1. Introduction

NQS (Networking Queueing System) has been installed on Universe (Cosmos and Eric) to control CPU-intensive jobs --- to ensure that all users get a fair share of the CPU resources available and that the systems do not "grind almost to a halt", as they have on occasion in the past.

The Network Queuing System (NQS) was originally developed by Sterling Software for National Aeronautics and Space Administration (NASA).

All CPU-intensive jobs which are run on Universe (Cosmos and Eric) must be submitted to the NQS (rather than run interactively or simply backgrounded). Jobs which are not submitted to the NQS may be killed without notice.

1.1. What's a CPU-intensive job?

1.2. How does it work (from a user point of view)?

Jobs to be run are "submitted" to a chosen queue. When resources become free the job is scheduled and run. Output from the job which would normally go to stdout is sent to a log file.

After submission the user can watch a job's progress by using the qstat command. In addition a user can hold a queued job (prevent it from being scheduled), releasing a held job, suspending a running job, resuming a suspended job, and delete a queued or running job.

...cont'snext...



About this document:

Produced from the SGML: /home/isd/public_html/_cosmeric/_nqs_user/_reml_grp/index.reml
On: 17/7/2003 at 18:7:12
Options: reml2 -i noindex -l long -o html -p multiple