(Click to open topic with navigation)

You are here: References > nitrosub Command

5.1 nitrosub Command

The nitrosub command lets users easily submit Nitro jobs without having to create their own user job scripts; thereby not requiring the users to modify bash shell scripts. The nitrosub command is designed to submit static or dynamic Nitro jobs.

This topic provides information on the nitrosub command and what system administrators need to do to configure the command for their system.

In this topic:

5.1.1 Modify the nitrosub Command for your System

When Nitro is first installed, system administrators need to modify the bin/nitrosub script for their specific resource manager and licensing model. Specifically,

  1. Uncomment the "_resource_manager" line for your resource manager
  2. Uncomment the "resouce_type" line for your licensing model's allocation (nodes or cores).
  3. If your system will be using dynamic jobs, set the "_dynamic_size" value to the number of resources to allocate to a dynamic job.

5.1.2 Command Parameters

The following table describes the parameters for the nitrosub command and how they affect a Nitro job submission.

The terms "node" and "processor" are mutually-exclusive.

Parameter Required

Description

 Syntax
Environment Variables No

Passes environment variables to Nitro via by the job scheduler and/or resource manager.

Users may specify multiple environment variable names using a comma-delimited list (square brackets indicate optional additional environment variable names).

 --env-var=xxx=nnn[,yyy=mmm[,zzz=ooo]]
Job Directory Path No

Sets the directory path the user desires the Nitro job to use for recording job information. This parameter is required for a Nitro job restart when a user or administrator cancels an executing Nitro job and must have the job directory value of the canceled Nitro job.

When not set, it is the default job directory Nitro creates.

The directory path must be accessible from a compute node and the job submission node.

 --job-dir=xxx
Job ID No

Sets the job ID the user desires to give the job. This parameter is required for a Nitro job restart when a user or administrator cancels an executing Nitro job and must have the job id value of the canceled Nitro job.

When not set, it is the default job ID Nitro creates.

 --job-id=xxx
Local Worker No

Indicates whether the Nitro coordinator should start a local worker on its resources.

The default is ‑‑local‑worker.

--local-worker

--no-local-worker
Resources Yes

Indicates the number of hosts (compute nodes/servers) or hardware cores/threads to be allocated to the Nitro job by the scheduler.

  • If for a static Nitro job, the quantity value (nn) is a positive decimal integer.
  • If for a dynamic Nitro job, the quantity range values (mm-nn) are positive decimal integers and "mm" must be less than "nn".

--resources=nn

--resources=mm-nn
Task File Path Yes

Specifies the path of the Nitro task file created by the user.

This path must be accessible from a compute node

--task-file=xxx
Wall Time Yes Specifies the time limit for the Nitro job's execution. The parameter's "xxx" value format depends on the job scheduler used. --wall-time=xxx

© 2017 Adaptive Computing