Kernel Samepage merging

[Prince_Philip]

Hi,
I have enabled KSM on my Worker Node.
Is there a way a container running in the Worker Node know whether KSM is enabled in the worker node?
thanks
Prince

[gabrielmougard]

To check this from within a container, you can mount the host’s /sys filesystem into the container using the lxc.mount.entry configuration option in the container’s configuration file. Once mounted, the container can read the /sys/kernel/mm/ksm/run file to determine the status of KSM on the host node.

lxc.mount.entry = /sys sys none bind,optional 0 0

Once the container is started, you can check the status of KSM on the host node by running the following command inside the container:

lxc shell <container> -- cat /sys/kernel/mm/ksm/run

If the output is 1 , then KSM is enabled on the host node. Can you try this on your side and tell me if this works for you ?

@tomp Do you agree with that ?

[Prince_Philip]

Thank you !!

[tomp (Thomas Parrott)]

I found this past post which may be of interest:

Running KSM on linux containers

Recently I’ve been doing an experiment to check the effectiveness of KSM(Kernel Same-page Merging) on containers I followed through the instructions of this tutorial I successfully built the ksm_preload tool and enabling it in the /etc/rc.local (I’m using ubuntu container) in the container and then I tried $ sudo ./ksm-wrapper lxc start container1 to launch multiple containers and check /sys/kernel/mm/ksm, but only full_scans has increased, the other remained unchange. there’s a guy claimed …

@amikhalitsyn is the mount that @gabrielmougard suggested above a security concern?

[amikhalitsyn (Aleksandr Mikhalitsyn)]

It can be a problem of course as we expose a host sysfs to a container.
As far as I know the KSM is not namespaced and we can read /sys/kernel/mm/ksm/run from any sysfs mount and it will give a host value. KSM is a completely independent with the namespaces/cgroups it works globally on the memory management subsystem level.