Even if it will not be used by most users for ECE4 runs, I think that it would be useful to have the possibility to run rdy2cpl in a non-parallel environment, or at least relax some limitations.
I have rdy2cpl installed in a non HPC where I also have an installation of ScriptEngine, pyoasis and open-mpi. I can run rdy2cpl using mpirun to execute the required number of mpi ranks. I just wanted to test the script_engine task defined in rdy2cpl, to get the same results as in an ECE4 experiment, but it failed for me with this error:
2023-09-08 12:23:56 ERROR [se.task:rdy2cpl.make_all <47f2c55bfa>] SLURM 'srun' command not found (rdy2cp needs a working SLURM environment)
2023-09-08 12:23:56 ERROR [se.instance.simplescriptengine] STOPPING SimpleScriptEngine due to task error in rdy2cpl.make_all <47f2c55bfa>
2023-09-08 12:23:56 ERROR [se.instance.simplescriptengine] For more debugging info, re-run with loglevel DEBUG
so it seems that it is not possible because I need slurm in my computer, which I do not have. It seems to me like a quite strong demand. In my opinion, relaxing this requirement would make rdy2cpl more usable, and it would be easier to implement some CI tests.
Going further, I think that it would be also great to have the possibility of running rdy2cpl command line (r2c) in a serial way, using a single task to generate all the oasis weights file sequentially.
Even if it will not be used by most users for ECE4 runs, I think that it would be useful to have the possibility to run rdy2cpl in a non-parallel environment, or at least relax some limitations. I have rdy2cpl installed in a non HPC where I also have an installation of ScriptEngine, pyoasis and open-mpi. I can run rdy2cpl using mpirun to execute the required number of mpi ranks. I just wanted to test the script_engine task defined in rdy2cpl, to get the same results as in an ECE4 experiment, but it failed for me with this error:
so it seems that it is not possible because I need slurm in my computer, which I do not have. It seems to me like a quite strong demand. In my opinion, relaxing this requirement would make rdy2cpl more usable, and it would be easier to implement some CI tests.
Going further, I think that it would be also great to have the possibility of running rdy2cpl command line (r2c) in a serial way, using a single task to generate all the oasis weights file sequentially.