Hi all,

I have installed udocker in our hcp infrastructure. I've been testing and I
found a very strange behaviour regarding memory comsumption. If a launch a
job with a memory reservation equal or higher to 4GB, any udocker container
can bypass this limit and use all the memory available in the node. With a
reservation lower than 4GB, it works fine (the container gets killed by
slurm if exceeds the limit). It's quiet weird.

Any experience on this?

Regards.

Reply via email to