This is sort of an odd request I think...
I would like to reduce the amount of traffic on the network when using HPC. I have my source files on a file server. When HPC gets the job it has to decide which resource (server) it is going to use. The solve process starts and the solution
is being performed by a third party software. As solution data becomes available it has to write back and forth to the file server. There may be other jobs running at the same time on other machines, and their project files would also be on the
file server. This means there will be lots of writing back and forth to and from the file server for all the jobs currently running.
If I could somehow know ahead of time which server it was going to use to solve I could move the project file over to that server, and then move it back to the file server in one chunk when it completes. I feel this would be more efficient since all
jobs would not be completing at the same time. I know I can allocate specific nodes for specific jobs, but I would like HPC to do this for me like it is supposed to. I just want to know if there is a way in the split second where the resources
get allocated that I can catch this with my script, and then move my file to that location.
Thank you for your assistance,
Craig
