different batch directives for case.run & case.st_archive

[henry_leroy_miller@noaa_gov]

Hi,working with a CIME enabled version of CESM (1.5_beta02), I've been able to get the FW5 case configured, built, and run on our NOAA Theia HPCS.  However, I get the same batch system directives for both the case.run and case.st_archive files.  Now that these two functions are separate jobs, what I'd like to be able to do is have the batch system directives be configured as shown here below, without manually editing these two files.for case.run:#PBS -q batch#PBS -l walltime=08:00:00#PBS -l procs={{ totalpes }}  (using syntax mix from config_batch.xml, {{ totalpes }} will evaluate to an integer) for case.st_archive:#PBS -q debug#PBS -l walltime=00:10:00#PBS -l procs=1#PBS -l vmem=7G all other PBS directives used are common to both.  My goal is to avoid having the researcher manually edit all these settings in one or the other file, to reduce errors.I've tried a syntax such as this, however it just puts both entries in both files:          -q batch           -q debug  I also tried using a section lifted from the section () in the batch_system definition section, howerver then the new case would not even create. is there a way to configure files in the CIME framework that one can specify differing batch system directives based upon the job name? thanks,
Roy  

 

[jedwards]

Hi Roy,You might consider updating to the latest model version where we've solved these problems, or with some effort you could backport the latest cime to your model.I made a couple of attempts to do what you want in that version and wasn't successful.   Sorry.

 

[jedwards]

I have a solution - it's not portable but should serve your needs on theia - remove -q directive from config_batch.xml and put it directly into the template.case.run and template.case.st_archive filesthese are located in directory cime_config/cesm/machines/ - add them right after the{{ batchdirectives }} line.  You should also be able to do this for walltime.

 

[henry_leroy_miller@noaa_gov]

Thanks jedwards!  for better or worse, I'm tied to this version of the model, as the scientific brain trust here and at NCAR have specified this version for the purposes of their specific research.  I'll investigate your suggested solution.  My other thought was to write a short bash/awk/sed or perl script to modify the generated case.st_archive file in the case directory. I'm always a little leary about just changing such files, as I don't know enough about the internals of the configuration and run framework to be sure that changes in these files will indeed control the running of the model.regards,Roy