The following topics describe commands that are used to manage job submission in Ansys RSM Cluster (ARC) setups:
Note: This is for reference or debugging purposes only. We do not recommend using these commands directly for cluster job submission. RSM is designed to integrate with Ansys RSM Cluster setups, eliminating the need to manually issue job submission commands.
The arcsubmit command in the ARC command directory is used to submit a job to an Ansys RSM Cluster (ARC). It will return a job ID that will be used to determine when the job is completed, or to cancel the job if needed.
Note: The arcsubmit command cannot be issued independently. It only works when issued via RSM.
Table 4.2: Options for 'arcsubmit'
| Option | Description | Default Setting |
|---|---|---|
| -n[umberExecUnits] [cores: ]2 | The number of cores used for the job. | [cores:]1 |
| -q[ueue] <queueName> | The name of the cluster queue to which the job will be submitted. | 'default' |
| -a[liasName] <testjob> | User-specified name to be used for the job, as needed for accounting or display | blank |
| -d[istributed] <fill | span> | Defines the distributed mode. For distributed jobs, this flag must be set to either “fill” or “span”. “Fill” means use all the cores on machine1 before using machine2. “Span” means use one core from every machine first. If not set, then the job will wait until it can run on a single machine. | None |
| -r[esource] m[emory]:25[b|kb|mb|gb|tb] | Suggests the maximum memory that the job will need to use. Default of 0 means “do not track”. | m:0mb |
| -r[esource] d[isk]:30[b|kb|mb|gb|tb] | Suggests the maximum disk space that the job will need to allocate. Default of 0 means “do not track”. | d:0mb |
| -r[esource] d[isk]:30[b|kb|mb|gb|tb],m[emory]:40[b|kb|mb|gb|tb] | Specifies the maximum resources (memory and disk space) that can be allocated to a job. If the resource needed to run a job is not available, the job will remain queued until it is able to run. | |
| -w[orkdir] /somewhere/else/ | The staging directory for job files. (This directory should be shared by all the execution nodes.) | current directory |
| -x[clusive] | Runs jobs in "exclusive" mode. This means that when a job is running, no other jobs can be scheduled on the same nodes that are assigned to this job. | not used |
| -e[rrorFile] FileName | Defines an alternate error file name. This is where stderr will be redirected from the cluster job. | <Job#>.Error |
| -o[utputFile] FileName | Defines an alternate output file name. This is where stdout will be redirected from the cluster job. | <Job#>.Output |
| -v[ariables] <All | var1:var2:> | Allows specification of environment variables that need to be passed to the job. They should be already set in the submission environment in order to be passed. The keyword “All” will pass almost all variables except some system-specific variables. | None |
The arcstatus command in the ARC command directory is used to retrieve and display the status of a job.
If the job ID is provided, the status of that specific job is retrieved.
arcstatus [jobId]
If no job ID is provided, the status of all running jobs is retrieved.
The following example shows the types of information that the arcstatus command retrieves:
C:\Program Files\ANSYS Inc\v242\RSM\ARC\tools\winx64>arcstatus
JobNumber Status UserName Cores Submit Time Queue Assigned Machines Info
===================================================================================================
1 Finished ANSYS\atester 2 11/12/2018 17:21:09 test atester1234:2 No ErrorsThe arckill command in the ARC command directory is used to cancel a running job. The job ID is provided as an argument:
arckill [jobId]