Canceled C-Move kill conquest Server

  • Hi,
    We just connected a carestream streaming server to our conquest PACS : it basicly querys conquest to retrieve the dicom as needed.


    We face strange behavior/bug when a c-move operation is cancelled in the middle of file forwarding.
    As soon as the operation is canceled, conquest service just stops. Everytime it is canceled. Otherwise it is working perfectly.
    I updated from conquest 1417 to 1419 without improvement.


    Here I join the log of the begining end the ending of the c-move :


    Any ideas are welcomed ?


    PS : I would like to use this first post to also thanks all the conquest team and community for the great work. Using conquest for many years now and can't count all the helps conquest brought to our clinics and by extension their patients.


    Sincerely
    Thibaut

  • Hi,


    start a debugger e.g. visual studio express edition. Attach the debugger to dgate(64). exe and then send the c-cancel. The debugger should pop up and show the error location.


    More info will be shown if I share the dgate.pdb file, but you can start without.


    Marcel

  • Here is the debugger result (sorry it's french) :
    Exception levée à 0x000000013F59AE7A dans dgate64.exe : 0xC0000005 : Violation d'accès lors de la lecture de l'emplacement 0x0000000001F5BD58.

  • Hi,


    during a move there are two connections to conquest, the client and the receiver. Do you know which one sends the cancel request. And can you please report the exact version on which you reported the crash? Otherwise I cannot locate the crash address.


    Marcel

  • Sorry for the delay, I was on vacation and missed the notification,
    The os is windows server 2008 R2.
    ConQuest DICOM server version 1.4.19. Date of this release: 20170208.
    I believe the one who cancel the request is the receiver.

  • Hi,


    here is 64 bits dgate.exe (replace dgate64.exe), do not change name, and its associated debug file. If you place both in the server folder the debugger should show a bit more detail at a crash.


    Marcel

  • Hi Thibaut,


    Here is a fix (32 bits dgate with debug information). Can you give it a try? Basically there were two paths where an assocation could be dropped without the read ahead thread being stopped. This code is quite old, but never encountered issues with it. Now it closes the thread and waits for it in both possible paths.


    Please test it.


    Marcel

Participate now!

Don’t have an account yet? Register yourself now and be a part of our community!