Submitting and Monitoring Ansys EM HPC Jobs

Jobs may be submitted to the Windows HPC Scheduler using any of the following methods:

Client Utilities from the Microsoft HPC Pack, must be installed on the submit host to use any of these methods to submit a job to a cluster. The Submit HPC Job dialog box will be unable to contact the cluster head node if the client utilities are not installed.

This document covers the first method. See the Microsoft documentation for information on the other three methods.

Jobs may be submitted from any Microsoft Windows host meeting the following requirements:

Job Monitoring

  1. Windows HPC Jobs may be monitored using the Monitor Job dialog box, which is brought up by the Tools > Job Management > Monitor Jobs... command. This dialog bos may also be brought up by checking the Begin monitoring this job now check box when a job is successfully submitting using the job submission dialog box. You can also select the Simulation tab of the ribbon and select the Monitor icon. You can monitor this job either automatically (by checking the option) or through the Tools> Job Management> Monitor Jobs... command. For more details, see Monitor Jobs window.

In addition to the above requirements to allow job monitoring the following is also necessary:

Cluster Configuration

Any job running on a Windows HPC Cluster that is distributed over multiple compute hosts requires network communication between processes running on these hosts. The cluster must be configured to allow this communication. Any firewall or other security software must be disabled or configured to allow communication between any of the compute hosts were a job could run.

Job Submission User Profile on Cluster Compute Nodes

In order for a job to run correctly, the submission user's profile must be accessible and properly initialized on the cluster compute nodes where the job runs. If the Ansoft/temp subdirectory of the user's "My Documents" directory does not exist or is not accessible on the compute cluster nodes where a job runs, the batchoptions for the job will not be processed correctly, resulting in job failure. One way to ensure that this directory is created on each compute host is for the submission user to login to each compute host and run the product GUI one time.