choderalab / clusterutils

Utilities for running parallel jobs with Torque/Moab and MPI
GNU General Public License v2.0
3 stars 3 forks source link

Fix control-port bug in SLURM? #10

Closed Lnaden closed 7 years ago

Lnaden commented 7 years ago

SLURM runs can fail when the --control-port is not the first node in nodelist. This ensures the first host is always the first host on the list to get around it. I can't 100% confirm this is the cause, but this does at least fix it.

Other changes:

Lnaden commented 7 years ago

@jchodera can you give this a quick look over to see if it looks good to you? I'l get this merged in and cut the conda-recipe release as well

jchodera commented 7 years ago

Looks good!