Translations:META-Farm/147/en

• Log into a cluster.
• Load the meta-farm module.

$ module load meta-farm

• Choose a name for a farm directory, e.g. Farm_name, and create it with the following command

$ farm_init.run  Farm_name

• This will also create a few important files inside the farm directory, some of which you will need to customize.
• Copy your executable and input files to the farm directory. (You may skip this step if you plan to use full paths everywhere.)
• Edit the table.dat file inside the farm directory. This is a text file describing one case (one independent computation) per line. For examples, see one or more of
  • single_case.sh
  • Example: Numbered input files (advanced)
  • Example: Input file must have the same name (advanced)
  • Using all the columns in the cases table explicitly (advanced)
• Modify the single_case.sh script if needed. In many cases you don't have to make any changes. For more information see one or more of
  • single_case.sh
  • STATUS and handling errors
  • Example: Input file must have the same name (advanced)
  • Using all the columns in the cases table explicitly (advanced)
• Modify the job_script.sh file to suit your needs as described at job_script.sh below. In particular, use a correct account name, and set an appropriate job runtime. For more about runtimes, see Estimating the runtime and number of metajobs.
• Inside the farm directory, execute

$ submit.run -1

for the one case per job (SIMPLE) mode, or

$ submit.run N

for the many cases per job (META) mode, where N is the number of metajobs to use. N should be significantly smaller than the total number of cases.