[Gerry Kan]

Howdy Foamers:

I am running a job with about 400+ subdomains.

Sometimes when the run went wrong and I want to start the run from scratch, I need to remove the processor??? directories in order to (quickly) remove the timestep data, and rebuild them using decomposePar. This actually takes quite some time because of the number of subdomains involved.

So this begs the question, whether there is a more convenient way to just remove the timestep files within the processor??? directories?

Thanks in advance, Gerry.

[HPE]

I would not delete `processor*` directories if you do not change the mesh.

- If you want to replace the `processor*/0` directory from the new `0.orig` directory:
-- `rm -rf processor*/<time> && restore0Dir -processor`
-- You need to source `/bin/tools/RunFunctions` to use `restore0Dir`.
- If you want to decompose fields only, and not the existing mesh:
-- `decomposePar -fields`
- If you want to perform decompositions in parallel:
-- `mpirun -np X redistributePar -decompose -parallel`

I am sure the number of methods reducing the computational effort can further be increased.

Hope it helps a bit.