On 10/16/18 3:38 AM, Bjørn-Helge Mevik wrote: > Just a tip: Make sure that the kernel has support for constraining swap > space. I believe we once had to reinstall one of our clusters once > because we had forgotten to check that.
I tried starting slurmd with -D -v -v -v and got: slurmd: debug: Reading cgroup.conf file /share/apps/slurm-18.08.1/16.04/etc/cgroup.conf slurmd: debug: Reading cgroup.conf file /share/apps/slurm-18.08.1/16.04/etc/cgroup.conf slurmd: debug: task/cgroup: now constraining jobs allocated cores slurmd: debug: task/cgroup/memory: total:64388M allowed:100%(enforced), swap:0%(enforced), max:100%(64388M) max+swap:0%(64388M) min:30M kmem:100%(64388M permissive) min:30M swappiness:0(set) slurmd: debug: task/cgroup: now constraining jobs allocated memory slurmd: debug: task/cgroup: unable to open /share/apps/slurm-18.08.1/16.04/etc/cgroup_allowed_devices_file.conf: No such file or directory slurmd: debug: task/cgroup: now constraining jobs allocated devices slurmd: debug: task/cgroup: loaded Looks pretty promising. Has anyone out there managed to get constrained swap working?