItem(@"TestData\LabTrans\ZeroHours\Fault.xml")]
    public void LabTransZeroHours()
    {
        var serviceMock = new EndpointsMock<Quant_IntMockAddresses>();
        serviceMock.SetupReceive(ep => ep.Rcv_Int_MX_Hours_Http)
            .SetupSendRequestAndReceiveResponse(ep => ep.Snd_Int_AX_Hours_Http);

        var testClient = serviceMock.CreateMessagingClient();
        testClient
            .Send(
                r => r.Rcv_Int_MX_Hours_Http,
                "PublishAXLABTRANS.xml",
                Encoding.UTF8,
                10,
                beforeSendAction: ctx => ctx.DebugInfo("Sending message in to Rcv_MX_Hours_Http")
            )
            .ReceiveRequestAndSendResponse(rp => rp.Snd_Int_AX_Hours_Http,
            responseSelector: rs => new StaticFileResponseSelector()
            {
                FilePath = "Fault.xml"
            },
           requestValidator: x => ValidateZeroHoursMessage(x));
    }

[svetvasilev]

Hi Ton, thanks for sharing this. I am glad to see that you are utilizing the new programming model. This is s clear indication that there is something fishy with the adapter and connection management when a connection through a given named pipe is closed and then tried re-open again. I have authored several unit tests around this issue and they are all passing just fine. I guess i miss something and have to dig a bit further. Will get back to you once i have something worked out. In the meantime could you please share how do you find the new programming model? Thanks, Kind regards, Svetoslav

…

On Wed, Nov 20, 2019 at 12:48 PM Sherjeel Ton ***@***.***> wrote: I have the same issue but with different setup. I have two [TestMethod] which runs sequencially (succeed and soap fault). Both are configured to hit the same endpoint on BizTalk. The first will always succeed while the second one will always fail, caused by timeout, this also applies when I change the order of the [TestMethod]'s. If I run them seperateley by manually activating them 1 by 1, both will succeed. [TestMethod] [DeploymentItem(@"TestData\LabTrans\Success\HoursServiceCreateResponse.xml")] [DeploymentItem(@"TestData\LabTrans\Success\PublishAXLABTRANS.xml")] public void LabTrans() { var serviceMock = new EndpointsMock<Quant_IntMockAddresses>(); serviceMock.SetupReceive(ep => ep.Rcv_Int_MX_Hours_Http) .SetupSendRequestAndReceiveResponse(ep => ep.Snd_Int_AX_Hours_Http); var testClient = serviceMock.CreateMessagingClient(); testClient.Send( r => r.Rcv_Int_MX_Hours_Http, "PublishAXLABTRANS.xml", Encoding.UTF8, 10, beforeSendAction: ctx => ctx.DebugInfo("Sending message in to Rcv_MX_Hours_Http") ) .ReceiveRequestAndSendResponse(rp => rp.Snd_Int_AX_Hours_Http, responseSelector: rs => new StaticFileResponseSelector() { FilePath = "HoursServiceCreateResponse.xml" }, requestValidator: x => ValidateOutMessage(x)); } [TestMethod] [DeploymentItem(@"TestData\LabTrans\ZeroHours\PublishAXLABTRANS.xml")] [DeploymentItem(@"TestData\LabTrans\ZeroHours\Fault.xml")] public void LabTransZeroHours() { var serviceMock = new EndpointsMock<Quant_IntMockAddresses>(); serviceMock.SetupReceive(ep => ep.Rcv_Int_MX_Hours_Http) .SetupSendRequestAndReceiveResponse(ep => ep.Snd_Int_AX_Hours_Http); var testClient = serviceMock.CreateMessagingClient(); testClient .Send( r => r.Rcv_Int_MX_Hours_Http, "PublishAXLABTRANS.xml", Encoding.UTF8, 10, beforeSendAction: ctx => ctx.DebugInfo("Sending message in to Rcv_MX_Hours_Http") ) .ReceiveRequestAndSendResponse(rp => rp.Snd_Int_AX_Hours_Http, responseSelector: rs => new StaticFileResponseSelector() { FilePath = "Fault.xml" }, requestValidator: x => ValidateZeroHoursMessage(x)); } — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#9?email_source=notifications&email_token=AGRNIWYBG6VABBPZ7NSHDPTQUUPZRA5CNFSM4E7OW6Z2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEERW3SY#issuecomment-555969995>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AGRNIW3S4LNBFTFOS4JIHLLQUUPZRANCNFSM4E7OW6ZQ> .

[Shille88]

Hi,
Thanks for your response. The new programming model is good, I like it.

So the same setup works on your machine, even if you use the function "Run All tests" through the test explorer?