There are a couple of options to control logging when submitting jobs to the RCS compute service. You can determine the naming, contents and location of log files using the following flags, which can either be used in the header of your job script (e.g. #PBS -j oe) or when submitting your job (e.g. qsub -j oe …):
-j oemerges the error (.e) and output (.o) log files into a single file (.o)-Nnames your job, causing your log files to use this name as a prefix rather than the default (i.e. filename of your job script)-eand-ocan be used to specify the exact filename (and optionally path) of your error and output log files, overriding the name of your job (if specified)
These flags can be used in combination, as shown by these examples:
| Arguments | Output file | Error file |
|---|---|---|
| None | job.pbs.sh.o1765125 |
job.pbs.sh.e1765125 |
-j oe |
job.pbs.sh.o1765125 |
None |
-N job1 |
job1.o1765125 |
job1.e1765125 |
-j oe -N job1 |
job1.o1765125 |
None |
-o o.log -e ~/e.log |
o.log |
~/e.log |
-j oe -o o.log |
o.log |
None |
Further resources
- Getting started on the RCS web pages
- RCS cluster scripts on GitHub