about torque, cluster job scheduler

What is torque, cluster job scheduler

The Torque is a HPC cluster job scheduler is an open-source scheduler based on the original PBS codebase, distrubuted resource manager Torque stands for Terascale Open-source Resource and QUEue Manager (TORQUE), it can be integrate with Maui scheduler or Moab Manager for best utilization, management of HPC jobs and cluster administration.

TORQUE cluster consists of one head node (Active/Passive) and many compute nodes. The head node runs the pbs_server deamon and the compute nodes run the pbs_mom daemon. 

Jobs can be submit using a .sh executable job script. where users can define the required numbers of compute nodes and number of cores (processors) etc. 

Master node is use to submit the compute jobs with the help of scheduler it allocate the jobs to free nodes on cluster. 

Example of PBS job script file :- 

#!/bin/bash

#PBS -l nodes=4:ppn=16 

#PBS -N job name 

#PBS -e error file

#PBS -o output file 

mpirun -f $PBS_NODEFILE ? np 64 ./a.out  

Some of the frequently used Torque commands. 

qsub submit a pbs job

qdel delete pbs job

qstat show status of pbs jobs

qstat -a list all jobs

qhold hold pbs jobs
  

qstat -q list all queues

qstat -au userid list jobs for userid

qstat -r list running jobs

qstat -Qf queue list full information about queue

pbsnodes -l list status of all compute nodes

pbsnodes -a view node status 

qstat -f job_id list full information about job_id

qstat -B list summary status of the job server