Skip to Main Content

The University of Tennessee

Newton header banner

Frequently Used Tools:



Home » Documentation » Using the Grid Engine » Cluster Queues

Cluster queues

The concept of a "queue" in the grid engine is a bit more general than in other batch-queue systems.
Cluster queues are logical groups of machines that provide certain resources and on which jobs can execute.
When a job is execute by the Grid Engine, it is placed into a queue which satisfies all the resource requirements of the job.
When you submit a job, you can either request a queue explicitly, or you can request specific resources (CPU time needed, memory needed, etc.) and allow the Grid Engine to determine which queue to use to execute the job (this method is preferred).

Here is a listing of the Newton cluster queues:

[Newton:sigma00 ~]$ qstat -g c
CLUSTER QUEUE                   CQLOAD   USED    RES  AVAIL  TOTAL aoACDS  cdsuE  
--------------------------------------------------------------------------------
dataxfer                          0.00      0      0      8      8      0      0 
long_chi                          0.69    236      0   1444   1728      0     48 
long_monster                      -NA-      0      0      0      0      0      0 
long_phi                          0.59    448      0    356    864      0     60 
long_rho                          0.48    326      0    234    576      0     16 
long_rho2                         0.26     17      0    111    192      0     64 
long_sigma                        0.88    899      0   1621   2568      0     48 
medium_chi                        0.69    965      0    715   1728      0     48 
medium_phi                        0.59     92      0    712    864      0     60 
medium_rho                        0.48    184      0    376    576      0     16 
medium_rho2                       0.26     53      0     75    192      0     64 
medium_sigma                      0.88    436      0   2084   2568      0     48 
short_chi                         0.69     32      0   1648   1728      0     48 
short_monster                     0.00      0      0     48     48      0      0 
short_phi                         0.59     48      0    756    864      0     60 
short_rho                         0.48      0      0    560    576      0     16 
short_rho2                        0.26      0      0    128    192      0     64 
short_sigma                       0.88     21      0   2499   2568      0     48 

The formula for Newton cluster queue names is <time limit>_<cluster>. For example, the queue "long_sigma" indicates that this queue has no job runtime limits and contains machines that are located in the "Sigma" cluster. Both these factors are covered in detail later. This queue has a total of 2568 CPU cores. A given compute node CPU can be available in multiple queues, but it can only be used by one queue at a time. For this reason, the "Avail" column does not necessarily indicate that a given queue is able to immediately execute a job.

Back to Using The Grid Engine