Serial (Single Processor) Batch Jobs

Any batch job that does not require multiple processors or multiprocessing libraries, such as OpenMP and MPI, is a serial batch job. Here is a sample serial batch script, runjob.single:

#!/bin/bash
#
# All lines starting with "#PBS" are PBS commands
#
# Request 1 node
#PBS -l nodes=1
#
# Set wall clock time to 0 hours, 15 minutes and 0 seconds
#PBS -l walltime=00:15:00

# cd to working directory
cd $PBS_O_WORKDIR
# name of executable
MYPROG="ssor"

# Run MYPROG
$MYPROG >& myoutput

# make sure to exit the script, else job may not finish properly
exit 0

cootie:~/batch-examples % qsub runjob.single
13952.hn003.nerf.bu.edu
cootie:~/batch-examples % qstat
Job id           Name             User             Time Use S Queue
---------------- ---------------- ---------------- -------- - -----
13952.hn003      runjob.single    kadin                   0 R dque 

Typically, a batch job generates at least two files: an output file (equivalent to STDOUT) and an error file (equivalent to STDERR). The naming convention is batch-script-name.XJOBID, where X is either "o" for output or "e" for error. JOBID is the unique ID of the job - in this case it is 13952. In this example, there is no error and hence the file runjob.single.e13952 contains 0 records. The output file, runjob.single.o13952, consists of three sections:

1. A header section (PBS Prologue) which lists information such as JOBID, user name and node list.
2. A body section which includes user output to STDOUT. Incidentally, the warning message

   Warning: no access to tty (Bad file descriptor).  
   Thus no job control in this shell.

   reminds that this is a remote batch job which can not be acted upon (such as ^Z for suspension).
3. A foot section (PBS Epilogue) which is similar to the header. A useful difference is the report of wallclock time towards the end.

----------------------------------------
Begin PBS Prologue Mon Oct 20 11:03:26 EDT 2003
Job ID:         13952.hn003.nerf.bu.edu
Username:       kadin
Group:          scv
Nodes:          nodem049
End PBS Prologue Mon Oct 20 11:03:26 EDT 2003
----------------------------------------
Warning: no access to tty (Bad file descriptor).
Thus no job control in this shell.
 The result =   1.000000  <== output of execution
----------------------------------------
Begin PBS Epilogue Mon Oct 20 11:03:32 EDT 2003
Job ID:         13952.hn003.nerf.bu.edu
Username:       kadin
Group:          scv
Job Name:       runjob.single
Session:        15012
Limits:         cput=24:00:00,neednodes=nodem049,walltime=00:15:00
Resources:      cput=00:00:00,mem=0kb,vmem=0kb,walltime=00:00:01
Queue:          dque
Account:
Nodes:          nodem049

Killing leftovers...

End PBS Epilogue Mon Oct 20 11:03:32 EDT 2003
----------------------------------------

The runjob.single batch script above is fairly representative of a typical PBS batch script. The name of the executable is explicitly defined. As a result, the script is specific for the executable.

Submitting a serial bach job without a script

There is an alternative way to submit a batch script without having to prepare a batch script. The user would instead run a script ssub which takes the user's executable name as its command line input. Internally, ssub will generate a specific batch script with the executable incorporated and submit to the batch queue. For example,

cootie % ssub ../mypath/executable-name

Additional PBS command line input may be passed on to ssub as a second command line input parameter:

cootie % ssub ../mypath/executable-name "-l walltime=00:30:00"

Because ssub is not a system script, you will need to copy it (left click to view and right-click to copy) into your local directory, like ~/bin. Remember to give this script an execute attribute with

cootie % chmod +x ssub