Forum Discussion

jens's avatar
jens
New Contributor
16 years ago

Set ThreadPool for MockServiceRunner

Hi,

is there any way to modify the thread pool available to a mock service when starting it via MockServiceRunner.bat?

We are testing a SOAP intermediary so we are using SoapUI for both the client and the server. Right now the bottleneck is not our intermediary but the mock service.

Webservices that have but a few operations perform very well (< 5ms Response time). But webservices with some more operations take considerably longer (200ms).  I presume this is caused by the QueuedThreadPool. We would like to increase the threadpool in order to overcome this limitation.

Best regards,

Jens
  • SmartBear_Suppo's avatar
    SmartBear_Suppo
    Icon for Alumni rankAlumni
    16 years ago
    Hi Jens,

    Although we could easily add a possibility for this I want to make sure that this is the actual cause of your problems, since the default seems to be 250 threads (judging by http://jetty.codehaus.org/jetty/jetty-6 ... dPool.html) which is quite a lot. I wonder:
    - which soapUI version are you using?
    - what kind of dispatcher are you using for your mockservice operations?
    - are there any scripts involved in your mock functionality?

    regards,

    /Ole
    eviware.com
  • jens's avatar
    jens
    New Contributor
    16 years ago
    Hi Ole,

    - the Version is 3.5 (free right now, pro soon).

    - The dispatcher choosen right now is just sequence and there exists just one response. We are planning to complicate things later on with scripting, though.

    - There are no scripts involved in the mock functionality (again, planning to).

    To illustrate the situation, please find a screenshot attached. The difference in response time between the services with just a few operations and those with more is staggering. This test run had the mock service as a direct communication partner configured. Our intermediary did not take part.

    As you can see, the response times inside the poorly performing mock service increases along the operators. Hence my guess concerning the thread pool.

    Best regards,

    Jens
  • jens's avatar
    jens
    New Contributor
    16 years ago
    Hi Ole,

    I am not able to send you the project file as the company I am working for considers the WSDLs to be confidential - sorry.

    Regards,

    Jens
  • SmartBear_Suppo's avatar
    SmartBear_Suppo
    Icon for Alumni rankAlumni
    16 years ago
    Hi Jens,

    ok, I've been trying to reproduce this with two different mockservices; one containing 10 mockoperations with one response each, the other containing just 2. Unfortunately I can't see any performance difference as you do; the performance decreases for both my mock services in line with the numbers you are showing and my profiling shows that most time is spent actually sending and receiving the request/response messages.

    Don't know how to move this further.. can you reproduce it with some publically available wsdl and send me your project? (for example with the amazon wsdl at http://soap.amazon.com/schemas2/AmazonWebServices.wsdl)

    regards,

    /Ole
    eviware.com
  • jens's avatar
    jens
    New Contributor
    16 years ago
    Hi Ole,

    sorry for the late response, the project needed attendance.

    I've tried reproducing the problem with the Amazon WSDL but wasn't able to. Infrastructure and Client/Server Setup where the same as with our WSDLs. As you reported already, there was no differrence between a mockservice with multiple operations and one with but one operation.

    So I still don't know why the problem arises but will have to find out myself. Thank you for your support Ole.

    Best regards

    Jens