kubernetes: Cgroups per qos not working

Is this a BUG REPORT or FEATURE REQUEST?:

/kind bug

What happened:

Setting a Resource requests and limits on a Pod other Pods running on the node where it was scheduled are moved to the cgroups created to serve the QoS request.

What you expected to happen:

Pods to stay in their assigned cgroups.

How to reproduce it (as minimally and precisely as possible):

kubelet running with the following flags:

--enforce-node-allocatable=pods
--cgroups-per-qos=true
--cgroup-root=/root/cgroup-here

Create a pod without resource reservations and then create another one with resource reservations. For example, using the following template:

apiVersion: v1
kind: Pod
metadata:
  name: nginx
  labels:
    app: nginx
spec:
  containers:
  - name: nginx
    image: nginx
    ports:
      - containerPort: 80
    resources:
      requests:
        memory: "32Mi"
        cpu: "100m"
      limits:
        memory: "32Mi"
        cpu: "100m"

Anything else we need to know?:

Environment:

• Kubernetes version (use kubectl version):

Server Version: version.Info{Major:"1", Minor:"12+", GitVersion:"v1.12.0-beta.0", GitCommit:"450ed5307355ef0e3d215f243b80178bd465c975", GitTreeState:"clean", BuildDate:"2018-08-14T21:03:09Z", GoVersion:"go1.10.3", Compiler:"gc", Platform:"linux/amd64"}

• Cloud provider or hardware configuration:

• OS (e.g. from /etc/os-release):

CentOS 7.5
CoreOS-1465

• Kernel (e.g. uname -a):

Linux d63a8d93-6617-4bdb-80ee-36499392246c 3.10.0-862.6.3.el7.x86_64 #1 SMP Tue Jun 26 16:32:21 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux

Linux 391887eb-14f8-4b7c-9498-41fc1f8c92cb 4.12.7-coreos #1 SMP Tue Aug 15 23:54:56 UTC 2017 x86_64 Intel(R) Xeon(R) CPU @ 2.20GHz GenuineIntel GNU/Linux

• Install tools:
• Others: