Home ยป pbs_submission_flags

pbs_submission_flags

Below is a list of the most common and useful directives.

 

Option System type Description
 -k  All  Send "stdout" and/or "stderr" to your home directory when the job runs

#PBS -k o    or
#PBS -k e    or
#PBS -koe
 
 -l  All Precedes a resource request e.g. processors, wallclock
-M   All Send an email messages to an alternative email address

#PBS -M me@myplace.thingy.com
-m  All  Send an email address when a job begins execution and/or ends or aborts

#PBS -m b
#PBS -m be
#PBS -m ba

etc
mem  Shared
Memory		 Specifies the amount of memory you need for a job. This is a requirement for ORAC where you may need a lot of memory but not many processors. Memory and processors are given out together on ORAC so if you need a high memory to processor ratio, remember to specify this directive.

#PBS -l mem=80gb
mpiprocs  Clusters  Number of processes per node on a cluster. This should equal number of processors on a node in most cases.

#PBS -l mpiprocs=4

 -N  All Give your job a unique name

#PBS -N galaxies1234
 -ncpus Shared
Memory		 The number of processors to use for a shared memory job. On ORAC this should always be a multiple of 4 as processors are allocated by blades, each with 4 processors.

#PBS ncpus=4
-r All Control whether or not jobs should automatically re-run from the start if the system crashes or is rebooted. Users with check points might not which this to happen.

#PBS -r n
#PBS -r y
select Clusters Number of compute nodes to use. Usually combined with the mpiprocs directive

#PBS -l select=2
 -V All Make sure that the environment in which the job runs is the same as the environment in which it was submitted.

#PBS -V
 walltime  All  The maximum time a job can run before being stopped. If not used a default of a few minutes is used. Use this flag to prevent jobs which go bad running for hundreds of hours. Format is HH:MM:SS

#PBS -l walltime=12:00:00

 

How can I make a PBS wait until another has completed before running ?

The trick is to make the subsequent job(s) depend on the first. In the following simple example, job 80 is not going to start until job 79 finishes (afterany means 79 just has to end, successfully or not) The second job has a line in its submission script that says 

#PBS -W depend=afterany:79

To show it works, I use a qstat -a

                                                            Req'd  Req'd   Elap
    Job ID          Username Queue    Jobname    SessID NDS TSK Memory Time  S Time
    --------------- -------- -------- ---------- ------ --- --- ------ ----- - -----
    79.corvus       testuser corvus   Depend01    15771   1   1 1800mb   --  R 00:02
    80.corvus       testuser corvus   Depend02      --    1   1 1800mb   --  H   --

The state "H" means hold .. i.e. waiting for something Obviously you'll need to figure out the PBS jobIDs as you go along ... if you use a script to submit the jobs, then capture the last job number and use it to create 3the correct depend= statement. The man qsub page is missing some of the important facts (at least on my terminal) so you might want to download the PDF user guide from sunfire01 (I emailed you about it after the rebuild of PBS) The missing details for depend are as follows:

 

    after:arg_list
                   This job may be scheduled for execution at any point after all jobs
                   in arg_list have started execution.

    afterok:arg_list
                   This job may be scheduled for execution only after all jobs in
                   arg_list have terminated with no errors. See "Warning about exit
                   status with csh" in EXIT STATUS.

    afternotok:arg_list
                   This job may be scheduled for execution only after all jobs in
                   arg_list have terminated with errors. See "Warning about exit status
                   with csh" in EXIT STATUS.

    afterany:arg_list
                   This job may be scheduled for execution after all jobs in
                   arg_list have terminated, with or without errors.

    before:arg_list
                   Jobs in arg_list may begin execution once this job has begun
                   execution.

    beforeok:arg_list
                   Jobs in arg_list may begin execution once this job terminates
                   without errors. See "Warning about exit status with csh" in
                   EXIT STATUS.

    beforenotok:arg_list
                   If this job terminates execution with errors, the jobs in arg_list
                   may begin. See "Warning about exit status with csh" in EXIT
                   STATUS.

    beforeany:arg_list
                   Jobs in arg_list may begin execution once this job terminates
                   execution, with or without errors.

    on:count
                   This job may be scheduled for execution after count depen-
                   dencies on other jobs have been satisfied. This type is used in
                   conjunction with one of the before types listed. count is an
                   integer greater than 0.