You may need to edit the tool, either to make it work with the specifities of your cluster, or just to add new features. Eitherway, please consider submitting some pull requests to the GitHub to share these edits with other users!
There are some example of intermediary files here.
Custom workload manager
All the code related to the workload manager is stored in the
GreenAlgorithms_workloadManager.py file. At the moment it is built for SLURM, but you can tailor it to different tools by editing this script.
You just need to make sure that you create a variable
WM.df_agg (the output of
WM.clean_logs_df()) similar to the example file here. Only the columns with a name ending with X are needed (the other ones are being used by the workload manager script). NB: the
pd.DataFrame has been converted to a csv to be included here.
If SLURM seem to be returning data that are not expected, it may be an issue with SLURM version. The tool has been tested with SLURM
20.11.9, the raw output of the
sacct command (this is the command pulling all the logs) can be checked here (here as a table); in the code, this is
WM.logs_raw, the output of