Easy Access to Queue Status

(Initially established as BC Policy FY07-05, 2 October 2008)

It is required that each allocated system at each center has available a command named "show_queues" which will display current queuing system information in the following format:

QUEUE INFORMATION:
                   Maximum            Jobs                Cores        Queue
Queue Name   Wall Time   Cores   Running  Pending   Running  Pending  Running
------------------------------------------------------------------------------
<queue #1>   <hhh:mm>    <cores>   <#>      <#>       <#>      <#>      <Y/N>
<queue #2>   <hhh:mm>    <cores>   <#>      <#>       <#>      <#>      <Y/N>


NODE USAGE INFORMATION:
Node Type        Cores Available    Cores Running    Cores Free
-----------------------------------------------------------------
<node type #1>       <#>                <#>              <#>
<node type #2>       <#>                <#>              <#>


ADDITIONAL NOTES:
------------------------------------------------------------------------------------------------------
<Center specific notes. E.g. Descriptions of CPH, small, medium, and large standard queues, debug queues, etc.>

The fields associated with each column in the QUEUE INFORMATION table are defined as:

• Queue Name - Name of the queue
• Maximum Wall Time - Allowed for a single job
• Maximum Cores - Number of cores allowed for a single job
• Jobs Running - Number of jobs currently running
• Jobs Pending - Number of jobs waiting to run
• Cores Running - Number of cores currently running jobs
• Cores Pending - Number of cores requested by jobs pending
• Queue Running - Yes/No field signifying whether this queue is currently running and accepting jobs.

The fields associated with each column in the NODE USAGE INFORMATION table are defined as:

• Node Type - Type of node (e.g. 4way, 16way, etc.)
• Cores Available - Number of cores available on the system
• Cores Running - Number of cores currently running jobs
• Cores Free - Number of cores that are currently not running jobs

Queue Status Report has limited usefulness in:

• Telling a user when their job will run.
• Picking a system on which to run jobs.
• Determining the best queue for a particular job.

Also to note:

• In some cases, resources available to a queue are also limited (especially related to background queue). So, Maximum Cores refers to a job, but there is also a maximum cores available for the entire queue.
• Certain centers have limits per user and/or per group. These limits are not displayed in this command's output. Users will need to refer to each center's queuing documentation to find these details.
• Resources are required for some jobs (e.g. Cobalt, Abaqus). Currently no centralized or unified source of this information exists.
• CPH is not represented in this format and neither are small, medium, and large queues like those used at ARSC. Both of these methods provide for different tradeoffs between maximum wall times vs. maximum cores.

Back to BC Policy FY10-06