Sign in
Microsoft.com
 
Microsoft
 
 
 

locked
An operation on a socket could not be performed because the system lacked sufficient buffer space or because a queue was full RRS feed

 > 

  • Question

  • Question
    Sign in to vote
    0
    Sign in to vote

    Dear all,

    We have integrated Windows HPC in our solution. Currently, our application does polling to HPC to get the job status in every 2 seconds. We receive the error "An operation on a socket could not be performed because the system lacked sufficient buffer space or because a queue was full.".  (snapshot is attached below). Is this issue due to consistent pooling to HPC? we also observe huge memory consumption (80%) due to this. Please help us.


    Puneet Sharma


    Monday, July 10, 2017 6:28 PM

Answers

  • Question
    Sign in to vote
    0
    Sign in to vote

    This looks like the usage pattern you're using with "poll job status" is causing port leaks. Could you share the code snip with us so that we can have a local retry?

    And an appropriate approach on "poll job status" is to: Register Job State Change Event + Polling (more than 2 seconds, for example, if there is no event get for the lastest 3 minutes), thus to lower the system load.

    Please also share the exact HPC Pack version you're using.

    Qiufang Shi


    Tuesday, July 11, 2017 2:10 AM

All replies

  • Question
    Sign in to vote
    0
    Sign in to vote

    This looks like the usage pattern you're using with "poll job status" is causing port leaks. Could you share the code snip with us so that we can have a local retry?

    And an appropriate approach on "poll job status" is to: Register Job State Change Event + Polling (more than 2 seconds, for example, if there is no event get for the lastest 3 minutes), thus to lower the system load.

    Please also share the exact HPC Pack version you're using.

    Qiufang Shi


    Tuesday, July 11, 2017 2:10 AM
  • Question
    Sign in to vote
    0
    Sign in to vote
    Yes Quifang, we are going to implement this hybrid approach. For now, we increase the polling time period to 10 seconds and things are looking fine. Thanks for the prompt response.

    Puneet Sharma

    Tuesday, July 11, 2017 9:04 PM