Hello,
I have a slurm job which needs to launch multiple tasks across the allocated hosts for the job.
My criteria is that most of the tasks need to be launched from within the main task launched
by slurm in the launch compute node.
So, if the allocated hosts are h1, h2 & h3 with h1 being the main launcher node then the
initial task, say launchTask, launched in h1 will need to launch RemoteTask1, RemoteTask2 in h2 & h3 at some
point during execution.
Can I use srun from inside launchTask to do so? If yes, what would be the syntax / args?
If no, then what alternative I have other than using rsh/ssh which mayn't be available in the cluster.
Thanks in advance!
Regards,
Bhaskar.