jails Ubuntu jail limits RAM?

Does the Ubuntu jail described in https://wiki.freebsd.org/LinuxJails limit access of the jail environment to RAM usage for processes run under it via chroot?

I'm trying to run something for which I expected the RAM usage to shoot up but it hardly went up by 1-2 GB when I was expecting it to go up much much higher
 
You can use rctl to limit resources for process, user, login class or jail.
Thanks - when I try to see it for the particular jail it says rctl present but disabled and I need to enable it - does that mean that the jail process doesn't have any limitations placed onto them?

Are there other ways for a jail process to be limited? I'm curious why my software run in a jail is underperforming my expectations
 
rctl/racct is disabled by default. This can be enabled eg. via kern.racct.enable. See rctl(4) for more info.

As far as I know a jail on a system with rctl/racct disabled is resource unlimited. Even if rctl/racct is enabled it is unconstrained until you setup limits for the jail. This is usually done via rctl(8) and rctl.conf(5).

Are there other ways for a jail process to be limited?
rctl is certainly the way to go for this.

I'm curious why my software run in a jail is underperforming my expectations
Adding limits is usually not expected to yield more performance :p
Performance not living up to expectations can have a multitude of reasons including incorrect expectations. Can you provide more details?
 
Adding limits is usually not expected to yield more performance :p
I'm trying to do the opposite actually - trying to figure if the jail environment is constrained in some way to figure out the hurdles in performance.

Would it be possible the /etc/fstab mounts somewhat play a role in it - if not directly as you mentioned.
Performance not living up to expectations can have a multitude of reasons including incorrect expectations. Can you provide more details?
Was trying to run an LLM model - the RAM usage was expected to go up significantly but it hardly went up - instead the cpu load was maxing out - this wasn't the expected behaviour - hence my question.
 
Back
Top