Resource Assignment Issue

Hello,

I reached out to support on Monday about this issue and haven't received a reply from them.

Has anyone else experienced problem with their lxc containers not restricting physical resources? For example: if I deploy a lxc machine with 1 vCPU and 1 GB of RAM, I can log into the machine and see 16 vCPU (8 core x 2 threads) and 32 GB of memory.

This happens regardless of the template OS I use (debian, ubuntu, centos, fedora, kali).

For that matter does anyone know how to get support to reply?

I see the same behavior. I haven't tried to use all the resources to see if it is really limiting them or not.

I'm not sure the limiting works.

I set all of my LXCs to 1 GB RAM and 1 vCPU.

If I check the resources using virsh (using my atlassian antlet as an example):

virsh -c lxc:/// atlassian

It shows 1 CPU and 1 GB RAM.

Within the LXC, the usual commands (top, free) all show info from the antsle itself.

But, if I actually use an application in the LXC (e.g. Confluence), it shows it has 2 GB allocated.

According to this:

https://archives.flockport.com/lxc-advanced-guide/

Resource limits would be enforced by cgroups, but, when you're in an LXC antlet, if you do:

df -h

It shows a filesystem mounted to /sys/fs/cgroup, but, on the antsle, there isn't a cgroup for each lxc antlet under that path.

So I don't think you can actually limit the resources of an LXC antlet.

Although, for me, that's the advantage of an LXC antlet -- It runs with the full capability of the antsle and I let the hypervisor worry about the resource allocations.

If I want to limit the resources, I create a KVM antlet.