using slurm batch system - Forum - Delft3D Flexible Mesh
Forum
- Home
- 2. Questions and troubleshooting
- 2.1 Questions about the software
- 2.1.6 Other
- using slurm batch system
using slurm batch system
michael schaferkotter, modified 17 Days ago.
using slurm batch system
Youngling Posts: 3 Join Date: 4/3/16 Recent Posts
having run coupled flow,wave, swan simulations using the Portable Batch System(PBS), a new machine, and a new batch system, SLURM, presents new challenges.
anyone using SLURM?
After DFLOWFM:: Init
DWAVES:: init
TMP_ESMF_RegridWeightGen_flow_source_*.nc are generated
the destination .nc is generated.
ESMF_RegridWeightGen_in_Delft3D-WAVE.sh calls
ESMF_RegridWeightGen with arguments.
esmf_sh.log shows the attempt to create the first weight file, but nothing happens except
the slurm log
shows:
srun: Job 358251 step creation temporarily disabled, retrying (Requested nodes are busy)srun: Job 358251 step creation still disabled, retrying (Requested nodes are busy)
under PBS,ESMF_RegridWeightGen_in_Delft3D-WAVE.sh
would execute (successfully)
mpirun -np 1 $regridexec $arguments >>esmf_sh.log
Note: the weights files can be generated either manually at the command line.
the slurm MPI users guide says that one does not have to compile mpich with slurm support.
any suggestions?
anyone using SLURM?
After DFLOWFM:: Init
DWAVES:: init
TMP_ESMF_RegridWeightGen_flow_source_*.nc are generated
the destination .nc is generated.
ESMF_RegridWeightGen_in_Delft3D-WAVE.sh calls
ESMF_RegridWeightGen with arguments.
esmf_sh.log shows the attempt to create the first weight file, but nothing happens except
the slurm log
shows:
srun: Job 358251 step creation temporarily disabled, retrying (Requested nodes are busy)srun: Job 358251 step creation still disabled, retrying (Requested nodes are busy)
under PBS,ESMF_RegridWeightGen_in_Delft3D-WAVE.sh
would execute (successfully)
mpirun -np 1 $regridexec $arguments >>esmf_sh.log
Note: the weights files can be generated either manually at the command line.
the slurm MPI users guide says that one does not have to compile mpich with slurm support.
any suggestions?
William Thompson, modified 11 Days ago.
RE: using slurm batch system
Youngling Post: 1 Join Date: 9/19/23 Recent Posts
thanks
michael schaferkotter, modified 10 Days ago.
RE: using slurm batch system
Youngling Posts: 3 Join Date: 4/3/16 Recent Posts
the same type of behavior, e.g. 'hanging' occurs with d3d simultation that involves waves.
...
Part V - Initialisation & checking input... Part VI - Initialisation & checking second part...
Part VII - Initialisation output...
just 'sits' there.
...
Part V - Initialisation & checking input... Part VI - Initialisation & checking second part...
Part VII - Initialisation output...
just 'sits' there.
michael schaferkotter, modified 9 Days ago.
RE: using slurm batch system
Youngling Posts: 3 Join Date: 4/3/16 Recent Posts
the remark about the d3d-simulation 'hanging' was because of an omission of an ampersand in the call in the run script. d3d coupling seems to work.
dfm is still hanging on the RegridWeightGen
dfm is still hanging on the RegridWeightGen