locked
JobStatusEventHandler internal communication mechanism RRS feed

  • Question

  • Dear all,

    Would anybody please explain to me what happens when I create a JobStatusEventHandler for some job. What kind of connection is being created internally? what ports are being opens? and how does head nodes push the job status to the event handler?

    If I have 100 jobs running and attached JobStatusEventHandler for all of them, how many ports are going to be opened? how scalable is this approach? 

    Thanks,

    Puneet



    • Edited by PuneetSharma035 Wednesday, September 27, 2017 8:05 PM formating
    Wednesday, September 27, 2017 8:04 PM

All replies

  • Hi Sharma,

      If you're using versions before HPC Pack 2016, it is .net remoting underneath. Otherwise it is WCF. And scheduler listened on a port on the registered client and pushing events to the client when needed.

      For your second question, 100 jobs running, there will be still one port opening. 100 jobs update shouldn't be problem. But please be noted that the event notification is not guaranteed when the headnode is very busy or under pressure as we don't have retry mechanism. And event only available for C# API, not available for REST API.

      Our job GUI is leveraging the Event System, thus if you find the GUI can handle it, your code should handle it as well.

     


    Qiufang Shi

    • Proposed as answer by Puneet035 Wednesday, October 11, 2017 2:19 AM
    Friday, September 29, 2017 6:16 AM