Termination of nsjail leaves cgroup entries in /sys/fs/cgroup/*/NSJAIL/NSJAIL.* about nsjail HOT 3 CLOSED

STANDALONE_ONCE should be fixed now.
For STANDALONE_EXECVE we cannot really do the cleanup, as in this case nsjail is replaced by the target binary on execve() call and no nsjail code is executed after this point. (This is totally 'Works as intended').

@mtis88 this is probably expected and documented as such

Which pre-existing cpu cgroup to use as a parent (default: 'NSJAIL')

Nsjail does not attempt to create the cgroup parent as this would in most cases require more privileges than nsjail is running with.

from nsjail.

Another but similar issue, is that using cgroups and STANDALONE_EXECVE mode always leaves /sys/fs/cgroup/pids/NSJAIL/NSJAIL.* not cleaned up.

from nsjail.

I have to manually create the NSJAIL directory under the relevant cgroup (cpu in my case), is this expected?

from nsjail.