none
task groups and "Fail the job if any task in the job fails" RRS feed

  • Question

  • I recently tried running a job that has 3 groups

    A -> B -> C that depend on each others. So in theory for any task of B, all the tasks of A have to be completed.

    I tried disabling "Fail the job if any task in the job fails" and strangely, HPC tried doing tasks from A, B, C despite the fact that couple of tasks in the group A failed. Is this an intended behavior?


    Erik Putrycz - Solution Architect - Apption Software
    Tuesday, August 30, 2011 1:33 PM

All replies

  • Hi Erik,

    Please check the solution, which I posted as a response to a similar question:

    http://social.microsoft.com/Forums/en-US/windowshpcdevs/thread/86822933-abf8-42dc-a27e-4c84aa351516

    Regards,
    Łukasz

    Saturday, September 10, 2011 8:23 PM
  • Hi Erik,

    Please check the solution, which I posted as a response to a similar question:

    http://social.microsoft.com/Forums/en-US/windowshpcdevs/thread/86822933-abf8-42dc-a27e-4c84aa351516

    Regards,
    Łukasz

    Hi Łukasz,

    Thanks for the link, but are there any chances this behavior could be changed in future versions? According to your documentation "All tasks in a group will complete before any tasks in the next group are started.", this sounds more like a bug to me...

    Thanks

    Erik.


    Erik Putrycz - Solution Architect - Apption Software
    Monday, September 12, 2011 2:40 PM
  • Hi Erik,

    Thank you for your feedback. I'll add your description of the problem to our bug database and it will be taken into account during the development of future versions.

    Also the mentioned thread suggested a workaround. Will it work for you, or is it not suitable for your scenario?

    Thanks,
    Łukasz

    Monday, September 12, 2011 3:22 PM
  • Hi Erik,

    Thank you for your feedback. I'll add your description of the problem to our bug database and it will be taken into account during the development of future versions.

    Also the mentioned thread suggested a workaround. Will it work for you, or is it not suitable for your scenario?

    Thanks,
    Łukasz

    The workaround looks fine but it would be a bit of a pain to setup, test and deploy partly because of the way I create the jobs... I rather just fail the job if any task fails for now.

    Thanks

    Erik


    Erik Putrycz - Solution Architect - Apption Software
    Monday, September 12, 2011 3:54 PM