quick and dirty lib for submitting jobs to slurm via python2/python3.
from slurmpy import Slurm
s = Slurm("job-name", {"account": "my-account", "partition": "my-parition"})
s.run("""
do
lots
of
stuff
""")
The above will submit the job to sbatch
automatically write the script to scripts/
and automatically write logs/{name}.err and logs/{name}.out. It will have today's
date in the log and script names.
The script to run() can also contain $variables
which are filled with the cmd_kwarg dict.
E.g. echo $name
could be filled with cmd_kwargs={'name': 'sally'}
A command can be tested (not sent to queue) by setting the _cmd
are to run
as e.g. "ls".
The default is sbatch
which submits jobs to slurm.
Each time slurmpy.Slurm().run()
is called, it returns the job-id of the submitted job. One or more
of these can be provided as arguments to a subsequent job:
s = Slurm()
s.run(..., depends_on=[job_id])
# or for multiple dependencies
s.run(..., depends_on=[job_id_1, job_id_2, job_id_3])
to indicate that this job should not run until the the job with job_id
(or ids job_id_1-3
) has finished successfully.
Additionally, SLURM allows you to define how a job depends on previous jobs e.g. should it run
only on success or always etc. For this slurmpy.Slurm().run()
provides an additional argument depends_how
:
s = Slurmp()
s.run(..., depends_on=[job_id], depends_how='afterany')
currently supported are the following options with "afterok" as default:
"after", "afterany", "afterburstbuffer", "aftercorr", "afternotok", "afterok", "expand"
More details can be found here: https://slurm.schedmd.com/sbatch.html in the --dependency
section.
pip install slurmpy --user