[slurm-users] What is the minimal configuration for a compute node

[Maximilian Ebert]

Dear list,

After configuring the head node it is usually recommended to copy the entire slurm.conf to each compute node to make the available in the cluster. While this generally works it can be quite tedious for more complex configurations with multi-level includes etc. Is there some minimal configuration for the slurm.conf on a compute to establish the communication with the head node and the other compute nodes? I am aware of configless but we structured the configuration in subfolders and only files next to slurm.conf can be copied / shared in configless.

Thanks for your help.

[Michael Gutteridge]

Hi

By "minimal config" I'm assuming you mean "just enough config to get the slurmd to run".  As far as I'm aware, you really need to have a complete and matching config on each of your daemons- like slurmd literally won't start with differing configs.  There is the "NO_CONF_HASH" debug flag to get around that, but that's certainly not intended for production operations.

It might be tedious, but you'll save yourself a lot of aggravation down the road by copying the full config.

 - Michael

[Bernstein, Noam CIV USN NRL (6393) Washington DC (USA)]

 it can be quite tedious for more complex configurations with multi-level includes

It's either identical or configless, as far as I know also. What about changing your subdirectories to filenames (e.g. slurm/paritions/bob.conf -> slurm.partitions.bob.conf), and then doing configless, or just "cp -r" or "scp -r" (depending on how you can get into the compute node directory structure, e.g. vnfs or something) the entire config directory tree?