[slurm-users] Limit resources on login node

[Miguel Gutiérrez Páez]

Hi all,

I'm quite new in this slurm world, and I have some (a lot!) of doubts.
I want to limit resources on my login node so users don't overload this
host. I have configure ulimit in this way:

cat custom.conf
*          soft    nproc     4096
root       soft    nproc     unlimited
@basic     soft    nproc     100
@basic        hard as 2000000

As you may guess, @basic is the normal group for my users, and I have
limited they only can reserve 2GB of virtual memory. But something weird
occurs. Everything seems to work fine, but a certain sbatch is failing.
This sbatch launches finally a java jar file, and fails with this error:

Error occurred during initialization of VM
Could not allocate metaspace: 1073741824 bytes

If I unlimit memory resources (by commenting the last line in custom.conf
file), the same sbatch works properly. A scontrol show job of a failed job
shows that the job was launched in a compute node, where there is no any
restriction about memory (or other) resource. So, the login node is the
only node I limit resources.  Why is failing the sbatch if the compute
nodes have no any restriction but the login one? I'm pretty sure the
problem is due last line of custom.conf file. If I comment that line or
increase the limit to 4GB for example, everything works fine. What am I
missing?

Thanks a lot!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.schedmd.com/pipermail/slurm-users/attachments/20180131/1c4562da/attachment.html>