[slurm-users] What is the minimal configuration for a compute node
michael.gutteridge at gmail.com
Thu Aug 24 14:34:33 UTC 2023
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
It might be tedious, but you'll save yourself a lot of aggravation down the
road by copying the full config.
On Tue, Aug 22, 2023 at 7:41 PM Maximilian Ebert <max.ebert at me.com> wrote:
> 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.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the slurm-users