Python SLURM Manager
- make sure psm.py is in your $PATH
- copy psm_config.py.template to psm_config.py in your project directory
- edit psm_config.py (especially your default partition)
- create ~/.psm/outputs (or whatever path you specified in your psm_config.py)
- run psm in your project directory
- If you use the default configuration, PSM will save job definitions in your project directory and your joblist globally in your user directory.
The left Pane shows a list of all jobs.
- enable automatic refresh
- delete current item
- clear all finished/cancelled jobs
- cancel current job
The middle pane shows a list of all defined jobs.
- execute current job definition
- create/edit/delete job (opens job definition window)
The right pane shows the output of the currently (left pane) selected job.
This window has four fields (from top to bottom):
This name will be shown in lists in PSM and as title in slurm. Names must be unique.
List of commands to execute. Use python format-style placeholders for parameters, e.g.
sleep {0}
sleep {1}
List of parameters given to each instance. These parameters are formatted into the command template. Separate multiple instances with linebreaks and multiple parameters with ,
, e.g.
1,1
1,10
10,10
Use
RUN=SCRIPTNAME
to use a script name that outputs parameters to stdout.
Global parameters are passed to SLURM to request e.g. additional memory or a certain number of CPUs. Separate multiple options with linebreaks and key/value pairs with =
, e.g.
mem-per-cpu=5000
num-tasks=2
For a list of possible options, refer to the SLURM manual.
This code is released under the MIT License (see LICENSE file for more information).