While this change was tested ok and usually works fine, sometimes SLURM complains about mv with the !() option (activated by shopt -s extglob ). I haven't been able to work out why.
/var/spool/slurmd/job4209876/slurm_script: line 69: syntax error near unexpected token `('
/var/spool/slurmd/job4209876/slurm_script: line 69: ` mv /scale_wlg_nobackup/filesets/nobackup/nesi00213/RunFolder/Cybershake/v24p9/Runs/HikHBaymax/HikHBaymax_REL61/LF/!(e3d.par|OutBin/e3d.par) $backup_directory # leave e3d.par alone in case we need to restart this run'
I am not religious to the solution. In fact, the proposed solution here is more conventional and should be more reliable.
More importantly, I enforced read_restart=0, if restarting is not intended.
Since we now re-use the e3d.par, depending on the run history, this field might have been previously updated to 1. If this field is 1, this forces EMOD3D to read LF/Restart folder. However if this is meant to be a fresh run and this field is accidentally 1, no such folder exists and EMOD3D will instantly fail.
Recent change kept e3d.par in LF directory by skipping it during "mv" operation. https://github.com/ucgmsim/slurm_gm_workflow/commit/c51afc89c7282a487ba7fbeec9edc5e937d88dea
While this change was tested ok and usually works fine, sometimes SLURM complains about mv with the
!()option (activated byshopt -s extglob). I haven't been able to work out why.I am not religious to the solution. In fact, the proposed solution here is more conventional and should be more reliable.
More importantly, I enforced read_restart=0, if restarting is not intended.
Since we now re-use the e3d.par, depending on the run history, this field might have been previously updated to 1. If this field is 1, this forces EMOD3D to read LF/Restart folder. However if this is meant to be a fresh run and this field is accidentally 1, no such folder exists and EMOD3D will instantly fail.