Create a Job

Applies To: Windows Compute Cluster Server 2003

Job creation is the first step in cluster computing. Although jobs may be created and submitted at the same time, creation and submission are separate actions. Job submission consists of placing the job in the job queue.

A newly created job is a job with a set of user-specified job options or their defaults. It may or may not have tasks associated with it.

Note

A job without tasks may be submitted to reserve resources for a given run time, during which time tasks can be added as required.

To create a new job

  1. Click File, then click Submit Job….

  2. Click the General tab.

  3. Enter a job name in the Job Name text box. The job name can be any combination of characters and is the name that will appear under the Name column in the job queue.

  4. Enter a project name in the Project Name text box if there is a project associated with the job.

  5. Select a job priority from the drop-down list. For a user without administrative rights, the priority options are Normal, BelowNormal, and Lowest. For an administrator, the options are Highest, AboveNormal, Normal, BelowNormal, and Lowest.

    Note

    The General tab has a Submit button and a Save As Template... button. Every tab in the Job Properties sheet has these buttons. Use these buttons only when you are finished filling out all the tabs.

  6. Click the Processors tab.

  7. Enter the minimum and maximum number of compute processors required for the job. If you intend to set minimum and maximum processors for a task as well, they must a subset of the minimum and maximum processors for the job.

    Note

    If the minimum number of processors is too low, the job will slow and may stop. If the maximum of processors is too high, resources may be wasted. If the minimum number of processors is too high, the start of the job may be delayed.

  8. If you want to set a run time, enter it in days, hours, and minutes. A run time is the maximum amount of time allocated to the job. If the job is not finished within its run time, it is cancelled.

    Note

    If the minimum run time is too low, the job may time out before completion. If the minimum run time is too high, the start of the job may be delayed.

  9. If you want the job to hold the nodes reserved to it past the completion of all tasks, check Run job for duration of run time or until canceled. This will allow you to run extra tasks on the nodes until the end of the job run time.

  10. Click the Tasks tab. For steps to create tasks, see Create a Task.

  11. If you are using a license activation filter for license scheduling, select the Licenses tab.Using the Add button, add the name of each license feature required and the number of tokens of that feature.

    Note

    If the license filter cannot match the feature name entered with the name of an available feature, the job will remain in the queued state at top of the queue and will not activate. In this case, it will need to be cancelled and requeued. For this reason, it is important to enter the feature names carefully.

  12. If you want to specify individual nodes for this job or choose nonexclusive use of nodes, click the Advanced tab.

  13. In the list of nodes, check each node that you want to be used for your job.

    Note

    Checking nodes in this list only narrows the list of candidate nodes to those checked. It does not ensure that any given node will actually be reserved for the job, unless the minimum number of processors requested for the job requires all the nodes selected.

  14. If you want to make your nodes available to other jobs, clear the Use the allocated nodes exclusively for this job option. This also means that other jobs with the same setting will share their nodes with your job.

    Note

    This option does not affect backfilling.

  15. If you want to save your job as a template, click Save As Template....

  16. Enter a template name. The template name does not have to be the same as the job name. Then click OK.

    Note

    The template is an XML file describing the job that you have just defined. Although the template will include the job name if you entered one, the job name and the template name are distinct. This is because the template is not considered a job but a blueprint for jobs of a certain description.

  17. If you want to submit your job now, click Submit.

Creating a job from the command line

To create a new job at the command line interface, type:

job new [standard_job_options] [/jobfile:<job_file>] [/scheduler:<host>]

Note

When job creation is complete, the job ID for the new job will appear.

Standard job options

Option Definition Maximum Characters

/jobname:<job_name>

Name of the job.

80

/numprocessors:<min_processors> or <min_processors>- <max_processors>

Minimum and maximum number of processors to be allocated. The default is one processor.

N/A

/askednodes:<node1>,<node2>,…<nodeN>

Specifies individual nodes eligible to be reserved by the job. By default, all nodes in the cluster are candidates.

2080

/exclusive: true | false

By default, a job has exclusive use of nodes reserved by it. If /exclusive: is set to false, unused processors on these reserved nodes are available to other jobs. This is reciprocal, making nodes reserved to other jobs available to this job if they have also been flagged as nonexclusive.

N/A

/priority:<priority_class>

Schedule priority class: Highest, AboveNormal, Normal, BelowNormal,or Lowest. Highest and AboveNormal are available only to administrative users. The default is Normal. Within a priority class, the job is placed on a first come, first served basis unless it is requeued. If requeued, it always goes to the top of its priority class.

N/A

/runtime:[[[days:<num>]hours:<num>]minutes: <num>| infinite]

Maximum run time in day-hour-minute format. The job will be cancelled rather than allowed to run past the maximum run time. Default is Infinite.

8

/rununtilcanceled: true | false

Flag indicating that the job will hold its resources until it is cancelled or until it reaches its run time limit. This allows additional tasks to be run on the nodes as long as the job is running.

N/A

/projectname:<project_name>

Name of a project, if any, to which the job belongs.

80

/license:<feature1>:<amt1> /license:<feature2>:<amt2> …/license:<featureN>:<amtN>

License features required to run the tasks in the job and number of tokens of each.

160

Additional job options

Option Definition

/scheduler:<host>

Submit a job to the scheduler running on the specified host. By default, the job is submitted to the scheduler on the local host.

/jobfile:<job_file>

Overwrite the contents of an existing job with the values in this job template file, except where a different value is explicitly set in the command line.