Einstein & Slurm on a small hpc cluster

Hello everyone,

I have a small Beowulf-type cluster at my disposal and would put it to some good use during idle times. I have used all my google ninja skills to find a good way to run boinc/Einstein@home under a load manager like slurm but couldn't really find anything.

Someone very knowledgable told me that in this community, already some people are using the spare capacities of their cluster (ATLAS?) so I thought I ask how you manage it.

For me, the best would be to submit a boinc job to slurm that runs indefinitely and gets suspended whenever a real user needs the nodes (I was thinking to simply use preemption (https://slurm.schedmd.com/preempt.html). All nodes are stateless and I would not like to run a boinc instance on every single node...

Does anyone have experience with doing it like that? If not, how is it done on your cluster? 

Not sure if this is the right forum to post this so please move me around as you see fit.

all the best

Johannes