INTRODUCTION: how to use SGE queuing system? 

(1) understanding queuing system 
Our queuing system has roughly two compartments. 
a. master node: it is for both log-in and job submission. (ex. mlet1) 
b. slave nodes: they receive and execute job scripts from the master node. 

(2) preparation 
In principle, we have to prepare a job script for each job. The job script has the following format. 
a. first line: login shell (by default, we use csh) 
b. other lines: commands that will be computed at slave node(s). 

--ex. job script-- 
#!/bin/csh 

hostname 
----- 

(3) submit 
ex. "qsub -cwd -l mem=1g test.sh" for QC(quantum chemistry) works 
"qsub -cwd -l mem=1g -l qc=0 test.sh" for MD jobs 

(4) check job status 
a. I the job is running, then one can find files that end with its job ID. They contain stderr & stdout information from slave nodes. 
b. qstat: command for checking the job status 
(qw: queue waiting, r: running, dr: deleting running job) 
c. qstat -g -c : command for showing cluster information such as available slots. 

(5) delete job 
qdel 
ex. qdel 10213 
Job ID can be found at output of ‘qstat’