Description of problem:
kubelet logs show the following message:
475] failed to find cgroups of kubelet - cpu and memory cgroup hierarchy not unified. cpu: /, memory: /system.slice/kubelet.service
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. install cluster
2. ssh to node
3. journalctl -u kubelet
kubelet does not appear to be in a cpu cgroup
kubelet should be in cpu cgroup
If you cat /proc/<pid>/cgroup for kubelet, we see following:
It appears that RHEL CoreOS is not enabling default cpu accounting in systemd.
CPU accounting should be enabled by default.
cat /etc/systemd/system.conf | grep CPUAccounting
In 3.x releases, we always enabled cpu accounting as a drop-in.
The drop-in is this
I guess we need the MCO to lay this down for us.
Ryan can you do this?
I'll look into it.
Oops. Sounds like we want to modify the system.conf within the rpm. Looking into it.
PR merged https://github.com/openshift/machine-config-operator/pull/581 and verified working.
systemctl show -all|grep -e 'DefaultCPUAccounting\|DefaultBlockIOAccounting\|DefaultMemoryAccounting'
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.