Currently, it is not possible to set weights for nodes. This can result in bad scheduling decisions by Slurm e.g. a job might be scheduled to a larger node than necessary if not incentivized otherwise by weights.
Also, node values have been set more explicitly. CPUs is now set to vcpus instead of relying on implicit computation. Boards is set to slurm default 1. ThreadsPerCore is set to slurm default 1. This should make it easier to read and avoid default value changes affecting clusters when the slurm version changes.
The default TmpDisk value for Slurm is 0. By setting it to a low dummy default workflows that query it for scheduling can still run. CloWM uses TmpDisk=20000 for that.
Currently, it is not possible to set weights for nodes. This can result in bad scheduling decisions by Slurm e.g. a job might be scheduled to a larger node than necessary if not incentivized otherwise by weights.
Also, node values have been set more explicitly. CPUs is now set to vcpus instead of relying on implicit computation. Boards is set to slurm default 1. ThreadsPerCore is set to slurm default 1. This should make it easier to read and avoid default value changes affecting clusters when the slurm version changes.
The default TmpDisk value for Slurm is 0. By setting it to a low dummy default workflows that query it for scheduling can still run. CloWM uses
TmpDisk=20000for that.