djk2120/tether

tether.sh

A utility to tether cesm cases

cheyenne.template

• template for each queued tether submission
• you will need to edit this
  • specify your project code
  • tell it where to expect tether.sh (tdir)
• tether.sh will search and replace for jobname/joblist/jobid
  • dont change those!

example/joblist.txt

• file to define your run sequence
• jobname,caseroot,helper_script,status
• all helper scripts must be executable!
  • if you don't need a helper script, put: none
• status progresses from waiting->queued->submitted
• joblist should be in order, with first job queued, and the rest waiting

helper scripts

• each script will run before its associated case
  • edit namelists, case.build, etc
• case must be created before calling tether
• do not include case.submit in the helper scripts
  • tether is responsible for submitting the case

tether.sh

• the workhorse script
• takes joblist and template as command line arguments
• e.g.
  • ./tether.sh joblist.txt cheyenne.template
• should not need to edit this script
• its job is to:
  • run the appropriate helper script
  • call ./submit.case --resubmit-immediate from the current case
  • queue the next case in line
  • update the statuses in joblist.txt

directory structure

• put your joblist and shell scripts together in one directory
  • need not be nested within this directory
• call tether.sh from within THAT directory
• e.g. for my example here:
  • cd example
  • ../tether.sh joblist.txt ../cheyenne.template

logs

• the first segment will write to stdout
• tethered segments will write to their own files: jobname.oXXX