I would suggest that you configure Slurm accounting as described here:

http://www.schedmd.com/slurmdocs/accounting.html

Quoting "Giulio V. de Musso" <[email protected]>:

>
> Hi
>
> I run some sbatches on my SLURM cluster. I need to know the execution time
> for each job. So if for example the job ID is 34 SLURM should create a
> file named 34.time in which SLURM writes the job's start and end time
> (timestamps).
>
> I was thinking to get system's time in Prolog and Epilog scripts. The
> problem is I don't know how to get the job ID in the Prolog and Epilog
> batches (so that I can write the timestamps in the $JobID.time file).
>
> Is there any way to make things done?
>
> Thanks
> Giulio
>

Reply via email to