Nodes in not ready state with RHEL/CentOS systems

[galal-hussein]

rke v0.0.7-dev

rke failed when deploying network/addon jobs because node were in not ready state, when describing nodes i found this error message:

kubectl describe nodes
Name:               node-01
Roles:              etcd,master,worker
..........
KubeletNotReady              Failed to start ContainerManager mkdir /sys/fs/cgroup/cpuacct,cpu: read-only file system,runtime network not ready

Adding "/sys/fs/cgroup:/sys/fs/cgroup:rw" to kubelet volumes solved the problem

[soumyalj]

Reproduced the issue in rke version v0.0.7-dev.
Tested with rke version v0.0.8-dev for RHEL and Centos hosts. Creation of cluster was successful. The nodes were in Ready state. Creation of pods also succeeded.

soumyas-MBP:rke soumya$ kubectl --kubeconfig .kube_config_cluster.yml get nodes
NAME             STATUS    AGE       VERSION
104.236.249.49   Ready     3m        v1.8.3-rancher3
45.55.165.202    Ready     3m        v1.8.3-rancher3